{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 8802,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00034083162917518747,
      "grad_norm": 57.59091708666561,
      "learning_rate": 0.0,
      "loss": 5.8168,
      "step": 1
    },
    {
      "epoch": 0.0006816632583503749,
      "grad_norm": 35.504484478569786,
      "learning_rate": 1.1350737797956868e-08,
      "loss": 4.9348,
      "step": 2
    },
    {
      "epoch": 0.0010224948875255625,
      "grad_norm": 36.74505399565994,
      "learning_rate": 2.2701475595913736e-08,
      "loss": 4.5229,
      "step": 3
    },
    {
      "epoch": 0.0013633265167007499,
      "grad_norm": 31.028204839735388,
      "learning_rate": 3.40522133938706e-08,
      "loss": 4.4453,
      "step": 4
    },
    {
      "epoch": 0.0017041581458759373,
      "grad_norm": 42.971982705186356,
      "learning_rate": 4.540295119182747e-08,
      "loss": 5.0187,
      "step": 5
    },
    {
      "epoch": 0.002044989775051125,
      "grad_norm": 48.37098858867094,
      "learning_rate": 5.6753688989784336e-08,
      "loss": 4.6632,
      "step": 6
    },
    {
      "epoch": 0.0023858214042263124,
      "grad_norm": 41.05736206565161,
      "learning_rate": 6.81044267877412e-08,
      "loss": 5.0529,
      "step": 7
    },
    {
      "epoch": 0.0027266530334014998,
      "grad_norm": 30.895923414240883,
      "learning_rate": 7.945516458569808e-08,
      "loss": 4.097,
      "step": 8
    },
    {
      "epoch": 0.003067484662576687,
      "grad_norm": 28.139464264677446,
      "learning_rate": 9.080590238365495e-08,
      "loss": 4.0045,
      "step": 9
    },
    {
      "epoch": 0.0034083162917518746,
      "grad_norm": 56.60422298760434,
      "learning_rate": 1.0215664018161182e-07,
      "loss": 5.0767,
      "step": 10
    },
    {
      "epoch": 0.003749147920927062,
      "grad_norm": 32.58747616018831,
      "learning_rate": 1.1350737797956867e-07,
      "loss": 4.2204,
      "step": 11
    },
    {
      "epoch": 0.00408997955010225,
      "grad_norm": 77.12020113980064,
      "learning_rate": 1.2485811577752554e-07,
      "loss": 4.7817,
      "step": 12
    },
    {
      "epoch": 0.004430811179277437,
      "grad_norm": 38.578010149366236,
      "learning_rate": 1.362088535754824e-07,
      "loss": 4.1481,
      "step": 13
    },
    {
      "epoch": 0.004771642808452625,
      "grad_norm": 31.6737289164945,
      "learning_rate": 1.4755959137343928e-07,
      "loss": 4.6646,
      "step": 14
    },
    {
      "epoch": 0.005112474437627812,
      "grad_norm": 41.31979076745021,
      "learning_rate": 1.5891032917139615e-07,
      "loss": 4.1875,
      "step": 15
    },
    {
      "epoch": 0.0054533060668029995,
      "grad_norm": 31.343917032598256,
      "learning_rate": 1.7026106696935302e-07,
      "loss": 4.6402,
      "step": 16
    },
    {
      "epoch": 0.005794137695978187,
      "grad_norm": 39.75586604612425,
      "learning_rate": 1.816118047673099e-07,
      "loss": 4.9286,
      "step": 17
    },
    {
      "epoch": 0.006134969325153374,
      "grad_norm": 46.170048454435346,
      "learning_rate": 1.9296254256526676e-07,
      "loss": 5.4478,
      "step": 18
    },
    {
      "epoch": 0.006475800954328562,
      "grad_norm": 50.34261010053117,
      "learning_rate": 2.0431328036322363e-07,
      "loss": 4.3283,
      "step": 19
    },
    {
      "epoch": 0.006816632583503749,
      "grad_norm": 34.45048910040835,
      "learning_rate": 2.156640181611805e-07,
      "loss": 4.4405,
      "step": 20
    },
    {
      "epoch": 0.007157464212678937,
      "grad_norm": 29.228334505764643,
      "learning_rate": 2.2701475595913734e-07,
      "loss": 3.7926,
      "step": 21
    },
    {
      "epoch": 0.007498295841854124,
      "grad_norm": 43.094353346124514,
      "learning_rate": 2.3836549375709421e-07,
      "loss": 5.2052,
      "step": 22
    },
    {
      "epoch": 0.007839127471029312,
      "grad_norm": 49.11282664435778,
      "learning_rate": 2.497162315550511e-07,
      "loss": 5.0791,
      "step": 23
    },
    {
      "epoch": 0.0081799591002045,
      "grad_norm": 32.96859290994525,
      "learning_rate": 2.61066969353008e-07,
      "loss": 4.5298,
      "step": 24
    },
    {
      "epoch": 0.008520790729379687,
      "grad_norm": 38.613399659302985,
      "learning_rate": 2.724177071509648e-07,
      "loss": 4.6864,
      "step": 25
    },
    {
      "epoch": 0.008861622358554875,
      "grad_norm": 36.93695516496745,
      "learning_rate": 2.837684449489217e-07,
      "loss": 4.3844,
      "step": 26
    },
    {
      "epoch": 0.009202453987730062,
      "grad_norm": 42.6721215467686,
      "learning_rate": 2.9511918274687856e-07,
      "loss": 5.1473,
      "step": 27
    },
    {
      "epoch": 0.00954328561690525,
      "grad_norm": 57.352269402058106,
      "learning_rate": 3.064699205448354e-07,
      "loss": 4.2618,
      "step": 28
    },
    {
      "epoch": 0.009884117246080437,
      "grad_norm": 55.365441413724994,
      "learning_rate": 3.178206583427923e-07,
      "loss": 5.0097,
      "step": 29
    },
    {
      "epoch": 0.010224948875255624,
      "grad_norm": 25.35316815868296,
      "learning_rate": 3.2917139614074915e-07,
      "loss": 3.9989,
      "step": 30
    },
    {
      "epoch": 0.010565780504430812,
      "grad_norm": 34.23714511262736,
      "learning_rate": 3.4052213393870604e-07,
      "loss": 4.2315,
      "step": 31
    },
    {
      "epoch": 0.010906612133605999,
      "grad_norm": 55.82043410873589,
      "learning_rate": 3.518728717366629e-07,
      "loss": 6.0994,
      "step": 32
    },
    {
      "epoch": 0.011247443762781187,
      "grad_norm": 48.619073461652675,
      "learning_rate": 3.632236095346198e-07,
      "loss": 4.8489,
      "step": 33
    },
    {
      "epoch": 0.011588275391956374,
      "grad_norm": 37.56783465614343,
      "learning_rate": 3.745743473325766e-07,
      "loss": 4.7175,
      "step": 34
    },
    {
      "epoch": 0.011929107021131561,
      "grad_norm": 50.21177513772905,
      "learning_rate": 3.859250851305335e-07,
      "loss": 5.3359,
      "step": 35
    },
    {
      "epoch": 0.012269938650306749,
      "grad_norm": 44.46325523693967,
      "learning_rate": 3.9727582292849036e-07,
      "loss": 5.4258,
      "step": 36
    },
    {
      "epoch": 0.012610770279481936,
      "grad_norm": 27.459314920084783,
      "learning_rate": 4.0862656072644726e-07,
      "loss": 3.8282,
      "step": 37
    },
    {
      "epoch": 0.012951601908657124,
      "grad_norm": 44.14694629863951,
      "learning_rate": 4.199772985244041e-07,
      "loss": 4.1953,
      "step": 38
    },
    {
      "epoch": 0.013292433537832311,
      "grad_norm": 38.907810004842695,
      "learning_rate": 4.31328036322361e-07,
      "loss": 4.8621,
      "step": 39
    },
    {
      "epoch": 0.013633265167007498,
      "grad_norm": 33.0510438495151,
      "learning_rate": 4.4267877412031784e-07,
      "loss": 4.236,
      "step": 40
    },
    {
      "epoch": 0.013974096796182686,
      "grad_norm": 33.22668720989239,
      "learning_rate": 4.540295119182747e-07,
      "loss": 4.5137,
      "step": 41
    },
    {
      "epoch": 0.014314928425357873,
      "grad_norm": 78.71170207820887,
      "learning_rate": 4.653802497162316e-07,
      "loss": 5.724,
      "step": 42
    },
    {
      "epoch": 0.01465576005453306,
      "grad_norm": 59.544779257903116,
      "learning_rate": 4.7673098751418843e-07,
      "loss": 4.6553,
      "step": 43
    },
    {
      "epoch": 0.014996591683708248,
      "grad_norm": 32.149479771440475,
      "learning_rate": 4.880817253121453e-07,
      "loss": 4.6192,
      "step": 44
    },
    {
      "epoch": 0.015337423312883436,
      "grad_norm": 32.345267433240735,
      "learning_rate": 4.994324631101022e-07,
      "loss": 4.1883,
      "step": 45
    },
    {
      "epoch": 0.015678254942058625,
      "grad_norm": 38.79489670330069,
      "learning_rate": 5.10783200908059e-07,
      "loss": 4.8076,
      "step": 46
    },
    {
      "epoch": 0.016019086571233812,
      "grad_norm": 40.16376866120035,
      "learning_rate": 5.22133938706016e-07,
      "loss": 5.3198,
      "step": 47
    },
    {
      "epoch": 0.016359918200409,
      "grad_norm": 44.24017598075992,
      "learning_rate": 5.334846765039728e-07,
      "loss": 4.7539,
      "step": 48
    },
    {
      "epoch": 0.016700749829584187,
      "grad_norm": 32.10280391859119,
      "learning_rate": 5.448354143019296e-07,
      "loss": 4.3121,
      "step": 49
    },
    {
      "epoch": 0.017041581458759374,
      "grad_norm": 36.76858463688012,
      "learning_rate": 5.561861520998865e-07,
      "loss": 3.928,
      "step": 50
    },
    {
      "epoch": 0.017382413087934562,
      "grad_norm": 24.226358240965236,
      "learning_rate": 5.675368898978434e-07,
      "loss": 3.7659,
      "step": 51
    },
    {
      "epoch": 0.01772324471710975,
      "grad_norm": 47.50595317038215,
      "learning_rate": 5.788876276958003e-07,
      "loss": 4.8861,
      "step": 52
    },
    {
      "epoch": 0.018064076346284937,
      "grad_norm": 36.86532166495527,
      "learning_rate": 5.902383654937571e-07,
      "loss": 4.6835,
      "step": 53
    },
    {
      "epoch": 0.018404907975460124,
      "grad_norm": 35.76471915382078,
      "learning_rate": 6.01589103291714e-07,
      "loss": 3.8985,
      "step": 54
    },
    {
      "epoch": 0.01874573960463531,
      "grad_norm": 29.80717279427096,
      "learning_rate": 6.129398410896708e-07,
      "loss": 4.3766,
      "step": 55
    },
    {
      "epoch": 0.0190865712338105,
      "grad_norm": 24.044222742554417,
      "learning_rate": 6.242905788876278e-07,
      "loss": 3.9505,
      "step": 56
    },
    {
      "epoch": 0.019427402862985686,
      "grad_norm": 30.03966098559725,
      "learning_rate": 6.356413166855846e-07,
      "loss": 4.0955,
      "step": 57
    },
    {
      "epoch": 0.019768234492160874,
      "grad_norm": 32.04685142968214,
      "learning_rate": 6.469920544835414e-07,
      "loss": 4.6116,
      "step": 58
    },
    {
      "epoch": 0.02010906612133606,
      "grad_norm": 41.7555405534856,
      "learning_rate": 6.583427922814983e-07,
      "loss": 5.0312,
      "step": 59
    },
    {
      "epoch": 0.02044989775051125,
      "grad_norm": 29.574044326552514,
      "learning_rate": 6.696935300794552e-07,
      "loss": 3.8183,
      "step": 60
    },
    {
      "epoch": 0.020790729379686436,
      "grad_norm": 35.383343479813675,
      "learning_rate": 6.810442678774121e-07,
      "loss": 4.0759,
      "step": 61
    },
    {
      "epoch": 0.021131561008861623,
      "grad_norm": 28.865446702738982,
      "learning_rate": 6.923950056753689e-07,
      "loss": 4.2317,
      "step": 62
    },
    {
      "epoch": 0.02147239263803681,
      "grad_norm": 55.21760634206572,
      "learning_rate": 7.037457434733258e-07,
      "loss": 5.0048,
      "step": 63
    },
    {
      "epoch": 0.021813224267211998,
      "grad_norm": 39.28656961750482,
      "learning_rate": 7.150964812712827e-07,
      "loss": 4.2169,
      "step": 64
    },
    {
      "epoch": 0.022154055896387186,
      "grad_norm": 37.67836056485393,
      "learning_rate": 7.264472190692396e-07,
      "loss": 4.6545,
      "step": 65
    },
    {
      "epoch": 0.022494887525562373,
      "grad_norm": 54.93570670381075,
      "learning_rate": 7.377979568671964e-07,
      "loss": 4.0702,
      "step": 66
    },
    {
      "epoch": 0.02283571915473756,
      "grad_norm": 31.657977048597918,
      "learning_rate": 7.491486946651533e-07,
      "loss": 4.2067,
      "step": 67
    },
    {
      "epoch": 0.023176550783912748,
      "grad_norm": 46.4832275644821,
      "learning_rate": 7.604994324631101e-07,
      "loss": 5.0296,
      "step": 68
    },
    {
      "epoch": 0.023517382413087935,
      "grad_norm": 18.514303480457592,
      "learning_rate": 7.71850170261067e-07,
      "loss": 3.7887,
      "step": 69
    },
    {
      "epoch": 0.023858214042263123,
      "grad_norm": 42.1389573294111,
      "learning_rate": 7.832009080590239e-07,
      "loss": 4.9366,
      "step": 70
    },
    {
      "epoch": 0.02419904567143831,
      "grad_norm": 40.10025378505633,
      "learning_rate": 7.945516458569807e-07,
      "loss": 4.1243,
      "step": 71
    },
    {
      "epoch": 0.024539877300613498,
      "grad_norm": 78.92088848716773,
      "learning_rate": 8.059023836549376e-07,
      "loss": 4.6885,
      "step": 72
    },
    {
      "epoch": 0.024880708929788685,
      "grad_norm": 26.317292561023415,
      "learning_rate": 8.172531214528945e-07,
      "loss": 4.0815,
      "step": 73
    },
    {
      "epoch": 0.025221540558963872,
      "grad_norm": 27.519678827637833,
      "learning_rate": 8.286038592508514e-07,
      "loss": 3.7988,
      "step": 74
    },
    {
      "epoch": 0.02556237218813906,
      "grad_norm": 35.81082589311988,
      "learning_rate": 8.399545970488082e-07,
      "loss": 4.2285,
      "step": 75
    },
    {
      "epoch": 0.025903203817314247,
      "grad_norm": 36.77375833062312,
      "learning_rate": 8.51305334846765e-07,
      "loss": 4.6566,
      "step": 76
    },
    {
      "epoch": 0.026244035446489435,
      "grad_norm": 29.703379585457384,
      "learning_rate": 8.62656072644722e-07,
      "loss": 3.7432,
      "step": 77
    },
    {
      "epoch": 0.026584867075664622,
      "grad_norm": 36.956499057588225,
      "learning_rate": 8.740068104426788e-07,
      "loss": 3.8635,
      "step": 78
    },
    {
      "epoch": 0.02692569870483981,
      "grad_norm": 48.83904439087653,
      "learning_rate": 8.853575482406357e-07,
      "loss": 4.6102,
      "step": 79
    },
    {
      "epoch": 0.027266530334014997,
      "grad_norm": 36.868465057690344,
      "learning_rate": 8.967082860385925e-07,
      "loss": 4.4035,
      "step": 80
    },
    {
      "epoch": 0.027607361963190184,
      "grad_norm": 24.28111065178101,
      "learning_rate": 9.080590238365494e-07,
      "loss": 3.5808,
      "step": 81
    },
    {
      "epoch": 0.02794819359236537,
      "grad_norm": 41.37504780475601,
      "learning_rate": 9.194097616345063e-07,
      "loss": 4.4832,
      "step": 82
    },
    {
      "epoch": 0.02828902522154056,
      "grad_norm": 38.62655025478103,
      "learning_rate": 9.307604994324632e-07,
      "loss": 4.2528,
      "step": 83
    },
    {
      "epoch": 0.028629856850715747,
      "grad_norm": 50.33164580552034,
      "learning_rate": 9.4211123723042e-07,
      "loss": 3.8279,
      "step": 84
    },
    {
      "epoch": 0.028970688479890934,
      "grad_norm": 27.87148676420042,
      "learning_rate": 9.534619750283769e-07,
      "loss": 4.2272,
      "step": 85
    },
    {
      "epoch": 0.02931152010906612,
      "grad_norm": 24.143089590660246,
      "learning_rate": 9.648127128263338e-07,
      "loss": 3.9701,
      "step": 86
    },
    {
      "epoch": 0.02965235173824131,
      "grad_norm": 28.328589618319814,
      "learning_rate": 9.761634506242906e-07,
      "loss": 3.7022,
      "step": 87
    },
    {
      "epoch": 0.029993183367416496,
      "grad_norm": 30.03347409403586,
      "learning_rate": 9.875141884222475e-07,
      "loss": 3.482,
      "step": 88
    },
    {
      "epoch": 0.030334014996591684,
      "grad_norm": 45.232130952103994,
      "learning_rate": 9.988649262202043e-07,
      "loss": 4.8248,
      "step": 89
    },
    {
      "epoch": 0.03067484662576687,
      "grad_norm": 60.78338113303006,
      "learning_rate": 1.0102156640181612e-06,
      "loss": 4.8718,
      "step": 90
    },
    {
      "epoch": 0.03101567825494206,
      "grad_norm": 30.229593062195566,
      "learning_rate": 1.021566401816118e-06,
      "loss": 3.9943,
      "step": 91
    },
    {
      "epoch": 0.03135650988411725,
      "grad_norm": 22.47447511492927,
      "learning_rate": 1.032917139614075e-06,
      "loss": 3.76,
      "step": 92
    },
    {
      "epoch": 0.03169734151329243,
      "grad_norm": 65.43300672545271,
      "learning_rate": 1.044267877412032e-06,
      "loss": 4.3451,
      "step": 93
    },
    {
      "epoch": 0.032038173142467624,
      "grad_norm": 50.831219968265835,
      "learning_rate": 1.0556186152099888e-06,
      "loss": 4.606,
      "step": 94
    },
    {
      "epoch": 0.03237900477164281,
      "grad_norm": 41.0184461017877,
      "learning_rate": 1.0669693530079456e-06,
      "loss": 4.0382,
      "step": 95
    },
    {
      "epoch": 0.032719836400818,
      "grad_norm": 240.12359152440698,
      "learning_rate": 1.0783200908059024e-06,
      "loss": 3.733,
      "step": 96
    },
    {
      "epoch": 0.03306066802999318,
      "grad_norm": 32.320952547613125,
      "learning_rate": 1.0896708286038593e-06,
      "loss": 3.6696,
      "step": 97
    },
    {
      "epoch": 0.033401499659168374,
      "grad_norm": 35.57153441622058,
      "learning_rate": 1.1010215664018161e-06,
      "loss": 4.4483,
      "step": 98
    },
    {
      "epoch": 0.03374233128834356,
      "grad_norm": 49.18224083229904,
      "learning_rate": 1.112372304199773e-06,
      "loss": 4.2309,
      "step": 99
    },
    {
      "epoch": 0.03408316291751875,
      "grad_norm": 21.29971713894319,
      "learning_rate": 1.1237230419977298e-06,
      "loss": 3.5143,
      "step": 100
    },
    {
      "epoch": 0.03442399454669393,
      "grad_norm": 26.171300853316573,
      "learning_rate": 1.1350737797956869e-06,
      "loss": 3.5543,
      "step": 101
    },
    {
      "epoch": 0.034764826175869123,
      "grad_norm": 44.086782963499445,
      "learning_rate": 1.1464245175936437e-06,
      "loss": 4.1766,
      "step": 102
    },
    {
      "epoch": 0.03510565780504431,
      "grad_norm": 30.896609732705638,
      "learning_rate": 1.1577752553916006e-06,
      "loss": 4.1972,
      "step": 103
    },
    {
      "epoch": 0.0354464894342195,
      "grad_norm": 37.03141053528742,
      "learning_rate": 1.1691259931895574e-06,
      "loss": 4.2849,
      "step": 104
    },
    {
      "epoch": 0.03578732106339468,
      "grad_norm": 59.63533148292523,
      "learning_rate": 1.1804767309875143e-06,
      "loss": 4.6286,
      "step": 105
    },
    {
      "epoch": 0.03612815269256987,
      "grad_norm": 20.955901686710448,
      "learning_rate": 1.191827468785471e-06,
      "loss": 3.4592,
      "step": 106
    },
    {
      "epoch": 0.03646898432174506,
      "grad_norm": 22.39571713768874,
      "learning_rate": 1.203178206583428e-06,
      "loss": 3.4562,
      "step": 107
    },
    {
      "epoch": 0.03680981595092025,
      "grad_norm": 29.79417253459404,
      "learning_rate": 1.2145289443813848e-06,
      "loss": 3.8712,
      "step": 108
    },
    {
      "epoch": 0.03715064758009543,
      "grad_norm": 32.09391274068315,
      "learning_rate": 1.2258796821793416e-06,
      "loss": 3.9436,
      "step": 109
    },
    {
      "epoch": 0.03749147920927062,
      "grad_norm": 36.11220965457796,
      "learning_rate": 1.2372304199772987e-06,
      "loss": 3.8297,
      "step": 110
    },
    {
      "epoch": 0.03783231083844581,
      "grad_norm": 76.90999993153758,
      "learning_rate": 1.2485811577752555e-06,
      "loss": 3.3499,
      "step": 111
    },
    {
      "epoch": 0.038173142467621,
      "grad_norm": 38.51303611362027,
      "learning_rate": 1.2599318955732124e-06,
      "loss": 4.1105,
      "step": 112
    },
    {
      "epoch": 0.03851397409679618,
      "grad_norm": 26.591400688979366,
      "learning_rate": 1.2712826333711692e-06,
      "loss": 3.9424,
      "step": 113
    },
    {
      "epoch": 0.03885480572597137,
      "grad_norm": 23.138254071007996,
      "learning_rate": 1.282633371169126e-06,
      "loss": 3.4484,
      "step": 114
    },
    {
      "epoch": 0.039195637355146556,
      "grad_norm": 36.38350498197628,
      "learning_rate": 1.293984108967083e-06,
      "loss": 3.9138,
      "step": 115
    },
    {
      "epoch": 0.03953646898432175,
      "grad_norm": 29.970820256623774,
      "learning_rate": 1.3053348467650397e-06,
      "loss": 3.567,
      "step": 116
    },
    {
      "epoch": 0.03987730061349693,
      "grad_norm": 22.310364536411072,
      "learning_rate": 1.3166855845629966e-06,
      "loss": 3.4333,
      "step": 117
    },
    {
      "epoch": 0.04021813224267212,
      "grad_norm": 29.707320103064053,
      "learning_rate": 1.3280363223609536e-06,
      "loss": 4.0237,
      "step": 118
    },
    {
      "epoch": 0.040558963871847306,
      "grad_norm": 23.363929093222016,
      "learning_rate": 1.3393870601589105e-06,
      "loss": 3.5037,
      "step": 119
    },
    {
      "epoch": 0.0408997955010225,
      "grad_norm": 28.76695583301503,
      "learning_rate": 1.3507377979568673e-06,
      "loss": 3.988,
      "step": 120
    },
    {
      "epoch": 0.04124062713019768,
      "grad_norm": 45.20691980022229,
      "learning_rate": 1.3620885357548242e-06,
      "loss": 4.9047,
      "step": 121
    },
    {
      "epoch": 0.04158145875937287,
      "grad_norm": 39.28631850070014,
      "learning_rate": 1.373439273552781e-06,
      "loss": 4.4538,
      "step": 122
    },
    {
      "epoch": 0.041922290388548056,
      "grad_norm": 31.922807313665288,
      "learning_rate": 1.3847900113507379e-06,
      "loss": 4.317,
      "step": 123
    },
    {
      "epoch": 0.04226312201772325,
      "grad_norm": 26.063870747424055,
      "learning_rate": 1.3961407491486947e-06,
      "loss": 3.2997,
      "step": 124
    },
    {
      "epoch": 0.04260395364689843,
      "grad_norm": 31.9355859080573,
      "learning_rate": 1.4074914869466515e-06,
      "loss": 3.768,
      "step": 125
    },
    {
      "epoch": 0.04294478527607362,
      "grad_norm": 33.89876388844045,
      "learning_rate": 1.4188422247446084e-06,
      "loss": 4.0892,
      "step": 126
    },
    {
      "epoch": 0.043285616905248805,
      "grad_norm": 23.89289470929762,
      "learning_rate": 1.4301929625425654e-06,
      "loss": 3.4874,
      "step": 127
    },
    {
      "epoch": 0.043626448534423996,
      "grad_norm": 20.54350929250958,
      "learning_rate": 1.4415437003405223e-06,
      "loss": 3.6092,
      "step": 128
    },
    {
      "epoch": 0.04396728016359918,
      "grad_norm": 36.61796838886808,
      "learning_rate": 1.4528944381384791e-06,
      "loss": 4.333,
      "step": 129
    },
    {
      "epoch": 0.04430811179277437,
      "grad_norm": 21.070851523451854,
      "learning_rate": 1.464245175936436e-06,
      "loss": 2.9482,
      "step": 130
    },
    {
      "epoch": 0.044648943421949555,
      "grad_norm": 24.326320234915727,
      "learning_rate": 1.4755959137343928e-06,
      "loss": 3.463,
      "step": 131
    },
    {
      "epoch": 0.044989775051124746,
      "grad_norm": 51.98180724699175,
      "learning_rate": 1.4869466515323497e-06,
      "loss": 4.7931,
      "step": 132
    },
    {
      "epoch": 0.04533060668029993,
      "grad_norm": 22.702762626979833,
      "learning_rate": 1.4982973893303065e-06,
      "loss": 3.8917,
      "step": 133
    },
    {
      "epoch": 0.04567143830947512,
      "grad_norm": 60.05517275809835,
      "learning_rate": 1.5096481271282633e-06,
      "loss": 5.0576,
      "step": 134
    },
    {
      "epoch": 0.046012269938650305,
      "grad_norm": 37.619537007165945,
      "learning_rate": 1.5209988649262202e-06,
      "loss": 4.4647,
      "step": 135
    },
    {
      "epoch": 0.046353101567825496,
      "grad_norm": 27.274330826725596,
      "learning_rate": 1.5323496027241772e-06,
      "loss": 3.7104,
      "step": 136
    },
    {
      "epoch": 0.04669393319700068,
      "grad_norm": 35.937893531424805,
      "learning_rate": 1.543700340522134e-06,
      "loss": 3.8603,
      "step": 137
    },
    {
      "epoch": 0.04703476482617587,
      "grad_norm": 23.848134320515413,
      "learning_rate": 1.555051078320091e-06,
      "loss": 3.8803,
      "step": 138
    },
    {
      "epoch": 0.047375596455351054,
      "grad_norm": 20.787555469755514,
      "learning_rate": 1.5664018161180478e-06,
      "loss": 3.3743,
      "step": 139
    },
    {
      "epoch": 0.047716428084526245,
      "grad_norm": 42.73142734647558,
      "learning_rate": 1.5777525539160046e-06,
      "loss": 5.0564,
      "step": 140
    },
    {
      "epoch": 0.04805725971370143,
      "grad_norm": 22.330600936219067,
      "learning_rate": 1.5891032917139615e-06,
      "loss": 3.0635,
      "step": 141
    },
    {
      "epoch": 0.04839809134287662,
      "grad_norm": 18.87190554235252,
      "learning_rate": 1.6004540295119183e-06,
      "loss": 3.7781,
      "step": 142
    },
    {
      "epoch": 0.048738922972051804,
      "grad_norm": 34.798453927798896,
      "learning_rate": 1.6118047673098751e-06,
      "loss": 3.9081,
      "step": 143
    },
    {
      "epoch": 0.049079754601226995,
      "grad_norm": 54.447807214847195,
      "learning_rate": 1.6231555051078322e-06,
      "loss": 4.1109,
      "step": 144
    },
    {
      "epoch": 0.04942058623040218,
      "grad_norm": 28.968541777599967,
      "learning_rate": 1.634506242905789e-06,
      "loss": 3.8615,
      "step": 145
    },
    {
      "epoch": 0.04976141785957737,
      "grad_norm": 47.44566987070836,
      "learning_rate": 1.6458569807037459e-06,
      "loss": 4.1951,
      "step": 146
    },
    {
      "epoch": 0.050102249488752554,
      "grad_norm": 28.43698837152208,
      "learning_rate": 1.6572077185017027e-06,
      "loss": 3.9544,
      "step": 147
    },
    {
      "epoch": 0.050443081117927745,
      "grad_norm": 25.542114310264832,
      "learning_rate": 1.6685584562996596e-06,
      "loss": 3.3524,
      "step": 148
    },
    {
      "epoch": 0.05078391274710293,
      "grad_norm": 38.53664218283615,
      "learning_rate": 1.6799091940976164e-06,
      "loss": 4.4971,
      "step": 149
    },
    {
      "epoch": 0.05112474437627812,
      "grad_norm": 25.138641206315906,
      "learning_rate": 1.6912599318955733e-06,
      "loss": 3.5213,
      "step": 150
    },
    {
      "epoch": 0.0514655760054533,
      "grad_norm": 21.143745115155863,
      "learning_rate": 1.70261066969353e-06,
      "loss": 3.3219,
      "step": 151
    },
    {
      "epoch": 0.051806407634628494,
      "grad_norm": 46.68950919557165,
      "learning_rate": 1.713961407491487e-06,
      "loss": 4.077,
      "step": 152
    },
    {
      "epoch": 0.05214723926380368,
      "grad_norm": 38.39616065933547,
      "learning_rate": 1.725312145289444e-06,
      "loss": 3.6285,
      "step": 153
    },
    {
      "epoch": 0.05248807089297887,
      "grad_norm": 24.216574675911076,
      "learning_rate": 1.7366628830874008e-06,
      "loss": 3.3737,
      "step": 154
    },
    {
      "epoch": 0.05282890252215405,
      "grad_norm": 37.23784866695728,
      "learning_rate": 1.7480136208853577e-06,
      "loss": 4.2556,
      "step": 155
    },
    {
      "epoch": 0.053169734151329244,
      "grad_norm": 47.363117515136565,
      "learning_rate": 1.7593643586833145e-06,
      "loss": 3.5448,
      "step": 156
    },
    {
      "epoch": 0.05351056578050443,
      "grad_norm": 40.1060800000987,
      "learning_rate": 1.7707150964812714e-06,
      "loss": 4.2298,
      "step": 157
    },
    {
      "epoch": 0.05385139740967962,
      "grad_norm": 29.02274262213852,
      "learning_rate": 1.7820658342792282e-06,
      "loss": 3.5804,
      "step": 158
    },
    {
      "epoch": 0.0541922290388548,
      "grad_norm": 61.39155334138686,
      "learning_rate": 1.793416572077185e-06,
      "loss": 4.2972,
      "step": 159
    },
    {
      "epoch": 0.054533060668029994,
      "grad_norm": 60.74957472369227,
      "learning_rate": 1.804767309875142e-06,
      "loss": 3.7249,
      "step": 160
    },
    {
      "epoch": 0.05487389229720518,
      "grad_norm": 55.75039324256153,
      "learning_rate": 1.8161180476730988e-06,
      "loss": 3.6803,
      "step": 161
    },
    {
      "epoch": 0.05521472392638037,
      "grad_norm": 26.772128245163103,
      "learning_rate": 1.8274687854710558e-06,
      "loss": 3.5408,
      "step": 162
    },
    {
      "epoch": 0.05555555555555555,
      "grad_norm": 64.26565595342448,
      "learning_rate": 1.8388195232690126e-06,
      "loss": 3.8707,
      "step": 163
    },
    {
      "epoch": 0.05589638718473074,
      "grad_norm": 27.285922054710852,
      "learning_rate": 1.8501702610669695e-06,
      "loss": 3.3474,
      "step": 164
    },
    {
      "epoch": 0.05623721881390593,
      "grad_norm": 38.94204926018756,
      "learning_rate": 1.8615209988649263e-06,
      "loss": 3.5277,
      "step": 165
    },
    {
      "epoch": 0.05657805044308112,
      "grad_norm": 35.44798550410898,
      "learning_rate": 1.8728717366628832e-06,
      "loss": 3.8651,
      "step": 166
    },
    {
      "epoch": 0.0569188820722563,
      "grad_norm": 28.45287269341736,
      "learning_rate": 1.88422247446084e-06,
      "loss": 3.5238,
      "step": 167
    },
    {
      "epoch": 0.05725971370143149,
      "grad_norm": 20.069880891877332,
      "learning_rate": 1.8955732122587969e-06,
      "loss": 3.1765,
      "step": 168
    },
    {
      "epoch": 0.05760054533060668,
      "grad_norm": 23.588129348117036,
      "learning_rate": 1.9069239500567537e-06,
      "loss": 3.1477,
      "step": 169
    },
    {
      "epoch": 0.05794137695978187,
      "grad_norm": 55.43248756046462,
      "learning_rate": 1.9182746878547106e-06,
      "loss": 3.8716,
      "step": 170
    },
    {
      "epoch": 0.05828220858895705,
      "grad_norm": 22.508382291578755,
      "learning_rate": 1.9296254256526676e-06,
      "loss": 3.6909,
      "step": 171
    },
    {
      "epoch": 0.05862304021813224,
      "grad_norm": 23.67494255734024,
      "learning_rate": 1.9409761634506242e-06,
      "loss": 3.7349,
      "step": 172
    },
    {
      "epoch": 0.058963871847307434,
      "grad_norm": 20.435206992531167,
      "learning_rate": 1.9523269012485813e-06,
      "loss": 3.7444,
      "step": 173
    },
    {
      "epoch": 0.05930470347648262,
      "grad_norm": 27.359592139227342,
      "learning_rate": 1.963677639046538e-06,
      "loss": 3.4833,
      "step": 174
    },
    {
      "epoch": 0.05964553510565781,
      "grad_norm": 41.9918720227676,
      "learning_rate": 1.975028376844495e-06,
      "loss": 4.2869,
      "step": 175
    },
    {
      "epoch": 0.05998636673483299,
      "grad_norm": 24.985571923477423,
      "learning_rate": 1.986379114642452e-06,
      "loss": 3.5109,
      "step": 176
    },
    {
      "epoch": 0.06032719836400818,
      "grad_norm": 18.932698454780553,
      "learning_rate": 1.9977298524404087e-06,
      "loss": 3.5433,
      "step": 177
    },
    {
      "epoch": 0.06066802999318337,
      "grad_norm": 28.394253800610453,
      "learning_rate": 2.0090805902383657e-06,
      "loss": 3.3059,
      "step": 178
    },
    {
      "epoch": 0.06100886162235856,
      "grad_norm": 28.765191774016984,
      "learning_rate": 2.0204313280363224e-06,
      "loss": 3.3615,
      "step": 179
    },
    {
      "epoch": 0.06134969325153374,
      "grad_norm": 21.100033059413505,
      "learning_rate": 2.0317820658342794e-06,
      "loss": 3.2112,
      "step": 180
    },
    {
      "epoch": 0.06169052488070893,
      "grad_norm": 55.569110386457744,
      "learning_rate": 2.043132803632236e-06,
      "loss": 3.3645,
      "step": 181
    },
    {
      "epoch": 0.06203135650988412,
      "grad_norm": 53.56201188920185,
      "learning_rate": 2.054483541430193e-06,
      "loss": 4.3261,
      "step": 182
    },
    {
      "epoch": 0.06237218813905931,
      "grad_norm": 25.61567407848872,
      "learning_rate": 2.06583427922815e-06,
      "loss": 3.6449,
      "step": 183
    },
    {
      "epoch": 0.0627130197682345,
      "grad_norm": 41.31805103738482,
      "learning_rate": 2.0771850170261068e-06,
      "loss": 3.7219,
      "step": 184
    },
    {
      "epoch": 0.06305385139740968,
      "grad_norm": 21.4036248134155,
      "learning_rate": 2.088535754824064e-06,
      "loss": 3.2239,
      "step": 185
    },
    {
      "epoch": 0.06339468302658487,
      "grad_norm": 31.385442932622837,
      "learning_rate": 2.0998864926220205e-06,
      "loss": 3.9935,
      "step": 186
    },
    {
      "epoch": 0.06373551465576005,
      "grad_norm": 54.646220204239526,
      "learning_rate": 2.1112372304199775e-06,
      "loss": 3.4902,
      "step": 187
    },
    {
      "epoch": 0.06407634628493525,
      "grad_norm": 27.29704402724134,
      "learning_rate": 2.122587968217934e-06,
      "loss": 3.4634,
      "step": 188
    },
    {
      "epoch": 0.06441717791411043,
      "grad_norm": 48.73053545881037,
      "learning_rate": 2.1339387060158912e-06,
      "loss": 3.5613,
      "step": 189
    },
    {
      "epoch": 0.06475800954328562,
      "grad_norm": 22.450334845074032,
      "learning_rate": 2.145289443813848e-06,
      "loss": 3.7506,
      "step": 190
    },
    {
      "epoch": 0.0650988411724608,
      "grad_norm": 30.202422660728864,
      "learning_rate": 2.156640181611805e-06,
      "loss": 3.1018,
      "step": 191
    },
    {
      "epoch": 0.065439672801636,
      "grad_norm": 31.806702268103646,
      "learning_rate": 2.167990919409762e-06,
      "loss": 3.5866,
      "step": 192
    },
    {
      "epoch": 0.06578050443081118,
      "grad_norm": 30.308620533555548,
      "learning_rate": 2.1793416572077186e-06,
      "loss": 3.505,
      "step": 193
    },
    {
      "epoch": 0.06612133605998637,
      "grad_norm": 20.514870763856617,
      "learning_rate": 2.1906923950056756e-06,
      "loss": 3.6375,
      "step": 194
    },
    {
      "epoch": 0.06646216768916155,
      "grad_norm": 45.348106445490934,
      "learning_rate": 2.2020431328036323e-06,
      "loss": 3.7995,
      "step": 195
    },
    {
      "epoch": 0.06680299931833675,
      "grad_norm": 23.13881631594261,
      "learning_rate": 2.2133938706015893e-06,
      "loss": 3.3924,
      "step": 196
    },
    {
      "epoch": 0.06714383094751193,
      "grad_norm": 64.19228314441158,
      "learning_rate": 2.224744608399546e-06,
      "loss": 3.4198,
      "step": 197
    },
    {
      "epoch": 0.06748466257668712,
      "grad_norm": 63.01653089024398,
      "learning_rate": 2.236095346197503e-06,
      "loss": 4.098,
      "step": 198
    },
    {
      "epoch": 0.0678254942058623,
      "grad_norm": 16.849431919196444,
      "learning_rate": 2.2474460839954596e-06,
      "loss": 3.4506,
      "step": 199
    },
    {
      "epoch": 0.0681663258350375,
      "grad_norm": 21.187719457127443,
      "learning_rate": 2.2587968217934167e-06,
      "loss": 3.2854,
      "step": 200
    },
    {
      "epoch": 0.06850715746421268,
      "grad_norm": 23.220234293356746,
      "learning_rate": 2.2701475595913738e-06,
      "loss": 3.4267,
      "step": 201
    },
    {
      "epoch": 0.06884798909338787,
      "grad_norm": 35.08164526356085,
      "learning_rate": 2.2814982973893304e-06,
      "loss": 3.8705,
      "step": 202
    },
    {
      "epoch": 0.06918882072256305,
      "grad_norm": 39.076967623239256,
      "learning_rate": 2.2928490351872874e-06,
      "loss": 4.2538,
      "step": 203
    },
    {
      "epoch": 0.06952965235173825,
      "grad_norm": 21.065106342686978,
      "learning_rate": 2.304199772985244e-06,
      "loss": 3.4893,
      "step": 204
    },
    {
      "epoch": 0.06987048398091343,
      "grad_norm": 44.39646906269018,
      "learning_rate": 2.315550510783201e-06,
      "loss": 3.2634,
      "step": 205
    },
    {
      "epoch": 0.07021131561008861,
      "grad_norm": 17.351855017282148,
      "learning_rate": 2.3269012485811578e-06,
      "loss": 2.9929,
      "step": 206
    },
    {
      "epoch": 0.0705521472392638,
      "grad_norm": 49.86759205744471,
      "learning_rate": 2.338251986379115e-06,
      "loss": 3.9951,
      "step": 207
    },
    {
      "epoch": 0.070892978868439,
      "grad_norm": 39.44895992366927,
      "learning_rate": 2.3496027241770714e-06,
      "loss": 4.055,
      "step": 208
    },
    {
      "epoch": 0.07123381049761418,
      "grad_norm": 26.558574772267797,
      "learning_rate": 2.3609534619750285e-06,
      "loss": 3.2005,
      "step": 209
    },
    {
      "epoch": 0.07157464212678936,
      "grad_norm": 23.124943203956146,
      "learning_rate": 2.3723041997729856e-06,
      "loss": 3.4104,
      "step": 210
    },
    {
      "epoch": 0.07191547375596455,
      "grad_norm": 23.54306669958745,
      "learning_rate": 2.383654937570942e-06,
      "loss": 3.3117,
      "step": 211
    },
    {
      "epoch": 0.07225630538513975,
      "grad_norm": 40.676904697274544,
      "learning_rate": 2.3950056753688992e-06,
      "loss": 3.9221,
      "step": 212
    },
    {
      "epoch": 0.07259713701431493,
      "grad_norm": 32.80530857035327,
      "learning_rate": 2.406356413166856e-06,
      "loss": 3.7483,
      "step": 213
    },
    {
      "epoch": 0.07293796864349011,
      "grad_norm": 29.281875034756915,
      "learning_rate": 2.417707150964813e-06,
      "loss": 3.3878,
      "step": 214
    },
    {
      "epoch": 0.0732788002726653,
      "grad_norm": 20.6540063172641,
      "learning_rate": 2.4290578887627696e-06,
      "loss": 3.557,
      "step": 215
    },
    {
      "epoch": 0.0736196319018405,
      "grad_norm": 46.435644939486906,
      "learning_rate": 2.4404086265607266e-06,
      "loss": 3.7634,
      "step": 216
    },
    {
      "epoch": 0.07396046353101568,
      "grad_norm": 25.126693434874223,
      "learning_rate": 2.4517593643586832e-06,
      "loss": 3.6086,
      "step": 217
    },
    {
      "epoch": 0.07430129516019086,
      "grad_norm": 21.810454348790316,
      "learning_rate": 2.4631101021566403e-06,
      "loss": 3.3155,
      "step": 218
    },
    {
      "epoch": 0.07464212678936605,
      "grad_norm": 53.09160991385009,
      "learning_rate": 2.4744608399545974e-06,
      "loss": 3.2462,
      "step": 219
    },
    {
      "epoch": 0.07498295841854125,
      "grad_norm": 32.501693254674706,
      "learning_rate": 2.485811577752554e-06,
      "loss": 3.431,
      "step": 220
    },
    {
      "epoch": 0.07532379004771643,
      "grad_norm": 21.935563148720192,
      "learning_rate": 2.497162315550511e-06,
      "loss": 3.1367,
      "step": 221
    },
    {
      "epoch": 0.07566462167689161,
      "grad_norm": 42.9385355500167,
      "learning_rate": 2.508513053348468e-06,
      "loss": 3.0902,
      "step": 222
    },
    {
      "epoch": 0.0760054533060668,
      "grad_norm": 21.246872285049584,
      "learning_rate": 2.5198637911464247e-06,
      "loss": 3.5296,
      "step": 223
    },
    {
      "epoch": 0.076346284935242,
      "grad_norm": 29.521731400093564,
      "learning_rate": 2.531214528944382e-06,
      "loss": 3.6759,
      "step": 224
    },
    {
      "epoch": 0.07668711656441718,
      "grad_norm": 32.155300381648196,
      "learning_rate": 2.5425652667423384e-06,
      "loss": 3.7502,
      "step": 225
    },
    {
      "epoch": 0.07702794819359236,
      "grad_norm": 25.90320006372729,
      "learning_rate": 2.5539160045402955e-06,
      "loss": 3.1994,
      "step": 226
    },
    {
      "epoch": 0.07736877982276755,
      "grad_norm": 27.789835238756513,
      "learning_rate": 2.565266742338252e-06,
      "loss": 3.1151,
      "step": 227
    },
    {
      "epoch": 0.07770961145194274,
      "grad_norm": 37.09222234028481,
      "learning_rate": 2.576617480136209e-06,
      "loss": 3.797,
      "step": 228
    },
    {
      "epoch": 0.07805044308111793,
      "grad_norm": 30.702416550199718,
      "learning_rate": 2.587968217934166e-06,
      "loss": 3.7105,
      "step": 229
    },
    {
      "epoch": 0.07839127471029311,
      "grad_norm": 30.61150157817821,
      "learning_rate": 2.599318955732123e-06,
      "loss": 3.1572,
      "step": 230
    },
    {
      "epoch": 0.0787321063394683,
      "grad_norm": 35.66433784679161,
      "learning_rate": 2.6106696935300795e-06,
      "loss": 4.1919,
      "step": 231
    },
    {
      "epoch": 0.0790729379686435,
      "grad_norm": 21.043134553681316,
      "learning_rate": 2.6220204313280365e-06,
      "loss": 3.1608,
      "step": 232
    },
    {
      "epoch": 0.07941376959781868,
      "grad_norm": 40.947832982778664,
      "learning_rate": 2.633371169125993e-06,
      "loss": 3.8684,
      "step": 233
    },
    {
      "epoch": 0.07975460122699386,
      "grad_norm": 18.946432759798558,
      "learning_rate": 2.6447219069239506e-06,
      "loss": 2.4338,
      "step": 234
    },
    {
      "epoch": 0.08009543285616905,
      "grad_norm": 26.017493191175593,
      "learning_rate": 2.6560726447219073e-06,
      "loss": 3.3031,
      "step": 235
    },
    {
      "epoch": 0.08043626448534424,
      "grad_norm": 30.10315485251379,
      "learning_rate": 2.6674233825198643e-06,
      "loss": 3.3657,
      "step": 236
    },
    {
      "epoch": 0.08077709611451943,
      "grad_norm": 29.722828331000297,
      "learning_rate": 2.678774120317821e-06,
      "loss": 2.8644,
      "step": 237
    },
    {
      "epoch": 0.08111792774369461,
      "grad_norm": 19.31839303443129,
      "learning_rate": 2.690124858115778e-06,
      "loss": 3.3219,
      "step": 238
    },
    {
      "epoch": 0.0814587593728698,
      "grad_norm": 25.037875988561936,
      "learning_rate": 2.7014755959137347e-06,
      "loss": 3.4312,
      "step": 239
    },
    {
      "epoch": 0.081799591002045,
      "grad_norm": 34.8407242241867,
      "learning_rate": 2.7128263337116917e-06,
      "loss": 4.1811,
      "step": 240
    },
    {
      "epoch": 0.08214042263122018,
      "grad_norm": 31.61384821916224,
      "learning_rate": 2.7241770715096483e-06,
      "loss": 3.506,
      "step": 241
    },
    {
      "epoch": 0.08248125426039536,
      "grad_norm": 23.9847619206856,
      "learning_rate": 2.7355278093076054e-06,
      "loss": 3.4181,
      "step": 242
    },
    {
      "epoch": 0.08282208588957055,
      "grad_norm": 26.209208439183612,
      "learning_rate": 2.746878547105562e-06,
      "loss": 3.7077,
      "step": 243
    },
    {
      "epoch": 0.08316291751874574,
      "grad_norm": 29.23625380361444,
      "learning_rate": 2.758229284903519e-06,
      "loss": 3.7317,
      "step": 244
    },
    {
      "epoch": 0.08350374914792093,
      "grad_norm": 63.439444386741464,
      "learning_rate": 2.7695800227014757e-06,
      "loss": 3.8577,
      "step": 245
    },
    {
      "epoch": 0.08384458077709611,
      "grad_norm": 30.947168607779197,
      "learning_rate": 2.7809307604994328e-06,
      "loss": 3.2762,
      "step": 246
    },
    {
      "epoch": 0.0841854124062713,
      "grad_norm": 26.502783210103775,
      "learning_rate": 2.7922814982973894e-06,
      "loss": 3.3473,
      "step": 247
    },
    {
      "epoch": 0.0845262440354465,
      "grad_norm": 21.072478849628332,
      "learning_rate": 2.8036322360953465e-06,
      "loss": 3.4363,
      "step": 248
    },
    {
      "epoch": 0.08486707566462168,
      "grad_norm": 18.283665968486943,
      "learning_rate": 2.814982973893303e-06,
      "loss": 2.9151,
      "step": 249
    },
    {
      "epoch": 0.08520790729379686,
      "grad_norm": 24.223399567532702,
      "learning_rate": 2.8263337116912606e-06,
      "loss": 3.4902,
      "step": 250
    },
    {
      "epoch": 0.08554873892297205,
      "grad_norm": 25.63036338595163,
      "learning_rate": 2.8376844494892168e-06,
      "loss": 3.2724,
      "step": 251
    },
    {
      "epoch": 0.08588957055214724,
      "grad_norm": 24.978610060838108,
      "learning_rate": 2.8490351872871742e-06,
      "loss": 3.0956,
      "step": 252
    },
    {
      "epoch": 0.08623040218132243,
      "grad_norm": 29.73342604601439,
      "learning_rate": 2.860385925085131e-06,
      "loss": 3.3138,
      "step": 253
    },
    {
      "epoch": 0.08657123381049761,
      "grad_norm": 46.082948063934055,
      "learning_rate": 2.871736662883088e-06,
      "loss": 3.6935,
      "step": 254
    },
    {
      "epoch": 0.0869120654396728,
      "grad_norm": 29.69501560376505,
      "learning_rate": 2.8830874006810446e-06,
      "loss": 3.5614,
      "step": 255
    },
    {
      "epoch": 0.08725289706884799,
      "grad_norm": 30.306980462216288,
      "learning_rate": 2.8944381384790016e-06,
      "loss": 3.6208,
      "step": 256
    },
    {
      "epoch": 0.08759372869802318,
      "grad_norm": 21.886077062127463,
      "learning_rate": 2.9057888762769583e-06,
      "loss": 3.3761,
      "step": 257
    },
    {
      "epoch": 0.08793456032719836,
      "grad_norm": 43.409720943322334,
      "learning_rate": 2.9171396140749153e-06,
      "loss": 3.1294,
      "step": 258
    },
    {
      "epoch": 0.08827539195637356,
      "grad_norm": 26.17275320291272,
      "learning_rate": 2.928490351872872e-06,
      "loss": 3.3809,
      "step": 259
    },
    {
      "epoch": 0.08861622358554874,
      "grad_norm": 25.218953730232002,
      "learning_rate": 2.939841089670829e-06,
      "loss": 3.4094,
      "step": 260
    },
    {
      "epoch": 0.08895705521472393,
      "grad_norm": 31.847086127505893,
      "learning_rate": 2.9511918274687856e-06,
      "loss": 3.7041,
      "step": 261
    },
    {
      "epoch": 0.08929788684389911,
      "grad_norm": 26.17363855057902,
      "learning_rate": 2.9625425652667427e-06,
      "loss": 3.5187,
      "step": 262
    },
    {
      "epoch": 0.08963871847307431,
      "grad_norm": 40.81006293997907,
      "learning_rate": 2.9738933030646993e-06,
      "loss": 3.5497,
      "step": 263
    },
    {
      "epoch": 0.08997955010224949,
      "grad_norm": 16.210255368203537,
      "learning_rate": 2.9852440408626564e-06,
      "loss": 3.1234,
      "step": 264
    },
    {
      "epoch": 0.09032038173142468,
      "grad_norm": 32.31431679645391,
      "learning_rate": 2.996594778660613e-06,
      "loss": 3.9211,
      "step": 265
    },
    {
      "epoch": 0.09066121336059986,
      "grad_norm": 52.38485761652781,
      "learning_rate": 3.00794551645857e-06,
      "loss": 3.1208,
      "step": 266
    },
    {
      "epoch": 0.09100204498977506,
      "grad_norm": 25.842788550648024,
      "learning_rate": 3.0192962542565267e-06,
      "loss": 3.1373,
      "step": 267
    },
    {
      "epoch": 0.09134287661895024,
      "grad_norm": 26.47039528163675,
      "learning_rate": 3.030646992054484e-06,
      "loss": 3.2163,
      "step": 268
    },
    {
      "epoch": 0.09168370824812543,
      "grad_norm": 49.89578481927084,
      "learning_rate": 3.0419977298524404e-06,
      "loss": 3.4747,
      "step": 269
    },
    {
      "epoch": 0.09202453987730061,
      "grad_norm": 26.78144936364749,
      "learning_rate": 3.053348467650398e-06,
      "loss": 3.1215,
      "step": 270
    },
    {
      "epoch": 0.09236537150647581,
      "grad_norm": 26.205871283433847,
      "learning_rate": 3.0646992054483545e-06,
      "loss": 3.479,
      "step": 271
    },
    {
      "epoch": 0.09270620313565099,
      "grad_norm": 20.426963660385688,
      "learning_rate": 3.0760499432463115e-06,
      "loss": 2.7827,
      "step": 272
    },
    {
      "epoch": 0.09304703476482618,
      "grad_norm": 20.55255174066778,
      "learning_rate": 3.087400681044268e-06,
      "loss": 3.1502,
      "step": 273
    },
    {
      "epoch": 0.09338786639400136,
      "grad_norm": 22.762808579770258,
      "learning_rate": 3.0987514188422252e-06,
      "loss": 3.4616,
      "step": 274
    },
    {
      "epoch": 0.09372869802317656,
      "grad_norm": 23.503165311579803,
      "learning_rate": 3.110102156640182e-06,
      "loss": 3.5193,
      "step": 275
    },
    {
      "epoch": 0.09406952965235174,
      "grad_norm": 24.717522772008035,
      "learning_rate": 3.121452894438139e-06,
      "loss": 3.3335,
      "step": 276
    },
    {
      "epoch": 0.09441036128152692,
      "grad_norm": 28.393887039982605,
      "learning_rate": 3.1328036322360955e-06,
      "loss": 3.1796,
      "step": 277
    },
    {
      "epoch": 0.09475119291070211,
      "grad_norm": 23.759519226531328,
      "learning_rate": 3.1441543700340526e-06,
      "loss": 3.258,
      "step": 278
    },
    {
      "epoch": 0.0950920245398773,
      "grad_norm": 18.610174926461337,
      "learning_rate": 3.1555051078320092e-06,
      "loss": 3.4963,
      "step": 279
    },
    {
      "epoch": 0.09543285616905249,
      "grad_norm": 25.30913908219719,
      "learning_rate": 3.1668558456299663e-06,
      "loss": 3.4497,
      "step": 280
    },
    {
      "epoch": 0.09577368779822767,
      "grad_norm": 26.517445169273053,
      "learning_rate": 3.178206583427923e-06,
      "loss": 3.5015,
      "step": 281
    },
    {
      "epoch": 0.09611451942740286,
      "grad_norm": 39.83056712344248,
      "learning_rate": 3.18955732122588e-06,
      "loss": 4.0058,
      "step": 282
    },
    {
      "epoch": 0.09645535105657806,
      "grad_norm": 41.23501814081815,
      "learning_rate": 3.2009080590238366e-06,
      "loss": 4.1369,
      "step": 283
    },
    {
      "epoch": 0.09679618268575324,
      "grad_norm": 23.469824760761757,
      "learning_rate": 3.2122587968217937e-06,
      "loss": 3.1914,
      "step": 284
    },
    {
      "epoch": 0.09713701431492842,
      "grad_norm": 49.76956345485541,
      "learning_rate": 3.2236095346197503e-06,
      "loss": 4.1823,
      "step": 285
    },
    {
      "epoch": 0.09747784594410361,
      "grad_norm": 30.724558910248216,
      "learning_rate": 3.2349602724177078e-06,
      "loss": 3.3774,
      "step": 286
    },
    {
      "epoch": 0.0978186775732788,
      "grad_norm": 26.867356753870524,
      "learning_rate": 3.2463110102156644e-06,
      "loss": 3.6886,
      "step": 287
    },
    {
      "epoch": 0.09815950920245399,
      "grad_norm": 24.641571116943812,
      "learning_rate": 3.2576617480136215e-06,
      "loss": 3.0539,
      "step": 288
    },
    {
      "epoch": 0.09850034083162917,
      "grad_norm": 40.606643400049876,
      "learning_rate": 3.269012485811578e-06,
      "loss": 3.3282,
      "step": 289
    },
    {
      "epoch": 0.09884117246080436,
      "grad_norm": 44.81791437815206,
      "learning_rate": 3.280363223609535e-06,
      "loss": 3.8095,
      "step": 290
    },
    {
      "epoch": 0.09918200408997956,
      "grad_norm": 38.30550736945801,
      "learning_rate": 3.2917139614074918e-06,
      "loss": 3.7563,
      "step": 291
    },
    {
      "epoch": 0.09952283571915474,
      "grad_norm": 18.68954613253475,
      "learning_rate": 3.303064699205449e-06,
      "loss": 3.1461,
      "step": 292
    },
    {
      "epoch": 0.09986366734832992,
      "grad_norm": 29.090557427814645,
      "learning_rate": 3.3144154370034055e-06,
      "loss": 3.4867,
      "step": 293
    },
    {
      "epoch": 0.10020449897750511,
      "grad_norm": 26.14332965390013,
      "learning_rate": 3.3257661748013625e-06,
      "loss": 3.1441,
      "step": 294
    },
    {
      "epoch": 0.1005453306066803,
      "grad_norm": 30.914503124504886,
      "learning_rate": 3.337116912599319e-06,
      "loss": 3.0591,
      "step": 295
    },
    {
      "epoch": 0.10088616223585549,
      "grad_norm": 20.008122622679846,
      "learning_rate": 3.348467650397276e-06,
      "loss": 3.3649,
      "step": 296
    },
    {
      "epoch": 0.10122699386503067,
      "grad_norm": 26.092632920095117,
      "learning_rate": 3.359818388195233e-06,
      "loss": 3.5644,
      "step": 297
    },
    {
      "epoch": 0.10156782549420586,
      "grad_norm": 28.63212438558321,
      "learning_rate": 3.37116912599319e-06,
      "loss": 3.522,
      "step": 298
    },
    {
      "epoch": 0.10190865712338106,
      "grad_norm": 51.509821948389956,
      "learning_rate": 3.3825198637911465e-06,
      "loss": 3.9337,
      "step": 299
    },
    {
      "epoch": 0.10224948875255624,
      "grad_norm": 17.989704453454173,
      "learning_rate": 3.3938706015891036e-06,
      "loss": 2.9136,
      "step": 300
    },
    {
      "epoch": 0.10259032038173142,
      "grad_norm": 16.958469658198492,
      "learning_rate": 3.40522133938706e-06,
      "loss": 2.8021,
      "step": 301
    },
    {
      "epoch": 0.1029311520109066,
      "grad_norm": 37.13133445590453,
      "learning_rate": 3.4165720771850177e-06,
      "loss": 3.51,
      "step": 302
    },
    {
      "epoch": 0.1032719836400818,
      "grad_norm": 22.57628452179918,
      "learning_rate": 3.427922814982974e-06,
      "loss": 3.239,
      "step": 303
    },
    {
      "epoch": 0.10361281526925699,
      "grad_norm": 53.717734359321454,
      "learning_rate": 3.4392735527809314e-06,
      "loss": 3.9258,
      "step": 304
    },
    {
      "epoch": 0.10395364689843217,
      "grad_norm": 28.042921340328537,
      "learning_rate": 3.450624290578888e-06,
      "loss": 3.2924,
      "step": 305
    },
    {
      "epoch": 0.10429447852760736,
      "grad_norm": 15.701072697803744,
      "learning_rate": 3.461975028376845e-06,
      "loss": 3.0604,
      "step": 306
    },
    {
      "epoch": 0.10463531015678255,
      "grad_norm": 23.21846497763572,
      "learning_rate": 3.4733257661748017e-06,
      "loss": 3.2767,
      "step": 307
    },
    {
      "epoch": 0.10497614178595774,
      "grad_norm": 29.370133747684566,
      "learning_rate": 3.4846765039727587e-06,
      "loss": 3.2234,
      "step": 308
    },
    {
      "epoch": 0.10531697341513292,
      "grad_norm": 22.764078085214955,
      "learning_rate": 3.4960272417707154e-06,
      "loss": 3.3851,
      "step": 309
    },
    {
      "epoch": 0.1056578050443081,
      "grad_norm": 23.31570283622387,
      "learning_rate": 3.5073779795686724e-06,
      "loss": 3.4126,
      "step": 310
    },
    {
      "epoch": 0.1059986366734833,
      "grad_norm": 29.42933626565418,
      "learning_rate": 3.518728717366629e-06,
      "loss": 3.9545,
      "step": 311
    },
    {
      "epoch": 0.10633946830265849,
      "grad_norm": 24.436425513801332,
      "learning_rate": 3.530079455164586e-06,
      "loss": 3.4889,
      "step": 312
    },
    {
      "epoch": 0.10668029993183367,
      "grad_norm": 46.79413102742708,
      "learning_rate": 3.5414301929625428e-06,
      "loss": 3.4972,
      "step": 313
    },
    {
      "epoch": 0.10702113156100886,
      "grad_norm": 49.65149493219804,
      "learning_rate": 3.5527809307605e-06,
      "loss": 3.3471,
      "step": 314
    },
    {
      "epoch": 0.10736196319018405,
      "grad_norm": 30.65486345374183,
      "learning_rate": 3.5641316685584564e-06,
      "loss": 3.3228,
      "step": 315
    },
    {
      "epoch": 0.10770279481935924,
      "grad_norm": 39.70390547324314,
      "learning_rate": 3.5754824063564135e-06,
      "loss": 3.3986,
      "step": 316
    },
    {
      "epoch": 0.10804362644853442,
      "grad_norm": 25.14289397698772,
      "learning_rate": 3.58683314415437e-06,
      "loss": 3.536,
      "step": 317
    },
    {
      "epoch": 0.1083844580777096,
      "grad_norm": 26.268354776849872,
      "learning_rate": 3.598183881952327e-06,
      "loss": 3.8046,
      "step": 318
    },
    {
      "epoch": 0.1087252897068848,
      "grad_norm": 29.361490369722848,
      "learning_rate": 3.609534619750284e-06,
      "loss": 3.9811,
      "step": 319
    },
    {
      "epoch": 0.10906612133605999,
      "grad_norm": 21.63712113123274,
      "learning_rate": 3.6208853575482413e-06,
      "loss": 3.3854,
      "step": 320
    },
    {
      "epoch": 0.10940695296523517,
      "grad_norm": 39.485896913323444,
      "learning_rate": 3.6322360953461975e-06,
      "loss": 3.2964,
      "step": 321
    },
    {
      "epoch": 0.10974778459441036,
      "grad_norm": 48.29856764640093,
      "learning_rate": 3.643586833144155e-06,
      "loss": 2.8457,
      "step": 322
    },
    {
      "epoch": 0.11008861622358555,
      "grad_norm": 22.446917758269773,
      "learning_rate": 3.6549375709421116e-06,
      "loss": 3.2262,
      "step": 323
    },
    {
      "epoch": 0.11042944785276074,
      "grad_norm": 29.00727455870548,
      "learning_rate": 3.6662883087400687e-06,
      "loss": 2.9774,
      "step": 324
    },
    {
      "epoch": 0.11077027948193592,
      "grad_norm": 39.75232173436055,
      "learning_rate": 3.6776390465380253e-06,
      "loss": 3.2752,
      "step": 325
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 32.58609950538057,
      "learning_rate": 3.6889897843359824e-06,
      "loss": 3.4757,
      "step": 326
    },
    {
      "epoch": 0.1114519427402863,
      "grad_norm": 26.42898198287206,
      "learning_rate": 3.700340522133939e-06,
      "loss": 3.7568,
      "step": 327
    },
    {
      "epoch": 0.11179277436946149,
      "grad_norm": 44.117489953335216,
      "learning_rate": 3.711691259931896e-06,
      "loss": 3.8,
      "step": 328
    },
    {
      "epoch": 0.11213360599863667,
      "grad_norm": 28.855118482398545,
      "learning_rate": 3.7230419977298527e-06,
      "loss": 3.7875,
      "step": 329
    },
    {
      "epoch": 0.11247443762781185,
      "grad_norm": 25.188994445570888,
      "learning_rate": 3.7343927355278097e-06,
      "loss": 3.3821,
      "step": 330
    },
    {
      "epoch": 0.11281526925698705,
      "grad_norm": 23.229563979127278,
      "learning_rate": 3.7457434733257664e-06,
      "loss": 3.3874,
      "step": 331
    },
    {
      "epoch": 0.11315610088616224,
      "grad_norm": 32.61922999831858,
      "learning_rate": 3.7570942111237234e-06,
      "loss": 3.0305,
      "step": 332
    },
    {
      "epoch": 0.11349693251533742,
      "grad_norm": 33.943960940937544,
      "learning_rate": 3.76844494892168e-06,
      "loss": 3.3457,
      "step": 333
    },
    {
      "epoch": 0.1138377641445126,
      "grad_norm": 23.924585672170455,
      "learning_rate": 3.779795686719637e-06,
      "loss": 3.4819,
      "step": 334
    },
    {
      "epoch": 0.1141785957736878,
      "grad_norm": 16.096262308733213,
      "learning_rate": 3.7911464245175937e-06,
      "loss": 2.8664,
      "step": 335
    },
    {
      "epoch": 0.11451942740286299,
      "grad_norm": 25.72421495803069,
      "learning_rate": 3.8024971623155508e-06,
      "loss": 3.0988,
      "step": 336
    },
    {
      "epoch": 0.11486025903203817,
      "grad_norm": 24.834298977273658,
      "learning_rate": 3.8138479001135074e-06,
      "loss": 3.3421,
      "step": 337
    },
    {
      "epoch": 0.11520109066121335,
      "grad_norm": 24.341257151220084,
      "learning_rate": 3.8251986379114645e-06,
      "loss": 3.4394,
      "step": 338
    },
    {
      "epoch": 0.11554192229038855,
      "grad_norm": 32.284610474230924,
      "learning_rate": 3.836549375709421e-06,
      "loss": 3.2953,
      "step": 339
    },
    {
      "epoch": 0.11588275391956374,
      "grad_norm": 37.77876695563503,
      "learning_rate": 3.847900113507379e-06,
      "loss": 3.3744,
      "step": 340
    },
    {
      "epoch": 0.11622358554873892,
      "grad_norm": 24.832338706859655,
      "learning_rate": 3.859250851305335e-06,
      "loss": 3.7709,
      "step": 341
    },
    {
      "epoch": 0.1165644171779141,
      "grad_norm": 28.61531715327053,
      "learning_rate": 3.870601589103292e-06,
      "loss": 2.8732,
      "step": 342
    },
    {
      "epoch": 0.1169052488070893,
      "grad_norm": 31.007115637423407,
      "learning_rate": 3.8819523269012485e-06,
      "loss": 3.3029,
      "step": 343
    },
    {
      "epoch": 0.11724608043626449,
      "grad_norm": 35.68255577401891,
      "learning_rate": 3.893303064699206e-06,
      "loss": 3.7628,
      "step": 344
    },
    {
      "epoch": 0.11758691206543967,
      "grad_norm": 31.44230178958902,
      "learning_rate": 3.904653802497163e-06,
      "loss": 2.9968,
      "step": 345
    },
    {
      "epoch": 0.11792774369461487,
      "grad_norm": 27.364618915202037,
      "learning_rate": 3.91600454029512e-06,
      "loss": 2.9121,
      "step": 346
    },
    {
      "epoch": 0.11826857532379005,
      "grad_norm": 26.529262353922267,
      "learning_rate": 3.927355278093076e-06,
      "loss": 3.7887,
      "step": 347
    },
    {
      "epoch": 0.11860940695296524,
      "grad_norm": 21.79537082343792,
      "learning_rate": 3.938706015891033e-06,
      "loss": 3.0358,
      "step": 348
    },
    {
      "epoch": 0.11895023858214042,
      "grad_norm": 26.67844714442187,
      "learning_rate": 3.95005675368899e-06,
      "loss": 3.0775,
      "step": 349
    },
    {
      "epoch": 0.11929107021131562,
      "grad_norm": 20.21042511085461,
      "learning_rate": 3.9614074914869474e-06,
      "loss": 3.1757,
      "step": 350
    },
    {
      "epoch": 0.1196319018404908,
      "grad_norm": 28.634986822772976,
      "learning_rate": 3.972758229284904e-06,
      "loss": 3.0582,
      "step": 351
    },
    {
      "epoch": 0.11997273346966598,
      "grad_norm": 27.506678526895456,
      "learning_rate": 3.984108967082861e-06,
      "loss": 3.6261,
      "step": 352
    },
    {
      "epoch": 0.12031356509884117,
      "grad_norm": 26.231454159795575,
      "learning_rate": 3.995459704880817e-06,
      "loss": 3.3565,
      "step": 353
    },
    {
      "epoch": 0.12065439672801637,
      "grad_norm": 37.16585870374642,
      "learning_rate": 4.006810442678775e-06,
      "loss": 3.9849,
      "step": 354
    },
    {
      "epoch": 0.12099522835719155,
      "grad_norm": 47.26712457487932,
      "learning_rate": 4.0181611804767314e-06,
      "loss": 3.8256,
      "step": 355
    },
    {
      "epoch": 0.12133605998636673,
      "grad_norm": 29.3737643737477,
      "learning_rate": 4.029511918274688e-06,
      "loss": 3.6536,
      "step": 356
    },
    {
      "epoch": 0.12167689161554192,
      "grad_norm": 22.68092648195418,
      "learning_rate": 4.040862656072645e-06,
      "loss": 3.0232,
      "step": 357
    },
    {
      "epoch": 0.12201772324471712,
      "grad_norm": 21.384119550256347,
      "learning_rate": 4.052213393870602e-06,
      "loss": 3.2522,
      "step": 358
    },
    {
      "epoch": 0.1223585548738923,
      "grad_norm": 25.061308891750326,
      "learning_rate": 4.063564131668559e-06,
      "loss": 3.3553,
      "step": 359
    },
    {
      "epoch": 0.12269938650306748,
      "grad_norm": 37.347807645377095,
      "learning_rate": 4.0749148694665155e-06,
      "loss": 3.8297,
      "step": 360
    },
    {
      "epoch": 0.12304021813224267,
      "grad_norm": 37.61184907576129,
      "learning_rate": 4.086265607264472e-06,
      "loss": 3.1193,
      "step": 361
    },
    {
      "epoch": 0.12338104976141787,
      "grad_norm": 25.871013292402292,
      "learning_rate": 4.0976163450624296e-06,
      "loss": 3.3804,
      "step": 362
    },
    {
      "epoch": 0.12372188139059305,
      "grad_norm": 15.759794149377655,
      "learning_rate": 4.108967082860386e-06,
      "loss": 3.2964,
      "step": 363
    },
    {
      "epoch": 0.12406271301976823,
      "grad_norm": 28.208752219308547,
      "learning_rate": 4.120317820658344e-06,
      "loss": 3.5275,
      "step": 364
    },
    {
      "epoch": 0.12440354464894342,
      "grad_norm": 22.88555566293089,
      "learning_rate": 4.1316685584563e-06,
      "loss": 3.3824,
      "step": 365
    },
    {
      "epoch": 0.12474437627811862,
      "grad_norm": 23.371507076425747,
      "learning_rate": 4.143019296254257e-06,
      "loss": 3.3156,
      "step": 366
    },
    {
      "epoch": 0.1250852079072938,
      "grad_norm": 39.601470648223604,
      "learning_rate": 4.1543700340522136e-06,
      "loss": 3.4735,
      "step": 367
    },
    {
      "epoch": 0.125426039536469,
      "grad_norm": 43.084315254484984,
      "learning_rate": 4.165720771850171e-06,
      "loss": 3.4301,
      "step": 368
    },
    {
      "epoch": 0.12576687116564417,
      "grad_norm": 25.60388581382586,
      "learning_rate": 4.177071509648128e-06,
      "loss": 3.3273,
      "step": 369
    },
    {
      "epoch": 0.12610770279481937,
      "grad_norm": 19.273449561518348,
      "learning_rate": 4.188422247446084e-06,
      "loss": 3.2389,
      "step": 370
    },
    {
      "epoch": 0.12644853442399454,
      "grad_norm": 21.658112109470267,
      "learning_rate": 4.199772985244041e-06,
      "loss": 3.0686,
      "step": 371
    },
    {
      "epoch": 0.12678936605316973,
      "grad_norm": 39.350072375815394,
      "learning_rate": 4.211123723041998e-06,
      "loss": 3.7478,
      "step": 372
    },
    {
      "epoch": 0.12713019768234493,
      "grad_norm": 23.2680315384974,
      "learning_rate": 4.222474460839955e-06,
      "loss": 3.6035,
      "step": 373
    },
    {
      "epoch": 0.1274710293115201,
      "grad_norm": 21.861461330391762,
      "learning_rate": 4.233825198637912e-06,
      "loss": 3.7554,
      "step": 374
    },
    {
      "epoch": 0.1278118609406953,
      "grad_norm": 35.21458035767287,
      "learning_rate": 4.245175936435868e-06,
      "loss": 3.4164,
      "step": 375
    },
    {
      "epoch": 0.1281526925698705,
      "grad_norm": 22.400518372108994,
      "learning_rate": 4.256526674233826e-06,
      "loss": 2.9454,
      "step": 376
    },
    {
      "epoch": 0.12849352419904567,
      "grad_norm": 23.00724277264511,
      "learning_rate": 4.2678774120317824e-06,
      "loss": 3.412,
      "step": 377
    },
    {
      "epoch": 0.12883435582822086,
      "grad_norm": 58.01941704410782,
      "learning_rate": 4.279228149829739e-06,
      "loss": 4.0357,
      "step": 378
    },
    {
      "epoch": 0.12917518745739603,
      "grad_norm": 29.36389101275093,
      "learning_rate": 4.290578887627696e-06,
      "loss": 3.2109,
      "step": 379
    },
    {
      "epoch": 0.12951601908657123,
      "grad_norm": 24.386984826179006,
      "learning_rate": 4.301929625425653e-06,
      "loss": 3.5581,
      "step": 380
    },
    {
      "epoch": 0.12985685071574643,
      "grad_norm": 14.693266308882807,
      "learning_rate": 4.31328036322361e-06,
      "loss": 3.0419,
      "step": 381
    },
    {
      "epoch": 0.1301976823449216,
      "grad_norm": 26.30594134368428,
      "learning_rate": 4.324631101021567e-06,
      "loss": 3.4317,
      "step": 382
    },
    {
      "epoch": 0.1305385139740968,
      "grad_norm": 33.68306371776399,
      "learning_rate": 4.335981838819524e-06,
      "loss": 3.491,
      "step": 383
    },
    {
      "epoch": 0.130879345603272,
      "grad_norm": 33.75437605588583,
      "learning_rate": 4.3473325766174805e-06,
      "loss": 3.7256,
      "step": 384
    },
    {
      "epoch": 0.13122017723244717,
      "grad_norm": 41.01389346741316,
      "learning_rate": 4.358683314415437e-06,
      "loss": 3.8598,
      "step": 385
    },
    {
      "epoch": 0.13156100886162236,
      "grad_norm": 34.23819532149877,
      "learning_rate": 4.370034052213395e-06,
      "loss": 4.1207,
      "step": 386
    },
    {
      "epoch": 0.13190184049079753,
      "grad_norm": 27.61097850678461,
      "learning_rate": 4.381384790011351e-06,
      "loss": 2.8575,
      "step": 387
    },
    {
      "epoch": 0.13224267211997273,
      "grad_norm": 23.45228675076355,
      "learning_rate": 4.392735527809308e-06,
      "loss": 3.726,
      "step": 388
    },
    {
      "epoch": 0.13258350374914793,
      "grad_norm": 22.0667682218867,
      "learning_rate": 4.4040862656072645e-06,
      "loss": 3.1915,
      "step": 389
    },
    {
      "epoch": 0.1329243353783231,
      "grad_norm": 23.865899938088276,
      "learning_rate": 4.415437003405222e-06,
      "loss": 3.0718,
      "step": 390
    },
    {
      "epoch": 0.1332651670074983,
      "grad_norm": 28.23058669748859,
      "learning_rate": 4.426787741203179e-06,
      "loss": 3.2846,
      "step": 391
    },
    {
      "epoch": 0.1336059986366735,
      "grad_norm": 25.031098279397305,
      "learning_rate": 4.438138479001135e-06,
      "loss": 3.1893,
      "step": 392
    },
    {
      "epoch": 0.13394683026584867,
      "grad_norm": 50.6705506623126,
      "learning_rate": 4.449489216799092e-06,
      "loss": 3.2688,
      "step": 393
    },
    {
      "epoch": 0.13428766189502386,
      "grad_norm": 40.111075010992806,
      "learning_rate": 4.460839954597049e-06,
      "loss": 3.4855,
      "step": 394
    },
    {
      "epoch": 0.13462849352419903,
      "grad_norm": 25.88438425427351,
      "learning_rate": 4.472190692395006e-06,
      "loss": 3.4344,
      "step": 395
    },
    {
      "epoch": 0.13496932515337423,
      "grad_norm": 22.77529309840879,
      "learning_rate": 4.483541430192963e-06,
      "loss": 3.2855,
      "step": 396
    },
    {
      "epoch": 0.13531015678254943,
      "grad_norm": 27.03185132668022,
      "learning_rate": 4.494892167990919e-06,
      "loss": 3.2646,
      "step": 397
    },
    {
      "epoch": 0.1356509884117246,
      "grad_norm": 23.450874970517727,
      "learning_rate": 4.506242905788877e-06,
      "loss": 3.2668,
      "step": 398
    },
    {
      "epoch": 0.1359918200408998,
      "grad_norm": 24.529120168665212,
      "learning_rate": 4.517593643586833e-06,
      "loss": 3.3584,
      "step": 399
    },
    {
      "epoch": 0.136332651670075,
      "grad_norm": 24.211555751848802,
      "learning_rate": 4.528944381384791e-06,
      "loss": 3.1627,
      "step": 400
    },
    {
      "epoch": 0.13667348329925016,
      "grad_norm": 55.91516904658619,
      "learning_rate": 4.5402951191827475e-06,
      "loss": 3.5566,
      "step": 401
    },
    {
      "epoch": 0.13701431492842536,
      "grad_norm": 138.19376733662983,
      "learning_rate": 4.551645856980704e-06,
      "loss": 3.6277,
      "step": 402
    },
    {
      "epoch": 0.13735514655760053,
      "grad_norm": 27.820527345360514,
      "learning_rate": 4.562996594778661e-06,
      "loss": 3.637,
      "step": 403
    },
    {
      "epoch": 0.13769597818677573,
      "grad_norm": 21.24471506694553,
      "learning_rate": 4.574347332576618e-06,
      "loss": 2.969,
      "step": 404
    },
    {
      "epoch": 0.13803680981595093,
      "grad_norm": 30.292261750606603,
      "learning_rate": 4.585698070374575e-06,
      "loss": 3.5344,
      "step": 405
    },
    {
      "epoch": 0.1383776414451261,
      "grad_norm": 41.426628831148186,
      "learning_rate": 4.5970488081725315e-06,
      "loss": 3.3624,
      "step": 406
    },
    {
      "epoch": 0.1387184730743013,
      "grad_norm": 32.028039365339126,
      "learning_rate": 4.608399545970488e-06,
      "loss": 3.7061,
      "step": 407
    },
    {
      "epoch": 0.1390593047034765,
      "grad_norm": 24.50239499978448,
      "learning_rate": 4.619750283768446e-06,
      "loss": 3.3865,
      "step": 408
    },
    {
      "epoch": 0.13940013633265166,
      "grad_norm": 28.232554442795003,
      "learning_rate": 4.631101021566402e-06,
      "loss": 3.6223,
      "step": 409
    },
    {
      "epoch": 0.13974096796182686,
      "grad_norm": 22.960684497905188,
      "learning_rate": 4.642451759364359e-06,
      "loss": 3.3113,
      "step": 410
    },
    {
      "epoch": 0.14008179959100203,
      "grad_norm": 47.42477507605025,
      "learning_rate": 4.6538024971623155e-06,
      "loss": 3.1159,
      "step": 411
    },
    {
      "epoch": 0.14042263122017723,
      "grad_norm": 26.396505662440386,
      "learning_rate": 4.665153234960273e-06,
      "loss": 3.9023,
      "step": 412
    },
    {
      "epoch": 0.14076346284935243,
      "grad_norm": 17.65391775520735,
      "learning_rate": 4.67650397275823e-06,
      "loss": 2.8528,
      "step": 413
    },
    {
      "epoch": 0.1411042944785276,
      "grad_norm": 20.859295117894572,
      "learning_rate": 4.687854710556186e-06,
      "loss": 3.0974,
      "step": 414
    },
    {
      "epoch": 0.1414451261077028,
      "grad_norm": 22.226854348839577,
      "learning_rate": 4.699205448354143e-06,
      "loss": 3.5092,
      "step": 415
    },
    {
      "epoch": 0.141785957736878,
      "grad_norm": 28.930763959644306,
      "learning_rate": 4.7105561861521e-06,
      "loss": 3.2303,
      "step": 416
    },
    {
      "epoch": 0.14212678936605316,
      "grad_norm": 26.66244288975185,
      "learning_rate": 4.721906923950057e-06,
      "loss": 3.4155,
      "step": 417
    },
    {
      "epoch": 0.14246762099522836,
      "grad_norm": 29.22823274241426,
      "learning_rate": 4.7332576617480145e-06,
      "loss": 3.6289,
      "step": 418
    },
    {
      "epoch": 0.14280845262440353,
      "grad_norm": 37.3151455049413,
      "learning_rate": 4.744608399545971e-06,
      "loss": 3.8072,
      "step": 419
    },
    {
      "epoch": 0.14314928425357873,
      "grad_norm": 22.914006429614716,
      "learning_rate": 4.755959137343928e-06,
      "loss": 3.575,
      "step": 420
    },
    {
      "epoch": 0.14349011588275393,
      "grad_norm": 40.32178960374478,
      "learning_rate": 4.767309875141884e-06,
      "loss": 3.8046,
      "step": 421
    },
    {
      "epoch": 0.1438309475119291,
      "grad_norm": 32.34788563941903,
      "learning_rate": 4.778660612939842e-06,
      "loss": 2.5498,
      "step": 422
    },
    {
      "epoch": 0.1441717791411043,
      "grad_norm": 44.54710274991431,
      "learning_rate": 4.7900113507377985e-06,
      "loss": 3.2741,
      "step": 423
    },
    {
      "epoch": 0.1445126107702795,
      "grad_norm": 23.897792650030343,
      "learning_rate": 4.801362088535755e-06,
      "loss": 3.3159,
      "step": 424
    },
    {
      "epoch": 0.14485344239945466,
      "grad_norm": 16.501880410246084,
      "learning_rate": 4.812712826333712e-06,
      "loss": 3.2318,
      "step": 425
    },
    {
      "epoch": 0.14519427402862986,
      "grad_norm": 20.025380338093218,
      "learning_rate": 4.824063564131669e-06,
      "loss": 3.7572,
      "step": 426
    },
    {
      "epoch": 0.14553510565780503,
      "grad_norm": 32.837418179281165,
      "learning_rate": 4.835414301929626e-06,
      "loss": 4.0379,
      "step": 427
    },
    {
      "epoch": 0.14587593728698023,
      "grad_norm": 64.13756407890655,
      "learning_rate": 4.8467650397275825e-06,
      "loss": 3.3553,
      "step": 428
    },
    {
      "epoch": 0.14621676891615543,
      "grad_norm": 23.19828159290423,
      "learning_rate": 4.858115777525539e-06,
      "loss": 3.4679,
      "step": 429
    },
    {
      "epoch": 0.1465576005453306,
      "grad_norm": 28.680765735465517,
      "learning_rate": 4.869466515323497e-06,
      "loss": 3.9917,
      "step": 430
    },
    {
      "epoch": 0.1468984321745058,
      "grad_norm": 38.239247613809745,
      "learning_rate": 4.880817253121453e-06,
      "loss": 3.2472,
      "step": 431
    },
    {
      "epoch": 0.147239263803681,
      "grad_norm": 33.19618272455649,
      "learning_rate": 4.892167990919411e-06,
      "loss": 3.6932,
      "step": 432
    },
    {
      "epoch": 0.14758009543285616,
      "grad_norm": 19.12148709003621,
      "learning_rate": 4.9035187287173665e-06,
      "loss": 3.0767,
      "step": 433
    },
    {
      "epoch": 0.14792092706203136,
      "grad_norm": 30.19396243550386,
      "learning_rate": 4.914869466515324e-06,
      "loss": 3.7404,
      "step": 434
    },
    {
      "epoch": 0.14826175869120656,
      "grad_norm": 25.728042791737526,
      "learning_rate": 4.926220204313281e-06,
      "loss": 3.4838,
      "step": 435
    },
    {
      "epoch": 0.14860259032038173,
      "grad_norm": 24.68965124795127,
      "learning_rate": 4.937570942111238e-06,
      "loss": 3.7406,
      "step": 436
    },
    {
      "epoch": 0.14894342194955693,
      "grad_norm": 18.52073798040104,
      "learning_rate": 4.948921679909195e-06,
      "loss": 3.2825,
      "step": 437
    },
    {
      "epoch": 0.1492842535787321,
      "grad_norm": 19.87282104858766,
      "learning_rate": 4.960272417707151e-06,
      "loss": 3.4925,
      "step": 438
    },
    {
      "epoch": 0.1496250852079073,
      "grad_norm": 20.312571415307406,
      "learning_rate": 4.971623155505108e-06,
      "loss": 3.4295,
      "step": 439
    },
    {
      "epoch": 0.1499659168370825,
      "grad_norm": 27.808130443588837,
      "learning_rate": 4.9829738933030655e-06,
      "loss": 3.0021,
      "step": 440
    },
    {
      "epoch": 0.15030674846625766,
      "grad_norm": 23.175410101640328,
      "learning_rate": 4.994324631101022e-06,
      "loss": 3.3455,
      "step": 441
    },
    {
      "epoch": 0.15064758009543286,
      "grad_norm": 23.629373808760466,
      "learning_rate": 5.005675368898979e-06,
      "loss": 3.1739,
      "step": 442
    },
    {
      "epoch": 0.15098841172460806,
      "grad_norm": 29.6399617419849,
      "learning_rate": 5.017026106696936e-06,
      "loss": 3.1252,
      "step": 443
    },
    {
      "epoch": 0.15132924335378323,
      "grad_norm": 21.172144751351954,
      "learning_rate": 5.028376844494892e-06,
      "loss": 3.3184,
      "step": 444
    },
    {
      "epoch": 0.15167007498295842,
      "grad_norm": 23.057357184728996,
      "learning_rate": 5.0397275822928495e-06,
      "loss": 3.5939,
      "step": 445
    },
    {
      "epoch": 0.1520109066121336,
      "grad_norm": 27.113252748657676,
      "learning_rate": 5.051078320090806e-06,
      "loss": 3.1781,
      "step": 446
    },
    {
      "epoch": 0.1523517382413088,
      "grad_norm": 36.367359236269095,
      "learning_rate": 5.062429057888764e-06,
      "loss": 2.9693,
      "step": 447
    },
    {
      "epoch": 0.152692569870484,
      "grad_norm": 25.30322545526685,
      "learning_rate": 5.073779795686719e-06,
      "loss": 3.2579,
      "step": 448
    },
    {
      "epoch": 0.15303340149965916,
      "grad_norm": 19.654449748345346,
      "learning_rate": 5.085130533484677e-06,
      "loss": 3.4974,
      "step": 449
    },
    {
      "epoch": 0.15337423312883436,
      "grad_norm": 43.26822056020597,
      "learning_rate": 5.096481271282634e-06,
      "loss": 3.2836,
      "step": 450
    },
    {
      "epoch": 0.15371506475800956,
      "grad_norm": 24.330062143852118,
      "learning_rate": 5.107832009080591e-06,
      "loss": 3.2408,
      "step": 451
    },
    {
      "epoch": 0.15405589638718473,
      "grad_norm": 19.133937994585818,
      "learning_rate": 5.119182746878547e-06,
      "loss": 3.0242,
      "step": 452
    },
    {
      "epoch": 0.15439672801635992,
      "grad_norm": 48.633685665009445,
      "learning_rate": 5.130533484676504e-06,
      "loss": 3.0877,
      "step": 453
    },
    {
      "epoch": 0.1547375596455351,
      "grad_norm": 26.514135494712963,
      "learning_rate": 5.141884222474462e-06,
      "loss": 3.0649,
      "step": 454
    },
    {
      "epoch": 0.1550783912747103,
      "grad_norm": 20.996290329763468,
      "learning_rate": 5.153234960272418e-06,
      "loss": 3.0978,
      "step": 455
    },
    {
      "epoch": 0.1554192229038855,
      "grad_norm": 61.16073139055613,
      "learning_rate": 5.164585698070375e-06,
      "loss": 3.8337,
      "step": 456
    },
    {
      "epoch": 0.15576005453306066,
      "grad_norm": 27.475616250411083,
      "learning_rate": 5.175936435868332e-06,
      "loss": 3.3255,
      "step": 457
    },
    {
      "epoch": 0.15610088616223586,
      "grad_norm": 24.15334960963796,
      "learning_rate": 5.187287173666289e-06,
      "loss": 2.8465,
      "step": 458
    },
    {
      "epoch": 0.15644171779141106,
      "grad_norm": 18.78017018473755,
      "learning_rate": 5.198637911464246e-06,
      "loss": 2.992,
      "step": 459
    },
    {
      "epoch": 0.15678254942058623,
      "grad_norm": 27.93688557776103,
      "learning_rate": 5.209988649262202e-06,
      "loss": 2.8597,
      "step": 460
    },
    {
      "epoch": 0.15712338104976142,
      "grad_norm": 27.112359502110564,
      "learning_rate": 5.221339387060159e-06,
      "loss": 3.4835,
      "step": 461
    },
    {
      "epoch": 0.1574642126789366,
      "grad_norm": 29.72278679250819,
      "learning_rate": 5.2326901248581164e-06,
      "loss": 3.5909,
      "step": 462
    },
    {
      "epoch": 0.1578050443081118,
      "grad_norm": 21.40637610650895,
      "learning_rate": 5.244040862656073e-06,
      "loss": 3.4594,
      "step": 463
    },
    {
      "epoch": 0.158145875937287,
      "grad_norm": 61.602496668607145,
      "learning_rate": 5.25539160045403e-06,
      "loss": 3.3699,
      "step": 464
    },
    {
      "epoch": 0.15848670756646216,
      "grad_norm": 28.840009983672616,
      "learning_rate": 5.266742338251986e-06,
      "loss": 3.2751,
      "step": 465
    },
    {
      "epoch": 0.15882753919563736,
      "grad_norm": 26.81365738265123,
      "learning_rate": 5.278093076049944e-06,
      "loss": 3.3301,
      "step": 466
    },
    {
      "epoch": 0.15916837082481256,
      "grad_norm": 27.925863765473796,
      "learning_rate": 5.289443813847901e-06,
      "loss": 3.4233,
      "step": 467
    },
    {
      "epoch": 0.15950920245398773,
      "grad_norm": 47.48414190894158,
      "learning_rate": 5.300794551645857e-06,
      "loss": 3.4316,
      "step": 468
    },
    {
      "epoch": 0.15985003408316292,
      "grad_norm": 70.64816157414505,
      "learning_rate": 5.3121452894438146e-06,
      "loss": 3.5034,
      "step": 469
    },
    {
      "epoch": 0.1601908657123381,
      "grad_norm": 20.114132320428418,
      "learning_rate": 5.323496027241771e-06,
      "loss": 3.1561,
      "step": 470
    },
    {
      "epoch": 0.1605316973415133,
      "grad_norm": 34.28986509374082,
      "learning_rate": 5.334846765039729e-06,
      "loss": 2.7342,
      "step": 471
    },
    {
      "epoch": 0.1608725289706885,
      "grad_norm": 27.39827876146415,
      "learning_rate": 5.3461975028376844e-06,
      "loss": 3.907,
      "step": 472
    },
    {
      "epoch": 0.16121336059986366,
      "grad_norm": 30.980595751698406,
      "learning_rate": 5.357548240635642e-06,
      "loss": 3.1856,
      "step": 473
    },
    {
      "epoch": 0.16155419222903886,
      "grad_norm": 20.34585793915918,
      "learning_rate": 5.3688989784335986e-06,
      "loss": 3.2871,
      "step": 474
    },
    {
      "epoch": 0.16189502385821405,
      "grad_norm": 29.613228431912475,
      "learning_rate": 5.380249716231556e-06,
      "loss": 3.464,
      "step": 475
    },
    {
      "epoch": 0.16223585548738922,
      "grad_norm": 16.529127002039914,
      "learning_rate": 5.391600454029512e-06,
      "loss": 3.372,
      "step": 476
    },
    {
      "epoch": 0.16257668711656442,
      "grad_norm": 25.36139656186472,
      "learning_rate": 5.402951191827469e-06,
      "loss": 2.934,
      "step": 477
    },
    {
      "epoch": 0.1629175187457396,
      "grad_norm": 32.8902531556671,
      "learning_rate": 5.414301929625426e-06,
      "loss": 3.6385,
      "step": 478
    },
    {
      "epoch": 0.1632583503749148,
      "grad_norm": 19.83815607578902,
      "learning_rate": 5.425652667423383e-06,
      "loss": 2.9722,
      "step": 479
    },
    {
      "epoch": 0.16359918200409,
      "grad_norm": 43.49204282546653,
      "learning_rate": 5.437003405221339e-06,
      "loss": 3.7345,
      "step": 480
    },
    {
      "epoch": 0.16394001363326516,
      "grad_norm": 37.19314324987305,
      "learning_rate": 5.448354143019297e-06,
      "loss": 3.5872,
      "step": 481
    },
    {
      "epoch": 0.16428084526244036,
      "grad_norm": 24.74836508973546,
      "learning_rate": 5.459704880817253e-06,
      "loss": 3.5791,
      "step": 482
    },
    {
      "epoch": 0.16462167689161555,
      "grad_norm": 21.958278571643792,
      "learning_rate": 5.471055618615211e-06,
      "loss": 3.1751,
      "step": 483
    },
    {
      "epoch": 0.16496250852079072,
      "grad_norm": 36.33409307960184,
      "learning_rate": 5.4824063564131666e-06,
      "loss": 3.2012,
      "step": 484
    },
    {
      "epoch": 0.16530334014996592,
      "grad_norm": 32.96927940111071,
      "learning_rate": 5.493757094211124e-06,
      "loss": 3.3522,
      "step": 485
    },
    {
      "epoch": 0.1656441717791411,
      "grad_norm": 25.06118256145394,
      "learning_rate": 5.5051078320090815e-06,
      "loss": 3.7034,
      "step": 486
    },
    {
      "epoch": 0.1659850034083163,
      "grad_norm": 22.543937631502462,
      "learning_rate": 5.516458569807038e-06,
      "loss": 3.4567,
      "step": 487
    },
    {
      "epoch": 0.1663258350374915,
      "grad_norm": 52.51359285146997,
      "learning_rate": 5.527809307604994e-06,
      "loss": 3.9333,
      "step": 488
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 26.021244247952673,
      "learning_rate": 5.539160045402951e-06,
      "loss": 3.5291,
      "step": 489
    },
    {
      "epoch": 0.16700749829584186,
      "grad_norm": 26.653321198038487,
      "learning_rate": 5.550510783200909e-06,
      "loss": 3.4493,
      "step": 490
    },
    {
      "epoch": 0.16734832992501705,
      "grad_norm": 29.705759265257434,
      "learning_rate": 5.5618615209988655e-06,
      "loss": 3.2509,
      "step": 491
    },
    {
      "epoch": 0.16768916155419222,
      "grad_norm": 23.62812249973601,
      "learning_rate": 5.573212258796822e-06,
      "loss": 2.8018,
      "step": 492
    },
    {
      "epoch": 0.16802999318336742,
      "grad_norm": 21.88425871582131,
      "learning_rate": 5.584562996594779e-06,
      "loss": 3.1126,
      "step": 493
    },
    {
      "epoch": 0.1683708248125426,
      "grad_norm": 24.546216850289547,
      "learning_rate": 5.595913734392736e-06,
      "loss": 2.7836,
      "step": 494
    },
    {
      "epoch": 0.1687116564417178,
      "grad_norm": 34.82866566566624,
      "learning_rate": 5.607264472190693e-06,
      "loss": 3.9109,
      "step": 495
    },
    {
      "epoch": 0.169052488070893,
      "grad_norm": 29.68630132767053,
      "learning_rate": 5.6186152099886495e-06,
      "loss": 3.282,
      "step": 496
    },
    {
      "epoch": 0.16939331970006816,
      "grad_norm": 36.31462653618816,
      "learning_rate": 5.629965947786606e-06,
      "loss": 3.0522,
      "step": 497
    },
    {
      "epoch": 0.16973415132924335,
      "grad_norm": 38.29860915861816,
      "learning_rate": 5.641316685584564e-06,
      "loss": 3.3108,
      "step": 498
    },
    {
      "epoch": 0.17007498295841855,
      "grad_norm": 20.82118546458167,
      "learning_rate": 5.652667423382521e-06,
      "loss": 3.0351,
      "step": 499
    },
    {
      "epoch": 0.17041581458759372,
      "grad_norm": 34.82508193823718,
      "learning_rate": 5.664018161180477e-06,
      "loss": 3.3756,
      "step": 500
    },
    {
      "epoch": 0.17075664621676892,
      "grad_norm": 23.454176736152267,
      "learning_rate": 5.6753688989784335e-06,
      "loss": 3.0129,
      "step": 501
    },
    {
      "epoch": 0.1710974778459441,
      "grad_norm": 30.301950651779755,
      "learning_rate": 5.686719636776391e-06,
      "loss": 3.9632,
      "step": 502
    },
    {
      "epoch": 0.1714383094751193,
      "grad_norm": 21.07094418329733,
      "learning_rate": 5.6980703745743485e-06,
      "loss": 3.1781,
      "step": 503
    },
    {
      "epoch": 0.17177914110429449,
      "grad_norm": 21.8762824271937,
      "learning_rate": 5.709421112372304e-06,
      "loss": 3.1337,
      "step": 504
    },
    {
      "epoch": 0.17211997273346966,
      "grad_norm": 21.282604205711603,
      "learning_rate": 5.720771850170262e-06,
      "loss": 3.0701,
      "step": 505
    },
    {
      "epoch": 0.17246080436264485,
      "grad_norm": 30.424910670065746,
      "learning_rate": 5.732122587968218e-06,
      "loss": 2.9431,
      "step": 506
    },
    {
      "epoch": 0.17280163599182005,
      "grad_norm": 30.62379258718881,
      "learning_rate": 5.743473325766176e-06,
      "loss": 3.6791,
      "step": 507
    },
    {
      "epoch": 0.17314246762099522,
      "grad_norm": 21.003050551550963,
      "learning_rate": 5.754824063564132e-06,
      "loss": 3.4311,
      "step": 508
    },
    {
      "epoch": 0.17348329925017042,
      "grad_norm": 60.192278779691044,
      "learning_rate": 5.766174801362089e-06,
      "loss": 3.8561,
      "step": 509
    },
    {
      "epoch": 0.1738241308793456,
      "grad_norm": 35.40379264262327,
      "learning_rate": 5.777525539160046e-06,
      "loss": 2.9516,
      "step": 510
    },
    {
      "epoch": 0.1741649625085208,
      "grad_norm": 18.36607888401351,
      "learning_rate": 5.788876276958003e-06,
      "loss": 2.714,
      "step": 511
    },
    {
      "epoch": 0.17450579413769599,
      "grad_norm": 21.387287581520276,
      "learning_rate": 5.800227014755959e-06,
      "loss": 3.139,
      "step": 512
    },
    {
      "epoch": 0.17484662576687116,
      "grad_norm": 21.61281850539856,
      "learning_rate": 5.8115777525539165e-06,
      "loss": 3.0792,
      "step": 513
    },
    {
      "epoch": 0.17518745739604635,
      "grad_norm": 21.06866955887101,
      "learning_rate": 5.822928490351873e-06,
      "loss": 3.354,
      "step": 514
    },
    {
      "epoch": 0.17552828902522155,
      "grad_norm": 22.685552267589994,
      "learning_rate": 5.834279228149831e-06,
      "loss": 2.9687,
      "step": 515
    },
    {
      "epoch": 0.17586912065439672,
      "grad_norm": 59.43098420090411,
      "learning_rate": 5.845629965947786e-06,
      "loss": 3.3576,
      "step": 516
    },
    {
      "epoch": 0.17620995228357192,
      "grad_norm": 35.530390004546845,
      "learning_rate": 5.856980703745744e-06,
      "loss": 3.2322,
      "step": 517
    },
    {
      "epoch": 0.17655078391274712,
      "grad_norm": 29.052100138777856,
      "learning_rate": 5.8683314415437005e-06,
      "loss": 3.7539,
      "step": 518
    },
    {
      "epoch": 0.1768916155419223,
      "grad_norm": 24.715777197638086,
      "learning_rate": 5.879682179341658e-06,
      "loss": 3.2554,
      "step": 519
    },
    {
      "epoch": 0.17723244717109748,
      "grad_norm": 25.946668369183694,
      "learning_rate": 5.891032917139614e-06,
      "loss": 3.4278,
      "step": 520
    },
    {
      "epoch": 0.17757327880027265,
      "grad_norm": 27.789637451281425,
      "learning_rate": 5.902383654937571e-06,
      "loss": 3.0673,
      "step": 521
    },
    {
      "epoch": 0.17791411042944785,
      "grad_norm": 19.579369224944006,
      "learning_rate": 5.913734392735529e-06,
      "loss": 3.4305,
      "step": 522
    },
    {
      "epoch": 0.17825494205862305,
      "grad_norm": 22.39483266932541,
      "learning_rate": 5.925085130533485e-06,
      "loss": 3.0534,
      "step": 523
    },
    {
      "epoch": 0.17859577368779822,
      "grad_norm": 44.237088149564514,
      "learning_rate": 5.936435868331442e-06,
      "loss": 3.0159,
      "step": 524
    },
    {
      "epoch": 0.17893660531697342,
      "grad_norm": 26.568403074077814,
      "learning_rate": 5.947786606129399e-06,
      "loss": 3.1667,
      "step": 525
    },
    {
      "epoch": 0.17927743694614862,
      "grad_norm": 25.173552082789396,
      "learning_rate": 5.959137343927356e-06,
      "loss": 3.4813,
      "step": 526
    },
    {
      "epoch": 0.1796182685753238,
      "grad_norm": 32.56248490034913,
      "learning_rate": 5.970488081725313e-06,
      "loss": 2.9497,
      "step": 527
    },
    {
      "epoch": 0.17995910020449898,
      "grad_norm": 23.249643200902003,
      "learning_rate": 5.981838819523269e-06,
      "loss": 3.5269,
      "step": 528
    },
    {
      "epoch": 0.18029993183367415,
      "grad_norm": 27.76675467274532,
      "learning_rate": 5.993189557321226e-06,
      "loss": 3.7243,
      "step": 529
    },
    {
      "epoch": 0.18064076346284935,
      "grad_norm": 69.52770018289618,
      "learning_rate": 6.0045402951191835e-06,
      "loss": 3.2684,
      "step": 530
    },
    {
      "epoch": 0.18098159509202455,
      "grad_norm": 48.14771226964696,
      "learning_rate": 6.01589103291714e-06,
      "loss": 3.1029,
      "step": 531
    },
    {
      "epoch": 0.18132242672119972,
      "grad_norm": 23.529496028409927,
      "learning_rate": 6.027241770715097e-06,
      "loss": 3.4752,
      "step": 532
    },
    {
      "epoch": 0.18166325835037492,
      "grad_norm": 37.56672172284196,
      "learning_rate": 6.038592508513053e-06,
      "loss": 3.5109,
      "step": 533
    },
    {
      "epoch": 0.18200408997955012,
      "grad_norm": 23.51891776769473,
      "learning_rate": 6.049943246311011e-06,
      "loss": 2.8566,
      "step": 534
    },
    {
      "epoch": 0.18234492160872529,
      "grad_norm": 23.582532803143764,
      "learning_rate": 6.061293984108968e-06,
      "loss": 3.4881,
      "step": 535
    },
    {
      "epoch": 0.18268575323790048,
      "grad_norm": 20.799796234302473,
      "learning_rate": 6.072644721906924e-06,
      "loss": 3.5924,
      "step": 536
    },
    {
      "epoch": 0.18302658486707565,
      "grad_norm": 19.19297848708376,
      "learning_rate": 6.083995459704881e-06,
      "loss": 3.5124,
      "step": 537
    },
    {
      "epoch": 0.18336741649625085,
      "grad_norm": 27.00735853185107,
      "learning_rate": 6.095346197502838e-06,
      "loss": 3.3367,
      "step": 538
    },
    {
      "epoch": 0.18370824812542605,
      "grad_norm": 21.534487939573243,
      "learning_rate": 6.106696935300796e-06,
      "loss": 3.1773,
      "step": 539
    },
    {
      "epoch": 0.18404907975460122,
      "grad_norm": 22.180992387010704,
      "learning_rate": 6.1180476730987515e-06,
      "loss": 3.1487,
      "step": 540
    },
    {
      "epoch": 0.18438991138377642,
      "grad_norm": 26.55602329672257,
      "learning_rate": 6.129398410896709e-06,
      "loss": 3.3879,
      "step": 541
    },
    {
      "epoch": 0.18473074301295161,
      "grad_norm": 18.020067503370125,
      "learning_rate": 6.140749148694666e-06,
      "loss": 3.2404,
      "step": 542
    },
    {
      "epoch": 0.18507157464212678,
      "grad_norm": 29.05586953676158,
      "learning_rate": 6.152099886492623e-06,
      "loss": 3.3288,
      "step": 543
    },
    {
      "epoch": 0.18541240627130198,
      "grad_norm": 23.348934970482244,
      "learning_rate": 6.163450624290579e-06,
      "loss": 3.2509,
      "step": 544
    },
    {
      "epoch": 0.18575323790047715,
      "grad_norm": 22.320472879988202,
      "learning_rate": 6.174801362088536e-06,
      "loss": 3.2028,
      "step": 545
    },
    {
      "epoch": 0.18609406952965235,
      "grad_norm": 32.254979285161916,
      "learning_rate": 6.186152099886493e-06,
      "loss": 3.8129,
      "step": 546
    },
    {
      "epoch": 0.18643490115882755,
      "grad_norm": 25.641025759100756,
      "learning_rate": 6.1975028376844505e-06,
      "loss": 3.6254,
      "step": 547
    },
    {
      "epoch": 0.18677573278800272,
      "grad_norm": 31.149406715257253,
      "learning_rate": 6.208853575482406e-06,
      "loss": 3.3846,
      "step": 548
    },
    {
      "epoch": 0.18711656441717792,
      "grad_norm": 65.36951879493435,
      "learning_rate": 6.220204313280364e-06,
      "loss": 2.8993,
      "step": 549
    },
    {
      "epoch": 0.18745739604635311,
      "grad_norm": 25.3019528954931,
      "learning_rate": 6.23155505107832e-06,
      "loss": 3.624,
      "step": 550
    },
    {
      "epoch": 0.18779822767552828,
      "grad_norm": 29.21740242879814,
      "learning_rate": 6.242905788876278e-06,
      "loss": 3.8837,
      "step": 551
    },
    {
      "epoch": 0.18813905930470348,
      "grad_norm": 32.63382112057123,
      "learning_rate": 6.254256526674234e-06,
      "loss": 3.5399,
      "step": 552
    },
    {
      "epoch": 0.18847989093387865,
      "grad_norm": 69.057475658061,
      "learning_rate": 6.265607264472191e-06,
      "loss": 3.7291,
      "step": 553
    },
    {
      "epoch": 0.18882072256305385,
      "grad_norm": 23.130018847626225,
      "learning_rate": 6.2769580022701486e-06,
      "loss": 3.2533,
      "step": 554
    },
    {
      "epoch": 0.18916155419222905,
      "grad_norm": 22.99519857056165,
      "learning_rate": 6.288308740068105e-06,
      "loss": 3.1907,
      "step": 555
    },
    {
      "epoch": 0.18950238582140422,
      "grad_norm": 27.710510992977078,
      "learning_rate": 6.299659477866061e-06,
      "loss": 3.5782,
      "step": 556
    },
    {
      "epoch": 0.18984321745057942,
      "grad_norm": 29.093663289144295,
      "learning_rate": 6.3110102156640185e-06,
      "loss": 3.2991,
      "step": 557
    },
    {
      "epoch": 0.1901840490797546,
      "grad_norm": 32.320285271160294,
      "learning_rate": 6.322360953461976e-06,
      "loss": 3.5297,
      "step": 558
    },
    {
      "epoch": 0.19052488070892978,
      "grad_norm": 25.98250722744346,
      "learning_rate": 6.3337116912599326e-06,
      "loss": 2.7344,
      "step": 559
    },
    {
      "epoch": 0.19086571233810498,
      "grad_norm": 28.863510743642383,
      "learning_rate": 6.345062429057889e-06,
      "loss": 3.8643,
      "step": 560
    },
    {
      "epoch": 0.19120654396728015,
      "grad_norm": 26.569021530505545,
      "learning_rate": 6.356413166855846e-06,
      "loss": 3.7556,
      "step": 561
    },
    {
      "epoch": 0.19154737559645535,
      "grad_norm": 35.03092170777891,
      "learning_rate": 6.367763904653803e-06,
      "loss": 3.9044,
      "step": 562
    },
    {
      "epoch": 0.19188820722563055,
      "grad_norm": 26.059202918954323,
      "learning_rate": 6.37911464245176e-06,
      "loss": 3.2767,
      "step": 563
    },
    {
      "epoch": 0.19222903885480572,
      "grad_norm": 31.5134549581871,
      "learning_rate": 6.390465380249717e-06,
      "loss": 3.6411,
      "step": 564
    },
    {
      "epoch": 0.19256987048398091,
      "grad_norm": 26.877243467440774,
      "learning_rate": 6.401816118047673e-06,
      "loss": 3.3555,
      "step": 565
    },
    {
      "epoch": 0.1929107021131561,
      "grad_norm": 31.846154366195382,
      "learning_rate": 6.413166855845631e-06,
      "loss": 4.0248,
      "step": 566
    },
    {
      "epoch": 0.19325153374233128,
      "grad_norm": 18.616499143339563,
      "learning_rate": 6.424517593643587e-06,
      "loss": 2.933,
      "step": 567
    },
    {
      "epoch": 0.19359236537150648,
      "grad_norm": 23.093805714330088,
      "learning_rate": 6.435868331441544e-06,
      "loss": 3.281,
      "step": 568
    },
    {
      "epoch": 0.19393319700068165,
      "grad_norm": 30.83559569423827,
      "learning_rate": 6.447219069239501e-06,
      "loss": 3.634,
      "step": 569
    },
    {
      "epoch": 0.19427402862985685,
      "grad_norm": 32.73706036350775,
      "learning_rate": 6.458569807037458e-06,
      "loss": 3.4743,
      "step": 570
    },
    {
      "epoch": 0.19461486025903205,
      "grad_norm": 74.24147204735313,
      "learning_rate": 6.4699205448354155e-06,
      "loss": 3.7428,
      "step": 571
    },
    {
      "epoch": 0.19495569188820722,
      "grad_norm": 52.37636479029868,
      "learning_rate": 6.481271282633371e-06,
      "loss": 3.8128,
      "step": 572
    },
    {
      "epoch": 0.19529652351738241,
      "grad_norm": 29.32971293970952,
      "learning_rate": 6.492622020431329e-06,
      "loss": 3.694,
      "step": 573
    },
    {
      "epoch": 0.1956373551465576,
      "grad_norm": 53.82014887024488,
      "learning_rate": 6.5039727582292854e-06,
      "loss": 3.7754,
      "step": 574
    },
    {
      "epoch": 0.19597818677573278,
      "grad_norm": 28.850003121925102,
      "learning_rate": 6.515323496027243e-06,
      "loss": 2.7847,
      "step": 575
    },
    {
      "epoch": 0.19631901840490798,
      "grad_norm": 22.637784128737593,
      "learning_rate": 6.526674233825199e-06,
      "loss": 3.358,
      "step": 576
    },
    {
      "epoch": 0.19665985003408315,
      "grad_norm": 35.61107167179696,
      "learning_rate": 6.538024971623156e-06,
      "loss": 3.2443,
      "step": 577
    },
    {
      "epoch": 0.19700068166325835,
      "grad_norm": 31.80317806655503,
      "learning_rate": 6.549375709421113e-06,
      "loss": 4.0864,
      "step": 578
    },
    {
      "epoch": 0.19734151329243355,
      "grad_norm": 56.91706624954299,
      "learning_rate": 6.56072644721907e-06,
      "loss": 3.4392,
      "step": 579
    },
    {
      "epoch": 0.19768234492160872,
      "grad_norm": 23.213226100015838,
      "learning_rate": 6.572077185017026e-06,
      "loss": 3.5135,
      "step": 580
    },
    {
      "epoch": 0.1980231765507839,
      "grad_norm": 24.026197653791765,
      "learning_rate": 6.5834279228149836e-06,
      "loss": 3.3591,
      "step": 581
    },
    {
      "epoch": 0.1983640081799591,
      "grad_norm": 18.62673726649923,
      "learning_rate": 6.59477866061294e-06,
      "loss": 3.3976,
      "step": 582
    },
    {
      "epoch": 0.19870483980913428,
      "grad_norm": 22.60551870479549,
      "learning_rate": 6.606129398410898e-06,
      "loss": 3.3827,
      "step": 583
    },
    {
      "epoch": 0.19904567143830948,
      "grad_norm": 18.652295961874174,
      "learning_rate": 6.6174801362088534e-06,
      "loss": 3.4108,
      "step": 584
    },
    {
      "epoch": 0.19938650306748465,
      "grad_norm": 41.949281859304556,
      "learning_rate": 6.628830874006811e-06,
      "loss": 3.76,
      "step": 585
    },
    {
      "epoch": 0.19972733469665985,
      "grad_norm": 39.01023364768664,
      "learning_rate": 6.6401816118047676e-06,
      "loss": 2.827,
      "step": 586
    },
    {
      "epoch": 0.20006816632583505,
      "grad_norm": 20.878516587572626,
      "learning_rate": 6.651532349602725e-06,
      "loss": 2.9216,
      "step": 587
    },
    {
      "epoch": 0.20040899795501022,
      "grad_norm": 21.146966529246615,
      "learning_rate": 6.662883087400681e-06,
      "loss": 3.6306,
      "step": 588
    },
    {
      "epoch": 0.2007498295841854,
      "grad_norm": 24.592433533071503,
      "learning_rate": 6.674233825198638e-06,
      "loss": 3.3781,
      "step": 589
    },
    {
      "epoch": 0.2010906612133606,
      "grad_norm": 24.98703287628922,
      "learning_rate": 6.685584562996596e-06,
      "loss": 3.2197,
      "step": 590
    },
    {
      "epoch": 0.20143149284253578,
      "grad_norm": 16.002604694344356,
      "learning_rate": 6.696935300794552e-06,
      "loss": 3.542,
      "step": 591
    },
    {
      "epoch": 0.20177232447171098,
      "grad_norm": 20.538071312675253,
      "learning_rate": 6.708286038592508e-06,
      "loss": 2.7127,
      "step": 592
    },
    {
      "epoch": 0.20211315610088615,
      "grad_norm": 20.030323124806916,
      "learning_rate": 6.719636776390466e-06,
      "loss": 3.1837,
      "step": 593
    },
    {
      "epoch": 0.20245398773006135,
      "grad_norm": 32.797764482647544,
      "learning_rate": 6.730987514188423e-06,
      "loss": 3.8178,
      "step": 594
    },
    {
      "epoch": 0.20279481935923654,
      "grad_norm": 22.904398668139148,
      "learning_rate": 6.74233825198638e-06,
      "loss": 3.4008,
      "step": 595
    },
    {
      "epoch": 0.20313565098841171,
      "grad_norm": 19.788930144434676,
      "learning_rate": 6.753688989784336e-06,
      "loss": 2.7548,
      "step": 596
    },
    {
      "epoch": 0.2034764826175869,
      "grad_norm": 31.324709717321397,
      "learning_rate": 6.765039727582293e-06,
      "loss": 3.23,
      "step": 597
    },
    {
      "epoch": 0.2038173142467621,
      "grad_norm": 22.780068383723776,
      "learning_rate": 6.7763904653802505e-06,
      "loss": 3.0165,
      "step": 598
    },
    {
      "epoch": 0.20415814587593728,
      "grad_norm": 29.15127501460754,
      "learning_rate": 6.787741203178207e-06,
      "loss": 3.2529,
      "step": 599
    },
    {
      "epoch": 0.20449897750511248,
      "grad_norm": 17.40007686762209,
      "learning_rate": 6.799091940976164e-06,
      "loss": 2.9501,
      "step": 600
    },
    {
      "epoch": 0.20483980913428765,
      "grad_norm": 35.4553165212774,
      "learning_rate": 6.81044267877412e-06,
      "loss": 3.2171,
      "step": 601
    },
    {
      "epoch": 0.20518064076346285,
      "grad_norm": 19.103009345198597,
      "learning_rate": 6.821793416572078e-06,
      "loss": 3.5418,
      "step": 602
    },
    {
      "epoch": 0.20552147239263804,
      "grad_norm": 16.255161420164146,
      "learning_rate": 6.833144154370035e-06,
      "loss": 2.8603,
      "step": 603
    },
    {
      "epoch": 0.2058623040218132,
      "grad_norm": 28.72726768967839,
      "learning_rate": 6.844494892167991e-06,
      "loss": 3.3756,
      "step": 604
    },
    {
      "epoch": 0.2062031356509884,
      "grad_norm": 24.086255412194006,
      "learning_rate": 6.855845629965948e-06,
      "loss": 2.87,
      "step": 605
    },
    {
      "epoch": 0.2065439672801636,
      "grad_norm": 22.93265280345955,
      "learning_rate": 6.867196367763905e-06,
      "loss": 3.0186,
      "step": 606
    },
    {
      "epoch": 0.20688479890933878,
      "grad_norm": 22.14819204158573,
      "learning_rate": 6.878547105561863e-06,
      "loss": 3.0624,
      "step": 607
    },
    {
      "epoch": 0.20722563053851398,
      "grad_norm": 28.744363040752805,
      "learning_rate": 6.8898978433598185e-06,
      "loss": 2.8562,
      "step": 608
    },
    {
      "epoch": 0.20756646216768918,
      "grad_norm": 25.410481577215627,
      "learning_rate": 6.901248581157776e-06,
      "loss": 3.752,
      "step": 609
    },
    {
      "epoch": 0.20790729379686435,
      "grad_norm": 15.134889637275402,
      "learning_rate": 6.912599318955733e-06,
      "loss": 2.6328,
      "step": 610
    },
    {
      "epoch": 0.20824812542603954,
      "grad_norm": 31.79976657329084,
      "learning_rate": 6.92395005675369e-06,
      "loss": 3.3946,
      "step": 611
    },
    {
      "epoch": 0.2085889570552147,
      "grad_norm": 27.687031231491343,
      "learning_rate": 6.935300794551646e-06,
      "loss": 3.0865,
      "step": 612
    },
    {
      "epoch": 0.2089297886843899,
      "grad_norm": 40.862916583217206,
      "learning_rate": 6.946651532349603e-06,
      "loss": 3.5715,
      "step": 613
    },
    {
      "epoch": 0.2092706203135651,
      "grad_norm": 18.75215254358727,
      "learning_rate": 6.95800227014756e-06,
      "loss": 3.1755,
      "step": 614
    },
    {
      "epoch": 0.20961145194274028,
      "grad_norm": 30.292898763706404,
      "learning_rate": 6.9693530079455175e-06,
      "loss": 3.7768,
      "step": 615
    },
    {
      "epoch": 0.20995228357191548,
      "grad_norm": 13.881501658004648,
      "learning_rate": 6.980703745743473e-06,
      "loss": 2.8044,
      "step": 616
    },
    {
      "epoch": 0.21029311520109067,
      "grad_norm": 21.092000392401346,
      "learning_rate": 6.992054483541431e-06,
      "loss": 3.1219,
      "step": 617
    },
    {
      "epoch": 0.21063394683026584,
      "grad_norm": 26.58209487672545,
      "learning_rate": 7.003405221339387e-06,
      "loss": 2.8802,
      "step": 618
    },
    {
      "epoch": 0.21097477845944104,
      "grad_norm": 31.660388117348845,
      "learning_rate": 7.014755959137345e-06,
      "loss": 3.2726,
      "step": 619
    },
    {
      "epoch": 0.2113156100886162,
      "grad_norm": 24.023857227036515,
      "learning_rate": 7.026106696935301e-06,
      "loss": 3.5171,
      "step": 620
    },
    {
      "epoch": 0.2116564417177914,
      "grad_norm": 26.673484087347017,
      "learning_rate": 7.037457434733258e-06,
      "loss": 3.0256,
      "step": 621
    },
    {
      "epoch": 0.2119972733469666,
      "grad_norm": 30.92921940135245,
      "learning_rate": 7.048808172531215e-06,
      "loss": 3.5411,
      "step": 622
    },
    {
      "epoch": 0.21233810497614178,
      "grad_norm": 38.60252627194752,
      "learning_rate": 7.060158910329172e-06,
      "loss": 3.759,
      "step": 623
    },
    {
      "epoch": 0.21267893660531698,
      "grad_norm": 31.10064524411684,
      "learning_rate": 7.071509648127128e-06,
      "loss": 3.7815,
      "step": 624
    },
    {
      "epoch": 0.21301976823449217,
      "grad_norm": 39.91766960239971,
      "learning_rate": 7.0828603859250855e-06,
      "loss": 3.4385,
      "step": 625
    },
    {
      "epoch": 0.21336059986366734,
      "grad_norm": 14.943101718627684,
      "learning_rate": 7.094211123723043e-06,
      "loss": 2.6464,
      "step": 626
    },
    {
      "epoch": 0.21370143149284254,
      "grad_norm": 19.70803536364196,
      "learning_rate": 7.105561861521e-06,
      "loss": 3.0411,
      "step": 627
    },
    {
      "epoch": 0.2140422631220177,
      "grad_norm": 37.62224083588128,
      "learning_rate": 7.116912599318956e-06,
      "loss": 3.8128,
      "step": 628
    },
    {
      "epoch": 0.2143830947511929,
      "grad_norm": 37.576809504593655,
      "learning_rate": 7.128263337116913e-06,
      "loss": 3.7837,
      "step": 629
    },
    {
      "epoch": 0.2147239263803681,
      "grad_norm": 22.55171677882715,
      "learning_rate": 7.13961407491487e-06,
      "loss": 3.1353,
      "step": 630
    },
    {
      "epoch": 0.21506475800954328,
      "grad_norm": 36.21815291111241,
      "learning_rate": 7.150964812712827e-06,
      "loss": 3.5351,
      "step": 631
    },
    {
      "epoch": 0.21540558963871848,
      "grad_norm": 39.692900148232155,
      "learning_rate": 7.162315550510784e-06,
      "loss": 3.2352,
      "step": 632
    },
    {
      "epoch": 0.21574642126789367,
      "grad_norm": 32.82540343283596,
      "learning_rate": 7.17366628830874e-06,
      "loss": 3.8771,
      "step": 633
    },
    {
      "epoch": 0.21608725289706884,
      "grad_norm": 24.560724045206307,
      "learning_rate": 7.185017026106698e-06,
      "loss": 3.3772,
      "step": 634
    },
    {
      "epoch": 0.21642808452624404,
      "grad_norm": 29.950185048820973,
      "learning_rate": 7.196367763904654e-06,
      "loss": 3.9657,
      "step": 635
    },
    {
      "epoch": 0.2167689161554192,
      "grad_norm": 93.5947988373926,
      "learning_rate": 7.207718501702611e-06,
      "loss": 4.0873,
      "step": 636
    },
    {
      "epoch": 0.2171097477845944,
      "grad_norm": 54.64434237343563,
      "learning_rate": 7.219069239500568e-06,
      "loss": 3.1608,
      "step": 637
    },
    {
      "epoch": 0.2174505794137696,
      "grad_norm": 52.679036813503096,
      "learning_rate": 7.230419977298525e-06,
      "loss": 3.366,
      "step": 638
    },
    {
      "epoch": 0.21779141104294478,
      "grad_norm": 50.88533586750164,
      "learning_rate": 7.241770715096483e-06,
      "loss": 4.246,
      "step": 639
    },
    {
      "epoch": 0.21813224267211997,
      "grad_norm": 100.43471071529694,
      "learning_rate": 7.253121452894438e-06,
      "loss": 3.4288,
      "step": 640
    },
    {
      "epoch": 0.21847307430129517,
      "grad_norm": 32.295720899661305,
      "learning_rate": 7.264472190692395e-06,
      "loss": 4.2704,
      "step": 641
    },
    {
      "epoch": 0.21881390593047034,
      "grad_norm": 45.96512548500402,
      "learning_rate": 7.2758229284903525e-06,
      "loss": 3.3498,
      "step": 642
    },
    {
      "epoch": 0.21915473755964554,
      "grad_norm": 28.28620841578828,
      "learning_rate": 7.28717366628831e-06,
      "loss": 3.3084,
      "step": 643
    },
    {
      "epoch": 0.2194955691888207,
      "grad_norm": 25.409620232475053,
      "learning_rate": 7.298524404086266e-06,
      "loss": 3.4964,
      "step": 644
    },
    {
      "epoch": 0.2198364008179959,
      "grad_norm": 22.428681971089635,
      "learning_rate": 7.309875141884223e-06,
      "loss": 3.2854,
      "step": 645
    },
    {
      "epoch": 0.2201772324471711,
      "grad_norm": 22.928748643237626,
      "learning_rate": 7.32122587968218e-06,
      "loss": 3.404,
      "step": 646
    },
    {
      "epoch": 0.22051806407634628,
      "grad_norm": 21.043022530781105,
      "learning_rate": 7.332576617480137e-06,
      "loss": 3.2107,
      "step": 647
    },
    {
      "epoch": 0.22085889570552147,
      "grad_norm": 19.716945006988414,
      "learning_rate": 7.343927355278093e-06,
      "loss": 3.1192,
      "step": 648
    },
    {
      "epoch": 0.22119972733469667,
      "grad_norm": 17.286542988844342,
      "learning_rate": 7.355278093076051e-06,
      "loss": 2.9502,
      "step": 649
    },
    {
      "epoch": 0.22154055896387184,
      "grad_norm": 26.21972544307366,
      "learning_rate": 7.366628830874007e-06,
      "loss": 3.8412,
      "step": 650
    },
    {
      "epoch": 0.22188139059304704,
      "grad_norm": 31.10776378157114,
      "learning_rate": 7.377979568671965e-06,
      "loss": 2.7104,
      "step": 651
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 13.920521221873775,
      "learning_rate": 7.3893303064699205e-06,
      "loss": 3.0405,
      "step": 652
    },
    {
      "epoch": 0.2225630538513974,
      "grad_norm": 26.30217501825908,
      "learning_rate": 7.400681044267878e-06,
      "loss": 3.2028,
      "step": 653
    },
    {
      "epoch": 0.2229038854805726,
      "grad_norm": 38.52663870658809,
      "learning_rate": 7.412031782065835e-06,
      "loss": 3.0826,
      "step": 654
    },
    {
      "epoch": 0.22324471710974778,
      "grad_norm": 20.949763358462004,
      "learning_rate": 7.423382519863792e-06,
      "loss": 3.3962,
      "step": 655
    },
    {
      "epoch": 0.22358554873892297,
      "grad_norm": 20.39882608622639,
      "learning_rate": 7.434733257661748e-06,
      "loss": 3.3253,
      "step": 656
    },
    {
      "epoch": 0.22392638036809817,
      "grad_norm": 22.28491514603502,
      "learning_rate": 7.446083995459705e-06,
      "loss": 3.3699,
      "step": 657
    },
    {
      "epoch": 0.22426721199727334,
      "grad_norm": 28.94865453737843,
      "learning_rate": 7.457434733257663e-06,
      "loss": 3.2308,
      "step": 658
    },
    {
      "epoch": 0.22460804362644854,
      "grad_norm": 21.38604474315535,
      "learning_rate": 7.4687854710556195e-06,
      "loss": 3.2837,
      "step": 659
    },
    {
      "epoch": 0.2249488752556237,
      "grad_norm": 21.38499784066535,
      "learning_rate": 7.480136208853575e-06,
      "loss": 3.1846,
      "step": 660
    },
    {
      "epoch": 0.2252897068847989,
      "grad_norm": 22.93379307642456,
      "learning_rate": 7.491486946651533e-06,
      "loss": 3.141,
      "step": 661
    },
    {
      "epoch": 0.2256305385139741,
      "grad_norm": 30.304881382249665,
      "learning_rate": 7.50283768444949e-06,
      "loss": 3.4798,
      "step": 662
    },
    {
      "epoch": 0.22597137014314927,
      "grad_norm": 31.36002944797405,
      "learning_rate": 7.514188422247447e-06,
      "loss": 3.7983,
      "step": 663
    },
    {
      "epoch": 0.22631220177232447,
      "grad_norm": 26.097087673755233,
      "learning_rate": 7.5255391600454035e-06,
      "loss": 3.499,
      "step": 664
    },
    {
      "epoch": 0.22665303340149967,
      "grad_norm": 20.57046037909741,
      "learning_rate": 7.53688989784336e-06,
      "loss": 3.2229,
      "step": 665
    },
    {
      "epoch": 0.22699386503067484,
      "grad_norm": 47.562878709860364,
      "learning_rate": 7.5482406356413176e-06,
      "loss": 3.7525,
      "step": 666
    },
    {
      "epoch": 0.22733469665985004,
      "grad_norm": 27.390594888468755,
      "learning_rate": 7.559591373439274e-06,
      "loss": 2.6644,
      "step": 667
    },
    {
      "epoch": 0.2276755282890252,
      "grad_norm": 25.47541088784383,
      "learning_rate": 7.570942111237231e-06,
      "loss": 3.6583,
      "step": 668
    },
    {
      "epoch": 0.2280163599182004,
      "grad_norm": 29.155300987307562,
      "learning_rate": 7.5822928490351875e-06,
      "loss": 3.5922,
      "step": 669
    },
    {
      "epoch": 0.2283571915473756,
      "grad_norm": 31.10451556232216,
      "learning_rate": 7.593643586833145e-06,
      "loss": 3.5067,
      "step": 670
    },
    {
      "epoch": 0.22869802317655077,
      "grad_norm": 18.149757333009948,
      "learning_rate": 7.6049943246311016e-06,
      "loss": 3.1868,
      "step": 671
    },
    {
      "epoch": 0.22903885480572597,
      "grad_norm": 27.50169359632286,
      "learning_rate": 7.616345062429058e-06,
      "loss": 3.6593,
      "step": 672
    },
    {
      "epoch": 0.22937968643490117,
      "grad_norm": 16.656298612085447,
      "learning_rate": 7.627695800227015e-06,
      "loss": 2.9243,
      "step": 673
    },
    {
      "epoch": 0.22972051806407634,
      "grad_norm": 35.99538537724704,
      "learning_rate": 7.639046538024972e-06,
      "loss": 3.5115,
      "step": 674
    },
    {
      "epoch": 0.23006134969325154,
      "grad_norm": 21.238927116764263,
      "learning_rate": 7.650397275822929e-06,
      "loss": 3.653,
      "step": 675
    },
    {
      "epoch": 0.2304021813224267,
      "grad_norm": 24.789517952536414,
      "learning_rate": 7.661748013620886e-06,
      "loss": 3.3872,
      "step": 676
    },
    {
      "epoch": 0.2307430129516019,
      "grad_norm": 55.74348345838816,
      "learning_rate": 7.673098751418842e-06,
      "loss": 3.7864,
      "step": 677
    },
    {
      "epoch": 0.2310838445807771,
      "grad_norm": 43.076035949013175,
      "learning_rate": 7.6844494892168e-06,
      "loss": 3.107,
      "step": 678
    },
    {
      "epoch": 0.23142467620995227,
      "grad_norm": 29.924170179780113,
      "learning_rate": 7.695800227014757e-06,
      "loss": 4.0193,
      "step": 679
    },
    {
      "epoch": 0.23176550783912747,
      "grad_norm": 29.953359971908668,
      "learning_rate": 7.707150964812714e-06,
      "loss": 3.7907,
      "step": 680
    },
    {
      "epoch": 0.23210633946830267,
      "grad_norm": 52.94006135299579,
      "learning_rate": 7.71850170261067e-06,
      "loss": 3.2253,
      "step": 681
    },
    {
      "epoch": 0.23244717109747784,
      "grad_norm": 28.263883812667867,
      "learning_rate": 7.729852440408627e-06,
      "loss": 2.833,
      "step": 682
    },
    {
      "epoch": 0.23278800272665304,
      "grad_norm": 30.089251094826754,
      "learning_rate": 7.741203178206584e-06,
      "loss": 3.4134,
      "step": 683
    },
    {
      "epoch": 0.2331288343558282,
      "grad_norm": 30.094731514110844,
      "learning_rate": 7.75255391600454e-06,
      "loss": 2.7839,
      "step": 684
    },
    {
      "epoch": 0.2334696659850034,
      "grad_norm": 29.72830250491126,
      "learning_rate": 7.763904653802497e-06,
      "loss": 3.7319,
      "step": 685
    },
    {
      "epoch": 0.2338104976141786,
      "grad_norm": 18.0735291631865,
      "learning_rate": 7.775255391600455e-06,
      "loss": 3.071,
      "step": 686
    },
    {
      "epoch": 0.23415132924335377,
      "grad_norm": 23.88613291196941,
      "learning_rate": 7.786606129398412e-06,
      "loss": 3.0693,
      "step": 687
    },
    {
      "epoch": 0.23449216087252897,
      "grad_norm": 29.573233621949626,
      "learning_rate": 7.797956867196369e-06,
      "loss": 3.0225,
      "step": 688
    },
    {
      "epoch": 0.23483299250170417,
      "grad_norm": 29.542537263618925,
      "learning_rate": 7.809307604994325e-06,
      "loss": 3.1858,
      "step": 689
    },
    {
      "epoch": 0.23517382413087934,
      "grad_norm": 18.216259742446347,
      "learning_rate": 7.820658342792282e-06,
      "loss": 2.8649,
      "step": 690
    },
    {
      "epoch": 0.23551465576005454,
      "grad_norm": 125.65672187700817,
      "learning_rate": 7.83200908059024e-06,
      "loss": 3.3388,
      "step": 691
    },
    {
      "epoch": 0.23585548738922973,
      "grad_norm": 17.380752037557542,
      "learning_rate": 7.843359818388195e-06,
      "loss": 3.0402,
      "step": 692
    },
    {
      "epoch": 0.2361963190184049,
      "grad_norm": 23.594368754879273,
      "learning_rate": 7.854710556186152e-06,
      "loss": 2.6934,
      "step": 693
    },
    {
      "epoch": 0.2365371506475801,
      "grad_norm": 31.50893297161558,
      "learning_rate": 7.86606129398411e-06,
      "loss": 3.1054,
      "step": 694
    },
    {
      "epoch": 0.23687798227675527,
      "grad_norm": 33.8795403124054,
      "learning_rate": 7.877412031782067e-06,
      "loss": 3.3545,
      "step": 695
    },
    {
      "epoch": 0.23721881390593047,
      "grad_norm": 44.769421658207584,
      "learning_rate": 7.888762769580023e-06,
      "loss": 3.7756,
      "step": 696
    },
    {
      "epoch": 0.23755964553510567,
      "grad_norm": 28.234180617869896,
      "learning_rate": 7.90011350737798e-06,
      "loss": 3.6009,
      "step": 697
    },
    {
      "epoch": 0.23790047716428084,
      "grad_norm": 15.334225224695182,
      "learning_rate": 7.911464245175937e-06,
      "loss": 3.5085,
      "step": 698
    },
    {
      "epoch": 0.23824130879345604,
      "grad_norm": 21.915567193808116,
      "learning_rate": 7.922814982973895e-06,
      "loss": 3.002,
      "step": 699
    },
    {
      "epoch": 0.23858214042263123,
      "grad_norm": 34.603261266247074,
      "learning_rate": 7.93416572077185e-06,
      "loss": 3.5473,
      "step": 700
    },
    {
      "epoch": 0.2389229720518064,
      "grad_norm": 21.1563570463816,
      "learning_rate": 7.945516458569808e-06,
      "loss": 3.3214,
      "step": 701
    },
    {
      "epoch": 0.2392638036809816,
      "grad_norm": 39.49400937014496,
      "learning_rate": 7.956867196367765e-06,
      "loss": 3.5622,
      "step": 702
    },
    {
      "epoch": 0.23960463531015677,
      "grad_norm": 21.322113611123598,
      "learning_rate": 7.968217934165721e-06,
      "loss": 3.2879,
      "step": 703
    },
    {
      "epoch": 0.23994546693933197,
      "grad_norm": 40.04452731635841,
      "learning_rate": 7.979568671963678e-06,
      "loss": 3.6135,
      "step": 704
    },
    {
      "epoch": 0.24028629856850717,
      "grad_norm": 28.92576996822211,
      "learning_rate": 7.990919409761635e-06,
      "loss": 3.1345,
      "step": 705
    },
    {
      "epoch": 0.24062713019768234,
      "grad_norm": 24.673789690307704,
      "learning_rate": 8.002270147559591e-06,
      "loss": 3.6303,
      "step": 706
    },
    {
      "epoch": 0.24096796182685754,
      "grad_norm": 35.41563319616739,
      "learning_rate": 8.01362088535755e-06,
      "loss": 3.6358,
      "step": 707
    },
    {
      "epoch": 0.24130879345603273,
      "grad_norm": 26.08047689137244,
      "learning_rate": 8.024971623155505e-06,
      "loss": 3.5254,
      "step": 708
    },
    {
      "epoch": 0.2416496250852079,
      "grad_norm": 17.559255018922954,
      "learning_rate": 8.036322360953463e-06,
      "loss": 3.0724,
      "step": 709
    },
    {
      "epoch": 0.2419904567143831,
      "grad_norm": 31.95682148618894,
      "learning_rate": 8.04767309875142e-06,
      "loss": 3.7461,
      "step": 710
    },
    {
      "epoch": 0.24233128834355827,
      "grad_norm": 40.925147043952414,
      "learning_rate": 8.059023836549376e-06,
      "loss": 3.0832,
      "step": 711
    },
    {
      "epoch": 0.24267211997273347,
      "grad_norm": 20.63874100758961,
      "learning_rate": 8.070374574347333e-06,
      "loss": 3.402,
      "step": 712
    },
    {
      "epoch": 0.24301295160190867,
      "grad_norm": 18.715683220270982,
      "learning_rate": 8.08172531214529e-06,
      "loss": 3.4639,
      "step": 713
    },
    {
      "epoch": 0.24335378323108384,
      "grad_norm": 44.20355440433992,
      "learning_rate": 8.093076049943248e-06,
      "loss": 3.0679,
      "step": 714
    },
    {
      "epoch": 0.24369461486025903,
      "grad_norm": 31.826772524627636,
      "learning_rate": 8.104426787741204e-06,
      "loss": 3.7664,
      "step": 715
    },
    {
      "epoch": 0.24403544648943423,
      "grad_norm": 27.998862877527973,
      "learning_rate": 8.115777525539161e-06,
      "loss": 3.0858,
      "step": 716
    },
    {
      "epoch": 0.2443762781186094,
      "grad_norm": 29.09471705736548,
      "learning_rate": 8.127128263337118e-06,
      "loss": 3.2368,
      "step": 717
    },
    {
      "epoch": 0.2447171097477846,
      "grad_norm": 21.88657292660979,
      "learning_rate": 8.138479001135074e-06,
      "loss": 3.1833,
      "step": 718
    },
    {
      "epoch": 0.24505794137695977,
      "grad_norm": 18.544996823988352,
      "learning_rate": 8.149829738933031e-06,
      "loss": 3.0913,
      "step": 719
    },
    {
      "epoch": 0.24539877300613497,
      "grad_norm": 17.94552616844981,
      "learning_rate": 8.161180476730988e-06,
      "loss": 3.4685,
      "step": 720
    },
    {
      "epoch": 0.24573960463531017,
      "grad_norm": 20.94315313031552,
      "learning_rate": 8.172531214528944e-06,
      "loss": 3.1671,
      "step": 721
    },
    {
      "epoch": 0.24608043626448534,
      "grad_norm": 24.495557032183594,
      "learning_rate": 8.183881952326902e-06,
      "loss": 3.8512,
      "step": 722
    },
    {
      "epoch": 0.24642126789366053,
      "grad_norm": 22.53006209557111,
      "learning_rate": 8.195232690124859e-06,
      "loss": 3.7313,
      "step": 723
    },
    {
      "epoch": 0.24676209952283573,
      "grad_norm": 20.002518382137144,
      "learning_rate": 8.206583427922816e-06,
      "loss": 3.6776,
      "step": 724
    },
    {
      "epoch": 0.2471029311520109,
      "grad_norm": 22.114017817319763,
      "learning_rate": 8.217934165720772e-06,
      "loss": 3.4949,
      "step": 725
    },
    {
      "epoch": 0.2474437627811861,
      "grad_norm": 25.04085620931253,
      "learning_rate": 8.229284903518729e-06,
      "loss": 3.2238,
      "step": 726
    },
    {
      "epoch": 0.24778459441036127,
      "grad_norm": 27.126486716814522,
      "learning_rate": 8.240635641316687e-06,
      "loss": 3.1024,
      "step": 727
    },
    {
      "epoch": 0.24812542603953647,
      "grad_norm": 40.4970720660853,
      "learning_rate": 8.251986379114642e-06,
      "loss": 3.5811,
      "step": 728
    },
    {
      "epoch": 0.24846625766871167,
      "grad_norm": 23.270458870747998,
      "learning_rate": 8.2633371169126e-06,
      "loss": 2.7018,
      "step": 729
    },
    {
      "epoch": 0.24880708929788684,
      "grad_norm": 21.563893288169957,
      "learning_rate": 8.274687854710557e-06,
      "loss": 3.2195,
      "step": 730
    },
    {
      "epoch": 0.24914792092706203,
      "grad_norm": 21.969457282311005,
      "learning_rate": 8.286038592508514e-06,
      "loss": 3.5316,
      "step": 731
    },
    {
      "epoch": 0.24948875255623723,
      "grad_norm": 29.888320629441765,
      "learning_rate": 8.29738933030647e-06,
      "loss": 3.9979,
      "step": 732
    },
    {
      "epoch": 0.2498295841854124,
      "grad_norm": 21.275171828484144,
      "learning_rate": 8.308740068104427e-06,
      "loss": 3.0122,
      "step": 733
    },
    {
      "epoch": 0.2501704158145876,
      "grad_norm": 23.558357711598134,
      "learning_rate": 8.320090805902384e-06,
      "loss": 2.9943,
      "step": 734
    },
    {
      "epoch": 0.2505112474437628,
      "grad_norm": 25.754048516259058,
      "learning_rate": 8.331441543700342e-06,
      "loss": 3.1752,
      "step": 735
    },
    {
      "epoch": 0.250852079072938,
      "grad_norm": 31.869324476375695,
      "learning_rate": 8.342792281498297e-06,
      "loss": 3.7288,
      "step": 736
    },
    {
      "epoch": 0.25119291070211314,
      "grad_norm": 22.02338431337976,
      "learning_rate": 8.354143019296255e-06,
      "loss": 3.1557,
      "step": 737
    },
    {
      "epoch": 0.25153374233128833,
      "grad_norm": 22.15417353419875,
      "learning_rate": 8.365493757094212e-06,
      "loss": 3.178,
      "step": 738
    },
    {
      "epoch": 0.25187457396046353,
      "grad_norm": 28.587881771287908,
      "learning_rate": 8.376844494892169e-06,
      "loss": 2.9669,
      "step": 739
    },
    {
      "epoch": 0.25221540558963873,
      "grad_norm": 21.660486901847246,
      "learning_rate": 8.388195232690125e-06,
      "loss": 2.7981,
      "step": 740
    },
    {
      "epoch": 0.25255623721881393,
      "grad_norm": 21.464901129614255,
      "learning_rate": 8.399545970488082e-06,
      "loss": 3.2154,
      "step": 741
    },
    {
      "epoch": 0.25289706884798907,
      "grad_norm": 20.81353706994248,
      "learning_rate": 8.410896708286039e-06,
      "loss": 3.2952,
      "step": 742
    },
    {
      "epoch": 0.25323790047716427,
      "grad_norm": 20.67379945345823,
      "learning_rate": 8.422247446083997e-06,
      "loss": 2.9638,
      "step": 743
    },
    {
      "epoch": 0.25357873210633947,
      "grad_norm": 25.2845816012546,
      "learning_rate": 8.433598183881952e-06,
      "loss": 3.0447,
      "step": 744
    },
    {
      "epoch": 0.25391956373551466,
      "grad_norm": 29.181569750568247,
      "learning_rate": 8.44494892167991e-06,
      "loss": 3.9157,
      "step": 745
    },
    {
      "epoch": 0.25426039536468986,
      "grad_norm": 50.306047658897015,
      "learning_rate": 8.456299659477867e-06,
      "loss": 3.4021,
      "step": 746
    },
    {
      "epoch": 0.254601226993865,
      "grad_norm": 37.95104095783337,
      "learning_rate": 8.467650397275823e-06,
      "loss": 3.7127,
      "step": 747
    },
    {
      "epoch": 0.2549420586230402,
      "grad_norm": 16.867235090739403,
      "learning_rate": 8.47900113507378e-06,
      "loss": 2.9545,
      "step": 748
    },
    {
      "epoch": 0.2552828902522154,
      "grad_norm": 24.044200320036865,
      "learning_rate": 8.490351872871737e-06,
      "loss": 3.1379,
      "step": 749
    },
    {
      "epoch": 0.2556237218813906,
      "grad_norm": 23.68826881389898,
      "learning_rate": 8.501702610669695e-06,
      "loss": 3.6071,
      "step": 750
    },
    {
      "epoch": 0.2559645535105658,
      "grad_norm": 26.072628285910994,
      "learning_rate": 8.513053348467652e-06,
      "loss": 3.1413,
      "step": 751
    },
    {
      "epoch": 0.256305385139741,
      "grad_norm": 24.718537369704446,
      "learning_rate": 8.524404086265608e-06,
      "loss": 2.9534,
      "step": 752
    },
    {
      "epoch": 0.25664621676891614,
      "grad_norm": 18.016555004330048,
      "learning_rate": 8.535754824063565e-06,
      "loss": 3.076,
      "step": 753
    },
    {
      "epoch": 0.25698704839809133,
      "grad_norm": 26.854752188618,
      "learning_rate": 8.547105561861521e-06,
      "loss": 3.079,
      "step": 754
    },
    {
      "epoch": 0.25732788002726653,
      "grad_norm": 25.230622917445956,
      "learning_rate": 8.558456299659478e-06,
      "loss": 3.8375,
      "step": 755
    },
    {
      "epoch": 0.25766871165644173,
      "grad_norm": 15.914855786142633,
      "learning_rate": 8.569807037457435e-06,
      "loss": 3.1441,
      "step": 756
    },
    {
      "epoch": 0.2580095432856169,
      "grad_norm": 26.56339053004718,
      "learning_rate": 8.581157775255391e-06,
      "loss": 3.2563,
      "step": 757
    },
    {
      "epoch": 0.25835037491479207,
      "grad_norm": 24.89552674321171,
      "learning_rate": 8.59250851305335e-06,
      "loss": 3.2297,
      "step": 758
    },
    {
      "epoch": 0.25869120654396727,
      "grad_norm": 37.25610626482077,
      "learning_rate": 8.603859250851306e-06,
      "loss": 3.7545,
      "step": 759
    },
    {
      "epoch": 0.25903203817314246,
      "grad_norm": 35.79330991591616,
      "learning_rate": 8.615209988649263e-06,
      "loss": 3.3511,
      "step": 760
    },
    {
      "epoch": 0.25937286980231766,
      "grad_norm": 27.87096759657452,
      "learning_rate": 8.62656072644722e-06,
      "loss": 3.4937,
      "step": 761
    },
    {
      "epoch": 0.25971370143149286,
      "grad_norm": 32.72143582203686,
      "learning_rate": 8.637911464245176e-06,
      "loss": 3.7593,
      "step": 762
    },
    {
      "epoch": 0.260054533060668,
      "grad_norm": 27.985840306272664,
      "learning_rate": 8.649262202043135e-06,
      "loss": 3.6257,
      "step": 763
    },
    {
      "epoch": 0.2603953646898432,
      "grad_norm": 26.048307456705928,
      "learning_rate": 8.66061293984109e-06,
      "loss": 3.3931,
      "step": 764
    },
    {
      "epoch": 0.2607361963190184,
      "grad_norm": 32.094776243865624,
      "learning_rate": 8.671963677639048e-06,
      "loss": 3.0766,
      "step": 765
    },
    {
      "epoch": 0.2610770279481936,
      "grad_norm": 33.190822738472185,
      "learning_rate": 8.683314415437004e-06,
      "loss": 3.2532,
      "step": 766
    },
    {
      "epoch": 0.2614178595773688,
      "grad_norm": 26.666305515026433,
      "learning_rate": 8.694665153234961e-06,
      "loss": 3.7685,
      "step": 767
    },
    {
      "epoch": 0.261758691206544,
      "grad_norm": 22.69569420911664,
      "learning_rate": 8.706015891032918e-06,
      "loss": 3.0247,
      "step": 768
    },
    {
      "epoch": 0.26209952283571913,
      "grad_norm": 36.2068461132267,
      "learning_rate": 8.717366628830874e-06,
      "loss": 3.5706,
      "step": 769
    },
    {
      "epoch": 0.26244035446489433,
      "grad_norm": 25.11675597353937,
      "learning_rate": 8.728717366628831e-06,
      "loss": 3.671,
      "step": 770
    },
    {
      "epoch": 0.26278118609406953,
      "grad_norm": 22.2460131010649,
      "learning_rate": 8.74006810442679e-06,
      "loss": 2.4091,
      "step": 771
    },
    {
      "epoch": 0.2631220177232447,
      "grad_norm": 27.106951476016487,
      "learning_rate": 8.751418842224744e-06,
      "loss": 3.6041,
      "step": 772
    },
    {
      "epoch": 0.2634628493524199,
      "grad_norm": 19.88883884697766,
      "learning_rate": 8.762769580022703e-06,
      "loss": 3.405,
      "step": 773
    },
    {
      "epoch": 0.26380368098159507,
      "grad_norm": 20.50368807237646,
      "learning_rate": 8.77412031782066e-06,
      "loss": 3.4647,
      "step": 774
    },
    {
      "epoch": 0.26414451261077027,
      "grad_norm": 23.126077222692306,
      "learning_rate": 8.785471055618616e-06,
      "loss": 3.9059,
      "step": 775
    },
    {
      "epoch": 0.26448534423994546,
      "grad_norm": 18.0655313534014,
      "learning_rate": 8.796821793416572e-06,
      "loss": 3.5183,
      "step": 776
    },
    {
      "epoch": 0.26482617586912066,
      "grad_norm": 23.140956328999053,
      "learning_rate": 8.808172531214529e-06,
      "loss": 3.1886,
      "step": 777
    },
    {
      "epoch": 0.26516700749829586,
      "grad_norm": 41.110278540451354,
      "learning_rate": 8.819523269012486e-06,
      "loss": 3.7806,
      "step": 778
    },
    {
      "epoch": 0.26550783912747106,
      "grad_norm": 28.585057531119773,
      "learning_rate": 8.830874006810444e-06,
      "loss": 3.5836,
      "step": 779
    },
    {
      "epoch": 0.2658486707566462,
      "grad_norm": 42.78866346651865,
      "learning_rate": 8.842224744608399e-06,
      "loss": 3.6422,
      "step": 780
    },
    {
      "epoch": 0.2661895023858214,
      "grad_norm": 43.254943897727856,
      "learning_rate": 8.853575482406357e-06,
      "loss": 2.6481,
      "step": 781
    },
    {
      "epoch": 0.2665303340149966,
      "grad_norm": 15.383032910503957,
      "learning_rate": 8.864926220204314e-06,
      "loss": 3.2938,
      "step": 782
    },
    {
      "epoch": 0.2668711656441718,
      "grad_norm": 17.603301832141664,
      "learning_rate": 8.87627695800227e-06,
      "loss": 3.2174,
      "step": 783
    },
    {
      "epoch": 0.267211997273347,
      "grad_norm": 29.20666670769123,
      "learning_rate": 8.887627695800227e-06,
      "loss": 3.4076,
      "step": 784
    },
    {
      "epoch": 0.26755282890252213,
      "grad_norm": 24.7537573499683,
      "learning_rate": 8.898978433598184e-06,
      "loss": 3.6132,
      "step": 785
    },
    {
      "epoch": 0.26789366053169733,
      "grad_norm": 73.8550198315664,
      "learning_rate": 8.910329171396142e-06,
      "loss": 3.0391,
      "step": 786
    },
    {
      "epoch": 0.26823449216087253,
      "grad_norm": 24.528172210733576,
      "learning_rate": 8.921679909194099e-06,
      "loss": 3.5183,
      "step": 787
    },
    {
      "epoch": 0.2685753237900477,
      "grad_norm": 21.478137367927445,
      "learning_rate": 8.933030646992055e-06,
      "loss": 3.3609,
      "step": 788
    },
    {
      "epoch": 0.2689161554192229,
      "grad_norm": 19.54909258835435,
      "learning_rate": 8.944381384790012e-06,
      "loss": 2.866,
      "step": 789
    },
    {
      "epoch": 0.26925698704839807,
      "grad_norm": 26.87166298365148,
      "learning_rate": 8.955732122587969e-06,
      "loss": 3.4004,
      "step": 790
    },
    {
      "epoch": 0.26959781867757326,
      "grad_norm": 21.262880902795793,
      "learning_rate": 8.967082860385925e-06,
      "loss": 3.1228,
      "step": 791
    },
    {
      "epoch": 0.26993865030674846,
      "grad_norm": 24.79413190238954,
      "learning_rate": 8.978433598183882e-06,
      "loss": 3.0429,
      "step": 792
    },
    {
      "epoch": 0.27027948193592366,
      "grad_norm": 24.395933576878956,
      "learning_rate": 8.989784335981839e-06,
      "loss": 3.4952,
      "step": 793
    },
    {
      "epoch": 0.27062031356509886,
      "grad_norm": 23.91475960949721,
      "learning_rate": 9.001135073779797e-06,
      "loss": 3.2471,
      "step": 794
    },
    {
      "epoch": 0.27096114519427406,
      "grad_norm": 22.337045614116192,
      "learning_rate": 9.012485811577754e-06,
      "loss": 3.4552,
      "step": 795
    },
    {
      "epoch": 0.2713019768234492,
      "grad_norm": 28.976045213872876,
      "learning_rate": 9.02383654937571e-06,
      "loss": 3.8317,
      "step": 796
    },
    {
      "epoch": 0.2716428084526244,
      "grad_norm": 26.878664981423082,
      "learning_rate": 9.035187287173667e-06,
      "loss": 2.769,
      "step": 797
    },
    {
      "epoch": 0.2719836400817996,
      "grad_norm": 21.092339081213616,
      "learning_rate": 9.046538024971623e-06,
      "loss": 3.1567,
      "step": 798
    },
    {
      "epoch": 0.2723244717109748,
      "grad_norm": 21.27115420968662,
      "learning_rate": 9.057888762769582e-06,
      "loss": 3.3388,
      "step": 799
    },
    {
      "epoch": 0.27266530334015,
      "grad_norm": 21.334859332764378,
      "learning_rate": 9.069239500567537e-06,
      "loss": 2.7804,
      "step": 800
    },
    {
      "epoch": 0.27300613496932513,
      "grad_norm": 20.086611157320537,
      "learning_rate": 9.080590238365495e-06,
      "loss": 3.1492,
      "step": 801
    },
    {
      "epoch": 0.27334696659850033,
      "grad_norm": 19.590253905148945,
      "learning_rate": 9.091940976163452e-06,
      "loss": 3.2286,
      "step": 802
    },
    {
      "epoch": 0.2736877982276755,
      "grad_norm": 23.377639096329997,
      "learning_rate": 9.103291713961408e-06,
      "loss": 3.06,
      "step": 803
    },
    {
      "epoch": 0.2740286298568507,
      "grad_norm": 23.25023518585151,
      "learning_rate": 9.114642451759365e-06,
      "loss": 3.3556,
      "step": 804
    },
    {
      "epoch": 0.2743694614860259,
      "grad_norm": 33.19832442235924,
      "learning_rate": 9.125993189557322e-06,
      "loss": 3.1801,
      "step": 805
    },
    {
      "epoch": 0.27471029311520107,
      "grad_norm": 23.267995647419923,
      "learning_rate": 9.137343927355278e-06,
      "loss": 2.8149,
      "step": 806
    },
    {
      "epoch": 0.27505112474437626,
      "grad_norm": 18.469309658464905,
      "learning_rate": 9.148694665153237e-06,
      "loss": 2.9582,
      "step": 807
    },
    {
      "epoch": 0.27539195637355146,
      "grad_norm": 20.6039674662935,
      "learning_rate": 9.160045402951191e-06,
      "loss": 3.4269,
      "step": 808
    },
    {
      "epoch": 0.27573278800272666,
      "grad_norm": 24.87941854774333,
      "learning_rate": 9.17139614074915e-06,
      "loss": 3.6045,
      "step": 809
    },
    {
      "epoch": 0.27607361963190186,
      "grad_norm": 39.90973855651112,
      "learning_rate": 9.182746878547106e-06,
      "loss": 4.0639,
      "step": 810
    },
    {
      "epoch": 0.27641445126107705,
      "grad_norm": 23.85839114983585,
      "learning_rate": 9.194097616345063e-06,
      "loss": 3.0495,
      "step": 811
    },
    {
      "epoch": 0.2767552828902522,
      "grad_norm": 26.638642085275748,
      "learning_rate": 9.20544835414302e-06,
      "loss": 3.8556,
      "step": 812
    },
    {
      "epoch": 0.2770961145194274,
      "grad_norm": 26.22603006923388,
      "learning_rate": 9.216799091940976e-06,
      "loss": 3.8839,
      "step": 813
    },
    {
      "epoch": 0.2774369461486026,
      "grad_norm": 28.38764329066922,
      "learning_rate": 9.228149829738935e-06,
      "loss": 3.6359,
      "step": 814
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 27.438949154535752,
      "learning_rate": 9.239500567536891e-06,
      "loss": 3.4608,
      "step": 815
    },
    {
      "epoch": 0.278118609406953,
      "grad_norm": 29.546208093223125,
      "learning_rate": 9.250851305334846e-06,
      "loss": 3.4507,
      "step": 816
    },
    {
      "epoch": 0.27845944103612813,
      "grad_norm": 36.0313703745622,
      "learning_rate": 9.262202043132805e-06,
      "loss": 3.3545,
      "step": 817
    },
    {
      "epoch": 0.27880027266530333,
      "grad_norm": 22.997000976953416,
      "learning_rate": 9.273552780930761e-06,
      "loss": 3.2352,
      "step": 818
    },
    {
      "epoch": 0.2791411042944785,
      "grad_norm": 19.414012546910165,
      "learning_rate": 9.284903518728718e-06,
      "loss": 2.737,
      "step": 819
    },
    {
      "epoch": 0.2794819359236537,
      "grad_norm": 31.5413361385869,
      "learning_rate": 9.296254256526674e-06,
      "loss": 3.2417,
      "step": 820
    },
    {
      "epoch": 0.2798227675528289,
      "grad_norm": 15.827403673405787,
      "learning_rate": 9.307604994324631e-06,
      "loss": 2.9977,
      "step": 821
    },
    {
      "epoch": 0.28016359918200406,
      "grad_norm": 20.82469606006344,
      "learning_rate": 9.31895573212259e-06,
      "loss": 3.2735,
      "step": 822
    },
    {
      "epoch": 0.28050443081117926,
      "grad_norm": 32.20046594480221,
      "learning_rate": 9.330306469920546e-06,
      "loss": 3.2921,
      "step": 823
    },
    {
      "epoch": 0.28084526244035446,
      "grad_norm": 26.362909913318603,
      "learning_rate": 9.341657207718503e-06,
      "loss": 3.5835,
      "step": 824
    },
    {
      "epoch": 0.28118609406952966,
      "grad_norm": 22.84990659985345,
      "learning_rate": 9.35300794551646e-06,
      "loss": 3.0717,
      "step": 825
    },
    {
      "epoch": 0.28152692569870486,
      "grad_norm": 27.2027280666498,
      "learning_rate": 9.364358683314416e-06,
      "loss": 3.5037,
      "step": 826
    },
    {
      "epoch": 0.28186775732788005,
      "grad_norm": 29.798198940830076,
      "learning_rate": 9.375709421112373e-06,
      "loss": 4.0117,
      "step": 827
    },
    {
      "epoch": 0.2822085889570552,
      "grad_norm": 20.642400723301353,
      "learning_rate": 9.38706015891033e-06,
      "loss": 3.429,
      "step": 828
    },
    {
      "epoch": 0.2825494205862304,
      "grad_norm": 32.04416312412705,
      "learning_rate": 9.398410896708286e-06,
      "loss": 3.2567,
      "step": 829
    },
    {
      "epoch": 0.2828902522154056,
      "grad_norm": 22.52104807528634,
      "learning_rate": 9.409761634506244e-06,
      "loss": 3.5261,
      "step": 830
    },
    {
      "epoch": 0.2832310838445808,
      "grad_norm": 48.220650181790575,
      "learning_rate": 9.4211123723042e-06,
      "loss": 3.5013,
      "step": 831
    },
    {
      "epoch": 0.283571915473756,
      "grad_norm": 21.05062457771934,
      "learning_rate": 9.432463110102157e-06,
      "loss": 3.3212,
      "step": 832
    },
    {
      "epoch": 0.28391274710293113,
      "grad_norm": 13.087044900477736,
      "learning_rate": 9.443813847900114e-06,
      "loss": 2.6118,
      "step": 833
    },
    {
      "epoch": 0.2842535787321063,
      "grad_norm": 36.01934218552081,
      "learning_rate": 9.45516458569807e-06,
      "loss": 3.0547,
      "step": 834
    },
    {
      "epoch": 0.2845944103612815,
      "grad_norm": 29.59387842751995,
      "learning_rate": 9.466515323496029e-06,
      "loss": 3.3447,
      "step": 835
    },
    {
      "epoch": 0.2849352419904567,
      "grad_norm": 30.59072072344251,
      "learning_rate": 9.477866061293984e-06,
      "loss": 3.8825,
      "step": 836
    },
    {
      "epoch": 0.2852760736196319,
      "grad_norm": 19.260491796504766,
      "learning_rate": 9.489216799091942e-06,
      "loss": 3.2133,
      "step": 837
    },
    {
      "epoch": 0.28561690524880706,
      "grad_norm": 20.63171930742733,
      "learning_rate": 9.500567536889899e-06,
      "loss": 3.2055,
      "step": 838
    },
    {
      "epoch": 0.28595773687798226,
      "grad_norm": 20.245988624402422,
      "learning_rate": 9.511918274687855e-06,
      "loss": 2.9113,
      "step": 839
    },
    {
      "epoch": 0.28629856850715746,
      "grad_norm": 28.38871079241072,
      "learning_rate": 9.523269012485812e-06,
      "loss": 3.4454,
      "step": 840
    },
    {
      "epoch": 0.28663940013633266,
      "grad_norm": 15.018151311228648,
      "learning_rate": 9.534619750283769e-06,
      "loss": 3.1854,
      "step": 841
    },
    {
      "epoch": 0.28698023176550785,
      "grad_norm": 23.20153117378376,
      "learning_rate": 9.545970488081725e-06,
      "loss": 3.5299,
      "step": 842
    },
    {
      "epoch": 0.28732106339468305,
      "grad_norm": 25.978778297406183,
      "learning_rate": 9.557321225879684e-06,
      "loss": 3.1834,
      "step": 843
    },
    {
      "epoch": 0.2876618950238582,
      "grad_norm": 14.633036434579006,
      "learning_rate": 9.568671963677639e-06,
      "loss": 3.0366,
      "step": 844
    },
    {
      "epoch": 0.2880027266530334,
      "grad_norm": 42.84383409980026,
      "learning_rate": 9.580022701475597e-06,
      "loss": 3.532,
      "step": 845
    },
    {
      "epoch": 0.2883435582822086,
      "grad_norm": 23.954226311640735,
      "learning_rate": 9.591373439273554e-06,
      "loss": 3.5636,
      "step": 846
    },
    {
      "epoch": 0.2886843899113838,
      "grad_norm": 27.974792253005543,
      "learning_rate": 9.60272417707151e-06,
      "loss": 4.2702,
      "step": 847
    },
    {
      "epoch": 0.289025221540559,
      "grad_norm": 25.27620756189969,
      "learning_rate": 9.614074914869467e-06,
      "loss": 3.5109,
      "step": 848
    },
    {
      "epoch": 0.2893660531697341,
      "grad_norm": 37.07533122664887,
      "learning_rate": 9.625425652667424e-06,
      "loss": 3.6019,
      "step": 849
    },
    {
      "epoch": 0.2897068847989093,
      "grad_norm": 18.141635217799177,
      "learning_rate": 9.636776390465382e-06,
      "loss": 3.1653,
      "step": 850
    },
    {
      "epoch": 0.2900477164280845,
      "grad_norm": 16.408363596755482,
      "learning_rate": 9.648127128263338e-06,
      "loss": 3.4963,
      "step": 851
    },
    {
      "epoch": 0.2903885480572597,
      "grad_norm": 21.487261945576247,
      "learning_rate": 9.659477866061293e-06,
      "loss": 3.7555,
      "step": 852
    },
    {
      "epoch": 0.2907293796864349,
      "grad_norm": 21.814628428118155,
      "learning_rate": 9.670828603859252e-06,
      "loss": 3.3467,
      "step": 853
    },
    {
      "epoch": 0.29107021131561006,
      "grad_norm": 20.258145801585798,
      "learning_rate": 9.682179341657208e-06,
      "loss": 3.4661,
      "step": 854
    },
    {
      "epoch": 0.29141104294478526,
      "grad_norm": 16.00319387546845,
      "learning_rate": 9.693530079455165e-06,
      "loss": 3.3164,
      "step": 855
    },
    {
      "epoch": 0.29175187457396046,
      "grad_norm": 25.689939710866422,
      "learning_rate": 9.704880817253122e-06,
      "loss": 3.4663,
      "step": 856
    },
    {
      "epoch": 0.29209270620313565,
      "grad_norm": 26.822581954837638,
      "learning_rate": 9.716231555051078e-06,
      "loss": 3.1245,
      "step": 857
    },
    {
      "epoch": 0.29243353783231085,
      "grad_norm": 16.073062107850657,
      "learning_rate": 9.727582292849037e-06,
      "loss": 3.213,
      "step": 858
    },
    {
      "epoch": 0.29277436946148605,
      "grad_norm": 14.2032218597238,
      "learning_rate": 9.738933030646993e-06,
      "loss": 2.9584,
      "step": 859
    },
    {
      "epoch": 0.2931152010906612,
      "grad_norm": 22.487763438790918,
      "learning_rate": 9.75028376844495e-06,
      "loss": 3.2377,
      "step": 860
    },
    {
      "epoch": 0.2934560327198364,
      "grad_norm": 33.2252626373012,
      "learning_rate": 9.761634506242906e-06,
      "loss": 3.0383,
      "step": 861
    },
    {
      "epoch": 0.2937968643490116,
      "grad_norm": 27.33347190209937,
      "learning_rate": 9.772985244040863e-06,
      "loss": 3.5986,
      "step": 862
    },
    {
      "epoch": 0.2941376959781868,
      "grad_norm": 19.633427008685462,
      "learning_rate": 9.784335981838821e-06,
      "loss": 3.4734,
      "step": 863
    },
    {
      "epoch": 0.294478527607362,
      "grad_norm": 34.268797429388925,
      "learning_rate": 9.795686719636776e-06,
      "loss": 3.5889,
      "step": 864
    },
    {
      "epoch": 0.2948193592365371,
      "grad_norm": 25.08698067517474,
      "learning_rate": 9.807037457434733e-06,
      "loss": 3.1935,
      "step": 865
    },
    {
      "epoch": 0.2951601908657123,
      "grad_norm": 17.221767039952358,
      "learning_rate": 9.818388195232691e-06,
      "loss": 2.6544,
      "step": 866
    },
    {
      "epoch": 0.2955010224948875,
      "grad_norm": 22.754572200380267,
      "learning_rate": 9.829738933030648e-06,
      "loss": 3.7041,
      "step": 867
    },
    {
      "epoch": 0.2958418541240627,
      "grad_norm": 21.84309905599954,
      "learning_rate": 9.841089670828605e-06,
      "loss": 3.744,
      "step": 868
    },
    {
      "epoch": 0.2961826857532379,
      "grad_norm": 30.591901639534964,
      "learning_rate": 9.852440408626561e-06,
      "loss": 4.1962,
      "step": 869
    },
    {
      "epoch": 0.2965235173824131,
      "grad_norm": 18.108910996630414,
      "learning_rate": 9.863791146424518e-06,
      "loss": 3.255,
      "step": 870
    },
    {
      "epoch": 0.29686434901158826,
      "grad_norm": 25.604291529883042,
      "learning_rate": 9.875141884222476e-06,
      "loss": 3.4427,
      "step": 871
    },
    {
      "epoch": 0.29720518064076346,
      "grad_norm": 23.01863541949127,
      "learning_rate": 9.886492622020431e-06,
      "loss": 3.4422,
      "step": 872
    },
    {
      "epoch": 0.29754601226993865,
      "grad_norm": 20.387655041498515,
      "learning_rate": 9.89784335981839e-06,
      "loss": 3.3177,
      "step": 873
    },
    {
      "epoch": 0.29788684389911385,
      "grad_norm": 27.19500461419769,
      "learning_rate": 9.909194097616346e-06,
      "loss": 3.5902,
      "step": 874
    },
    {
      "epoch": 0.29822767552828905,
      "grad_norm": 17.8403695225531,
      "learning_rate": 9.920544835414303e-06,
      "loss": 3.223,
      "step": 875
    },
    {
      "epoch": 0.2985685071574642,
      "grad_norm": 29.3436181441884,
      "learning_rate": 9.93189557321226e-06,
      "loss": 3.3883,
      "step": 876
    },
    {
      "epoch": 0.2989093387866394,
      "grad_norm": 24.24646161719558,
      "learning_rate": 9.943246311010216e-06,
      "loss": 3.4357,
      "step": 877
    },
    {
      "epoch": 0.2992501704158146,
      "grad_norm": 34.524104002392875,
      "learning_rate": 9.954597048808173e-06,
      "loss": 3.4251,
      "step": 878
    },
    {
      "epoch": 0.2995910020449898,
      "grad_norm": 20.988087700741243,
      "learning_rate": 9.965947786606131e-06,
      "loss": 3.3922,
      "step": 879
    },
    {
      "epoch": 0.299931833674165,
      "grad_norm": 25.41350666440404,
      "learning_rate": 9.977298524404086e-06,
      "loss": 3.0033,
      "step": 880
    },
    {
      "epoch": 0.3002726653033401,
      "grad_norm": 47.64637761170976,
      "learning_rate": 9.988649262202044e-06,
      "loss": 3.7157,
      "step": 881
    },
    {
      "epoch": 0.3006134969325153,
      "grad_norm": 25.563287557803758,
      "learning_rate": 1e-05,
      "loss": 3.6178,
      "step": 882
    },
    {
      "epoch": 0.3009543285616905,
      "grad_norm": 19.621235606484483,
      "learning_rate": 9.999999606740049e-06,
      "loss": 3.6751,
      "step": 883
    },
    {
      "epoch": 0.3012951601908657,
      "grad_norm": 16.973824092381143,
      "learning_rate": 9.999998426960256e-06,
      "loss": 3.3814,
      "step": 884
    },
    {
      "epoch": 0.3016359918200409,
      "grad_norm": 22.449226600292146,
      "learning_rate": 9.999996460660805e-06,
      "loss": 3.4696,
      "step": 885
    },
    {
      "epoch": 0.3019768234492161,
      "grad_norm": 19.527031863127327,
      "learning_rate": 9.999993707842007e-06,
      "loss": 3.3002,
      "step": 886
    },
    {
      "epoch": 0.30231765507839126,
      "grad_norm": 44.945285037916015,
      "learning_rate": 9.999990168504296e-06,
      "loss": 3.6489,
      "step": 887
    },
    {
      "epoch": 0.30265848670756645,
      "grad_norm": 24.973038773254128,
      "learning_rate": 9.999985842648228e-06,
      "loss": 2.759,
      "step": 888
    },
    {
      "epoch": 0.30299931833674165,
      "grad_norm": 25.900935310061467,
      "learning_rate": 9.999980730274481e-06,
      "loss": 3.1809,
      "step": 889
    },
    {
      "epoch": 0.30334014996591685,
      "grad_norm": 23.144770439947493,
      "learning_rate": 9.999974831383862e-06,
      "loss": 3.5488,
      "step": 890
    },
    {
      "epoch": 0.30368098159509205,
      "grad_norm": 25.83270721382352,
      "learning_rate": 9.9999681459773e-06,
      "loss": 3.8451,
      "step": 891
    },
    {
      "epoch": 0.3040218132242672,
      "grad_norm": 17.536764377187176,
      "learning_rate": 9.999960674055844e-06,
      "loss": 3.1129,
      "step": 892
    },
    {
      "epoch": 0.3043626448534424,
      "grad_norm": 35.28931840289613,
      "learning_rate": 9.999952415620671e-06,
      "loss": 3.8822,
      "step": 893
    },
    {
      "epoch": 0.3047034764826176,
      "grad_norm": 20.49798965442751,
      "learning_rate": 9.999943370673078e-06,
      "loss": 3.2214,
      "step": 894
    },
    {
      "epoch": 0.3050443081117928,
      "grad_norm": 23.574449554632288,
      "learning_rate": 9.99993353921449e-06,
      "loss": 3.107,
      "step": 895
    },
    {
      "epoch": 0.305385139740968,
      "grad_norm": 32.37301565899288,
      "learning_rate": 9.999922921246453e-06,
      "loss": 3.6538,
      "step": 896
    },
    {
      "epoch": 0.3057259713701431,
      "grad_norm": 26.78641335981452,
      "learning_rate": 9.999911516770634e-06,
      "loss": 3.8318,
      "step": 897
    },
    {
      "epoch": 0.3060668029993183,
      "grad_norm": 32.1313320443883,
      "learning_rate": 9.999899325788834e-06,
      "loss": 3.5537,
      "step": 898
    },
    {
      "epoch": 0.3064076346284935,
      "grad_norm": 15.858855823165637,
      "learning_rate": 9.999886348302966e-06,
      "loss": 2.7203,
      "step": 899
    },
    {
      "epoch": 0.3067484662576687,
      "grad_norm": 21.89642099857721,
      "learning_rate": 9.99987258431507e-06,
      "loss": 3.198,
      "step": 900
    },
    {
      "epoch": 0.3070892978868439,
      "grad_norm": 20.50897435051379,
      "learning_rate": 9.999858033827313e-06,
      "loss": 3.2204,
      "step": 901
    },
    {
      "epoch": 0.3074301295160191,
      "grad_norm": 19.39467710260482,
      "learning_rate": 9.999842696841987e-06,
      "loss": 3.341,
      "step": 902
    },
    {
      "epoch": 0.30777096114519426,
      "grad_norm": 31.022873251801197,
      "learning_rate": 9.999826573361498e-06,
      "loss": 3.4892,
      "step": 903
    },
    {
      "epoch": 0.30811179277436945,
      "grad_norm": 24.855787856412118,
      "learning_rate": 9.999809663388386e-06,
      "loss": 3.2009,
      "step": 904
    },
    {
      "epoch": 0.30845262440354465,
      "grad_norm": 23.546815568395473,
      "learning_rate": 9.999791966925313e-06,
      "loss": 3.4443,
      "step": 905
    },
    {
      "epoch": 0.30879345603271985,
      "grad_norm": 18.5829572103646,
      "learning_rate": 9.999773483975061e-06,
      "loss": 3.5261,
      "step": 906
    },
    {
      "epoch": 0.30913428766189505,
      "grad_norm": 36.28576401451518,
      "learning_rate": 9.999754214540536e-06,
      "loss": 3.1505,
      "step": 907
    },
    {
      "epoch": 0.3094751192910702,
      "grad_norm": 16.2909535599977,
      "learning_rate": 9.99973415862477e-06,
      "loss": 2.6659,
      "step": 908
    },
    {
      "epoch": 0.3098159509202454,
      "grad_norm": 60.199884046060866,
      "learning_rate": 9.999713316230919e-06,
      "loss": 3.0563,
      "step": 909
    },
    {
      "epoch": 0.3101567825494206,
      "grad_norm": 19.312308109819764,
      "learning_rate": 9.999691687362258e-06,
      "loss": 3.7053,
      "step": 910
    },
    {
      "epoch": 0.3104976141785958,
      "grad_norm": 31.80731808113836,
      "learning_rate": 9.999669272022194e-06,
      "loss": 3.8509,
      "step": 911
    },
    {
      "epoch": 0.310838445807771,
      "grad_norm": 26.015822544054206,
      "learning_rate": 9.99964607021425e-06,
      "loss": 3.6209,
      "step": 912
    },
    {
      "epoch": 0.3111792774369461,
      "grad_norm": 24.694038163948807,
      "learning_rate": 9.99962208194208e-06,
      "loss": 3.4148,
      "step": 913
    },
    {
      "epoch": 0.3115201090661213,
      "grad_norm": 50.94538943517919,
      "learning_rate": 9.99959730720945e-06,
      "loss": 3.7067,
      "step": 914
    },
    {
      "epoch": 0.3118609406952965,
      "grad_norm": 19.830575864042817,
      "learning_rate": 9.999571746020264e-06,
      "loss": 3.4052,
      "step": 915
    },
    {
      "epoch": 0.3122017723244717,
      "grad_norm": 38.78441278709003,
      "learning_rate": 9.999545398378538e-06,
      "loss": 3.3276,
      "step": 916
    },
    {
      "epoch": 0.3125426039536469,
      "grad_norm": 17.306379258605357,
      "learning_rate": 9.99951826428842e-06,
      "loss": 3.1559,
      "step": 917
    },
    {
      "epoch": 0.3128834355828221,
      "grad_norm": 24.38371517237406,
      "learning_rate": 9.999490343754177e-06,
      "loss": 3.2217,
      "step": 918
    },
    {
      "epoch": 0.31322426721199725,
      "grad_norm": 21.742398537496943,
      "learning_rate": 9.999461636780201e-06,
      "loss": 2.9896,
      "step": 919
    },
    {
      "epoch": 0.31356509884117245,
      "grad_norm": 17.992825848211677,
      "learning_rate": 9.999432143371008e-06,
      "loss": 3.5454,
      "step": 920
    },
    {
      "epoch": 0.31390593047034765,
      "grad_norm": 16.7574433279547,
      "learning_rate": 9.999401863531237e-06,
      "loss": 3.3121,
      "step": 921
    },
    {
      "epoch": 0.31424676209952285,
      "grad_norm": 23.917050259437865,
      "learning_rate": 9.999370797265653e-06,
      "loss": 3.2039,
      "step": 922
    },
    {
      "epoch": 0.31458759372869805,
      "grad_norm": 16.061624809714367,
      "learning_rate": 9.99933894457914e-06,
      "loss": 3.3809,
      "step": 923
    },
    {
      "epoch": 0.3149284253578732,
      "grad_norm": 16.239659569723802,
      "learning_rate": 9.999306305476709e-06,
      "loss": 2.8577,
      "step": 924
    },
    {
      "epoch": 0.3152692569870484,
      "grad_norm": 20.16704851148129,
      "learning_rate": 9.999272879963497e-06,
      "loss": 3.2859,
      "step": 925
    },
    {
      "epoch": 0.3156100886162236,
      "grad_norm": 40.40588521156014,
      "learning_rate": 9.999238668044758e-06,
      "loss": 3.2831,
      "step": 926
    },
    {
      "epoch": 0.3159509202453988,
      "grad_norm": 16.895512865929234,
      "learning_rate": 9.999203669725877e-06,
      "loss": 3.2776,
      "step": 927
    },
    {
      "epoch": 0.316291751874574,
      "grad_norm": 25.61638382704822,
      "learning_rate": 9.999167885012357e-06,
      "loss": 3.6473,
      "step": 928
    },
    {
      "epoch": 0.3166325835037491,
      "grad_norm": 26.154040418672754,
      "learning_rate": 9.99913131390983e-06,
      "loss": 3.4091,
      "step": 929
    },
    {
      "epoch": 0.3169734151329243,
      "grad_norm": 27.239936767315605,
      "learning_rate": 9.999093956424044e-06,
      "loss": 3.4799,
      "step": 930
    },
    {
      "epoch": 0.3173142467620995,
      "grad_norm": 18.036041660163733,
      "learning_rate": 9.99905581256088e-06,
      "loss": 3.6383,
      "step": 931
    },
    {
      "epoch": 0.3176550783912747,
      "grad_norm": 22.143897398083077,
      "learning_rate": 9.999016882326338e-06,
      "loss": 3.7367,
      "step": 932
    },
    {
      "epoch": 0.3179959100204499,
      "grad_norm": 19.8129549041829,
      "learning_rate": 9.998977165726537e-06,
      "loss": 3.3933,
      "step": 933
    },
    {
      "epoch": 0.3183367416496251,
      "grad_norm": 20.218098067652143,
      "learning_rate": 9.998936662767732e-06,
      "loss": 3.4046,
      "step": 934
    },
    {
      "epoch": 0.31867757327880025,
      "grad_norm": 32.175965291488076,
      "learning_rate": 9.998895373456287e-06,
      "loss": 3.7336,
      "step": 935
    },
    {
      "epoch": 0.31901840490797545,
      "grad_norm": 15.444811121662028,
      "learning_rate": 9.998853297798703e-06,
      "loss": 3.3616,
      "step": 936
    },
    {
      "epoch": 0.31935923653715065,
      "grad_norm": 19.117116529845763,
      "learning_rate": 9.998810435801594e-06,
      "loss": 3.1808,
      "step": 937
    },
    {
      "epoch": 0.31970006816632585,
      "grad_norm": 17.520032265709165,
      "learning_rate": 9.998766787471705e-06,
      "loss": 3.2491,
      "step": 938
    },
    {
      "epoch": 0.32004089979550104,
      "grad_norm": 25.868935992525177,
      "learning_rate": 9.998722352815902e-06,
      "loss": 3.3832,
      "step": 939
    },
    {
      "epoch": 0.3203817314246762,
      "grad_norm": 18.08707943829483,
      "learning_rate": 9.998677131841175e-06,
      "loss": 3.0646,
      "step": 940
    },
    {
      "epoch": 0.3207225630538514,
      "grad_norm": 21.300425761289215,
      "learning_rate": 9.998631124554634e-06,
      "loss": 3.1914,
      "step": 941
    },
    {
      "epoch": 0.3210633946830266,
      "grad_norm": 21.950415459436282,
      "learning_rate": 9.99858433096352e-06,
      "loss": 2.8182,
      "step": 942
    },
    {
      "epoch": 0.3214042263122018,
      "grad_norm": 18.096751988766083,
      "learning_rate": 9.998536751075191e-06,
      "loss": 3.1606,
      "step": 943
    },
    {
      "epoch": 0.321745057941377,
      "grad_norm": 16.529596246647436,
      "learning_rate": 9.998488384897136e-06,
      "loss": 3.3708,
      "step": 944
    },
    {
      "epoch": 0.3220858895705521,
      "grad_norm": 38.82901907644897,
      "learning_rate": 9.998439232436958e-06,
      "loss": 3.7399,
      "step": 945
    },
    {
      "epoch": 0.3224267211997273,
      "grad_norm": 22.553126085750584,
      "learning_rate": 9.998389293702393e-06,
      "loss": 3.1169,
      "step": 946
    },
    {
      "epoch": 0.3227675528289025,
      "grad_norm": 19.868806926552562,
      "learning_rate": 9.998338568701293e-06,
      "loss": 2.963,
      "step": 947
    },
    {
      "epoch": 0.3231083844580777,
      "grad_norm": 21.484504978071534,
      "learning_rate": 9.998287057441641e-06,
      "loss": 3.3558,
      "step": 948
    },
    {
      "epoch": 0.3234492160872529,
      "grad_norm": 27.179275940719307,
      "learning_rate": 9.998234759931536e-06,
      "loss": 3.6229,
      "step": 949
    },
    {
      "epoch": 0.3237900477164281,
      "grad_norm": 26.407650505664417,
      "learning_rate": 9.998181676179205e-06,
      "loss": 3.9817,
      "step": 950
    },
    {
      "epoch": 0.32413087934560325,
      "grad_norm": 17.195563626263972,
      "learning_rate": 9.998127806193002e-06,
      "loss": 3.219,
      "step": 951
    },
    {
      "epoch": 0.32447171097477845,
      "grad_norm": 17.99645647069595,
      "learning_rate": 9.998073149981398e-06,
      "loss": 3.0451,
      "step": 952
    },
    {
      "epoch": 0.32481254260395365,
      "grad_norm": 20.64305654598908,
      "learning_rate": 9.998017707552992e-06,
      "loss": 2.9757,
      "step": 953
    },
    {
      "epoch": 0.32515337423312884,
      "grad_norm": 23.418652641809388,
      "learning_rate": 9.997961478916504e-06,
      "loss": 3.383,
      "step": 954
    },
    {
      "epoch": 0.32549420586230404,
      "grad_norm": 16.556362490162357,
      "learning_rate": 9.997904464080779e-06,
      "loss": 3.1277,
      "step": 955
    },
    {
      "epoch": 0.3258350374914792,
      "grad_norm": 13.392588666148562,
      "learning_rate": 9.997846663054786e-06,
      "loss": 2.9564,
      "step": 956
    },
    {
      "epoch": 0.3261758691206544,
      "grad_norm": 25.41985879051928,
      "learning_rate": 9.99778807584762e-06,
      "loss": 3.5963,
      "step": 957
    },
    {
      "epoch": 0.3265167007498296,
      "grad_norm": 19.8072567300104,
      "learning_rate": 9.997728702468493e-06,
      "loss": 3.6759,
      "step": 958
    },
    {
      "epoch": 0.3268575323790048,
      "grad_norm": 19.67160482760699,
      "learning_rate": 9.997668542926747e-06,
      "loss": 3.6193,
      "step": 959
    },
    {
      "epoch": 0.32719836400818,
      "grad_norm": 22.235335556172085,
      "learning_rate": 9.997607597231845e-06,
      "loss": 3.7544,
      "step": 960
    },
    {
      "epoch": 0.3275391956373552,
      "grad_norm": 25.072060075868396,
      "learning_rate": 9.997545865393373e-06,
      "loss": 3.1102,
      "step": 961
    },
    {
      "epoch": 0.3278800272665303,
      "grad_norm": 35.87604644050223,
      "learning_rate": 9.997483347421042e-06,
      "loss": 3.5712,
      "step": 962
    },
    {
      "epoch": 0.3282208588957055,
      "grad_norm": 18.47470784682449,
      "learning_rate": 9.997420043324687e-06,
      "loss": 3.1538,
      "step": 963
    },
    {
      "epoch": 0.3285616905248807,
      "grad_norm": 22.297708871385638,
      "learning_rate": 9.997355953114265e-06,
      "loss": 3.5527,
      "step": 964
    },
    {
      "epoch": 0.3289025221540559,
      "grad_norm": 23.264492087543953,
      "learning_rate": 9.997291076799858e-06,
      "loss": 3.1725,
      "step": 965
    },
    {
      "epoch": 0.3292433537832311,
      "grad_norm": 18.24215283505048,
      "learning_rate": 9.997225414391672e-06,
      "loss": 3.2788,
      "step": 966
    },
    {
      "epoch": 0.32958418541240625,
      "grad_norm": 16.200895207611723,
      "learning_rate": 9.997158965900035e-06,
      "loss": 3.3053,
      "step": 967
    },
    {
      "epoch": 0.32992501704158145,
      "grad_norm": 21.705154378363925,
      "learning_rate": 9.9970917313354e-06,
      "loss": 3.3499,
      "step": 968
    },
    {
      "epoch": 0.33026584867075665,
      "grad_norm": 32.12027799792602,
      "learning_rate": 9.997023710708345e-06,
      "loss": 4.0311,
      "step": 969
    },
    {
      "epoch": 0.33060668029993184,
      "grad_norm": 21.03514301831359,
      "learning_rate": 9.996954904029567e-06,
      "loss": 2.8821,
      "step": 970
    },
    {
      "epoch": 0.33094751192910704,
      "grad_norm": 38.19882425236756,
      "learning_rate": 9.996885311309892e-06,
      "loss": 3.8124,
      "step": 971
    },
    {
      "epoch": 0.3312883435582822,
      "grad_norm": 27.310042903493017,
      "learning_rate": 9.996814932560267e-06,
      "loss": 3.9093,
      "step": 972
    },
    {
      "epoch": 0.3316291751874574,
      "grad_norm": 28.072751961723945,
      "learning_rate": 9.99674376779176e-06,
      "loss": 3.3036,
      "step": 973
    },
    {
      "epoch": 0.3319700068166326,
      "grad_norm": 31.62206738581211,
      "learning_rate": 9.996671817015566e-06,
      "loss": 4.0746,
      "step": 974
    },
    {
      "epoch": 0.3323108384458078,
      "grad_norm": 26.650979341128934,
      "learning_rate": 9.996599080243006e-06,
      "loss": 3.2166,
      "step": 975
    },
    {
      "epoch": 0.332651670074983,
      "grad_norm": 28.973313632973767,
      "learning_rate": 9.996525557485522e-06,
      "loss": 3.7695,
      "step": 976
    },
    {
      "epoch": 0.3329925017041582,
      "grad_norm": 22.597452411932505,
      "learning_rate": 9.996451248754677e-06,
      "loss": 3.8569,
      "step": 977
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 22.838360096890366,
      "learning_rate": 9.99637615406216e-06,
      "loss": 3.4959,
      "step": 978
    },
    {
      "epoch": 0.3336741649625085,
      "grad_norm": 22.617214888476887,
      "learning_rate": 9.996300273419785e-06,
      "loss": 2.9747,
      "step": 979
    },
    {
      "epoch": 0.3340149965916837,
      "grad_norm": 19.244857113410877,
      "learning_rate": 9.996223606839488e-06,
      "loss": 3.7922,
      "step": 980
    },
    {
      "epoch": 0.3343558282208589,
      "grad_norm": 14.825928009453055,
      "learning_rate": 9.996146154333328e-06,
      "loss": 2.9686,
      "step": 981
    },
    {
      "epoch": 0.3346966598500341,
      "grad_norm": 22.373970259865008,
      "learning_rate": 9.996067915913489e-06,
      "loss": 4.0649,
      "step": 982
    },
    {
      "epoch": 0.33503749147920925,
      "grad_norm": 23.846791622542362,
      "learning_rate": 9.99598889159228e-06,
      "loss": 3.6298,
      "step": 983
    },
    {
      "epoch": 0.33537832310838445,
      "grad_norm": 24.198705449201203,
      "learning_rate": 9.995909081382131e-06,
      "loss": 2.9951,
      "step": 984
    },
    {
      "epoch": 0.33571915473755964,
      "grad_norm": 19.89617163897643,
      "learning_rate": 9.995828485295596e-06,
      "loss": 3.22,
      "step": 985
    },
    {
      "epoch": 0.33605998636673484,
      "grad_norm": 15.527251838920543,
      "learning_rate": 9.995747103345352e-06,
      "loss": 3.3982,
      "step": 986
    },
    {
      "epoch": 0.33640081799591004,
      "grad_norm": 22.942905569699253,
      "learning_rate": 9.9956649355442e-06,
      "loss": 3.5283,
      "step": 987
    },
    {
      "epoch": 0.3367416496250852,
      "grad_norm": 20.707743760280447,
      "learning_rate": 9.99558198190507e-06,
      "loss": 3.2543,
      "step": 988
    },
    {
      "epoch": 0.3370824812542604,
      "grad_norm": 23.227905746589638,
      "learning_rate": 9.995498242441006e-06,
      "loss": 3.1755,
      "step": 989
    },
    {
      "epoch": 0.3374233128834356,
      "grad_norm": 23.975459987042164,
      "learning_rate": 9.995413717165183e-06,
      "loss": 3.8023,
      "step": 990
    },
    {
      "epoch": 0.3377641445126108,
      "grad_norm": 48.91309212932905,
      "learning_rate": 9.995328406090899e-06,
      "loss": 3.3713,
      "step": 991
    },
    {
      "epoch": 0.338104976141786,
      "grad_norm": 19.734018733981276,
      "learning_rate": 9.995242309231568e-06,
      "loss": 3.48,
      "step": 992
    },
    {
      "epoch": 0.33844580777096117,
      "grad_norm": 27.019759588525194,
      "learning_rate": 9.995155426600738e-06,
      "loss": 3.5682,
      "step": 993
    },
    {
      "epoch": 0.3387866394001363,
      "grad_norm": 19.897336860144968,
      "learning_rate": 9.995067758212076e-06,
      "loss": 3.1228,
      "step": 994
    },
    {
      "epoch": 0.3391274710293115,
      "grad_norm": 20.203001419003115,
      "learning_rate": 9.99497930407937e-06,
      "loss": 3.4332,
      "step": 995
    },
    {
      "epoch": 0.3394683026584867,
      "grad_norm": 17.63602661928461,
      "learning_rate": 9.994890064216538e-06,
      "loss": 3.6235,
      "step": 996
    },
    {
      "epoch": 0.3398091342876619,
      "grad_norm": 31.562702523443928,
      "learning_rate": 9.994800038637612e-06,
      "loss": 3.1123,
      "step": 997
    },
    {
      "epoch": 0.3401499659168371,
      "grad_norm": 13.812592026331682,
      "learning_rate": 9.99470922735676e-06,
      "loss": 2.7344,
      "step": 998
    },
    {
      "epoch": 0.34049079754601225,
      "grad_norm": 18.92897380675409,
      "learning_rate": 9.994617630388261e-06,
      "loss": 2.6105,
      "step": 999
    },
    {
      "epoch": 0.34083162917518744,
      "grad_norm": 32.44286782331771,
      "learning_rate": 9.994525247746528e-06,
      "loss": 4.0815,
      "step": 1000
    },
    {
      "epoch": 0.34117246080436264,
      "grad_norm": 22.505353032860704,
      "learning_rate": 9.99443207944609e-06,
      "loss": 4.1015,
      "step": 1001
    },
    {
      "epoch": 0.34151329243353784,
      "grad_norm": 18.629999817064817,
      "learning_rate": 9.994338125501606e-06,
      "loss": 3.4006,
      "step": 1002
    },
    {
      "epoch": 0.34185412406271304,
      "grad_norm": 19.31263091644902,
      "learning_rate": 9.994243385927853e-06,
      "loss": 3.2789,
      "step": 1003
    },
    {
      "epoch": 0.3421949556918882,
      "grad_norm": 17.14956943836541,
      "learning_rate": 9.994147860739734e-06,
      "loss": 3.3262,
      "step": 1004
    },
    {
      "epoch": 0.3425357873210634,
      "grad_norm": 19.93820155464246,
      "learning_rate": 9.994051549952278e-06,
      "loss": 3.3257,
      "step": 1005
    },
    {
      "epoch": 0.3428766189502386,
      "grad_norm": 28.41769026316256,
      "learning_rate": 9.99395445358063e-06,
      "loss": 2.9444,
      "step": 1006
    },
    {
      "epoch": 0.3432174505794138,
      "grad_norm": 16.77933358036674,
      "learning_rate": 9.993856571640068e-06,
      "loss": 3.5764,
      "step": 1007
    },
    {
      "epoch": 0.34355828220858897,
      "grad_norm": 18.47257520337231,
      "learning_rate": 9.993757904145987e-06,
      "loss": 3.3175,
      "step": 1008
    },
    {
      "epoch": 0.34389911383776417,
      "grad_norm": 17.752840754367245,
      "learning_rate": 9.99365845111391e-06,
      "loss": 3.0823,
      "step": 1009
    },
    {
      "epoch": 0.3442399454669393,
      "grad_norm": 18.95202423672722,
      "learning_rate": 9.99355821255948e-06,
      "loss": 3.6742,
      "step": 1010
    },
    {
      "epoch": 0.3445807770961145,
      "grad_norm": 32.286025275265054,
      "learning_rate": 9.993457188498464e-06,
      "loss": 2.6442,
      "step": 1011
    },
    {
      "epoch": 0.3449216087252897,
      "grad_norm": 21.20956190627594,
      "learning_rate": 9.993355378946754e-06,
      "loss": 3.069,
      "step": 1012
    },
    {
      "epoch": 0.3452624403544649,
      "grad_norm": 25.048873055573395,
      "learning_rate": 9.993252783920367e-06,
      "loss": 3.5511,
      "step": 1013
    },
    {
      "epoch": 0.3456032719836401,
      "grad_norm": 14.639342274086404,
      "learning_rate": 9.99314940343544e-06,
      "loss": 3.1379,
      "step": 1014
    },
    {
      "epoch": 0.34594410361281525,
      "grad_norm": 13.795425339303037,
      "learning_rate": 9.993045237508234e-06,
      "loss": 2.9424,
      "step": 1015
    },
    {
      "epoch": 0.34628493524199044,
      "grad_norm": 21.74192949293576,
      "learning_rate": 9.992940286155138e-06,
      "loss": 3.9335,
      "step": 1016
    },
    {
      "epoch": 0.34662576687116564,
      "grad_norm": 25.38892144547374,
      "learning_rate": 9.992834549392656e-06,
      "loss": 3.3673,
      "step": 1017
    },
    {
      "epoch": 0.34696659850034084,
      "grad_norm": 19.772981590340688,
      "learning_rate": 9.992728027237427e-06,
      "loss": 2.8217,
      "step": 1018
    },
    {
      "epoch": 0.34730743012951604,
      "grad_norm": 29.92520567380896,
      "learning_rate": 9.992620719706204e-06,
      "loss": 3.7031,
      "step": 1019
    },
    {
      "epoch": 0.3476482617586912,
      "grad_norm": 13.75533107804516,
      "learning_rate": 9.992512626815866e-06,
      "loss": 2.8398,
      "step": 1020
    },
    {
      "epoch": 0.3479890933878664,
      "grad_norm": 24.141401577707605,
      "learning_rate": 9.992403748583418e-06,
      "loss": 3.4206,
      "step": 1021
    },
    {
      "epoch": 0.3483299250170416,
      "grad_norm": 26.208936344467155,
      "learning_rate": 9.992294085025988e-06,
      "loss": 3.9491,
      "step": 1022
    },
    {
      "epoch": 0.3486707566462168,
      "grad_norm": 26.052242924689345,
      "learning_rate": 9.992183636160823e-06,
      "loss": 3.1935,
      "step": 1023
    },
    {
      "epoch": 0.34901158827539197,
      "grad_norm": 20.918931925482756,
      "learning_rate": 9.992072402005302e-06,
      "loss": 2.7452,
      "step": 1024
    },
    {
      "epoch": 0.34935241990456717,
      "grad_norm": 17.18020871842603,
      "learning_rate": 9.991960382576919e-06,
      "loss": 2.9875,
      "step": 1025
    },
    {
      "epoch": 0.3496932515337423,
      "grad_norm": 21.298433274009927,
      "learning_rate": 9.991847577893295e-06,
      "loss": 3.9876,
      "step": 1026
    },
    {
      "epoch": 0.3500340831629175,
      "grad_norm": 24.12213225605878,
      "learning_rate": 9.991733987972177e-06,
      "loss": 3.6153,
      "step": 1027
    },
    {
      "epoch": 0.3503749147920927,
      "grad_norm": 16.416349390066298,
      "learning_rate": 9.99161961283143e-06,
      "loss": 3.0713,
      "step": 1028
    },
    {
      "epoch": 0.3507157464212679,
      "grad_norm": 22.960648603800205,
      "learning_rate": 9.99150445248905e-06,
      "loss": 3.1487,
      "step": 1029
    },
    {
      "epoch": 0.3510565780504431,
      "grad_norm": 16.85458561326823,
      "learning_rate": 9.991388506963147e-06,
      "loss": 3.1732,
      "step": 1030
    },
    {
      "epoch": 0.35139740967961824,
      "grad_norm": 23.96798306310886,
      "learning_rate": 9.991271776271962e-06,
      "loss": 3.7223,
      "step": 1031
    },
    {
      "epoch": 0.35173824130879344,
      "grad_norm": 22.993647256411176,
      "learning_rate": 9.99115426043386e-06,
      "loss": 3.5592,
      "step": 1032
    },
    {
      "epoch": 0.35207907293796864,
      "grad_norm": 17.488522252618193,
      "learning_rate": 9.991035959467322e-06,
      "loss": 3.0401,
      "step": 1033
    },
    {
      "epoch": 0.35241990456714384,
      "grad_norm": 19.23403241155621,
      "learning_rate": 9.99091687339096e-06,
      "loss": 3.666,
      "step": 1034
    },
    {
      "epoch": 0.35276073619631904,
      "grad_norm": 17.482386548088027,
      "learning_rate": 9.990797002223507e-06,
      "loss": 3.0659,
      "step": 1035
    },
    {
      "epoch": 0.35310156782549423,
      "grad_norm": 20.728426574282373,
      "learning_rate": 9.990676345983818e-06,
      "loss": 3.2289,
      "step": 1036
    },
    {
      "epoch": 0.3534423994546694,
      "grad_norm": 22.51524640075904,
      "learning_rate": 9.990554904690873e-06,
      "loss": 3.35,
      "step": 1037
    },
    {
      "epoch": 0.3537832310838446,
      "grad_norm": 19.724930831463066,
      "learning_rate": 9.990432678363776e-06,
      "loss": 3.6873,
      "step": 1038
    },
    {
      "epoch": 0.35412406271301977,
      "grad_norm": 14.408123350344335,
      "learning_rate": 9.990309667021752e-06,
      "loss": 3.4883,
      "step": 1039
    },
    {
      "epoch": 0.35446489434219497,
      "grad_norm": 28.788276423474827,
      "learning_rate": 9.990185870684152e-06,
      "loss": 4.1664,
      "step": 1040
    },
    {
      "epoch": 0.35480572597137017,
      "grad_norm": 15.51585339856858,
      "learning_rate": 9.990061289370449e-06,
      "loss": 3.2768,
      "step": 1041
    },
    {
      "epoch": 0.3551465576005453,
      "grad_norm": 19.502031202402918,
      "learning_rate": 9.989935923100243e-06,
      "loss": 3.0676,
      "step": 1042
    },
    {
      "epoch": 0.3554873892297205,
      "grad_norm": 16.767071111411063,
      "learning_rate": 9.989809771893251e-06,
      "loss": 3.0945,
      "step": 1043
    },
    {
      "epoch": 0.3558282208588957,
      "grad_norm": 22.446917422773613,
      "learning_rate": 9.98968283576932e-06,
      "loss": 3.2116,
      "step": 1044
    },
    {
      "epoch": 0.3561690524880709,
      "grad_norm": 23.20060235834063,
      "learning_rate": 9.989555114748413e-06,
      "loss": 2.9539,
      "step": 1045
    },
    {
      "epoch": 0.3565098841172461,
      "grad_norm": 16.578719503266097,
      "learning_rate": 9.989426608850626e-06,
      "loss": 3.4242,
      "step": 1046
    },
    {
      "epoch": 0.35685071574642124,
      "grad_norm": 22.287035591965946,
      "learning_rate": 9.989297318096172e-06,
      "loss": 3.1724,
      "step": 1047
    },
    {
      "epoch": 0.35719154737559644,
      "grad_norm": 19.32694437642856,
      "learning_rate": 9.98916724250539e-06,
      "loss": 3.409,
      "step": 1048
    },
    {
      "epoch": 0.35753237900477164,
      "grad_norm": 21.36577471081659,
      "learning_rate": 9.989036382098738e-06,
      "loss": 3.9806,
      "step": 1049
    },
    {
      "epoch": 0.35787321063394684,
      "grad_norm": 19.76611956750765,
      "learning_rate": 9.988904736896803e-06,
      "loss": 3.3786,
      "step": 1050
    },
    {
      "epoch": 0.35821404226312203,
      "grad_norm": 15.92363338018686,
      "learning_rate": 9.988772306920293e-06,
      "loss": 2.9802,
      "step": 1051
    },
    {
      "epoch": 0.35855487389229723,
      "grad_norm": 27.428430977679923,
      "learning_rate": 9.98863909219004e-06,
      "loss": 4.3364,
      "step": 1052
    },
    {
      "epoch": 0.3588957055214724,
      "grad_norm": 18.41003612845418,
      "learning_rate": 9.988505092727e-06,
      "loss": 3.0822,
      "step": 1053
    },
    {
      "epoch": 0.3592365371506476,
      "grad_norm": 19.71649056196144,
      "learning_rate": 9.988370308552251e-06,
      "loss": 3.2499,
      "step": 1054
    },
    {
      "epoch": 0.35957736877982277,
      "grad_norm": 23.964632173348072,
      "learning_rate": 9.988234739686995e-06,
      "loss": 3.7395,
      "step": 1055
    },
    {
      "epoch": 0.35991820040899797,
      "grad_norm": 16.992650836957864,
      "learning_rate": 9.988098386152557e-06,
      "loss": 2.8445,
      "step": 1056
    },
    {
      "epoch": 0.36025903203817317,
      "grad_norm": 25.05561620732673,
      "learning_rate": 9.987961247970386e-06,
      "loss": 3.0535,
      "step": 1057
    },
    {
      "epoch": 0.3605998636673483,
      "grad_norm": 20.454041507850672,
      "learning_rate": 9.987823325162056e-06,
      "loss": 3.2925,
      "step": 1058
    },
    {
      "epoch": 0.3609406952965235,
      "grad_norm": 22.82011166604307,
      "learning_rate": 9.98768461774926e-06,
      "loss": 3.0006,
      "step": 1059
    },
    {
      "epoch": 0.3612815269256987,
      "grad_norm": 27.941917653723237,
      "learning_rate": 9.987545125753818e-06,
      "loss": 3.9587,
      "step": 1060
    },
    {
      "epoch": 0.3616223585548739,
      "grad_norm": 19.601863530182797,
      "learning_rate": 9.987404849197677e-06,
      "loss": 3.2454,
      "step": 1061
    },
    {
      "epoch": 0.3619631901840491,
      "grad_norm": 13.508322420607161,
      "learning_rate": 9.987263788102897e-06,
      "loss": 3.0896,
      "step": 1062
    },
    {
      "epoch": 0.36230402181322424,
      "grad_norm": 16.586454953095092,
      "learning_rate": 9.98712194249167e-06,
      "loss": 3.906,
      "step": 1063
    },
    {
      "epoch": 0.36264485344239944,
      "grad_norm": 17.593708759555597,
      "learning_rate": 9.98697931238631e-06,
      "loss": 2.8176,
      "step": 1064
    },
    {
      "epoch": 0.36298568507157464,
      "grad_norm": 22.713781142112314,
      "learning_rate": 9.986835897809252e-06,
      "loss": 3.1157,
      "step": 1065
    },
    {
      "epoch": 0.36332651670074984,
      "grad_norm": 19.645757694658144,
      "learning_rate": 9.986691698783055e-06,
      "loss": 3.1626,
      "step": 1066
    },
    {
      "epoch": 0.36366734832992503,
      "grad_norm": 15.009012236390939,
      "learning_rate": 9.986546715330402e-06,
      "loss": 3.3271,
      "step": 1067
    },
    {
      "epoch": 0.36400817995910023,
      "grad_norm": 18.96198096082133,
      "learning_rate": 9.986400947474102e-06,
      "loss": 3.4214,
      "step": 1068
    },
    {
      "epoch": 0.3643490115882754,
      "grad_norm": 19.166306887422248,
      "learning_rate": 9.986254395237083e-06,
      "loss": 3.1294,
      "step": 1069
    },
    {
      "epoch": 0.36468984321745057,
      "grad_norm": 18.936331885375957,
      "learning_rate": 9.986107058642398e-06,
      "loss": 3.638,
      "step": 1070
    },
    {
      "epoch": 0.36503067484662577,
      "grad_norm": 18.86850166421172,
      "learning_rate": 9.985958937713226e-06,
      "loss": 3.5961,
      "step": 1071
    },
    {
      "epoch": 0.36537150647580097,
      "grad_norm": 17.558983996613158,
      "learning_rate": 9.985810032472862e-06,
      "loss": 3.3082,
      "step": 1072
    },
    {
      "epoch": 0.36571233810497616,
      "grad_norm": 20.467214295716197,
      "learning_rate": 9.985660342944735e-06,
      "loss": 3.6779,
      "step": 1073
    },
    {
      "epoch": 0.3660531697341513,
      "grad_norm": 33.3392009170554,
      "learning_rate": 9.985509869152388e-06,
      "loss": 3.5264,
      "step": 1074
    },
    {
      "epoch": 0.3663940013633265,
      "grad_norm": 23.43209530143601,
      "learning_rate": 9.985358611119494e-06,
      "loss": 3.1104,
      "step": 1075
    },
    {
      "epoch": 0.3667348329925017,
      "grad_norm": 19.71275364543631,
      "learning_rate": 9.985206568869843e-06,
      "loss": 3.2861,
      "step": 1076
    },
    {
      "epoch": 0.3670756646216769,
      "grad_norm": 27.445077004842894,
      "learning_rate": 9.985053742427355e-06,
      "loss": 3.3212,
      "step": 1077
    },
    {
      "epoch": 0.3674164962508521,
      "grad_norm": 20.625561695016522,
      "learning_rate": 9.984900131816068e-06,
      "loss": 3.8432,
      "step": 1078
    },
    {
      "epoch": 0.36775732788002724,
      "grad_norm": 18.33900375240487,
      "learning_rate": 9.984745737060147e-06,
      "loss": 3.4131,
      "step": 1079
    },
    {
      "epoch": 0.36809815950920244,
      "grad_norm": 34.10523488220508,
      "learning_rate": 9.984590558183879e-06,
      "loss": 3.0082,
      "step": 1080
    },
    {
      "epoch": 0.36843899113837764,
      "grad_norm": 23.47722378492024,
      "learning_rate": 9.984434595211673e-06,
      "loss": 3.5834,
      "step": 1081
    },
    {
      "epoch": 0.36877982276755283,
      "grad_norm": 19.92961329084212,
      "learning_rate": 9.984277848168062e-06,
      "loss": 2.7864,
      "step": 1082
    },
    {
      "epoch": 0.36912065439672803,
      "grad_norm": 18.068075194691808,
      "learning_rate": 9.984120317077707e-06,
      "loss": 3.014,
      "step": 1083
    },
    {
      "epoch": 0.36946148602590323,
      "grad_norm": 43.0074893297257,
      "learning_rate": 9.983962001965383e-06,
      "loss": 3.4474,
      "step": 1084
    },
    {
      "epoch": 0.36980231765507837,
      "grad_norm": 18.353886047426464,
      "learning_rate": 9.983802902855998e-06,
      "loss": 3.0148,
      "step": 1085
    },
    {
      "epoch": 0.37014314928425357,
      "grad_norm": 19.82067375153483,
      "learning_rate": 9.983643019774577e-06,
      "loss": 3.4038,
      "step": 1086
    },
    {
      "epoch": 0.37048398091342877,
      "grad_norm": 14.983948240272273,
      "learning_rate": 9.983482352746269e-06,
      "loss": 3.4074,
      "step": 1087
    },
    {
      "epoch": 0.37082481254260397,
      "grad_norm": 36.498563058136014,
      "learning_rate": 9.983320901796349e-06,
      "loss": 3.5127,
      "step": 1088
    },
    {
      "epoch": 0.37116564417177916,
      "grad_norm": 20.19239493846256,
      "learning_rate": 9.983158666950215e-06,
      "loss": 3.1834,
      "step": 1089
    },
    {
      "epoch": 0.3715064758009543,
      "grad_norm": 17.543431272890693,
      "learning_rate": 9.982995648233385e-06,
      "loss": 3.2767,
      "step": 1090
    },
    {
      "epoch": 0.3718473074301295,
      "grad_norm": 15.76336839049142,
      "learning_rate": 9.982831845671503e-06,
      "loss": 3.0666,
      "step": 1091
    },
    {
      "epoch": 0.3721881390593047,
      "grad_norm": 18.608142862155752,
      "learning_rate": 9.982667259290336e-06,
      "loss": 3.4996,
      "step": 1092
    },
    {
      "epoch": 0.3725289706884799,
      "grad_norm": 15.082139354553401,
      "learning_rate": 9.982501889115774e-06,
      "loss": 3.0387,
      "step": 1093
    },
    {
      "epoch": 0.3728698023176551,
      "grad_norm": 20.769727737871346,
      "learning_rate": 9.982335735173832e-06,
      "loss": 3.4275,
      "step": 1094
    },
    {
      "epoch": 0.37321063394683024,
      "grad_norm": 20.48396914150774,
      "learning_rate": 9.982168797490644e-06,
      "loss": 3.883,
      "step": 1095
    },
    {
      "epoch": 0.37355146557600544,
      "grad_norm": 19.425064877229985,
      "learning_rate": 9.982001076092471e-06,
      "loss": 3.1877,
      "step": 1096
    },
    {
      "epoch": 0.37389229720518063,
      "grad_norm": 16.570002116038122,
      "learning_rate": 9.981832571005698e-06,
      "loss": 3.4833,
      "step": 1097
    },
    {
      "epoch": 0.37423312883435583,
      "grad_norm": 16.114235220459534,
      "learning_rate": 9.98166328225683e-06,
      "loss": 3.4179,
      "step": 1098
    },
    {
      "epoch": 0.37457396046353103,
      "grad_norm": 25.08823715116121,
      "learning_rate": 9.981493209872498e-06,
      "loss": 3.8198,
      "step": 1099
    },
    {
      "epoch": 0.37491479209270623,
      "grad_norm": 16.35297073757911,
      "learning_rate": 9.98132235387945e-06,
      "loss": 3.023,
      "step": 1100
    },
    {
      "epoch": 0.37525562372188137,
      "grad_norm": 22.283598192950166,
      "learning_rate": 9.98115071430457e-06,
      "loss": 3.5575,
      "step": 1101
    },
    {
      "epoch": 0.37559645535105657,
      "grad_norm": 29.156705265778815,
      "learning_rate": 9.980978291174851e-06,
      "loss": 3.362,
      "step": 1102
    },
    {
      "epoch": 0.37593728698023177,
      "grad_norm": 24.999208422625884,
      "learning_rate": 9.98080508451742e-06,
      "loss": 3.1936,
      "step": 1103
    },
    {
      "epoch": 0.37627811860940696,
      "grad_norm": 46.42766859682043,
      "learning_rate": 9.980631094359523e-06,
      "loss": 3.7643,
      "step": 1104
    },
    {
      "epoch": 0.37661895023858216,
      "grad_norm": 19.306817013925766,
      "learning_rate": 9.980456320728526e-06,
      "loss": 3.0389,
      "step": 1105
    },
    {
      "epoch": 0.3769597818677573,
      "grad_norm": 20.47327378173784,
      "learning_rate": 9.980280763651925e-06,
      "loss": 3.407,
      "step": 1106
    },
    {
      "epoch": 0.3773006134969325,
      "grad_norm": 27.978346320783228,
      "learning_rate": 9.980104423157334e-06,
      "loss": 3.1507,
      "step": 1107
    },
    {
      "epoch": 0.3776414451261077,
      "grad_norm": 18.358841119529536,
      "learning_rate": 9.979927299272493e-06,
      "loss": 3.3895,
      "step": 1108
    },
    {
      "epoch": 0.3779822767552829,
      "grad_norm": 17.326984252146055,
      "learning_rate": 9.979749392025264e-06,
      "loss": 3.1683,
      "step": 1109
    },
    {
      "epoch": 0.3783231083844581,
      "grad_norm": 18.364077076491938,
      "learning_rate": 9.97957070144363e-06,
      "loss": 3.653,
      "step": 1110
    },
    {
      "epoch": 0.37866394001363324,
      "grad_norm": 19.982112407445076,
      "learning_rate": 9.979391227555706e-06,
      "loss": 4.0101,
      "step": 1111
    },
    {
      "epoch": 0.37900477164280844,
      "grad_norm": 26.75860803747121,
      "learning_rate": 9.979210970389718e-06,
      "loss": 3.4242,
      "step": 1112
    },
    {
      "epoch": 0.37934560327198363,
      "grad_norm": 42.1185977821696,
      "learning_rate": 9.979029929974022e-06,
      "loss": 3.1769,
      "step": 1113
    },
    {
      "epoch": 0.37968643490115883,
      "grad_norm": 20.095407942551564,
      "learning_rate": 9.9788481063371e-06,
      "loss": 3.0247,
      "step": 1114
    },
    {
      "epoch": 0.38002726653033403,
      "grad_norm": 35.16746100125832,
      "learning_rate": 9.97866549950755e-06,
      "loss": 3.4061,
      "step": 1115
    },
    {
      "epoch": 0.3803680981595092,
      "grad_norm": 18.887572618894907,
      "learning_rate": 9.978482109514098e-06,
      "loss": 3.5824,
      "step": 1116
    },
    {
      "epoch": 0.38070892978868437,
      "grad_norm": 16.1929320972325,
      "learning_rate": 9.978297936385593e-06,
      "loss": 3.2194,
      "step": 1117
    },
    {
      "epoch": 0.38104976141785957,
      "grad_norm": 24.367166620844337,
      "learning_rate": 9.978112980151005e-06,
      "loss": 3.2417,
      "step": 1118
    },
    {
      "epoch": 0.38139059304703476,
      "grad_norm": 18.627421366288406,
      "learning_rate": 9.977927240839428e-06,
      "loss": 2.8761,
      "step": 1119
    },
    {
      "epoch": 0.38173142467620996,
      "grad_norm": 24.181655497000293,
      "learning_rate": 9.97774071848008e-06,
      "loss": 4.2616,
      "step": 1120
    },
    {
      "epoch": 0.38207225630538516,
      "grad_norm": 24.902860714299575,
      "learning_rate": 9.9775534131023e-06,
      "loss": 3.7441,
      "step": 1121
    },
    {
      "epoch": 0.3824130879345603,
      "grad_norm": 15.899223906028,
      "learning_rate": 9.977365324735557e-06,
      "loss": 2.8298,
      "step": 1122
    },
    {
      "epoch": 0.3827539195637355,
      "grad_norm": 35.296482765797435,
      "learning_rate": 9.977176453409433e-06,
      "loss": 3.0591,
      "step": 1123
    },
    {
      "epoch": 0.3830947511929107,
      "grad_norm": 46.64766714505639,
      "learning_rate": 9.97698679915364e-06,
      "loss": 3.0709,
      "step": 1124
    },
    {
      "epoch": 0.3834355828220859,
      "grad_norm": 16.297870507727698,
      "learning_rate": 9.976796361998012e-06,
      "loss": 3.1372,
      "step": 1125
    },
    {
      "epoch": 0.3837764144512611,
      "grad_norm": 16.298386037359844,
      "learning_rate": 9.976605141972502e-06,
      "loss": 3.4797,
      "step": 1126
    },
    {
      "epoch": 0.3841172460804363,
      "grad_norm": 22.4298983907977,
      "learning_rate": 9.976413139107195e-06,
      "loss": 3.3933,
      "step": 1127
    },
    {
      "epoch": 0.38445807770961143,
      "grad_norm": 14.223846316844636,
      "learning_rate": 9.97622035343229e-06,
      "loss": 3.0967,
      "step": 1128
    },
    {
      "epoch": 0.38479890933878663,
      "grad_norm": 13.061440596225498,
      "learning_rate": 9.976026784978116e-06,
      "loss": 2.8874,
      "step": 1129
    },
    {
      "epoch": 0.38513974096796183,
      "grad_norm": 22.189187265980017,
      "learning_rate": 9.97583243377512e-06,
      "loss": 3.0119,
      "step": 1130
    },
    {
      "epoch": 0.38548057259713703,
      "grad_norm": 21.628301764070514,
      "learning_rate": 9.975637299853872e-06,
      "loss": 3.3413,
      "step": 1131
    },
    {
      "epoch": 0.3858214042263122,
      "grad_norm": 16.888657742961943,
      "learning_rate": 9.975441383245073e-06,
      "loss": 3.1451,
      "step": 1132
    },
    {
      "epoch": 0.38616223585548737,
      "grad_norm": 14.914431617443563,
      "learning_rate": 9.975244683979536e-06,
      "loss": 3.0768,
      "step": 1133
    },
    {
      "epoch": 0.38650306748466257,
      "grad_norm": 23.70323853397031,
      "learning_rate": 9.975047202088206e-06,
      "loss": 2.8694,
      "step": 1134
    },
    {
      "epoch": 0.38684389911383776,
      "grad_norm": 17.137650281738782,
      "learning_rate": 9.974848937602146e-06,
      "loss": 3.8108,
      "step": 1135
    },
    {
      "epoch": 0.38718473074301296,
      "grad_norm": 16.805433532685782,
      "learning_rate": 9.974649890552546e-06,
      "loss": 3.1146,
      "step": 1136
    },
    {
      "epoch": 0.38752556237218816,
      "grad_norm": 20.51841487254913,
      "learning_rate": 9.974450060970715e-06,
      "loss": 3.8989,
      "step": 1137
    },
    {
      "epoch": 0.3878663940013633,
      "grad_norm": 15.191953214370274,
      "learning_rate": 9.974249448888084e-06,
      "loss": 3.4339,
      "step": 1138
    },
    {
      "epoch": 0.3882072256305385,
      "grad_norm": 23.729340425477773,
      "learning_rate": 9.974048054336218e-06,
      "loss": 3.3732,
      "step": 1139
    },
    {
      "epoch": 0.3885480572597137,
      "grad_norm": 23.53579880755216,
      "learning_rate": 9.97384587734679e-06,
      "loss": 3.2826,
      "step": 1140
    },
    {
      "epoch": 0.3888888888888889,
      "grad_norm": 27.01249958206964,
      "learning_rate": 9.973642917951604e-06,
      "loss": 3.4409,
      "step": 1141
    },
    {
      "epoch": 0.3892297205180641,
      "grad_norm": 16.932066428977826,
      "learning_rate": 9.97343917618259e-06,
      "loss": 3.154,
      "step": 1142
    },
    {
      "epoch": 0.3895705521472393,
      "grad_norm": 17.046706001021196,
      "learning_rate": 9.973234652071797e-06,
      "loss": 3.1687,
      "step": 1143
    },
    {
      "epoch": 0.38991138377641443,
      "grad_norm": 14.264721413914925,
      "learning_rate": 9.973029345651395e-06,
      "loss": 2.7162,
      "step": 1144
    },
    {
      "epoch": 0.39025221540558963,
      "grad_norm": 35.9278860429445,
      "learning_rate": 9.972823256953679e-06,
      "loss": 2.7275,
      "step": 1145
    },
    {
      "epoch": 0.39059304703476483,
      "grad_norm": 18.266933682048528,
      "learning_rate": 9.97261638601107e-06,
      "loss": 3.4257,
      "step": 1146
    },
    {
      "epoch": 0.39093387866394,
      "grad_norm": 15.886323661873147,
      "learning_rate": 9.97240873285611e-06,
      "loss": 3.4741,
      "step": 1147
    },
    {
      "epoch": 0.3912747102931152,
      "grad_norm": 25.055164552576443,
      "learning_rate": 9.972200297521459e-06,
      "loss": 3.3787,
      "step": 1148
    },
    {
      "epoch": 0.39161554192229037,
      "grad_norm": 37.54278055158264,
      "learning_rate": 9.971991080039912e-06,
      "loss": 3.7619,
      "step": 1149
    },
    {
      "epoch": 0.39195637355146556,
      "grad_norm": 23.64788019628668,
      "learning_rate": 9.971781080444372e-06,
      "loss": 2.8984,
      "step": 1150
    },
    {
      "epoch": 0.39229720518064076,
      "grad_norm": 15.349091148105563,
      "learning_rate": 9.97157029876788e-06,
      "loss": 3.3605,
      "step": 1151
    },
    {
      "epoch": 0.39263803680981596,
      "grad_norm": 24.62082065806219,
      "learning_rate": 9.971358735043586e-06,
      "loss": 3.7627,
      "step": 1152
    },
    {
      "epoch": 0.39297886843899116,
      "grad_norm": 19.19083650158772,
      "learning_rate": 9.971146389304775e-06,
      "loss": 3.2797,
      "step": 1153
    },
    {
      "epoch": 0.3933197000681663,
      "grad_norm": 25.65918065270858,
      "learning_rate": 9.970933261584848e-06,
      "loss": 3.3647,
      "step": 1154
    },
    {
      "epoch": 0.3936605316973415,
      "grad_norm": 26.15407115935983,
      "learning_rate": 9.97071935191733e-06,
      "loss": 3.6857,
      "step": 1155
    },
    {
      "epoch": 0.3940013633265167,
      "grad_norm": 26.932113974397037,
      "learning_rate": 9.97050466033587e-06,
      "loss": 3.0976,
      "step": 1156
    },
    {
      "epoch": 0.3943421949556919,
      "grad_norm": 19.21007095761264,
      "learning_rate": 9.970289186874242e-06,
      "loss": 2.8935,
      "step": 1157
    },
    {
      "epoch": 0.3946830265848671,
      "grad_norm": 17.625942220545294,
      "learning_rate": 9.970072931566339e-06,
      "loss": 2.8462,
      "step": 1158
    },
    {
      "epoch": 0.3950238582140423,
      "grad_norm": 29.5502537398197,
      "learning_rate": 9.96985589444618e-06,
      "loss": 3.1347,
      "step": 1159
    },
    {
      "epoch": 0.39536468984321743,
      "grad_norm": 23.923542769331835,
      "learning_rate": 9.969638075547904e-06,
      "loss": 3.9263,
      "step": 1160
    },
    {
      "epoch": 0.39570552147239263,
      "grad_norm": 25.52929156483782,
      "learning_rate": 9.969419474905775e-06,
      "loss": 3.4313,
      "step": 1161
    },
    {
      "epoch": 0.3960463531015678,
      "grad_norm": 19.947560843630296,
      "learning_rate": 9.969200092554182e-06,
      "loss": 3.5263,
      "step": 1162
    },
    {
      "epoch": 0.396387184730743,
      "grad_norm": 24.390865720884936,
      "learning_rate": 9.968979928527632e-06,
      "loss": 3.9256,
      "step": 1163
    },
    {
      "epoch": 0.3967280163599182,
      "grad_norm": 17.451106354969276,
      "learning_rate": 9.96875898286076e-06,
      "loss": 3.0229,
      "step": 1164
    },
    {
      "epoch": 0.39706884798909337,
      "grad_norm": 20.588576688527375,
      "learning_rate": 9.96853725558832e-06,
      "loss": 3.699,
      "step": 1165
    },
    {
      "epoch": 0.39740967961826856,
      "grad_norm": 29.712170549512468,
      "learning_rate": 9.96831474674519e-06,
      "loss": 3.8406,
      "step": 1166
    },
    {
      "epoch": 0.39775051124744376,
      "grad_norm": 16.2421190833622,
      "learning_rate": 9.968091456366374e-06,
      "loss": 2.9316,
      "step": 1167
    },
    {
      "epoch": 0.39809134287661896,
      "grad_norm": 29.979478379399776,
      "learning_rate": 9.967867384486994e-06,
      "loss": 3.6148,
      "step": 1168
    },
    {
      "epoch": 0.39843217450579416,
      "grad_norm": 15.006749803392585,
      "learning_rate": 9.967642531142298e-06,
      "loss": 3.3437,
      "step": 1169
    },
    {
      "epoch": 0.3987730061349693,
      "grad_norm": 17.17188861440722,
      "learning_rate": 9.967416896367659e-06,
      "loss": 3.1986,
      "step": 1170
    },
    {
      "epoch": 0.3991138377641445,
      "grad_norm": 15.382429578297627,
      "learning_rate": 9.967190480198568e-06,
      "loss": 3.0635,
      "step": 1171
    },
    {
      "epoch": 0.3994546693933197,
      "grad_norm": 21.75481244313121,
      "learning_rate": 9.96696328267064e-06,
      "loss": 3.1002,
      "step": 1172
    },
    {
      "epoch": 0.3997955010224949,
      "grad_norm": 21.65600401874912,
      "learning_rate": 9.966735303819615e-06,
      "loss": 3.4696,
      "step": 1173
    },
    {
      "epoch": 0.4001363326516701,
      "grad_norm": 26.20538907480564,
      "learning_rate": 9.966506543681355e-06,
      "loss": 3.6298,
      "step": 1174
    },
    {
      "epoch": 0.4004771642808453,
      "grad_norm": 15.384027607111575,
      "learning_rate": 9.966277002291846e-06,
      "loss": 3.1741,
      "step": 1175
    },
    {
      "epoch": 0.40081799591002043,
      "grad_norm": 21.575583361437022,
      "learning_rate": 9.966046679687196e-06,
      "loss": 3.4558,
      "step": 1176
    },
    {
      "epoch": 0.40115882753919563,
      "grad_norm": 14.93224192590806,
      "learning_rate": 9.965815575903633e-06,
      "loss": 3.1491,
      "step": 1177
    },
    {
      "epoch": 0.4014996591683708,
      "grad_norm": 18.266480124060873,
      "learning_rate": 9.965583690977513e-06,
      "loss": 3.271,
      "step": 1178
    },
    {
      "epoch": 0.401840490797546,
      "grad_norm": 18.15014878812278,
      "learning_rate": 9.96535102494531e-06,
      "loss": 3.4557,
      "step": 1179
    },
    {
      "epoch": 0.4021813224267212,
      "grad_norm": 14.78591868993229,
      "learning_rate": 9.965117577843627e-06,
      "loss": 3.5044,
      "step": 1180
    },
    {
      "epoch": 0.40252215405589636,
      "grad_norm": 18.006239556464017,
      "learning_rate": 9.964883349709182e-06,
      "loss": 3.432,
      "step": 1181
    },
    {
      "epoch": 0.40286298568507156,
      "grad_norm": 15.5006342128743,
      "learning_rate": 9.964648340578825e-06,
      "loss": 3.2959,
      "step": 1182
    },
    {
      "epoch": 0.40320381731424676,
      "grad_norm": 22.29590921288553,
      "learning_rate": 9.964412550489518e-06,
      "loss": 2.9379,
      "step": 1183
    },
    {
      "epoch": 0.40354464894342196,
      "grad_norm": 14.666222181379753,
      "learning_rate": 9.964175979478356e-06,
      "loss": 3.3488,
      "step": 1184
    },
    {
      "epoch": 0.40388548057259716,
      "grad_norm": 27.572666016295813,
      "learning_rate": 9.96393862758255e-06,
      "loss": 3.4957,
      "step": 1185
    },
    {
      "epoch": 0.4042263122017723,
      "grad_norm": 26.041482139458992,
      "learning_rate": 9.963700494839439e-06,
      "loss": 3.5245,
      "step": 1186
    },
    {
      "epoch": 0.4045671438309475,
      "grad_norm": 10.721662597377698,
      "learning_rate": 9.963461581286479e-06,
      "loss": 2.7889,
      "step": 1187
    },
    {
      "epoch": 0.4049079754601227,
      "grad_norm": 21.171672107576963,
      "learning_rate": 9.963221886961256e-06,
      "loss": 3.3197,
      "step": 1188
    },
    {
      "epoch": 0.4052488070892979,
      "grad_norm": 16.45736871802216,
      "learning_rate": 9.962981411901472e-06,
      "loss": 2.6047,
      "step": 1189
    },
    {
      "epoch": 0.4055896387184731,
      "grad_norm": 18.754869212454427,
      "learning_rate": 9.962740156144955e-06,
      "loss": 3.74,
      "step": 1190
    },
    {
      "epoch": 0.4059304703476483,
      "grad_norm": 19.258123340232142,
      "learning_rate": 9.962498119729657e-06,
      "loss": 3.4674,
      "step": 1191
    },
    {
      "epoch": 0.40627130197682343,
      "grad_norm": 22.752948788345012,
      "learning_rate": 9.962255302693649e-06,
      "loss": 3.315,
      "step": 1192
    },
    {
      "epoch": 0.4066121336059986,
      "grad_norm": 15.21069862633653,
      "learning_rate": 9.962011705075129e-06,
      "loss": 3.1081,
      "step": 1193
    },
    {
      "epoch": 0.4069529652351738,
      "grad_norm": 23.76549426568509,
      "learning_rate": 9.961767326912415e-06,
      "loss": 3.1769,
      "step": 1194
    },
    {
      "epoch": 0.407293796864349,
      "grad_norm": 29.168571396164477,
      "learning_rate": 9.961522168243948e-06,
      "loss": 3.4966,
      "step": 1195
    },
    {
      "epoch": 0.4076346284935242,
      "grad_norm": 19.822056281598023,
      "learning_rate": 9.961276229108294e-06,
      "loss": 3.4679,
      "step": 1196
    },
    {
      "epoch": 0.40797546012269936,
      "grad_norm": 19.875313755070206,
      "learning_rate": 9.96102950954414e-06,
      "loss": 3.2736,
      "step": 1197
    },
    {
      "epoch": 0.40831629175187456,
      "grad_norm": 17.07387588528351,
      "learning_rate": 9.960782009590294e-06,
      "loss": 3.2129,
      "step": 1198
    },
    {
      "epoch": 0.40865712338104976,
      "grad_norm": 42.11518279910929,
      "learning_rate": 9.96053372928569e-06,
      "loss": 3.3457,
      "step": 1199
    },
    {
      "epoch": 0.40899795501022496,
      "grad_norm": 29.04058917696115,
      "learning_rate": 9.960284668669386e-06,
      "loss": 3.9032,
      "step": 1200
    },
    {
      "epoch": 0.40933878663940015,
      "grad_norm": 18.127108684787718,
      "learning_rate": 9.960034827780557e-06,
      "loss": 3.0157,
      "step": 1201
    },
    {
      "epoch": 0.4096796182685753,
      "grad_norm": 27.258651527999735,
      "learning_rate": 9.959784206658502e-06,
      "loss": 3.168,
      "step": 1202
    },
    {
      "epoch": 0.4100204498977505,
      "grad_norm": 18.857935641290723,
      "learning_rate": 9.95953280534265e-06,
      "loss": 3.3551,
      "step": 1203
    },
    {
      "epoch": 0.4103612815269257,
      "grad_norm": 34.289783103871564,
      "learning_rate": 9.959280623872546e-06,
      "loss": 3.1036,
      "step": 1204
    },
    {
      "epoch": 0.4107021131561009,
      "grad_norm": 24.660154376713834,
      "learning_rate": 9.959027662287856e-06,
      "loss": 3.571,
      "step": 1205
    },
    {
      "epoch": 0.4110429447852761,
      "grad_norm": 28.81366649594681,
      "learning_rate": 9.958773920628375e-06,
      "loss": 3.7016,
      "step": 1206
    },
    {
      "epoch": 0.4113837764144513,
      "grad_norm": 18.909304606844096,
      "learning_rate": 9.958519398934017e-06,
      "loss": 3.4561,
      "step": 1207
    },
    {
      "epoch": 0.4117246080436264,
      "grad_norm": 31.51682313397142,
      "learning_rate": 9.95826409724482e-06,
      "loss": 3.748,
      "step": 1208
    },
    {
      "epoch": 0.4120654396728016,
      "grad_norm": 17.73897980575181,
      "learning_rate": 9.95800801560094e-06,
      "loss": 3.4925,
      "step": 1209
    },
    {
      "epoch": 0.4124062713019768,
      "grad_norm": 30.239087492543515,
      "learning_rate": 9.957751154042663e-06,
      "loss": 3.6837,
      "step": 1210
    },
    {
      "epoch": 0.412747102931152,
      "grad_norm": 18.016705569432823,
      "learning_rate": 9.957493512610394e-06,
      "loss": 3.2729,
      "step": 1211
    },
    {
      "epoch": 0.4130879345603272,
      "grad_norm": 16.39644097297851,
      "learning_rate": 9.95723509134466e-06,
      "loss": 2.9552,
      "step": 1212
    },
    {
      "epoch": 0.41342876618950236,
      "grad_norm": 29.66237232711792,
      "learning_rate": 9.956975890286115e-06,
      "loss": 3.465,
      "step": 1213
    },
    {
      "epoch": 0.41376959781867756,
      "grad_norm": 14.694052660780379,
      "learning_rate": 9.956715909475528e-06,
      "loss": 3.0025,
      "step": 1214
    },
    {
      "epoch": 0.41411042944785276,
      "grad_norm": 24.375854368229295,
      "learning_rate": 9.956455148953797e-06,
      "loss": 3.6427,
      "step": 1215
    },
    {
      "epoch": 0.41445126107702795,
      "grad_norm": 17.762791502644774,
      "learning_rate": 9.956193608761941e-06,
      "loss": 3.5354,
      "step": 1216
    },
    {
      "epoch": 0.41479209270620315,
      "grad_norm": 20.639002739532813,
      "learning_rate": 9.955931288941101e-06,
      "loss": 3.4592,
      "step": 1217
    },
    {
      "epoch": 0.41513292433537835,
      "grad_norm": 15.908862691975889,
      "learning_rate": 9.95566818953254e-06,
      "loss": 3.1109,
      "step": 1218
    },
    {
      "epoch": 0.4154737559645535,
      "grad_norm": 15.752169728877808,
      "learning_rate": 9.955404310577648e-06,
      "loss": 3.2173,
      "step": 1219
    },
    {
      "epoch": 0.4158145875937287,
      "grad_norm": 28.653289453367275,
      "learning_rate": 9.955139652117929e-06,
      "loss": 3.6138,
      "step": 1220
    },
    {
      "epoch": 0.4161554192229039,
      "grad_norm": 24.058059072534068,
      "learning_rate": 9.954874214195018e-06,
      "loss": 3.6496,
      "step": 1221
    },
    {
      "epoch": 0.4164962508520791,
      "grad_norm": 19.41644537365908,
      "learning_rate": 9.954607996850669e-06,
      "loss": 4.0455,
      "step": 1222
    },
    {
      "epoch": 0.4168370824812543,
      "grad_norm": 16.583559865747322,
      "learning_rate": 9.95434100012676e-06,
      "loss": 3.4023,
      "step": 1223
    },
    {
      "epoch": 0.4171779141104294,
      "grad_norm": 34.37938987958627,
      "learning_rate": 9.954073224065288e-06,
      "loss": 3.2806,
      "step": 1224
    },
    {
      "epoch": 0.4175187457396046,
      "grad_norm": 20.956859628261356,
      "learning_rate": 9.953804668708379e-06,
      "loss": 3.123,
      "step": 1225
    },
    {
      "epoch": 0.4178595773687798,
      "grad_norm": 25.40207370172699,
      "learning_rate": 9.953535334098273e-06,
      "loss": 3.4604,
      "step": 1226
    },
    {
      "epoch": 0.418200408997955,
      "grad_norm": 19.580034035730385,
      "learning_rate": 9.953265220277341e-06,
      "loss": 3.3837,
      "step": 1227
    },
    {
      "epoch": 0.4185412406271302,
      "grad_norm": 20.274286627222697,
      "learning_rate": 9.952994327288072e-06,
      "loss": 3.4538,
      "step": 1228
    },
    {
      "epoch": 0.41888207225630536,
      "grad_norm": 21.000560022001828,
      "learning_rate": 9.952722655173079e-06,
      "loss": 3.4361,
      "step": 1229
    },
    {
      "epoch": 0.41922290388548056,
      "grad_norm": 16.163294430206808,
      "learning_rate": 9.952450203975097e-06,
      "loss": 3.0352,
      "step": 1230
    },
    {
      "epoch": 0.41956373551465576,
      "grad_norm": 12.96291220419135,
      "learning_rate": 9.952176973736983e-06,
      "loss": 2.5976,
      "step": 1231
    },
    {
      "epoch": 0.41990456714383095,
      "grad_norm": 20.173337170746663,
      "learning_rate": 9.951902964501717e-06,
      "loss": 3.7217,
      "step": 1232
    },
    {
      "epoch": 0.42024539877300615,
      "grad_norm": 14.254582149545545,
      "learning_rate": 9.951628176312403e-06,
      "loss": 3.1739,
      "step": 1233
    },
    {
      "epoch": 0.42058623040218135,
      "grad_norm": 20.32513125576033,
      "learning_rate": 9.951352609212264e-06,
      "loss": 3.3297,
      "step": 1234
    },
    {
      "epoch": 0.4209270620313565,
      "grad_norm": 13.410238802540723,
      "learning_rate": 9.95107626324465e-06,
      "loss": 3.1863,
      "step": 1235
    },
    {
      "epoch": 0.4212678936605317,
      "grad_norm": 14.14263142417603,
      "learning_rate": 9.950799138453033e-06,
      "loss": 3.0327,
      "step": 1236
    },
    {
      "epoch": 0.4216087252897069,
      "grad_norm": 25.265025431958858,
      "learning_rate": 9.950521234881001e-06,
      "loss": 3.0474,
      "step": 1237
    },
    {
      "epoch": 0.4219495569188821,
      "grad_norm": 24.27652017043735,
      "learning_rate": 9.950242552572272e-06,
      "loss": 3.2635,
      "step": 1238
    },
    {
      "epoch": 0.4222903885480573,
      "grad_norm": 12.159838230143585,
      "learning_rate": 9.949963091570684e-06,
      "loss": 2.6814,
      "step": 1239
    },
    {
      "epoch": 0.4226312201772324,
      "grad_norm": 16.303265990147615,
      "learning_rate": 9.949682851920198e-06,
      "loss": 3.4281,
      "step": 1240
    },
    {
      "epoch": 0.4229720518064076,
      "grad_norm": 13.423285503936983,
      "learning_rate": 9.949401833664896e-06,
      "loss": 2.7545,
      "step": 1241
    },
    {
      "epoch": 0.4233128834355828,
      "grad_norm": 12.987657885589574,
      "learning_rate": 9.949120036848983e-06,
      "loss": 2.7399,
      "step": 1242
    },
    {
      "epoch": 0.423653715064758,
      "grad_norm": 16.19256076803677,
      "learning_rate": 9.948837461516786e-06,
      "loss": 3.1973,
      "step": 1243
    },
    {
      "epoch": 0.4239945466939332,
      "grad_norm": 19.030230950402757,
      "learning_rate": 9.948554107712756e-06,
      "loss": 2.4848,
      "step": 1244
    },
    {
      "epoch": 0.42433537832310836,
      "grad_norm": 21.02031776470229,
      "learning_rate": 9.948269975481466e-06,
      "loss": 2.8758,
      "step": 1245
    },
    {
      "epoch": 0.42467620995228356,
      "grad_norm": 17.34334703785008,
      "learning_rate": 9.947985064867612e-06,
      "loss": 3.4508,
      "step": 1246
    },
    {
      "epoch": 0.42501704158145875,
      "grad_norm": 29.087033257442627,
      "learning_rate": 9.947699375916009e-06,
      "loss": 3.3941,
      "step": 1247
    },
    {
      "epoch": 0.42535787321063395,
      "grad_norm": 20.83494120974773,
      "learning_rate": 9.9474129086716e-06,
      "loss": 3.2492,
      "step": 1248
    },
    {
      "epoch": 0.42569870483980915,
      "grad_norm": 24.470379155611795,
      "learning_rate": 9.947125663179444e-06,
      "loss": 2.8744,
      "step": 1249
    },
    {
      "epoch": 0.42603953646898435,
      "grad_norm": 21.22617400484804,
      "learning_rate": 9.94683763948473e-06,
      "loss": 3.4913,
      "step": 1250
    },
    {
      "epoch": 0.4263803680981595,
      "grad_norm": 27.574729562831525,
      "learning_rate": 9.946548837632762e-06,
      "loss": 3.7022,
      "step": 1251
    },
    {
      "epoch": 0.4267211997273347,
      "grad_norm": 16.395532414861783,
      "learning_rate": 9.94625925766897e-06,
      "loss": 3.145,
      "step": 1252
    },
    {
      "epoch": 0.4270620313565099,
      "grad_norm": 21.46362741262201,
      "learning_rate": 9.945968899638908e-06,
      "loss": 3.705,
      "step": 1253
    },
    {
      "epoch": 0.4274028629856851,
      "grad_norm": 18.92734035964414,
      "learning_rate": 9.945677763588248e-06,
      "loss": 3.6316,
      "step": 1254
    },
    {
      "epoch": 0.4277436946148603,
      "grad_norm": 12.637277191977182,
      "learning_rate": 9.94538584956279e-06,
      "loss": 2.7837,
      "step": 1255
    },
    {
      "epoch": 0.4280845262440354,
      "grad_norm": 16.256621494279322,
      "learning_rate": 9.94509315760845e-06,
      "loss": 3.1757,
      "step": 1256
    },
    {
      "epoch": 0.4284253578732106,
      "grad_norm": 18.874930993252317,
      "learning_rate": 9.944799687771272e-06,
      "loss": 3.2242,
      "step": 1257
    },
    {
      "epoch": 0.4287661895023858,
      "grad_norm": 14.882314364715942,
      "learning_rate": 9.94450544009742e-06,
      "loss": 3.342,
      "step": 1258
    },
    {
      "epoch": 0.429107021131561,
      "grad_norm": 16.883415933766166,
      "learning_rate": 9.944210414633178e-06,
      "loss": 3.0105,
      "step": 1259
    },
    {
      "epoch": 0.4294478527607362,
      "grad_norm": 20.86418630286833,
      "learning_rate": 9.943914611424956e-06,
      "loss": 3.6061,
      "step": 1260
    },
    {
      "epoch": 0.42978868438991136,
      "grad_norm": 25.392180719892604,
      "learning_rate": 9.943618030519285e-06,
      "loss": 3.7127,
      "step": 1261
    },
    {
      "epoch": 0.43012951601908656,
      "grad_norm": 24.60540995441095,
      "learning_rate": 9.943320671962818e-06,
      "loss": 3.6651,
      "step": 1262
    },
    {
      "epoch": 0.43047034764826175,
      "grad_norm": 19.45498135661641,
      "learning_rate": 9.943022535802332e-06,
      "loss": 3.1034,
      "step": 1263
    },
    {
      "epoch": 0.43081117927743695,
      "grad_norm": 17.77692078217039,
      "learning_rate": 9.942723622084722e-06,
      "loss": 3.2344,
      "step": 1264
    },
    {
      "epoch": 0.43115201090661215,
      "grad_norm": 32.42464635095363,
      "learning_rate": 9.942423930857013e-06,
      "loss": 2.9709,
      "step": 1265
    },
    {
      "epoch": 0.43149284253578735,
      "grad_norm": 15.556021872160075,
      "learning_rate": 9.942123462166344e-06,
      "loss": 3.4175,
      "step": 1266
    },
    {
      "epoch": 0.4318336741649625,
      "grad_norm": 13.955911581250279,
      "learning_rate": 9.94182221605998e-06,
      "loss": 3.042,
      "step": 1267
    },
    {
      "epoch": 0.4321745057941377,
      "grad_norm": 16.86786396289203,
      "learning_rate": 9.941520192585312e-06,
      "loss": 3.3289,
      "step": 1268
    },
    {
      "epoch": 0.4325153374233129,
      "grad_norm": 22.964851697215803,
      "learning_rate": 9.941217391789844e-06,
      "loss": 3.7592,
      "step": 1269
    },
    {
      "epoch": 0.4328561690524881,
      "grad_norm": 24.517079363772993,
      "learning_rate": 9.940913813721212e-06,
      "loss": 3.5793,
      "step": 1270
    },
    {
      "epoch": 0.4331970006816633,
      "grad_norm": 18.78949794075681,
      "learning_rate": 9.940609458427166e-06,
      "loss": 3.7287,
      "step": 1271
    },
    {
      "epoch": 0.4335378323108384,
      "grad_norm": 21.565742124473232,
      "learning_rate": 9.940304325955588e-06,
      "loss": 3.5396,
      "step": 1272
    },
    {
      "epoch": 0.4338786639400136,
      "grad_norm": 19.680444183507277,
      "learning_rate": 9.93999841635447e-06,
      "loss": 2.7035,
      "step": 1273
    },
    {
      "epoch": 0.4342194955691888,
      "grad_norm": 22.108099453826366,
      "learning_rate": 9.93969172967194e-06,
      "loss": 3.1808,
      "step": 1274
    },
    {
      "epoch": 0.434560327198364,
      "grad_norm": 25.223980953538707,
      "learning_rate": 9.939384265956234e-06,
      "loss": 3.6471,
      "step": 1275
    },
    {
      "epoch": 0.4349011588275392,
      "grad_norm": 20.273502464218655,
      "learning_rate": 9.939076025255721e-06,
      "loss": 3.1385,
      "step": 1276
    },
    {
      "epoch": 0.43524199045671436,
      "grad_norm": 16.593079925893687,
      "learning_rate": 9.938767007618889e-06,
      "loss": 3.469,
      "step": 1277
    },
    {
      "epoch": 0.43558282208588955,
      "grad_norm": 23.33648361810804,
      "learning_rate": 9.938457213094345e-06,
      "loss": 3.266,
      "step": 1278
    },
    {
      "epoch": 0.43592365371506475,
      "grad_norm": 20.581816660634118,
      "learning_rate": 9.938146641730824e-06,
      "loss": 3.5982,
      "step": 1279
    },
    {
      "epoch": 0.43626448534423995,
      "grad_norm": 13.341607087665903,
      "learning_rate": 9.937835293577178e-06,
      "loss": 2.9119,
      "step": 1280
    },
    {
      "epoch": 0.43660531697341515,
      "grad_norm": 12.713325164074323,
      "learning_rate": 9.937523168682384e-06,
      "loss": 2.9269,
      "step": 1281
    },
    {
      "epoch": 0.43694614860259035,
      "grad_norm": 15.077218369847607,
      "learning_rate": 9.93721026709554e-06,
      "loss": 3.2999,
      "step": 1282
    },
    {
      "epoch": 0.4372869802317655,
      "grad_norm": 25.765249453743216,
      "learning_rate": 9.936896588865867e-06,
      "loss": 3.2603,
      "step": 1283
    },
    {
      "epoch": 0.4376278118609407,
      "grad_norm": 15.18273195464565,
      "learning_rate": 9.93658213404271e-06,
      "loss": 3.0309,
      "step": 1284
    },
    {
      "epoch": 0.4379686434901159,
      "grad_norm": 18.45777383135803,
      "learning_rate": 9.93626690267553e-06,
      "loss": 3.263,
      "step": 1285
    },
    {
      "epoch": 0.4383094751192911,
      "grad_norm": 15.052442524061176,
      "learning_rate": 9.935950894813916e-06,
      "loss": 3.4803,
      "step": 1286
    },
    {
      "epoch": 0.4386503067484663,
      "grad_norm": 24.46358566489536,
      "learning_rate": 9.935634110507577e-06,
      "loss": 3.5294,
      "step": 1287
    },
    {
      "epoch": 0.4389911383776414,
      "grad_norm": 15.401733581329813,
      "learning_rate": 9.935316549806347e-06,
      "loss": 3.159,
      "step": 1288
    },
    {
      "epoch": 0.4393319700068166,
      "grad_norm": 15.17432376586215,
      "learning_rate": 9.934998212760177e-06,
      "loss": 3.2565,
      "step": 1289
    },
    {
      "epoch": 0.4396728016359918,
      "grad_norm": 20.23005109547779,
      "learning_rate": 9.93467909941914e-06,
      "loss": 3.2633,
      "step": 1290
    },
    {
      "epoch": 0.440013633265167,
      "grad_norm": 17.34527047308293,
      "learning_rate": 9.93435920983344e-06,
      "loss": 3.2543,
      "step": 1291
    },
    {
      "epoch": 0.4403544648943422,
      "grad_norm": 23.654481789461396,
      "learning_rate": 9.934038544053393e-06,
      "loss": 3.4398,
      "step": 1292
    },
    {
      "epoch": 0.44069529652351735,
      "grad_norm": 17.30524358878623,
      "learning_rate": 9.933717102129442e-06,
      "loss": 3.568,
      "step": 1293
    },
    {
      "epoch": 0.44103612815269255,
      "grad_norm": 40.73849045675947,
      "learning_rate": 9.933394884112149e-06,
      "loss": 3.3131,
      "step": 1294
    },
    {
      "epoch": 0.44137695978186775,
      "grad_norm": 13.48233386384235,
      "learning_rate": 9.933071890052203e-06,
      "loss": 2.8618,
      "step": 1295
    },
    {
      "epoch": 0.44171779141104295,
      "grad_norm": 19.3739197822594,
      "learning_rate": 9.93274812000041e-06,
      "loss": 3.4381,
      "step": 1296
    },
    {
      "epoch": 0.44205862304021815,
      "grad_norm": 22.505236846709582,
      "learning_rate": 9.932423574007704e-06,
      "loss": 3.5259,
      "step": 1297
    },
    {
      "epoch": 0.44239945466939334,
      "grad_norm": 18.166863152816198,
      "learning_rate": 9.932098252125134e-06,
      "loss": 3.3791,
      "step": 1298
    },
    {
      "epoch": 0.4427402862985685,
      "grad_norm": 18.67870889470471,
      "learning_rate": 9.931772154403876e-06,
      "loss": 2.9438,
      "step": 1299
    },
    {
      "epoch": 0.4430811179277437,
      "grad_norm": 19.365169792685528,
      "learning_rate": 9.931445280895225e-06,
      "loss": 3.3447,
      "step": 1300
    },
    {
      "epoch": 0.4434219495569189,
      "grad_norm": 17.479787423043756,
      "learning_rate": 9.931117631650599e-06,
      "loss": 3.0086,
      "step": 1301
    },
    {
      "epoch": 0.4437627811860941,
      "grad_norm": 26.78678862461852,
      "learning_rate": 9.930789206721543e-06,
      "loss": 3.806,
      "step": 1302
    },
    {
      "epoch": 0.4441036128152693,
      "grad_norm": 17.294408071302772,
      "learning_rate": 9.930460006159715e-06,
      "loss": 3.4311,
      "step": 1303
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 25.566965330074048,
      "learning_rate": 9.9301300300169e-06,
      "loss": 3.4183,
      "step": 1304
    },
    {
      "epoch": 0.4447852760736196,
      "grad_norm": 18.541284454599168,
      "learning_rate": 9.929799278345006e-06,
      "loss": 3.278,
      "step": 1305
    },
    {
      "epoch": 0.4451261077027948,
      "grad_norm": 23.860202309718726,
      "learning_rate": 9.929467751196062e-06,
      "loss": 3.5605,
      "step": 1306
    },
    {
      "epoch": 0.44546693933197,
      "grad_norm": 29.168442049675523,
      "learning_rate": 9.929135448622219e-06,
      "loss": 3.9469,
      "step": 1307
    },
    {
      "epoch": 0.4458077709611452,
      "grad_norm": 17.205703915947037,
      "learning_rate": 9.928802370675746e-06,
      "loss": 3.0094,
      "step": 1308
    },
    {
      "epoch": 0.4461486025903204,
      "grad_norm": 26.886089682180632,
      "learning_rate": 9.92846851740904e-06,
      "loss": 3.5325,
      "step": 1309
    },
    {
      "epoch": 0.44648943421949555,
      "grad_norm": 20.16688695287021,
      "learning_rate": 9.928133888874617e-06,
      "loss": 3.2827,
      "step": 1310
    },
    {
      "epoch": 0.44683026584867075,
      "grad_norm": 25.08825652133083,
      "learning_rate": 9.927798485125116e-06,
      "loss": 3.3124,
      "step": 1311
    },
    {
      "epoch": 0.44717109747784595,
      "grad_norm": 15.701382375370306,
      "learning_rate": 9.927462306213297e-06,
      "loss": 3.428,
      "step": 1312
    },
    {
      "epoch": 0.44751192910702114,
      "grad_norm": 35.59945615178777,
      "learning_rate": 9.927125352192043e-06,
      "loss": 3.5128,
      "step": 1313
    },
    {
      "epoch": 0.44785276073619634,
      "grad_norm": 25.47516493333518,
      "learning_rate": 9.926787623114355e-06,
      "loss": 3.9966,
      "step": 1314
    },
    {
      "epoch": 0.4481935923653715,
      "grad_norm": 20.202065410919403,
      "learning_rate": 9.926449119033362e-06,
      "loss": 2.9776,
      "step": 1315
    },
    {
      "epoch": 0.4485344239945467,
      "grad_norm": 22.287644530407693,
      "learning_rate": 9.926109840002312e-06,
      "loss": 3.188,
      "step": 1316
    },
    {
      "epoch": 0.4488752556237219,
      "grad_norm": 18.11901462327546,
      "learning_rate": 9.925769786074576e-06,
      "loss": 3.6658,
      "step": 1317
    },
    {
      "epoch": 0.4492160872528971,
      "grad_norm": 21.4696781001905,
      "learning_rate": 9.925428957303642e-06,
      "loss": 3.4313,
      "step": 1318
    },
    {
      "epoch": 0.4495569188820723,
      "grad_norm": 19.117122410707882,
      "learning_rate": 9.925087353743128e-06,
      "loss": 3.5096,
      "step": 1319
    },
    {
      "epoch": 0.4498977505112474,
      "grad_norm": 15.662249673353935,
      "learning_rate": 9.924744975446764e-06,
      "loss": 3.1163,
      "step": 1320
    },
    {
      "epoch": 0.4502385821404226,
      "grad_norm": 21.035204487644233,
      "learning_rate": 9.924401822468415e-06,
      "loss": 3.1549,
      "step": 1321
    },
    {
      "epoch": 0.4505794137695978,
      "grad_norm": 34.8105610965106,
      "learning_rate": 9.924057894862056e-06,
      "loss": 3.8211,
      "step": 1322
    },
    {
      "epoch": 0.450920245398773,
      "grad_norm": 37.73434150267029,
      "learning_rate": 9.923713192681788e-06,
      "loss": 3.4141,
      "step": 1323
    },
    {
      "epoch": 0.4512610770279482,
      "grad_norm": 17.599889165856055,
      "learning_rate": 9.923367715981833e-06,
      "loss": 3.5111,
      "step": 1324
    },
    {
      "epoch": 0.4516019086571234,
      "grad_norm": 14.264125432332277,
      "learning_rate": 9.923021464816539e-06,
      "loss": 3.2514,
      "step": 1325
    },
    {
      "epoch": 0.45194274028629855,
      "grad_norm": 15.59071385296851,
      "learning_rate": 9.922674439240371e-06,
      "loss": 3.1986,
      "step": 1326
    },
    {
      "epoch": 0.45228357191547375,
      "grad_norm": 20.320557409623483,
      "learning_rate": 9.922326639307918e-06,
      "loss": 3.5042,
      "step": 1327
    },
    {
      "epoch": 0.45262440354464895,
      "grad_norm": 22.541848392931826,
      "learning_rate": 9.92197806507389e-06,
      "loss": 3.2427,
      "step": 1328
    },
    {
      "epoch": 0.45296523517382414,
      "grad_norm": 17.28874126222633,
      "learning_rate": 9.921628716593117e-06,
      "loss": 3.4786,
      "step": 1329
    },
    {
      "epoch": 0.45330606680299934,
      "grad_norm": 19.401473749708646,
      "learning_rate": 9.921278593920559e-06,
      "loss": 3.0076,
      "step": 1330
    },
    {
      "epoch": 0.4536468984321745,
      "grad_norm": 17.413771332026283,
      "learning_rate": 9.920927697111284e-06,
      "loss": 2.9912,
      "step": 1331
    },
    {
      "epoch": 0.4539877300613497,
      "grad_norm": 25.359905424252783,
      "learning_rate": 9.920576026220495e-06,
      "loss": 2.7707,
      "step": 1332
    },
    {
      "epoch": 0.4543285616905249,
      "grad_norm": 13.150117482106511,
      "learning_rate": 9.920223581303507e-06,
      "loss": 3.0638,
      "step": 1333
    },
    {
      "epoch": 0.4546693933197001,
      "grad_norm": 24.601313577883428,
      "learning_rate": 9.919870362415766e-06,
      "loss": 3.2811,
      "step": 1334
    },
    {
      "epoch": 0.4550102249488753,
      "grad_norm": 15.748082223093167,
      "learning_rate": 9.919516369612831e-06,
      "loss": 3.1356,
      "step": 1335
    },
    {
      "epoch": 0.4553510565780504,
      "grad_norm": 16.528480071282623,
      "learning_rate": 9.91916160295039e-06,
      "loss": 3.227,
      "step": 1336
    },
    {
      "epoch": 0.4556918882072256,
      "grad_norm": 22.250393507997394,
      "learning_rate": 9.918806062484244e-06,
      "loss": 3.7354,
      "step": 1337
    },
    {
      "epoch": 0.4560327198364008,
      "grad_norm": 21.78567235559386,
      "learning_rate": 9.918449748270324e-06,
      "loss": 3.6751,
      "step": 1338
    },
    {
      "epoch": 0.456373551465576,
      "grad_norm": 30.451089582685576,
      "learning_rate": 9.918092660364679e-06,
      "loss": 3.7318,
      "step": 1339
    },
    {
      "epoch": 0.4567143830947512,
      "grad_norm": 17.949110567841906,
      "learning_rate": 9.917734798823484e-06,
      "loss": 3.3207,
      "step": 1340
    },
    {
      "epoch": 0.4570552147239264,
      "grad_norm": 16.53405692982974,
      "learning_rate": 9.917376163703026e-06,
      "loss": 3.3129,
      "step": 1341
    },
    {
      "epoch": 0.45739604635310155,
      "grad_norm": 16.807318821262047,
      "learning_rate": 9.917016755059723e-06,
      "loss": 2.7985,
      "step": 1342
    },
    {
      "epoch": 0.45773687798227675,
      "grad_norm": 22.112568998286715,
      "learning_rate": 9.91665657295011e-06,
      "loss": 3.3293,
      "step": 1343
    },
    {
      "epoch": 0.45807770961145194,
      "grad_norm": 14.48657604825247,
      "learning_rate": 9.916295617430848e-06,
      "loss": 3.4248,
      "step": 1344
    },
    {
      "epoch": 0.45841854124062714,
      "grad_norm": 67.40001023944164,
      "learning_rate": 9.915933888558713e-06,
      "loss": 3.4829,
      "step": 1345
    },
    {
      "epoch": 0.45875937286980234,
      "grad_norm": 26.792792050176892,
      "learning_rate": 9.915571386390609e-06,
      "loss": 3.1143,
      "step": 1346
    },
    {
      "epoch": 0.4591002044989775,
      "grad_norm": 28.15838079189428,
      "learning_rate": 9.915208110983559e-06,
      "loss": 3.9373,
      "step": 1347
    },
    {
      "epoch": 0.4594410361281527,
      "grad_norm": 20.727032006927367,
      "learning_rate": 9.914844062394704e-06,
      "loss": 3.2286,
      "step": 1348
    },
    {
      "epoch": 0.4597818677573279,
      "grad_norm": 15.113211676238658,
      "learning_rate": 9.914479240681316e-06,
      "loss": 3.2959,
      "step": 1349
    },
    {
      "epoch": 0.4601226993865031,
      "grad_norm": 21.705596314464017,
      "learning_rate": 9.914113645900779e-06,
      "loss": 3.7463,
      "step": 1350
    },
    {
      "epoch": 0.4604635310156783,
      "grad_norm": 21.084638991505766,
      "learning_rate": 9.913747278110603e-06,
      "loss": 3.5771,
      "step": 1351
    },
    {
      "epoch": 0.4608043626448534,
      "grad_norm": 23.957372807547685,
      "learning_rate": 9.91338013736842e-06,
      "loss": 3.727,
      "step": 1352
    },
    {
      "epoch": 0.4611451942740286,
      "grad_norm": 39.556333682076506,
      "learning_rate": 9.913012223731982e-06,
      "loss": 3.1418,
      "step": 1353
    },
    {
      "epoch": 0.4614860259032038,
      "grad_norm": 22.205602204419737,
      "learning_rate": 9.912643537259165e-06,
      "loss": 3.2436,
      "step": 1354
    },
    {
      "epoch": 0.461826857532379,
      "grad_norm": 21.36507621385849,
      "learning_rate": 9.912274078007962e-06,
      "loss": 3.1771,
      "step": 1355
    },
    {
      "epoch": 0.4621676891615542,
      "grad_norm": 20.060217181849783,
      "learning_rate": 9.911903846036493e-06,
      "loss": 3.5251,
      "step": 1356
    },
    {
      "epoch": 0.4625085207907294,
      "grad_norm": 25.521153560730564,
      "learning_rate": 9.911532841402996e-06,
      "loss": 3.7046,
      "step": 1357
    },
    {
      "epoch": 0.46284935241990455,
      "grad_norm": 31.489152081960793,
      "learning_rate": 9.91116106416583e-06,
      "loss": 3.7034,
      "step": 1358
    },
    {
      "epoch": 0.46319018404907975,
      "grad_norm": 17.914329065923788,
      "learning_rate": 9.910788514383479e-06,
      "loss": 3.1938,
      "step": 1359
    },
    {
      "epoch": 0.46353101567825494,
      "grad_norm": 21.32971450454713,
      "learning_rate": 9.910415192114547e-06,
      "loss": 3.4244,
      "step": 1360
    },
    {
      "epoch": 0.46387184730743014,
      "grad_norm": 22.42196832689618,
      "learning_rate": 9.910041097417756e-06,
      "loss": 3.7876,
      "step": 1361
    },
    {
      "epoch": 0.46421267893660534,
      "grad_norm": 26.009610911776434,
      "learning_rate": 9.909666230351955e-06,
      "loss": 3.48,
      "step": 1362
    },
    {
      "epoch": 0.4645535105657805,
      "grad_norm": 15.387337542674189,
      "learning_rate": 9.909290590976113e-06,
      "loss": 3.1548,
      "step": 1363
    },
    {
      "epoch": 0.4648943421949557,
      "grad_norm": 25.984563999303468,
      "learning_rate": 9.908914179349319e-06,
      "loss": 3.3753,
      "step": 1364
    },
    {
      "epoch": 0.4652351738241309,
      "grad_norm": 18.073034947004505,
      "learning_rate": 9.908536995530783e-06,
      "loss": 3.6673,
      "step": 1365
    },
    {
      "epoch": 0.4655760054533061,
      "grad_norm": 16.377488868717066,
      "learning_rate": 9.908159039579836e-06,
      "loss": 3.3102,
      "step": 1366
    },
    {
      "epoch": 0.46591683708248127,
      "grad_norm": 25.923084902838944,
      "learning_rate": 9.907780311555935e-06,
      "loss": 3.3404,
      "step": 1367
    },
    {
      "epoch": 0.4662576687116564,
      "grad_norm": 18.955340516919737,
      "learning_rate": 9.907400811518653e-06,
      "loss": 3.5062,
      "step": 1368
    },
    {
      "epoch": 0.4665985003408316,
      "grad_norm": 17.810659979917745,
      "learning_rate": 9.90702053952769e-06,
      "loss": 2.989,
      "step": 1369
    },
    {
      "epoch": 0.4669393319700068,
      "grad_norm": 19.09820863891688,
      "learning_rate": 9.90663949564286e-06,
      "loss": 3.4141,
      "step": 1370
    },
    {
      "epoch": 0.467280163599182,
      "grad_norm": 27.99861527734652,
      "learning_rate": 9.906257679924107e-06,
      "loss": 4.0354,
      "step": 1371
    },
    {
      "epoch": 0.4676209952283572,
      "grad_norm": 21.84215755578167,
      "learning_rate": 9.90587509243149e-06,
      "loss": 3.8505,
      "step": 1372
    },
    {
      "epoch": 0.4679618268575324,
      "grad_norm": 22.90212309683293,
      "learning_rate": 9.905491733225193e-06,
      "loss": 3.3077,
      "step": 1373
    },
    {
      "epoch": 0.46830265848670755,
      "grad_norm": 26.84186688749521,
      "learning_rate": 9.905107602365517e-06,
      "loss": 3.4043,
      "step": 1374
    },
    {
      "epoch": 0.46864349011588274,
      "grad_norm": 20.19973158812703,
      "learning_rate": 9.90472269991289e-06,
      "loss": 3.4343,
      "step": 1375
    },
    {
      "epoch": 0.46898432174505794,
      "grad_norm": 20.831814005167,
      "learning_rate": 9.904337025927857e-06,
      "loss": 3.6319,
      "step": 1376
    },
    {
      "epoch": 0.46932515337423314,
      "grad_norm": 13.616022184031408,
      "learning_rate": 9.903950580471086e-06,
      "loss": 3.4311,
      "step": 1377
    },
    {
      "epoch": 0.46966598500340834,
      "grad_norm": 11.477658281098835,
      "learning_rate": 9.903563363603368e-06,
      "loss": 3.325,
      "step": 1378
    },
    {
      "epoch": 0.4700068166325835,
      "grad_norm": 23.36313277614423,
      "learning_rate": 9.903175375385614e-06,
      "loss": 4.231,
      "step": 1379
    },
    {
      "epoch": 0.4703476482617587,
      "grad_norm": 20.87664480952871,
      "learning_rate": 9.902786615878855e-06,
      "loss": 3.9094,
      "step": 1380
    },
    {
      "epoch": 0.4706884798909339,
      "grad_norm": 18.599177192921395,
      "learning_rate": 9.902397085144244e-06,
      "loss": 3.1055,
      "step": 1381
    },
    {
      "epoch": 0.4710293115201091,
      "grad_norm": 24.296852677979636,
      "learning_rate": 9.902006783243054e-06,
      "loss": 3.104,
      "step": 1382
    },
    {
      "epoch": 0.47137014314928427,
      "grad_norm": 19.774954923500378,
      "learning_rate": 9.901615710236685e-06,
      "loss": 3.022,
      "step": 1383
    },
    {
      "epoch": 0.47171097477845947,
      "grad_norm": 28.611433073881983,
      "learning_rate": 9.901223866186653e-06,
      "loss": 3.4924,
      "step": 1384
    },
    {
      "epoch": 0.4720518064076346,
      "grad_norm": 16.775051047976614,
      "learning_rate": 9.900831251154596e-06,
      "loss": 3.1574,
      "step": 1385
    },
    {
      "epoch": 0.4723926380368098,
      "grad_norm": 28.288287060215296,
      "learning_rate": 9.900437865202275e-06,
      "loss": 3.9469,
      "step": 1386
    },
    {
      "epoch": 0.472733469665985,
      "grad_norm": 15.315391439242383,
      "learning_rate": 9.90004370839157e-06,
      "loss": 3.2715,
      "step": 1387
    },
    {
      "epoch": 0.4730743012951602,
      "grad_norm": 14.692688934693367,
      "learning_rate": 9.899648780784482e-06,
      "loss": 3.0429,
      "step": 1388
    },
    {
      "epoch": 0.4734151329243354,
      "grad_norm": 18.05068175738741,
      "learning_rate": 9.899253082443138e-06,
      "loss": 3.0802,
      "step": 1389
    },
    {
      "epoch": 0.47375596455351054,
      "grad_norm": 12.279897547837207,
      "learning_rate": 9.89885661342978e-06,
      "loss": 3.021,
      "step": 1390
    },
    {
      "epoch": 0.47409679618268574,
      "grad_norm": 16.66204913977151,
      "learning_rate": 9.898459373806776e-06,
      "loss": 3.3303,
      "step": 1391
    },
    {
      "epoch": 0.47443762781186094,
      "grad_norm": 14.044315460655591,
      "learning_rate": 9.898061363636614e-06,
      "loss": 2.9206,
      "step": 1392
    },
    {
      "epoch": 0.47477845944103614,
      "grad_norm": 13.775958690207847,
      "learning_rate": 9.8976625829819e-06,
      "loss": 3.0537,
      "step": 1393
    },
    {
      "epoch": 0.47511929107021134,
      "grad_norm": 11.751090229703895,
      "learning_rate": 9.897263031905366e-06,
      "loss": 2.9268,
      "step": 1394
    },
    {
      "epoch": 0.4754601226993865,
      "grad_norm": 12.284157146171538,
      "learning_rate": 9.896862710469862e-06,
      "loss": 3.0942,
      "step": 1395
    },
    {
      "epoch": 0.4758009543285617,
      "grad_norm": 30.458505678821325,
      "learning_rate": 9.89646161873836e-06,
      "loss": 4.0367,
      "step": 1396
    },
    {
      "epoch": 0.4761417859577369,
      "grad_norm": 23.239595257237596,
      "learning_rate": 9.896059756773954e-06,
      "loss": 3.4346,
      "step": 1397
    },
    {
      "epoch": 0.47648261758691207,
      "grad_norm": 20.03817938709607,
      "learning_rate": 9.89565712463986e-06,
      "loss": 3.2087,
      "step": 1398
    },
    {
      "epoch": 0.47682344921608727,
      "grad_norm": 21.70792339959685,
      "learning_rate": 9.89525372239941e-06,
      "loss": 3.0152,
      "step": 1399
    },
    {
      "epoch": 0.47716428084526247,
      "grad_norm": 19.245943232007658,
      "learning_rate": 9.89484955011606e-06,
      "loss": 2.8573,
      "step": 1400
    },
    {
      "epoch": 0.4775051124744376,
      "grad_norm": 13.327048925542654,
      "learning_rate": 9.894444607853394e-06,
      "loss": 2.7419,
      "step": 1401
    },
    {
      "epoch": 0.4778459441036128,
      "grad_norm": 16.194779038932634,
      "learning_rate": 9.894038895675106e-06,
      "loss": 3.0109,
      "step": 1402
    },
    {
      "epoch": 0.478186775732788,
      "grad_norm": 28.37837825908542,
      "learning_rate": 9.89363241364502e-06,
      "loss": 3.7048,
      "step": 1403
    },
    {
      "epoch": 0.4785276073619632,
      "grad_norm": 14.162264883308696,
      "learning_rate": 9.893225161827073e-06,
      "loss": 2.9237,
      "step": 1404
    },
    {
      "epoch": 0.4788684389911384,
      "grad_norm": 15.687705330633051,
      "learning_rate": 9.892817140285328e-06,
      "loss": 3.1812,
      "step": 1405
    },
    {
      "epoch": 0.47920927062031354,
      "grad_norm": 25.017223446779024,
      "learning_rate": 9.892408349083972e-06,
      "loss": 3.1272,
      "step": 1406
    },
    {
      "epoch": 0.47955010224948874,
      "grad_norm": 10.889274858598391,
      "learning_rate": 9.891998788287307e-06,
      "loss": 2.7111,
      "step": 1407
    },
    {
      "epoch": 0.47989093387866394,
      "grad_norm": 15.705354735066164,
      "learning_rate": 9.891588457959758e-06,
      "loss": 3.4863,
      "step": 1408
    },
    {
      "epoch": 0.48023176550783914,
      "grad_norm": 20.39521788141351,
      "learning_rate": 9.89117735816587e-06,
      "loss": 3.6017,
      "step": 1409
    },
    {
      "epoch": 0.48057259713701433,
      "grad_norm": 28.357659717505072,
      "learning_rate": 9.890765488970317e-06,
      "loss": 3.2948,
      "step": 1410
    },
    {
      "epoch": 0.4809134287661895,
      "grad_norm": 15.352861623318987,
      "learning_rate": 9.89035285043788e-06,
      "loss": 3.2627,
      "step": 1411
    },
    {
      "epoch": 0.4812542603953647,
      "grad_norm": 23.589323729570992,
      "learning_rate": 9.889939442633475e-06,
      "loss": 3.1794,
      "step": 1412
    },
    {
      "epoch": 0.4815950920245399,
      "grad_norm": 17.489197024291006,
      "learning_rate": 9.88952526562213e-06,
      "loss": 2.9441,
      "step": 1413
    },
    {
      "epoch": 0.48193592365371507,
      "grad_norm": 27.005934948484622,
      "learning_rate": 9.889110319468995e-06,
      "loss": 3.4281,
      "step": 1414
    },
    {
      "epoch": 0.48227675528289027,
      "grad_norm": 18.539069799377945,
      "learning_rate": 9.888694604239345e-06,
      "loss": 3.3327,
      "step": 1415
    },
    {
      "epoch": 0.48261758691206547,
      "grad_norm": 24.261681720297897,
      "learning_rate": 9.888278119998573e-06,
      "loss": 3.3178,
      "step": 1416
    },
    {
      "epoch": 0.4829584185412406,
      "grad_norm": 24.319891138396788,
      "learning_rate": 9.887860866812193e-06,
      "loss": 4.0272,
      "step": 1417
    },
    {
      "epoch": 0.4832992501704158,
      "grad_norm": 22.29386927965848,
      "learning_rate": 9.887442844745843e-06,
      "loss": 3.8549,
      "step": 1418
    },
    {
      "epoch": 0.483640081799591,
      "grad_norm": 18.95886003716071,
      "learning_rate": 9.887024053865276e-06,
      "loss": 3.5539,
      "step": 1419
    },
    {
      "epoch": 0.4839809134287662,
      "grad_norm": 23.006481879020416,
      "learning_rate": 9.886604494236371e-06,
      "loss": 3.5862,
      "step": 1420
    },
    {
      "epoch": 0.4843217450579414,
      "grad_norm": 19.959376117192033,
      "learning_rate": 9.886184165925128e-06,
      "loss": 3.3702,
      "step": 1421
    },
    {
      "epoch": 0.48466257668711654,
      "grad_norm": 21.78534514799325,
      "learning_rate": 9.885763068997664e-06,
      "loss": 3.4926,
      "step": 1422
    },
    {
      "epoch": 0.48500340831629174,
      "grad_norm": 22.862353893347308,
      "learning_rate": 9.88534120352022e-06,
      "loss": 3.1608,
      "step": 1423
    },
    {
      "epoch": 0.48534423994546694,
      "grad_norm": 21.208608006438084,
      "learning_rate": 9.884918569559157e-06,
      "loss": 3.3871,
      "step": 1424
    },
    {
      "epoch": 0.48568507157464214,
      "grad_norm": 16.006091098304996,
      "learning_rate": 9.88449516718096e-06,
      "loss": 3.2517,
      "step": 1425
    },
    {
      "epoch": 0.48602590320381733,
      "grad_norm": 18.700120989882965,
      "learning_rate": 9.884070996452226e-06,
      "loss": 3.2115,
      "step": 1426
    },
    {
      "epoch": 0.4863667348329925,
      "grad_norm": 17.811188371987623,
      "learning_rate": 9.883646057439681e-06,
      "loss": 3.5495,
      "step": 1427
    },
    {
      "epoch": 0.4867075664621677,
      "grad_norm": 20.59931611426034,
      "learning_rate": 9.883220350210173e-06,
      "loss": 3.3206,
      "step": 1428
    },
    {
      "epoch": 0.48704839809134287,
      "grad_norm": 12.741973722614254,
      "learning_rate": 9.882793874830665e-06,
      "loss": 2.5837,
      "step": 1429
    },
    {
      "epoch": 0.48738922972051807,
      "grad_norm": 47.36273251105926,
      "learning_rate": 9.882366631368243e-06,
      "loss": 3.8457,
      "step": 1430
    },
    {
      "epoch": 0.48773006134969327,
      "grad_norm": 18.268341945462026,
      "learning_rate": 9.881938619890114e-06,
      "loss": 3.4347,
      "step": 1431
    },
    {
      "epoch": 0.48807089297886846,
      "grad_norm": 12.112871820639358,
      "learning_rate": 9.881509840463605e-06,
      "loss": 2.9547,
      "step": 1432
    },
    {
      "epoch": 0.4884117246080436,
      "grad_norm": 20.771223372065126,
      "learning_rate": 9.881080293156166e-06,
      "loss": 3.7989,
      "step": 1433
    },
    {
      "epoch": 0.4887525562372188,
      "grad_norm": 14.504243169247511,
      "learning_rate": 9.880649978035368e-06,
      "loss": 3.1705,
      "step": 1434
    },
    {
      "epoch": 0.489093387866394,
      "grad_norm": 15.276379574992415,
      "learning_rate": 9.880218895168901e-06,
      "loss": 3.3698,
      "step": 1435
    },
    {
      "epoch": 0.4894342194955692,
      "grad_norm": 17.82870641841164,
      "learning_rate": 9.879787044624571e-06,
      "loss": 3.5095,
      "step": 1436
    },
    {
      "epoch": 0.4897750511247444,
      "grad_norm": 21.299159259589153,
      "learning_rate": 9.879354426470316e-06,
      "loss": 4.0162,
      "step": 1437
    },
    {
      "epoch": 0.49011588275391954,
      "grad_norm": 21.787565423265086,
      "learning_rate": 9.878921040774186e-06,
      "loss": 3.6665,
      "step": 1438
    },
    {
      "epoch": 0.49045671438309474,
      "grad_norm": 27.661342782102746,
      "learning_rate": 9.878486887604354e-06,
      "loss": 3.646,
      "step": 1439
    },
    {
      "epoch": 0.49079754601226994,
      "grad_norm": 14.464310133780947,
      "learning_rate": 9.878051967029114e-06,
      "loss": 3.6752,
      "step": 1440
    },
    {
      "epoch": 0.49113837764144513,
      "grad_norm": 30.97829951958,
      "learning_rate": 9.877616279116882e-06,
      "loss": 3.7221,
      "step": 1441
    },
    {
      "epoch": 0.49147920927062033,
      "grad_norm": 19.343023703250715,
      "learning_rate": 9.877179823936192e-06,
      "loss": 3.6507,
      "step": 1442
    },
    {
      "epoch": 0.4918200408997955,
      "grad_norm": 19.769706148319564,
      "learning_rate": 9.876742601555702e-06,
      "loss": 3.6298,
      "step": 1443
    },
    {
      "epoch": 0.49216087252897067,
      "grad_norm": 20.691737109203775,
      "learning_rate": 9.876304612044186e-06,
      "loss": 3.5661,
      "step": 1444
    },
    {
      "epoch": 0.49250170415814587,
      "grad_norm": 16.84617621272282,
      "learning_rate": 9.875865855470543e-06,
      "loss": 3.3524,
      "step": 1445
    },
    {
      "epoch": 0.49284253578732107,
      "grad_norm": 20.04025309842055,
      "learning_rate": 9.875426331903792e-06,
      "loss": 3.3857,
      "step": 1446
    },
    {
      "epoch": 0.49318336741649627,
      "grad_norm": 19.643142115625395,
      "learning_rate": 9.874986041413072e-06,
      "loss": 3.2521,
      "step": 1447
    },
    {
      "epoch": 0.49352419904567146,
      "grad_norm": 19.431213328412806,
      "learning_rate": 9.87454498406764e-06,
      "loss": 3.5349,
      "step": 1448
    },
    {
      "epoch": 0.4938650306748466,
      "grad_norm": 29.118623230362207,
      "learning_rate": 9.874103159936879e-06,
      "loss": 3.9259,
      "step": 1449
    },
    {
      "epoch": 0.4942058623040218,
      "grad_norm": 14.462070653294717,
      "learning_rate": 9.873660569090286e-06,
      "loss": 2.8797,
      "step": 1450
    },
    {
      "epoch": 0.494546693933197,
      "grad_norm": 19.506563844788015,
      "learning_rate": 9.873217211597487e-06,
      "loss": 3.4225,
      "step": 1451
    },
    {
      "epoch": 0.4948875255623722,
      "grad_norm": 22.058615911209994,
      "learning_rate": 9.87277308752822e-06,
      "loss": 3.5253,
      "step": 1452
    },
    {
      "epoch": 0.4952283571915474,
      "grad_norm": 20.084471800069092,
      "learning_rate": 9.87232819695235e-06,
      "loss": 3.6815,
      "step": 1453
    },
    {
      "epoch": 0.49556918882072254,
      "grad_norm": 21.763220959022675,
      "learning_rate": 9.871882539939858e-06,
      "loss": 3.6426,
      "step": 1454
    },
    {
      "epoch": 0.49591002044989774,
      "grad_norm": 19.387041357241856,
      "learning_rate": 9.871436116560848e-06,
      "loss": 3.3174,
      "step": 1455
    },
    {
      "epoch": 0.49625085207907293,
      "grad_norm": 18.401665987160555,
      "learning_rate": 9.870988926885547e-06,
      "loss": 3.4638,
      "step": 1456
    },
    {
      "epoch": 0.49659168370824813,
      "grad_norm": 20.669051354377007,
      "learning_rate": 9.870540970984295e-06,
      "loss": 3.5606,
      "step": 1457
    },
    {
      "epoch": 0.49693251533742333,
      "grad_norm": 10.786420616393665,
      "learning_rate": 9.87009224892756e-06,
      "loss": 2.5839,
      "step": 1458
    },
    {
      "epoch": 0.4972733469665985,
      "grad_norm": 19.38191499159506,
      "learning_rate": 9.86964276078593e-06,
      "loss": 2.9767,
      "step": 1459
    },
    {
      "epoch": 0.49761417859577367,
      "grad_norm": 24.004253224907508,
      "learning_rate": 9.869192506630108e-06,
      "loss": 3.342,
      "step": 1460
    },
    {
      "epoch": 0.49795501022494887,
      "grad_norm": 16.6465397107678,
      "learning_rate": 9.86874148653092e-06,
      "loss": 3.0304,
      "step": 1461
    },
    {
      "epoch": 0.49829584185412407,
      "grad_norm": 22.146288602806564,
      "learning_rate": 9.868289700559316e-06,
      "loss": 3.2289,
      "step": 1462
    },
    {
      "epoch": 0.49863667348329926,
      "grad_norm": 19.843855545461913,
      "learning_rate": 9.867837148786362e-06,
      "loss": 3.3373,
      "step": 1463
    },
    {
      "epoch": 0.49897750511247446,
      "grad_norm": 30.334947623252884,
      "learning_rate": 9.867383831283247e-06,
      "loss": 3.615,
      "step": 1464
    },
    {
      "epoch": 0.4993183367416496,
      "grad_norm": 14.015059432838122,
      "learning_rate": 9.86692974812128e-06,
      "loss": 2.9441,
      "step": 1465
    },
    {
      "epoch": 0.4996591683708248,
      "grad_norm": 22.641918064432012,
      "learning_rate": 9.866474899371889e-06,
      "loss": 3.0232,
      "step": 1466
    },
    {
      "epoch": 0.5,
      "grad_norm": 20.60022919078122,
      "learning_rate": 9.866019285106623e-06,
      "loss": 3.8762,
      "step": 1467
    },
    {
      "epoch": 0.5003408316291752,
      "grad_norm": 13.503331653647683,
      "learning_rate": 9.865562905397155e-06,
      "loss": 2.933,
      "step": 1468
    },
    {
      "epoch": 0.5006816632583504,
      "grad_norm": 30.069052317624163,
      "learning_rate": 9.865105760315273e-06,
      "loss": 3.1253,
      "step": 1469
    },
    {
      "epoch": 0.5010224948875256,
      "grad_norm": 18.274629341007465,
      "learning_rate": 9.864647849932885e-06,
      "loss": 3.7296,
      "step": 1470
    },
    {
      "epoch": 0.5013633265167008,
      "grad_norm": 18.943607546875764,
      "learning_rate": 9.864189174322028e-06,
      "loss": 3.1164,
      "step": 1471
    },
    {
      "epoch": 0.501704158145876,
      "grad_norm": 37.02505145895467,
      "learning_rate": 9.86372973355485e-06,
      "loss": 3.4828,
      "step": 1472
    },
    {
      "epoch": 0.5020449897750511,
      "grad_norm": 22.501411635369788,
      "learning_rate": 9.863269527703625e-06,
      "loss": 3.0979,
      "step": 1473
    },
    {
      "epoch": 0.5023858214042263,
      "grad_norm": 15.40704590719538,
      "learning_rate": 9.862808556840742e-06,
      "loss": 3.2368,
      "step": 1474
    },
    {
      "epoch": 0.5027266530334015,
      "grad_norm": 22.31105162704072,
      "learning_rate": 9.862346821038715e-06,
      "loss": 3.9659,
      "step": 1475
    },
    {
      "epoch": 0.5030674846625767,
      "grad_norm": 19.821397992069933,
      "learning_rate": 9.861884320370177e-06,
      "loss": 3.134,
      "step": 1476
    },
    {
      "epoch": 0.5034083162917519,
      "grad_norm": 39.097673616084386,
      "learning_rate": 9.861421054907882e-06,
      "loss": 3.525,
      "step": 1477
    },
    {
      "epoch": 0.5037491479209271,
      "grad_norm": 25.01720808251774,
      "learning_rate": 9.860957024724704e-06,
      "loss": 3.3522,
      "step": 1478
    },
    {
      "epoch": 0.5040899795501023,
      "grad_norm": 17.576717191222194,
      "learning_rate": 9.860492229893634e-06,
      "loss": 3.6972,
      "step": 1479
    },
    {
      "epoch": 0.5044308111792775,
      "grad_norm": 12.626978261019117,
      "learning_rate": 9.86002667048779e-06,
      "loss": 2.67,
      "step": 1480
    },
    {
      "epoch": 0.5047716428084527,
      "grad_norm": 23.247648062535642,
      "learning_rate": 9.859560346580402e-06,
      "loss": 3.2598,
      "step": 1481
    },
    {
      "epoch": 0.5051124744376279,
      "grad_norm": 18.456353109544878,
      "learning_rate": 9.859093258244826e-06,
      "loss": 3.3604,
      "step": 1482
    },
    {
      "epoch": 0.505453306066803,
      "grad_norm": 17.37281536993216,
      "learning_rate": 9.85862540555454e-06,
      "loss": 3.2297,
      "step": 1483
    },
    {
      "epoch": 0.5057941376959781,
      "grad_norm": 14.417659205408224,
      "learning_rate": 9.858156788583135e-06,
      "loss": 2.8988,
      "step": 1484
    },
    {
      "epoch": 0.5061349693251533,
      "grad_norm": 16.169807907080862,
      "learning_rate": 9.857687407404328e-06,
      "loss": 2.994,
      "step": 1485
    },
    {
      "epoch": 0.5064758009543285,
      "grad_norm": 12.561483755633231,
      "learning_rate": 9.857217262091954e-06,
      "loss": 2.802,
      "step": 1486
    },
    {
      "epoch": 0.5068166325835037,
      "grad_norm": 18.227128515839404,
      "learning_rate": 9.85674635271997e-06,
      "loss": 3.2977,
      "step": 1487
    },
    {
      "epoch": 0.5071574642126789,
      "grad_norm": 37.93226663069863,
      "learning_rate": 9.856274679362451e-06,
      "loss": 3.6507,
      "step": 1488
    },
    {
      "epoch": 0.5074982958418541,
      "grad_norm": 15.159854356352005,
      "learning_rate": 9.855802242093594e-06,
      "loss": 3.0808,
      "step": 1489
    },
    {
      "epoch": 0.5078391274710293,
      "grad_norm": 22.517404815996226,
      "learning_rate": 9.855329040987714e-06,
      "loss": 2.9795,
      "step": 1490
    },
    {
      "epoch": 0.5081799591002045,
      "grad_norm": 16.760231990468657,
      "learning_rate": 9.854855076119247e-06,
      "loss": 3.231,
      "step": 1491
    },
    {
      "epoch": 0.5085207907293797,
      "grad_norm": 17.90418969245724,
      "learning_rate": 9.854380347562752e-06,
      "loss": 3.0522,
      "step": 1492
    },
    {
      "epoch": 0.5088616223585549,
      "grad_norm": 18.20400947887226,
      "learning_rate": 9.853904855392903e-06,
      "loss": 3.7319,
      "step": 1493
    },
    {
      "epoch": 0.50920245398773,
      "grad_norm": 18.64045895352656,
      "learning_rate": 9.853428599684498e-06,
      "loss": 3.5467,
      "step": 1494
    },
    {
      "epoch": 0.5095432856169052,
      "grad_norm": 19.134270141609278,
      "learning_rate": 9.852951580512455e-06,
      "loss": 3.0835,
      "step": 1495
    },
    {
      "epoch": 0.5098841172460804,
      "grad_norm": 14.70362076084693,
      "learning_rate": 9.85247379795181e-06,
      "loss": 2.3288,
      "step": 1496
    },
    {
      "epoch": 0.5102249488752556,
      "grad_norm": 24.71588228051837,
      "learning_rate": 9.851995252077718e-06,
      "loss": 3.1787,
      "step": 1497
    },
    {
      "epoch": 0.5105657805044308,
      "grad_norm": 16.264941947327777,
      "learning_rate": 9.85151594296546e-06,
      "loss": 3.5998,
      "step": 1498
    },
    {
      "epoch": 0.510906612133606,
      "grad_norm": 21.2281478996706,
      "learning_rate": 9.85103587069043e-06,
      "loss": 3.518,
      "step": 1499
    },
    {
      "epoch": 0.5112474437627812,
      "grad_norm": 14.837927787896014,
      "learning_rate": 9.850555035328147e-06,
      "loss": 3.0262,
      "step": 1500
    },
    {
      "epoch": 0.5115882753919564,
      "grad_norm": 16.933181713093916,
      "learning_rate": 9.850073436954248e-06,
      "loss": 3.1886,
      "step": 1501
    },
    {
      "epoch": 0.5119291070211316,
      "grad_norm": 24.801652949051785,
      "learning_rate": 9.849591075644491e-06,
      "loss": 3.6613,
      "step": 1502
    },
    {
      "epoch": 0.5122699386503068,
      "grad_norm": 16.215628558427305,
      "learning_rate": 9.849107951474754e-06,
      "loss": 3.2878,
      "step": 1503
    },
    {
      "epoch": 0.512610770279482,
      "grad_norm": 21.94114164491225,
      "learning_rate": 9.84862406452103e-06,
      "loss": 3.3205,
      "step": 1504
    },
    {
      "epoch": 0.5129516019086571,
      "grad_norm": 18.429871782390133,
      "learning_rate": 9.848139414859441e-06,
      "loss": 3.5262,
      "step": 1505
    },
    {
      "epoch": 0.5132924335378323,
      "grad_norm": 13.077439367623349,
      "learning_rate": 9.847654002566223e-06,
      "loss": 3.2146,
      "step": 1506
    },
    {
      "epoch": 0.5136332651670075,
      "grad_norm": 18.6750115742223,
      "learning_rate": 9.847167827717732e-06,
      "loss": 3.3218,
      "step": 1507
    },
    {
      "epoch": 0.5139740967961827,
      "grad_norm": 14.960405424219784,
      "learning_rate": 9.846680890390448e-06,
      "loss": 3.2697,
      "step": 1508
    },
    {
      "epoch": 0.5143149284253579,
      "grad_norm": 16.557908136676847,
      "learning_rate": 9.846193190660963e-06,
      "loss": 3.6454,
      "step": 1509
    },
    {
      "epoch": 0.5146557600545331,
      "grad_norm": 39.35260863696515,
      "learning_rate": 9.845704728606002e-06,
      "loss": 2.9334,
      "step": 1510
    },
    {
      "epoch": 0.5149965916837083,
      "grad_norm": 22.924413332295792,
      "learning_rate": 9.845215504302395e-06,
      "loss": 2.9324,
      "step": 1511
    },
    {
      "epoch": 0.5153374233128835,
      "grad_norm": 17.340290195211917,
      "learning_rate": 9.844725517827103e-06,
      "loss": 3.2726,
      "step": 1512
    },
    {
      "epoch": 0.5156782549420587,
      "grad_norm": 29.529971047192003,
      "learning_rate": 9.844234769257199e-06,
      "loss": 3.7938,
      "step": 1513
    },
    {
      "epoch": 0.5160190865712339,
      "grad_norm": 14.147812617472812,
      "learning_rate": 9.843743258669884e-06,
      "loss": 3.0679,
      "step": 1514
    },
    {
      "epoch": 0.516359918200409,
      "grad_norm": 21.02012687391246,
      "learning_rate": 9.843250986142472e-06,
      "loss": 3.503,
      "step": 1515
    },
    {
      "epoch": 0.5167007498295841,
      "grad_norm": 12.28752844832706,
      "learning_rate": 9.842757951752399e-06,
      "loss": 2.9297,
      "step": 1516
    },
    {
      "epoch": 0.5170415814587593,
      "grad_norm": 17.766570490289503,
      "learning_rate": 9.842264155577224e-06,
      "loss": 3.4262,
      "step": 1517
    },
    {
      "epoch": 0.5173824130879345,
      "grad_norm": 25.15880555586446,
      "learning_rate": 9.84176959769462e-06,
      "loss": 2.9639,
      "step": 1518
    },
    {
      "epoch": 0.5177232447171097,
      "grad_norm": 25.657478030853603,
      "learning_rate": 9.841274278182385e-06,
      "loss": 3.6734,
      "step": 1519
    },
    {
      "epoch": 0.5180640763462849,
      "grad_norm": 42.24240323478798,
      "learning_rate": 9.840778197118432e-06,
      "loss": 3.4465,
      "step": 1520
    },
    {
      "epoch": 0.5184049079754601,
      "grad_norm": 18.860358432617478,
      "learning_rate": 9.8402813545808e-06,
      "loss": 3.4771,
      "step": 1521
    },
    {
      "epoch": 0.5187457396046353,
      "grad_norm": 17.424658832712826,
      "learning_rate": 9.839783750647644e-06,
      "loss": 3.3126,
      "step": 1522
    },
    {
      "epoch": 0.5190865712338105,
      "grad_norm": 14.195987245475699,
      "learning_rate": 9.839285385397236e-06,
      "loss": 2.7686,
      "step": 1523
    },
    {
      "epoch": 0.5194274028629857,
      "grad_norm": 31.809597811474323,
      "learning_rate": 9.838786258907975e-06,
      "loss": 3.1601,
      "step": 1524
    },
    {
      "epoch": 0.5197682344921609,
      "grad_norm": 16.758837510060033,
      "learning_rate": 9.838286371258373e-06,
      "loss": 3.3856,
      "step": 1525
    },
    {
      "epoch": 0.520109066121336,
      "grad_norm": 18.154773271648114,
      "learning_rate": 9.837785722527064e-06,
      "loss": 3.2396,
      "step": 1526
    },
    {
      "epoch": 0.5204498977505112,
      "grad_norm": 17.572187491039667,
      "learning_rate": 9.837284312792804e-06,
      "loss": 3.6367,
      "step": 1527
    },
    {
      "epoch": 0.5207907293796864,
      "grad_norm": 13.945193262325507,
      "learning_rate": 9.836782142134464e-06,
      "loss": 2.9765,
      "step": 1528
    },
    {
      "epoch": 0.5211315610088616,
      "grad_norm": 13.001474106279495,
      "learning_rate": 9.83627921063104e-06,
      "loss": 3.1411,
      "step": 1529
    },
    {
      "epoch": 0.5214723926380368,
      "grad_norm": 16.96862665021584,
      "learning_rate": 9.835775518361641e-06,
      "loss": 3.025,
      "step": 1530
    },
    {
      "epoch": 0.521813224267212,
      "grad_norm": 19.074402296883825,
      "learning_rate": 9.835271065405505e-06,
      "loss": 3.3979,
      "step": 1531
    },
    {
      "epoch": 0.5221540558963872,
      "grad_norm": 15.646227508659338,
      "learning_rate": 9.834765851841982e-06,
      "loss": 2.9912,
      "step": 1532
    },
    {
      "epoch": 0.5224948875255624,
      "grad_norm": 18.987463863690124,
      "learning_rate": 9.834259877750546e-06,
      "loss": 3.4337,
      "step": 1533
    },
    {
      "epoch": 0.5228357191547376,
      "grad_norm": 30.158094864341944,
      "learning_rate": 9.833753143210786e-06,
      "loss": 3.855,
      "step": 1534
    },
    {
      "epoch": 0.5231765507839128,
      "grad_norm": 18.092710190090514,
      "learning_rate": 9.833245648302413e-06,
      "loss": 3.0971,
      "step": 1535
    },
    {
      "epoch": 0.523517382413088,
      "grad_norm": 32.45604763067535,
      "learning_rate": 9.832737393105263e-06,
      "loss": 3.8787,
      "step": 1536
    },
    {
      "epoch": 0.5238582140422631,
      "grad_norm": 16.516283628170143,
      "learning_rate": 9.83222837769928e-06,
      "loss": 3.3238,
      "step": 1537
    },
    {
      "epoch": 0.5241990456714383,
      "grad_norm": 25.143736025965485,
      "learning_rate": 9.83171860216454e-06,
      "loss": 3.5386,
      "step": 1538
    },
    {
      "epoch": 0.5245398773006135,
      "grad_norm": 18.05832553524198,
      "learning_rate": 9.831208066581227e-06,
      "loss": 3.4275,
      "step": 1539
    },
    {
      "epoch": 0.5248807089297887,
      "grad_norm": 26.338051961660387,
      "learning_rate": 9.830696771029655e-06,
      "loss": 3.3948,
      "step": 1540
    },
    {
      "epoch": 0.5252215405589639,
      "grad_norm": 17.566181895848874,
      "learning_rate": 9.83018471559025e-06,
      "loss": 3.462,
      "step": 1541
    },
    {
      "epoch": 0.5255623721881391,
      "grad_norm": 21.42940509503575,
      "learning_rate": 9.829671900343562e-06,
      "loss": 3.7183,
      "step": 1542
    },
    {
      "epoch": 0.5259032038173143,
      "grad_norm": 17.171087361218188,
      "learning_rate": 9.829158325370259e-06,
      "loss": 3.0027,
      "step": 1543
    },
    {
      "epoch": 0.5262440354464895,
      "grad_norm": 17.702420294529105,
      "learning_rate": 9.828643990751125e-06,
      "loss": 3.3262,
      "step": 1544
    },
    {
      "epoch": 0.5265848670756647,
      "grad_norm": 17.030466302876533,
      "learning_rate": 9.828128896567071e-06,
      "loss": 3.294,
      "step": 1545
    },
    {
      "epoch": 0.5269256987048399,
      "grad_norm": 18.00971013034927,
      "learning_rate": 9.827613042899123e-06,
      "loss": 3.3253,
      "step": 1546
    },
    {
      "epoch": 0.527266530334015,
      "grad_norm": 24.180215833475874,
      "learning_rate": 9.827096429828424e-06,
      "loss": 4.0187,
      "step": 1547
    },
    {
      "epoch": 0.5276073619631901,
      "grad_norm": 18.102718395705804,
      "learning_rate": 9.826579057436243e-06,
      "loss": 3.4088,
      "step": 1548
    },
    {
      "epoch": 0.5279481935923653,
      "grad_norm": 15.126789210535533,
      "learning_rate": 9.826060925803962e-06,
      "loss": 3.0297,
      "step": 1549
    },
    {
      "epoch": 0.5282890252215405,
      "grad_norm": 22.121026224817925,
      "learning_rate": 9.825542035013085e-06,
      "loss": 3.3939,
      "step": 1550
    },
    {
      "epoch": 0.5286298568507157,
      "grad_norm": 16.5307827554448,
      "learning_rate": 9.825022385145238e-06,
      "loss": 3.746,
      "step": 1551
    },
    {
      "epoch": 0.5289706884798909,
      "grad_norm": 19.56978130074802,
      "learning_rate": 9.824501976282162e-06,
      "loss": 3.5714,
      "step": 1552
    },
    {
      "epoch": 0.5293115201090661,
      "grad_norm": 19.721439059861737,
      "learning_rate": 9.823980808505718e-06,
      "loss": 3.8662,
      "step": 1553
    },
    {
      "epoch": 0.5296523517382413,
      "grad_norm": 15.168238495971588,
      "learning_rate": 9.82345888189789e-06,
      "loss": 3.1397,
      "step": 1554
    },
    {
      "epoch": 0.5299931833674165,
      "grad_norm": 47.281193660631445,
      "learning_rate": 9.822936196540782e-06,
      "loss": 3.9795,
      "step": 1555
    },
    {
      "epoch": 0.5303340149965917,
      "grad_norm": 18.81058424729779,
      "learning_rate": 9.822412752516609e-06,
      "loss": 3.5447,
      "step": 1556
    },
    {
      "epoch": 0.5306748466257669,
      "grad_norm": 16.89956120198298,
      "learning_rate": 9.821888549907713e-06,
      "loss": 3.1807,
      "step": 1557
    },
    {
      "epoch": 0.5310156782549421,
      "grad_norm": 11.907844232291687,
      "learning_rate": 9.821363588796553e-06,
      "loss": 3.174,
      "step": 1558
    },
    {
      "epoch": 0.5313565098841172,
      "grad_norm": 25.06354999859159,
      "learning_rate": 9.82083786926571e-06,
      "loss": 3.3657,
      "step": 1559
    },
    {
      "epoch": 0.5316973415132924,
      "grad_norm": 14.734146377192172,
      "learning_rate": 9.820311391397877e-06,
      "loss": 2.7441,
      "step": 1560
    },
    {
      "epoch": 0.5320381731424676,
      "grad_norm": 18.222051045016897,
      "learning_rate": 9.819784155275874e-06,
      "loss": 3.8756,
      "step": 1561
    },
    {
      "epoch": 0.5323790047716428,
      "grad_norm": 27.797341406212208,
      "learning_rate": 9.819256160982637e-06,
      "loss": 4.098,
      "step": 1562
    },
    {
      "epoch": 0.532719836400818,
      "grad_norm": 15.668848026683467,
      "learning_rate": 9.818727408601223e-06,
      "loss": 3.4028,
      "step": 1563
    },
    {
      "epoch": 0.5330606680299932,
      "grad_norm": 24.36842197197313,
      "learning_rate": 9.818197898214805e-06,
      "loss": 3.6023,
      "step": 1564
    },
    {
      "epoch": 0.5334014996591684,
      "grad_norm": 13.090273545397695,
      "learning_rate": 9.817667629906677e-06,
      "loss": 3.5448,
      "step": 1565
    },
    {
      "epoch": 0.5337423312883436,
      "grad_norm": 16.03664822972654,
      "learning_rate": 9.817136603760251e-06,
      "loss": 3.1292,
      "step": 1566
    },
    {
      "epoch": 0.5340831629175188,
      "grad_norm": 25.23053729849264,
      "learning_rate": 9.816604819859064e-06,
      "loss": 3.5512,
      "step": 1567
    },
    {
      "epoch": 0.534423994546694,
      "grad_norm": 32.18112277262398,
      "learning_rate": 9.816072278286763e-06,
      "loss": 3.3344,
      "step": 1568
    },
    {
      "epoch": 0.5347648261758691,
      "grad_norm": 17.52753073065037,
      "learning_rate": 9.815538979127122e-06,
      "loss": 3.5201,
      "step": 1569
    },
    {
      "epoch": 0.5351056578050443,
      "grad_norm": 23.728957317153238,
      "learning_rate": 9.81500492246403e-06,
      "loss": 3.1171,
      "step": 1570
    },
    {
      "epoch": 0.5354464894342195,
      "grad_norm": 19.303051827633745,
      "learning_rate": 9.814470108381498e-06,
      "loss": 3.3116,
      "step": 1571
    },
    {
      "epoch": 0.5357873210633947,
      "grad_norm": 23.890154273842533,
      "learning_rate": 9.813934536963649e-06,
      "loss": 3.2899,
      "step": 1572
    },
    {
      "epoch": 0.5361281526925699,
      "grad_norm": 20.29406598367854,
      "learning_rate": 9.813398208294737e-06,
      "loss": 3.3675,
      "step": 1573
    },
    {
      "epoch": 0.5364689843217451,
      "grad_norm": 16.29425805986339,
      "learning_rate": 9.812861122459125e-06,
      "loss": 3.311,
      "step": 1574
    },
    {
      "epoch": 0.5368098159509203,
      "grad_norm": 13.901055057801692,
      "learning_rate": 9.812323279541298e-06,
      "loss": 3.2979,
      "step": 1575
    },
    {
      "epoch": 0.5371506475800955,
      "grad_norm": 13.890331908546461,
      "learning_rate": 9.811784679625865e-06,
      "loss": 3.4154,
      "step": 1576
    },
    {
      "epoch": 0.5374914792092707,
      "grad_norm": 22.712261059068037,
      "learning_rate": 9.811245322797546e-06,
      "loss": 3.3621,
      "step": 1577
    },
    {
      "epoch": 0.5378323108384458,
      "grad_norm": 29.640060640746466,
      "learning_rate": 9.810705209141184e-06,
      "loss": 3.4965,
      "step": 1578
    },
    {
      "epoch": 0.538173142467621,
      "grad_norm": 23.373623314191434,
      "learning_rate": 9.810164338741745e-06,
      "loss": 3.6503,
      "step": 1579
    },
    {
      "epoch": 0.5385139740967961,
      "grad_norm": 20.277328983627505,
      "learning_rate": 9.809622711684306e-06,
      "loss": 3.6415,
      "step": 1580
    },
    {
      "epoch": 0.5388548057259713,
      "grad_norm": 16.228906071471375,
      "learning_rate": 9.809080328054067e-06,
      "loss": 3.2832,
      "step": 1581
    },
    {
      "epoch": 0.5391956373551465,
      "grad_norm": 24.747337601596403,
      "learning_rate": 9.80853718793635e-06,
      "loss": 3.7074,
      "step": 1582
    },
    {
      "epoch": 0.5395364689843217,
      "grad_norm": 18.25615337952711,
      "learning_rate": 9.807993291416594e-06,
      "loss": 3.0295,
      "step": 1583
    },
    {
      "epoch": 0.5398773006134969,
      "grad_norm": 15.967954945786703,
      "learning_rate": 9.80744863858035e-06,
      "loss": 3.2497,
      "step": 1584
    },
    {
      "epoch": 0.5402181322426721,
      "grad_norm": 15.864065329577292,
      "learning_rate": 9.8069032295133e-06,
      "loss": 3.3424,
      "step": 1585
    },
    {
      "epoch": 0.5405589638718473,
      "grad_norm": 16.642382938159773,
      "learning_rate": 9.806357064301234e-06,
      "loss": 3.2591,
      "step": 1586
    },
    {
      "epoch": 0.5408997955010225,
      "grad_norm": 18.09584582423569,
      "learning_rate": 9.805810143030072e-06,
      "loss": 3.7017,
      "step": 1587
    },
    {
      "epoch": 0.5412406271301977,
      "grad_norm": 14.616271752845991,
      "learning_rate": 9.805262465785844e-06,
      "loss": 3.3582,
      "step": 1588
    },
    {
      "epoch": 0.5415814587593729,
      "grad_norm": 19.149403414723615,
      "learning_rate": 9.804714032654698e-06,
      "loss": 3.3646,
      "step": 1589
    },
    {
      "epoch": 0.5419222903885481,
      "grad_norm": 20.19994017303491,
      "learning_rate": 9.804164843722912e-06,
      "loss": 3.91,
      "step": 1590
    },
    {
      "epoch": 0.5422631220177232,
      "grad_norm": 15.226810264078326,
      "learning_rate": 9.80361489907687e-06,
      "loss": 3.214,
      "step": 1591
    },
    {
      "epoch": 0.5426039536468984,
      "grad_norm": 15.17206853449658,
      "learning_rate": 9.803064198803082e-06,
      "loss": 3.0032,
      "step": 1592
    },
    {
      "epoch": 0.5429447852760736,
      "grad_norm": 29.046443515653625,
      "learning_rate": 9.802512742988174e-06,
      "loss": 3.0798,
      "step": 1593
    },
    {
      "epoch": 0.5432856169052488,
      "grad_norm": 19.0569842627887,
      "learning_rate": 9.801960531718898e-06,
      "loss": 2.999,
      "step": 1594
    },
    {
      "epoch": 0.543626448534424,
      "grad_norm": 20.15938099234503,
      "learning_rate": 9.801407565082111e-06,
      "loss": 3.2121,
      "step": 1595
    },
    {
      "epoch": 0.5439672801635992,
      "grad_norm": 13.379928051203606,
      "learning_rate": 9.800853843164803e-06,
      "loss": 2.9685,
      "step": 1596
    },
    {
      "epoch": 0.5443081117927744,
      "grad_norm": 18.915001167614477,
      "learning_rate": 9.800299366054074e-06,
      "loss": 3.5305,
      "step": 1597
    },
    {
      "epoch": 0.5446489434219496,
      "grad_norm": 16.927476667317567,
      "learning_rate": 9.799744133837146e-06,
      "loss": 3.5487,
      "step": 1598
    },
    {
      "epoch": 0.5449897750511248,
      "grad_norm": 17.355929917410133,
      "learning_rate": 9.799188146601357e-06,
      "loss": 3.5321,
      "step": 1599
    },
    {
      "epoch": 0.5453306066803,
      "grad_norm": 21.346532564630852,
      "learning_rate": 9.798631404434171e-06,
      "loss": 2.7368,
      "step": 1600
    },
    {
      "epoch": 0.5456714383094751,
      "grad_norm": 19.357865155176892,
      "learning_rate": 9.79807390742316e-06,
      "loss": 3.3909,
      "step": 1601
    },
    {
      "epoch": 0.5460122699386503,
      "grad_norm": 12.361284368511987,
      "learning_rate": 9.797515655656026e-06,
      "loss": 3.1964,
      "step": 1602
    },
    {
      "epoch": 0.5463531015678255,
      "grad_norm": 38.44915539245129,
      "learning_rate": 9.79695664922058e-06,
      "loss": 3.3785,
      "step": 1603
    },
    {
      "epoch": 0.5466939331970007,
      "grad_norm": 19.400168352681344,
      "learning_rate": 9.796396888204758e-06,
      "loss": 2.1677,
      "step": 1604
    },
    {
      "epoch": 0.5470347648261759,
      "grad_norm": 14.555306366973157,
      "learning_rate": 9.795836372696611e-06,
      "loss": 3.1132,
      "step": 1605
    },
    {
      "epoch": 0.547375596455351,
      "grad_norm": 20.565165187085686,
      "learning_rate": 9.795275102784313e-06,
      "loss": 3.7085,
      "step": 1606
    },
    {
      "epoch": 0.5477164280845263,
      "grad_norm": 18.878151539880257,
      "learning_rate": 9.794713078556153e-06,
      "loss": 3.5641,
      "step": 1607
    },
    {
      "epoch": 0.5480572597137015,
      "grad_norm": 21.45197725002221,
      "learning_rate": 9.794150300100538e-06,
      "loss": 3.5332,
      "step": 1608
    },
    {
      "epoch": 0.5483980913428766,
      "grad_norm": 14.858367164159114,
      "learning_rate": 9.793586767505995e-06,
      "loss": 3.2428,
      "step": 1609
    },
    {
      "epoch": 0.5487389229720518,
      "grad_norm": 17.453274361208347,
      "learning_rate": 9.793022480861174e-06,
      "loss": 3.053,
      "step": 1610
    },
    {
      "epoch": 0.549079754601227,
      "grad_norm": 19.211320918476435,
      "learning_rate": 9.792457440254836e-06,
      "loss": 3.1129,
      "step": 1611
    },
    {
      "epoch": 0.5494205862304021,
      "grad_norm": 38.2449969216942,
      "learning_rate": 9.791891645775865e-06,
      "loss": 3.5408,
      "step": 1612
    },
    {
      "epoch": 0.5497614178595773,
      "grad_norm": 20.699184455481394,
      "learning_rate": 9.791325097513261e-06,
      "loss": 3.0889,
      "step": 1613
    },
    {
      "epoch": 0.5501022494887525,
      "grad_norm": 19.530840221111806,
      "learning_rate": 9.790757795556146e-06,
      "loss": 3.4241,
      "step": 1614
    },
    {
      "epoch": 0.5504430811179277,
      "grad_norm": 15.293559535215353,
      "learning_rate": 9.790189739993761e-06,
      "loss": 3.6441,
      "step": 1615
    },
    {
      "epoch": 0.5507839127471029,
      "grad_norm": 18.303314235913128,
      "learning_rate": 9.78962093091546e-06,
      "loss": 3.6989,
      "step": 1616
    },
    {
      "epoch": 0.5511247443762781,
      "grad_norm": 25.819985272536595,
      "learning_rate": 9.789051368410723e-06,
      "loss": 3.1663,
      "step": 1617
    },
    {
      "epoch": 0.5514655760054533,
      "grad_norm": 23.275192276313707,
      "learning_rate": 9.788481052569139e-06,
      "loss": 3.0462,
      "step": 1618
    },
    {
      "epoch": 0.5518064076346285,
      "grad_norm": 16.92218748119021,
      "learning_rate": 9.787909983480422e-06,
      "loss": 3.1385,
      "step": 1619
    },
    {
      "epoch": 0.5521472392638037,
      "grad_norm": 24.323514583929494,
      "learning_rate": 9.787338161234408e-06,
      "loss": 3.8555,
      "step": 1620
    },
    {
      "epoch": 0.5524880708929789,
      "grad_norm": 19.062689736285673,
      "learning_rate": 9.786765585921043e-06,
      "loss": 3.0739,
      "step": 1621
    },
    {
      "epoch": 0.5528289025221541,
      "grad_norm": 27.021687146844428,
      "learning_rate": 9.786192257630396e-06,
      "loss": 3.6903,
      "step": 1622
    },
    {
      "epoch": 0.5531697341513292,
      "grad_norm": 17.20370716966874,
      "learning_rate": 9.785618176452654e-06,
      "loss": 3.0834,
      "step": 1623
    },
    {
      "epoch": 0.5535105657805044,
      "grad_norm": 26.117489100936556,
      "learning_rate": 9.785043342478122e-06,
      "loss": 3.9895,
      "step": 1624
    },
    {
      "epoch": 0.5538513974096796,
      "grad_norm": 23.96701015593615,
      "learning_rate": 9.784467755797225e-06,
      "loss": 3.9757,
      "step": 1625
    },
    {
      "epoch": 0.5541922290388548,
      "grad_norm": 13.47531519438833,
      "learning_rate": 9.783891416500503e-06,
      "loss": 3.0538,
      "step": 1626
    },
    {
      "epoch": 0.55453306066803,
      "grad_norm": 23.184839114031046,
      "learning_rate": 9.783314324678618e-06,
      "loss": 3.9616,
      "step": 1627
    },
    {
      "epoch": 0.5548738922972052,
      "grad_norm": 15.896665085648067,
      "learning_rate": 9.78273648042235e-06,
      "loss": 3.0429,
      "step": 1628
    },
    {
      "epoch": 0.5552147239263804,
      "grad_norm": 16.584292409597342,
      "learning_rate": 9.782157883822593e-06,
      "loss": 3.9655,
      "step": 1629
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 15.414700296677145,
      "learning_rate": 9.781578534970366e-06,
      "loss": 3.0915,
      "step": 1630
    },
    {
      "epoch": 0.5558963871847308,
      "grad_norm": 13.927641787411831,
      "learning_rate": 9.780998433956798e-06,
      "loss": 3.2449,
      "step": 1631
    },
    {
      "epoch": 0.556237218813906,
      "grad_norm": 16.050433572300463,
      "learning_rate": 9.780417580873147e-06,
      "loss": 4.1973,
      "step": 1632
    },
    {
      "epoch": 0.5565780504430811,
      "grad_norm": 20.662314740514987,
      "learning_rate": 9.779835975810781e-06,
      "loss": 3.355,
      "step": 1633
    },
    {
      "epoch": 0.5569188820722563,
      "grad_norm": 17.368809628416585,
      "learning_rate": 9.779253618861187e-06,
      "loss": 3.2864,
      "step": 1634
    },
    {
      "epoch": 0.5572597137014315,
      "grad_norm": 16.489728292250785,
      "learning_rate": 9.778670510115975e-06,
      "loss": 3.3656,
      "step": 1635
    },
    {
      "epoch": 0.5576005453306067,
      "grad_norm": 19.610326881908954,
      "learning_rate": 9.778086649666866e-06,
      "loss": 3.6221,
      "step": 1636
    },
    {
      "epoch": 0.5579413769597819,
      "grad_norm": 17.115994637490854,
      "learning_rate": 9.777502037605711e-06,
      "loss": 3.4412,
      "step": 1637
    },
    {
      "epoch": 0.558282208588957,
      "grad_norm": 20.381716377262936,
      "learning_rate": 9.776916674024464e-06,
      "loss": 3.4456,
      "step": 1638
    },
    {
      "epoch": 0.5586230402181322,
      "grad_norm": 16.555081429676417,
      "learning_rate": 9.77633055901521e-06,
      "loss": 3.2284,
      "step": 1639
    },
    {
      "epoch": 0.5589638718473074,
      "grad_norm": 15.506779095145,
      "learning_rate": 9.775743692670145e-06,
      "loss": 2.9177,
      "step": 1640
    },
    {
      "epoch": 0.5593047034764826,
      "grad_norm": 17.833966397344103,
      "learning_rate": 9.775156075081585e-06,
      "loss": 2.8882,
      "step": 1641
    },
    {
      "epoch": 0.5596455351056578,
      "grad_norm": 25.061388202303085,
      "learning_rate": 9.774567706341968e-06,
      "loss": 3.9267,
      "step": 1642
    },
    {
      "epoch": 0.559986366734833,
      "grad_norm": 14.087858683556812,
      "learning_rate": 9.773978586543843e-06,
      "loss": 3.4599,
      "step": 1643
    },
    {
      "epoch": 0.5603271983640081,
      "grad_norm": 19.296110892844645,
      "learning_rate": 9.773388715779882e-06,
      "loss": 3.5348,
      "step": 1644
    },
    {
      "epoch": 0.5606680299931833,
      "grad_norm": 18.03541357821088,
      "learning_rate": 9.772798094142874e-06,
      "loss": 3.6253,
      "step": 1645
    },
    {
      "epoch": 0.5610088616223585,
      "grad_norm": 17.928299268217412,
      "learning_rate": 9.772206721725726e-06,
      "loss": 3.3037,
      "step": 1646
    },
    {
      "epoch": 0.5613496932515337,
      "grad_norm": 16.528704855402648,
      "learning_rate": 9.771614598621465e-06,
      "loss": 3.325,
      "step": 1647
    },
    {
      "epoch": 0.5616905248807089,
      "grad_norm": 19.788673728801015,
      "learning_rate": 9.771021724923233e-06,
      "loss": 3.4014,
      "step": 1648
    },
    {
      "epoch": 0.5620313565098841,
      "grad_norm": 19.55693390714591,
      "learning_rate": 9.77042810072429e-06,
      "loss": 3.1488,
      "step": 1649
    },
    {
      "epoch": 0.5623721881390593,
      "grad_norm": 23.384242118423504,
      "learning_rate": 9.769833726118018e-06,
      "loss": 3.3793,
      "step": 1650
    },
    {
      "epoch": 0.5627130197682345,
      "grad_norm": 22.117570414815415,
      "learning_rate": 9.769238601197913e-06,
      "loss": 3.0375,
      "step": 1651
    },
    {
      "epoch": 0.5630538513974097,
      "grad_norm": 13.507446397030291,
      "learning_rate": 9.76864272605759e-06,
      "loss": 3.101,
      "step": 1652
    },
    {
      "epoch": 0.5633946830265849,
      "grad_norm": 26.700571920440346,
      "learning_rate": 9.768046100790783e-06,
      "loss": 3.3218,
      "step": 1653
    },
    {
      "epoch": 0.5637355146557601,
      "grad_norm": 19.532855088842652,
      "learning_rate": 9.767448725491345e-06,
      "loss": 3.0882,
      "step": 1654
    },
    {
      "epoch": 0.5640763462849352,
      "grad_norm": 22.099595054955273,
      "learning_rate": 9.766850600253243e-06,
      "loss": 3.6108,
      "step": 1655
    },
    {
      "epoch": 0.5644171779141104,
      "grad_norm": 17.943772869233804,
      "learning_rate": 9.766251725170567e-06,
      "loss": 3.2113,
      "step": 1656
    },
    {
      "epoch": 0.5647580095432856,
      "grad_norm": 16.16315135674204,
      "learning_rate": 9.765652100337523e-06,
      "loss": 3.1791,
      "step": 1657
    },
    {
      "epoch": 0.5650988411724608,
      "grad_norm": 20.867805006684456,
      "learning_rate": 9.765051725848431e-06,
      "loss": 3.0327,
      "step": 1658
    },
    {
      "epoch": 0.565439672801636,
      "grad_norm": 14.641070241892788,
      "learning_rate": 9.764450601797733e-06,
      "loss": 2.8684,
      "step": 1659
    },
    {
      "epoch": 0.5657805044308112,
      "grad_norm": 31.414163938664256,
      "learning_rate": 9.76384872827999e-06,
      "loss": 3.7878,
      "step": 1660
    },
    {
      "epoch": 0.5661213360599864,
      "grad_norm": 32.29708684015511,
      "learning_rate": 9.763246105389878e-06,
      "loss": 3.6638,
      "step": 1661
    },
    {
      "epoch": 0.5664621676891616,
      "grad_norm": 19.243617720153274,
      "learning_rate": 9.762642733222192e-06,
      "loss": 3.6802,
      "step": 1662
    },
    {
      "epoch": 0.5668029993183368,
      "grad_norm": 27.12063490116175,
      "learning_rate": 9.762038611871847e-06,
      "loss": 3.0304,
      "step": 1663
    },
    {
      "epoch": 0.567143830947512,
      "grad_norm": 21.952715801497042,
      "learning_rate": 9.76143374143387e-06,
      "loss": 3.8243,
      "step": 1664
    },
    {
      "epoch": 0.5674846625766872,
      "grad_norm": 15.855859610447718,
      "learning_rate": 9.760828122003412e-06,
      "loss": 3.3323,
      "step": 1665
    },
    {
      "epoch": 0.5678254942058623,
      "grad_norm": 12.418121967709316,
      "learning_rate": 9.760221753675738e-06,
      "loss": 2.9369,
      "step": 1666
    },
    {
      "epoch": 0.5681663258350375,
      "grad_norm": 22.82029166558021,
      "learning_rate": 9.759614636546233e-06,
      "loss": 3.5121,
      "step": 1667
    },
    {
      "epoch": 0.5685071574642127,
      "grad_norm": 19.00558604869277,
      "learning_rate": 9.759006770710399e-06,
      "loss": 3.1048,
      "step": 1668
    },
    {
      "epoch": 0.5688479890933879,
      "grad_norm": 36.26538696488828,
      "learning_rate": 9.758398156263856e-06,
      "loss": 3.7706,
      "step": 1669
    },
    {
      "epoch": 0.569188820722563,
      "grad_norm": 24.521662502237394,
      "learning_rate": 9.757788793302338e-06,
      "loss": 2.6241,
      "step": 1670
    },
    {
      "epoch": 0.5695296523517382,
      "grad_norm": 20.35130945863565,
      "learning_rate": 9.757178681921706e-06,
      "loss": 3.8115,
      "step": 1671
    },
    {
      "epoch": 0.5698704839809134,
      "grad_norm": 32.99574492072187,
      "learning_rate": 9.75656782221793e-06,
      "loss": 3.2726,
      "step": 1672
    },
    {
      "epoch": 0.5702113156100886,
      "grad_norm": 17.88265634427725,
      "learning_rate": 9.755956214287098e-06,
      "loss": 3.0049,
      "step": 1673
    },
    {
      "epoch": 0.5705521472392638,
      "grad_norm": 21.091673452877927,
      "learning_rate": 9.755343858225426e-06,
      "loss": 3.6076,
      "step": 1674
    },
    {
      "epoch": 0.570892978868439,
      "grad_norm": 18.57965269474806,
      "learning_rate": 9.75473075412923e-06,
      "loss": 3.5048,
      "step": 1675
    },
    {
      "epoch": 0.5712338104976141,
      "grad_norm": 17.066930905625384,
      "learning_rate": 9.754116902094963e-06,
      "loss": 3.011,
      "step": 1676
    },
    {
      "epoch": 0.5715746421267893,
      "grad_norm": 17.37129729613841,
      "learning_rate": 9.75350230221918e-06,
      "loss": 3.4308,
      "step": 1677
    },
    {
      "epoch": 0.5719154737559645,
      "grad_norm": 20.868200149225153,
      "learning_rate": 9.752886954598564e-06,
      "loss": 3.2967,
      "step": 1678
    },
    {
      "epoch": 0.5722563053851397,
      "grad_norm": 21.98096232576628,
      "learning_rate": 9.752270859329908e-06,
      "loss": 3.4418,
      "step": 1679
    },
    {
      "epoch": 0.5725971370143149,
      "grad_norm": 32.345992688775816,
      "learning_rate": 9.75165401651013e-06,
      "loss": 3.6531,
      "step": 1680
    },
    {
      "epoch": 0.5729379686434901,
      "grad_norm": 23.80895702210845,
      "learning_rate": 9.751036426236258e-06,
      "loss": 3.4389,
      "step": 1681
    },
    {
      "epoch": 0.5732788002726653,
      "grad_norm": 21.042913556074335,
      "learning_rate": 9.750418088605445e-06,
      "loss": 3.5839,
      "step": 1682
    },
    {
      "epoch": 0.5736196319018405,
      "grad_norm": 20.073834587778606,
      "learning_rate": 9.749799003714954e-06,
      "loss": 3.4739,
      "step": 1683
    },
    {
      "epoch": 0.5739604635310157,
      "grad_norm": 20.11950536380492,
      "learning_rate": 9.749179171662174e-06,
      "loss": 2.9713,
      "step": 1684
    },
    {
      "epoch": 0.5743012951601909,
      "grad_norm": 17.153768407921994,
      "learning_rate": 9.748558592544604e-06,
      "loss": 2.7397,
      "step": 1685
    },
    {
      "epoch": 0.5746421267893661,
      "grad_norm": 15.685230551478982,
      "learning_rate": 9.747937266459867e-06,
      "loss": 3.3199,
      "step": 1686
    },
    {
      "epoch": 0.5749829584185412,
      "grad_norm": 28.347930590695363,
      "learning_rate": 9.747315193505695e-06,
      "loss": 3.1641,
      "step": 1687
    },
    {
      "epoch": 0.5753237900477164,
      "grad_norm": 16.78628184433956,
      "learning_rate": 9.746692373779945e-06,
      "loss": 3.3041,
      "step": 1688
    },
    {
      "epoch": 0.5756646216768916,
      "grad_norm": 15.638622825433908,
      "learning_rate": 9.74606880738059e-06,
      "loss": 3.5301,
      "step": 1689
    },
    {
      "epoch": 0.5760054533060668,
      "grad_norm": 16.660191092037085,
      "learning_rate": 9.745444494405718e-06,
      "loss": 3.1102,
      "step": 1690
    },
    {
      "epoch": 0.576346284935242,
      "grad_norm": 15.145293729868367,
      "learning_rate": 9.744819434953537e-06,
      "loss": 3.1792,
      "step": 1691
    },
    {
      "epoch": 0.5766871165644172,
      "grad_norm": 14.26129323429016,
      "learning_rate": 9.74419362912237e-06,
      "loss": 3.2652,
      "step": 1692
    },
    {
      "epoch": 0.5770279481935924,
      "grad_norm": 15.897537768864254,
      "learning_rate": 9.74356707701066e-06,
      "loss": 3.2117,
      "step": 1693
    },
    {
      "epoch": 0.5773687798227676,
      "grad_norm": 14.695673139894016,
      "learning_rate": 9.742939778716965e-06,
      "loss": 3.0266,
      "step": 1694
    },
    {
      "epoch": 0.5777096114519428,
      "grad_norm": 13.649979372066875,
      "learning_rate": 9.742311734339964e-06,
      "loss": 3.133,
      "step": 1695
    },
    {
      "epoch": 0.578050443081118,
      "grad_norm": 15.658604536453211,
      "learning_rate": 9.741682943978447e-06,
      "loss": 3.7167,
      "step": 1696
    },
    {
      "epoch": 0.5783912747102932,
      "grad_norm": 11.94210990787602,
      "learning_rate": 9.741053407731329e-06,
      "loss": 3.2525,
      "step": 1697
    },
    {
      "epoch": 0.5787321063394683,
      "grad_norm": 16.630800247906436,
      "learning_rate": 9.740423125697638e-06,
      "loss": 3.897,
      "step": 1698
    },
    {
      "epoch": 0.5790729379686435,
      "grad_norm": 21.933370652759564,
      "learning_rate": 9.739792097976517e-06,
      "loss": 4.1034,
      "step": 1699
    },
    {
      "epoch": 0.5794137695978187,
      "grad_norm": 16.283444476659973,
      "learning_rate": 9.739160324667231e-06,
      "loss": 3.1147,
      "step": 1700
    },
    {
      "epoch": 0.5797546012269938,
      "grad_norm": 20.222291723604275,
      "learning_rate": 9.73852780586916e-06,
      "loss": 3.6643,
      "step": 1701
    },
    {
      "epoch": 0.580095432856169,
      "grad_norm": 34.86842990658669,
      "learning_rate": 9.737894541681804e-06,
      "loss": 3.4645,
      "step": 1702
    },
    {
      "epoch": 0.5804362644853442,
      "grad_norm": 10.131899107706852,
      "learning_rate": 9.737260532204775e-06,
      "loss": 3.0308,
      "step": 1703
    },
    {
      "epoch": 0.5807770961145194,
      "grad_norm": 27.301909822686365,
      "learning_rate": 9.736625777537807e-06,
      "loss": 3.4509,
      "step": 1704
    },
    {
      "epoch": 0.5811179277436946,
      "grad_norm": 21.188859726753112,
      "learning_rate": 9.735990277780747e-06,
      "loss": 3.6515,
      "step": 1705
    },
    {
      "epoch": 0.5814587593728698,
      "grad_norm": 20.646469722038926,
      "learning_rate": 9.735354033033564e-06,
      "loss": 3.4244,
      "step": 1706
    },
    {
      "epoch": 0.581799591002045,
      "grad_norm": 19.111392591774955,
      "learning_rate": 9.734717043396342e-06,
      "loss": 3.2992,
      "step": 1707
    },
    {
      "epoch": 0.5821404226312201,
      "grad_norm": 12.818916658618486,
      "learning_rate": 9.734079308969282e-06,
      "loss": 2.795,
      "step": 1708
    },
    {
      "epoch": 0.5824812542603953,
      "grad_norm": 17.863965712204635,
      "learning_rate": 9.7334408298527e-06,
      "loss": 3.0267,
      "step": 1709
    },
    {
      "epoch": 0.5828220858895705,
      "grad_norm": 18.371143372987238,
      "learning_rate": 9.732801606147033e-06,
      "loss": 3.3958,
      "step": 1710
    },
    {
      "epoch": 0.5831629175187457,
      "grad_norm": 25.73759906154846,
      "learning_rate": 9.732161637952834e-06,
      "loss": 3.5781,
      "step": 1711
    },
    {
      "epoch": 0.5835037491479209,
      "grad_norm": 16.656303230555313,
      "learning_rate": 9.73152092537077e-06,
      "loss": 3.1206,
      "step": 1712
    },
    {
      "epoch": 0.5838445807770961,
      "grad_norm": 16.052805562940392,
      "learning_rate": 9.73087946850163e-06,
      "loss": 3.0211,
      "step": 1713
    },
    {
      "epoch": 0.5841854124062713,
      "grad_norm": 19.82719606798507,
      "learning_rate": 9.730237267446318e-06,
      "loss": 3.5862,
      "step": 1714
    },
    {
      "epoch": 0.5845262440354465,
      "grad_norm": 15.362003875352956,
      "learning_rate": 9.729594322305853e-06,
      "loss": 3.4804,
      "step": 1715
    },
    {
      "epoch": 0.5848670756646217,
      "grad_norm": 19.706397312138662,
      "learning_rate": 9.728950633181375e-06,
      "loss": 3.9479,
      "step": 1716
    },
    {
      "epoch": 0.5852079072937969,
      "grad_norm": 29.31642775584269,
      "learning_rate": 9.728306200174136e-06,
      "loss": 3.3872,
      "step": 1717
    },
    {
      "epoch": 0.5855487389229721,
      "grad_norm": 19.672143710946138,
      "learning_rate": 9.72766102338551e-06,
      "loss": 3.1713,
      "step": 1718
    },
    {
      "epoch": 0.5858895705521472,
      "grad_norm": 14.978996288300168,
      "learning_rate": 9.727015102916986e-06,
      "loss": 3.314,
      "step": 1719
    },
    {
      "epoch": 0.5862304021813224,
      "grad_norm": 12.535090190387233,
      "learning_rate": 9.726368438870169e-06,
      "loss": 2.9904,
      "step": 1720
    },
    {
      "epoch": 0.5865712338104976,
      "grad_norm": 29.853944149723763,
      "learning_rate": 9.72572103134678e-06,
      "loss": 3.5488,
      "step": 1721
    },
    {
      "epoch": 0.5869120654396728,
      "grad_norm": 12.568757566148179,
      "learning_rate": 9.725072880448663e-06,
      "loss": 3.2826,
      "step": 1722
    },
    {
      "epoch": 0.587252897068848,
      "grad_norm": 19.895205566469492,
      "learning_rate": 9.724423986277772e-06,
      "loss": 3.7687,
      "step": 1723
    },
    {
      "epoch": 0.5875937286980232,
      "grad_norm": 19.463887104307048,
      "learning_rate": 9.72377434893618e-06,
      "loss": 3.1531,
      "step": 1724
    },
    {
      "epoch": 0.5879345603271984,
      "grad_norm": 16.69781491599921,
      "learning_rate": 9.72312396852608e-06,
      "loss": 2.9758,
      "step": 1725
    },
    {
      "epoch": 0.5882753919563736,
      "grad_norm": 19.317821633422565,
      "learning_rate": 9.722472845149777e-06,
      "loss": 3.4119,
      "step": 1726
    },
    {
      "epoch": 0.5886162235855488,
      "grad_norm": 24.64048825929287,
      "learning_rate": 9.721820978909698e-06,
      "loss": 4.0101,
      "step": 1727
    },
    {
      "epoch": 0.588957055214724,
      "grad_norm": 13.085506404682956,
      "learning_rate": 9.72116836990838e-06,
      "loss": 2.7017,
      "step": 1728
    },
    {
      "epoch": 0.5892978868438992,
      "grad_norm": 15.047291198385711,
      "learning_rate": 9.720515018248483e-06,
      "loss": 3.4557,
      "step": 1729
    },
    {
      "epoch": 0.5896387184730743,
      "grad_norm": 17.695315801435573,
      "learning_rate": 9.719860924032786e-06,
      "loss": 3.3825,
      "step": 1730
    },
    {
      "epoch": 0.5899795501022495,
      "grad_norm": 33.46290133391076,
      "learning_rate": 9.719206087364174e-06,
      "loss": 3.4395,
      "step": 1731
    },
    {
      "epoch": 0.5903203817314246,
      "grad_norm": 18.58268265990649,
      "learning_rate": 9.718550508345661e-06,
      "loss": 3.1963,
      "step": 1732
    },
    {
      "epoch": 0.5906612133605998,
      "grad_norm": 10.030358439421084,
      "learning_rate": 9.717894187080368e-06,
      "loss": 3.3816,
      "step": 1733
    },
    {
      "epoch": 0.591002044989775,
      "grad_norm": 14.628639121194412,
      "learning_rate": 9.717237123671539e-06,
      "loss": 3.1266,
      "step": 1734
    },
    {
      "epoch": 0.5913428766189502,
      "grad_norm": 15.485227376649549,
      "learning_rate": 9.716579318222532e-06,
      "loss": 3.4757,
      "step": 1735
    },
    {
      "epoch": 0.5916837082481254,
      "grad_norm": 23.133060374852793,
      "learning_rate": 9.715920770836822e-06,
      "loss": 2.8721,
      "step": 1736
    },
    {
      "epoch": 0.5920245398773006,
      "grad_norm": 12.010613608909534,
      "learning_rate": 9.715261481618004e-06,
      "loss": 3.0459,
      "step": 1737
    },
    {
      "epoch": 0.5923653715064758,
      "grad_norm": 15.108066919578581,
      "learning_rate": 9.714601450669781e-06,
      "loss": 2.9643,
      "step": 1738
    },
    {
      "epoch": 0.592706203135651,
      "grad_norm": 21.970325386998258,
      "learning_rate": 9.713940678095985e-06,
      "loss": 3.2193,
      "step": 1739
    },
    {
      "epoch": 0.5930470347648262,
      "grad_norm": 22.137009450507044,
      "learning_rate": 9.713279164000552e-06,
      "loss": 3.9195,
      "step": 1740
    },
    {
      "epoch": 0.5933878663940013,
      "grad_norm": 18.48329499315672,
      "learning_rate": 9.712616908487549e-06,
      "loss": 3.3258,
      "step": 1741
    },
    {
      "epoch": 0.5937286980231765,
      "grad_norm": 15.285310420037039,
      "learning_rate": 9.711953911661142e-06,
      "loss": 3.0607,
      "step": 1742
    },
    {
      "epoch": 0.5940695296523517,
      "grad_norm": 22.484314381025495,
      "learning_rate": 9.711290173625629e-06,
      "loss": 3.1433,
      "step": 1743
    },
    {
      "epoch": 0.5944103612815269,
      "grad_norm": 16.560494292257584,
      "learning_rate": 9.710625694485419e-06,
      "loss": 3.1543,
      "step": 1744
    },
    {
      "epoch": 0.5947511929107021,
      "grad_norm": 21.143120713135982,
      "learning_rate": 9.709960474345033e-06,
      "loss": 2.9393,
      "step": 1745
    },
    {
      "epoch": 0.5950920245398773,
      "grad_norm": 27.503473944805915,
      "learning_rate": 9.709294513309116e-06,
      "loss": 3.3063,
      "step": 1746
    },
    {
      "epoch": 0.5954328561690525,
      "grad_norm": 17.087838545492417,
      "learning_rate": 9.708627811482425e-06,
      "loss": 3.3532,
      "step": 1747
    },
    {
      "epoch": 0.5957736877982277,
      "grad_norm": 16.011383297670417,
      "learning_rate": 9.707960368969835e-06,
      "loss": 3.3669,
      "step": 1748
    },
    {
      "epoch": 0.5961145194274029,
      "grad_norm": 15.585304441511918,
      "learning_rate": 9.707292185876338e-06,
      "loss": 3.5645,
      "step": 1749
    },
    {
      "epoch": 0.5964553510565781,
      "grad_norm": 14.90797392911369,
      "learning_rate": 9.706623262307043e-06,
      "loss": 3.2119,
      "step": 1750
    },
    {
      "epoch": 0.5967961826857532,
      "grad_norm": 34.46728319386735,
      "learning_rate": 9.70595359836717e-06,
      "loss": 3.8667,
      "step": 1751
    },
    {
      "epoch": 0.5971370143149284,
      "grad_norm": 17.926165341434796,
      "learning_rate": 9.705283194162064e-06,
      "loss": 2.8577,
      "step": 1752
    },
    {
      "epoch": 0.5974778459441036,
      "grad_norm": 17.72102081500821,
      "learning_rate": 9.704612049797182e-06,
      "loss": 3.2888,
      "step": 1753
    },
    {
      "epoch": 0.5978186775732788,
      "grad_norm": 19.131007534154147,
      "learning_rate": 9.703940165378095e-06,
      "loss": 3.2549,
      "step": 1754
    },
    {
      "epoch": 0.598159509202454,
      "grad_norm": 15.403960353858936,
      "learning_rate": 9.703267541010495e-06,
      "loss": 3.6266,
      "step": 1755
    },
    {
      "epoch": 0.5985003408316292,
      "grad_norm": 18.141761538154697,
      "learning_rate": 9.702594176800189e-06,
      "loss": 3.4513,
      "step": 1756
    },
    {
      "epoch": 0.5988411724608044,
      "grad_norm": 27.417494811955905,
      "learning_rate": 9.7019200728531e-06,
      "loss": 4.0869,
      "step": 1757
    },
    {
      "epoch": 0.5991820040899796,
      "grad_norm": 15.733982780764096,
      "learning_rate": 9.701245229275264e-06,
      "loss": 3.214,
      "step": 1758
    },
    {
      "epoch": 0.5995228357191548,
      "grad_norm": 40.084443808419486,
      "learning_rate": 9.70056964617284e-06,
      "loss": 3.089,
      "step": 1759
    },
    {
      "epoch": 0.59986366734833,
      "grad_norm": 16.707991751760506,
      "learning_rate": 9.699893323652098e-06,
      "loss": 3.6205,
      "step": 1760
    },
    {
      "epoch": 0.6002044989775052,
      "grad_norm": 17.560822308231135,
      "learning_rate": 9.699216261819426e-06,
      "loss": 3.1029,
      "step": 1761
    },
    {
      "epoch": 0.6005453306066802,
      "grad_norm": 16.31764355915314,
      "learning_rate": 9.698538460781333e-06,
      "loss": 3.5345,
      "step": 1762
    },
    {
      "epoch": 0.6008861622358554,
      "grad_norm": 18.47959433778221,
      "learning_rate": 9.697859920644432e-06,
      "loss": 4.0105,
      "step": 1763
    },
    {
      "epoch": 0.6012269938650306,
      "grad_norm": 19.528025694746837,
      "learning_rate": 9.697180641515467e-06,
      "loss": 3.3635,
      "step": 1764
    },
    {
      "epoch": 0.6015678254942058,
      "grad_norm": 26.649094177420068,
      "learning_rate": 9.696500623501289e-06,
      "loss": 4.0405,
      "step": 1765
    },
    {
      "epoch": 0.601908657123381,
      "grad_norm": 27.399155621177428,
      "learning_rate": 9.695819866708867e-06,
      "loss": 4.0702,
      "step": 1766
    },
    {
      "epoch": 0.6022494887525562,
      "grad_norm": 14.324142637352084,
      "learning_rate": 9.695138371245286e-06,
      "loss": 3.1551,
      "step": 1767
    },
    {
      "epoch": 0.6025903203817314,
      "grad_norm": 18.0018605933936,
      "learning_rate": 9.69445613721775e-06,
      "loss": 2.9436,
      "step": 1768
    },
    {
      "epoch": 0.6029311520109066,
      "grad_norm": 18.371189287082153,
      "learning_rate": 9.693773164733576e-06,
      "loss": 3.62,
      "step": 1769
    },
    {
      "epoch": 0.6032719836400818,
      "grad_norm": 20.075763829815855,
      "learning_rate": 9.693089453900198e-06,
      "loss": 3.9432,
      "step": 1770
    },
    {
      "epoch": 0.603612815269257,
      "grad_norm": 12.512502426447503,
      "learning_rate": 9.692405004825166e-06,
      "loss": 3.2894,
      "step": 1771
    },
    {
      "epoch": 0.6039536468984322,
      "grad_norm": 29.790802309400387,
      "learning_rate": 9.691719817616148e-06,
      "loss": 3.9221,
      "step": 1772
    },
    {
      "epoch": 0.6042944785276073,
      "grad_norm": 11.676445592325315,
      "learning_rate": 9.691033892380927e-06,
      "loss": 3.0217,
      "step": 1773
    },
    {
      "epoch": 0.6046353101567825,
      "grad_norm": 19.58354859574178,
      "learning_rate": 9.6903472292274e-06,
      "loss": 3.1649,
      "step": 1774
    },
    {
      "epoch": 0.6049761417859577,
      "grad_norm": 19.214764802630345,
      "learning_rate": 9.689659828263582e-06,
      "loss": 3.6177,
      "step": 1775
    },
    {
      "epoch": 0.6053169734151329,
      "grad_norm": 33.215468784956435,
      "learning_rate": 9.688971689597605e-06,
      "loss": 2.9406,
      "step": 1776
    },
    {
      "epoch": 0.6056578050443081,
      "grad_norm": 34.4364586749498,
      "learning_rate": 9.688282813337717e-06,
      "loss": 3.0985,
      "step": 1777
    },
    {
      "epoch": 0.6059986366734833,
      "grad_norm": 14.018101748450775,
      "learning_rate": 9.687593199592278e-06,
      "loss": 3.2192,
      "step": 1778
    },
    {
      "epoch": 0.6063394683026585,
      "grad_norm": 18.332921175345987,
      "learning_rate": 9.686902848469768e-06,
      "loss": 3.1068,
      "step": 1779
    },
    {
      "epoch": 0.6066802999318337,
      "grad_norm": 15.899824386929613,
      "learning_rate": 9.686211760078783e-06,
      "loss": 2.9783,
      "step": 1780
    },
    {
      "epoch": 0.6070211315610089,
      "grad_norm": 14.28342884102595,
      "learning_rate": 9.685519934528033e-06,
      "loss": 3.387,
      "step": 1781
    },
    {
      "epoch": 0.6073619631901841,
      "grad_norm": 12.998057687259088,
      "learning_rate": 9.684827371926346e-06,
      "loss": 3.1405,
      "step": 1782
    },
    {
      "epoch": 0.6077027948193592,
      "grad_norm": 14.952210105424163,
      "learning_rate": 9.684134072382664e-06,
      "loss": 3.1433,
      "step": 1783
    },
    {
      "epoch": 0.6080436264485344,
      "grad_norm": 18.73245511395826,
      "learning_rate": 9.683440036006046e-06,
      "loss": 3.1966,
      "step": 1784
    },
    {
      "epoch": 0.6083844580777096,
      "grad_norm": 23.69203177551261,
      "learning_rate": 9.682745262905667e-06,
      "loss": 4.037,
      "step": 1785
    },
    {
      "epoch": 0.6087252897068848,
      "grad_norm": 56.65339131637852,
      "learning_rate": 9.682049753190816e-06,
      "loss": 3.2339,
      "step": 1786
    },
    {
      "epoch": 0.60906612133606,
      "grad_norm": 19.33934175243023,
      "learning_rate": 9.681353506970901e-06,
      "loss": 3.2614,
      "step": 1787
    },
    {
      "epoch": 0.6094069529652352,
      "grad_norm": 23.18493226215609,
      "learning_rate": 9.680656524355444e-06,
      "loss": 2.9967,
      "step": 1788
    },
    {
      "epoch": 0.6097477845944104,
      "grad_norm": 34.775263290492724,
      "learning_rate": 9.679958805454084e-06,
      "loss": 3.2045,
      "step": 1789
    },
    {
      "epoch": 0.6100886162235856,
      "grad_norm": 16.414338872007665,
      "learning_rate": 9.679260350376574e-06,
      "loss": 3.4499,
      "step": 1790
    },
    {
      "epoch": 0.6104294478527608,
      "grad_norm": 17.019092895899327,
      "learning_rate": 9.678561159232784e-06,
      "loss": 4.0108,
      "step": 1791
    },
    {
      "epoch": 0.610770279481936,
      "grad_norm": 13.156088882837388,
      "learning_rate": 9.677861232132699e-06,
      "loss": 2.7811,
      "step": 1792
    },
    {
      "epoch": 0.6111111111111112,
      "grad_norm": 16.883566441252388,
      "learning_rate": 9.67716056918642e-06,
      "loss": 3.5357,
      "step": 1793
    },
    {
      "epoch": 0.6114519427402862,
      "grad_norm": 17.174598067804375,
      "learning_rate": 9.676459170504165e-06,
      "loss": 3.7759,
      "step": 1794
    },
    {
      "epoch": 0.6117927743694614,
      "grad_norm": 13.273156481569139,
      "learning_rate": 9.675757036196268e-06,
      "loss": 3.397,
      "step": 1795
    },
    {
      "epoch": 0.6121336059986366,
      "grad_norm": 13.898923199502358,
      "learning_rate": 9.675054166373174e-06,
      "loss": 3.2752,
      "step": 1796
    },
    {
      "epoch": 0.6124744376278118,
      "grad_norm": 18.641911625176398,
      "learning_rate": 9.67435056114545e-06,
      "loss": 3.0756,
      "step": 1797
    },
    {
      "epoch": 0.612815269256987,
      "grad_norm": 31.34538033378389,
      "learning_rate": 9.673646220623776e-06,
      "loss": 3.3097,
      "step": 1798
    },
    {
      "epoch": 0.6131561008861622,
      "grad_norm": 20.537599329393217,
      "learning_rate": 9.672941144918946e-06,
      "loss": 3.3971,
      "step": 1799
    },
    {
      "epoch": 0.6134969325153374,
      "grad_norm": 15.152559174975213,
      "learning_rate": 9.672235334141873e-06,
      "loss": 3.3698,
      "step": 1800
    },
    {
      "epoch": 0.6138377641445126,
      "grad_norm": 14.108301168258606,
      "learning_rate": 9.671528788403582e-06,
      "loss": 2.8024,
      "step": 1801
    },
    {
      "epoch": 0.6141785957736878,
      "grad_norm": 14.314416365442304,
      "learning_rate": 9.670821507815215e-06,
      "loss": 2.8505,
      "step": 1802
    },
    {
      "epoch": 0.614519427402863,
      "grad_norm": 13.746173824030882,
      "learning_rate": 9.670113492488033e-06,
      "loss": 2.7899,
      "step": 1803
    },
    {
      "epoch": 0.6148602590320382,
      "grad_norm": 19.33679944487343,
      "learning_rate": 9.669404742533409e-06,
      "loss": 3.1968,
      "step": 1804
    },
    {
      "epoch": 0.6152010906612133,
      "grad_norm": 17.55018859703818,
      "learning_rate": 9.668695258062828e-06,
      "loss": 3.4387,
      "step": 1805
    },
    {
      "epoch": 0.6155419222903885,
      "grad_norm": 16.886446110418067,
      "learning_rate": 9.667985039187902e-06,
      "loss": 2.6743,
      "step": 1806
    },
    {
      "epoch": 0.6158827539195637,
      "grad_norm": 27.700302233006546,
      "learning_rate": 9.667274086020343e-06,
      "loss": 3.5538,
      "step": 1807
    },
    {
      "epoch": 0.6162235855487389,
      "grad_norm": 30.244565436605427,
      "learning_rate": 9.666562398671995e-06,
      "loss": 3.2588,
      "step": 1808
    },
    {
      "epoch": 0.6165644171779141,
      "grad_norm": 20.26335838019999,
      "learning_rate": 9.665849977254802e-06,
      "loss": 3.6427,
      "step": 1809
    },
    {
      "epoch": 0.6169052488070893,
      "grad_norm": 27.77334602516442,
      "learning_rate": 9.665136821880834e-06,
      "loss": 3.2375,
      "step": 1810
    },
    {
      "epoch": 0.6172460804362645,
      "grad_norm": 16.44531524760947,
      "learning_rate": 9.664422932662275e-06,
      "loss": 2.9253,
      "step": 1811
    },
    {
      "epoch": 0.6175869120654397,
      "grad_norm": 21.80302981711445,
      "learning_rate": 9.66370830971142e-06,
      "loss": 3.4555,
      "step": 1812
    },
    {
      "epoch": 0.6179277436946149,
      "grad_norm": 15.616377422993686,
      "learning_rate": 9.662992953140682e-06,
      "loss": 3.3133,
      "step": 1813
    },
    {
      "epoch": 0.6182685753237901,
      "grad_norm": 20.962309783746115,
      "learning_rate": 9.662276863062592e-06,
      "loss": 2.797,
      "step": 1814
    },
    {
      "epoch": 0.6186094069529653,
      "grad_norm": 15.978376943741383,
      "learning_rate": 9.66156003958979e-06,
      "loss": 3.4829,
      "step": 1815
    },
    {
      "epoch": 0.6189502385821404,
      "grad_norm": 16.717870616965534,
      "learning_rate": 9.660842482835039e-06,
      "loss": 3.1759,
      "step": 1816
    },
    {
      "epoch": 0.6192910702113156,
      "grad_norm": 17.18510611054113,
      "learning_rate": 9.660124192911208e-06,
      "loss": 3.5615,
      "step": 1817
    },
    {
      "epoch": 0.6196319018404908,
      "grad_norm": 20.349384926759715,
      "learning_rate": 9.659405169931296e-06,
      "loss": 3.9533,
      "step": 1818
    },
    {
      "epoch": 0.619972733469666,
      "grad_norm": 19.748165624345436,
      "learning_rate": 9.658685414008397e-06,
      "loss": 3.469,
      "step": 1819
    },
    {
      "epoch": 0.6203135650988412,
      "grad_norm": 15.27514433153184,
      "learning_rate": 9.65796492525574e-06,
      "loss": 3.3842,
      "step": 1820
    },
    {
      "epoch": 0.6206543967280164,
      "grad_norm": 12.577268722615022,
      "learning_rate": 9.65724370378666e-06,
      "loss": 3.2128,
      "step": 1821
    },
    {
      "epoch": 0.6209952283571916,
      "grad_norm": 16.347382691662606,
      "learning_rate": 9.656521749714603e-06,
      "loss": 3.3777,
      "step": 1822
    },
    {
      "epoch": 0.6213360599863668,
      "grad_norm": 17.84220432191084,
      "learning_rate": 9.655799063153139e-06,
      "loss": 3.0493,
      "step": 1823
    },
    {
      "epoch": 0.621676891615542,
      "grad_norm": 30.497852188661568,
      "learning_rate": 9.65507564421595e-06,
      "loss": 3.4249,
      "step": 1824
    },
    {
      "epoch": 0.6220177232447172,
      "grad_norm": 16.67352836233098,
      "learning_rate": 9.654351493016831e-06,
      "loss": 3.2943,
      "step": 1825
    },
    {
      "epoch": 0.6223585548738922,
      "grad_norm": 14.207902036529346,
      "learning_rate": 9.653626609669694e-06,
      "loss": 3.4879,
      "step": 1826
    },
    {
      "epoch": 0.6226993865030674,
      "grad_norm": 25.47878905480449,
      "learning_rate": 9.652900994288565e-06,
      "loss": 3.5075,
      "step": 1827
    },
    {
      "epoch": 0.6230402181322426,
      "grad_norm": 22.910564255233158,
      "learning_rate": 9.652174646987588e-06,
      "loss": 3.2844,
      "step": 1828
    },
    {
      "epoch": 0.6233810497614178,
      "grad_norm": 12.849090796188992,
      "learning_rate": 9.651447567881021e-06,
      "loss": 3.1157,
      "step": 1829
    },
    {
      "epoch": 0.623721881390593,
      "grad_norm": 12.735103601478624,
      "learning_rate": 9.650719757083233e-06,
      "loss": 3.1012,
      "step": 1830
    },
    {
      "epoch": 0.6240627130197682,
      "grad_norm": 15.699814846927211,
      "learning_rate": 9.649991214708716e-06,
      "loss": 2.9129,
      "step": 1831
    },
    {
      "epoch": 0.6244035446489434,
      "grad_norm": 36.060754340209265,
      "learning_rate": 9.64926194087207e-06,
      "loss": 2.7859,
      "step": 1832
    },
    {
      "epoch": 0.6247443762781186,
      "grad_norm": 16.53714020497567,
      "learning_rate": 9.648531935688012e-06,
      "loss": 3.119,
      "step": 1833
    },
    {
      "epoch": 0.6250852079072938,
      "grad_norm": 39.18307371935072,
      "learning_rate": 9.647801199271376e-06,
      "loss": 2.9805,
      "step": 1834
    },
    {
      "epoch": 0.625426039536469,
      "grad_norm": 16.97595558283937,
      "learning_rate": 9.647069731737111e-06,
      "loss": 3.8182,
      "step": 1835
    },
    {
      "epoch": 0.6257668711656442,
      "grad_norm": 9.860569606805274,
      "learning_rate": 9.646337533200277e-06,
      "loss": 2.5504,
      "step": 1836
    },
    {
      "epoch": 0.6261077027948193,
      "grad_norm": 15.685724181421113,
      "learning_rate": 9.645604603776053e-06,
      "loss": 3.2027,
      "step": 1837
    },
    {
      "epoch": 0.6264485344239945,
      "grad_norm": 17.138342701064246,
      "learning_rate": 9.644870943579732e-06,
      "loss": 3.3941,
      "step": 1838
    },
    {
      "epoch": 0.6267893660531697,
      "grad_norm": 24.120690620969757,
      "learning_rate": 9.644136552726721e-06,
      "loss": 3.4363,
      "step": 1839
    },
    {
      "epoch": 0.6271301976823449,
      "grad_norm": 12.887324635803548,
      "learning_rate": 9.643401431332542e-06,
      "loss": 3.5322,
      "step": 1840
    },
    {
      "epoch": 0.6274710293115201,
      "grad_norm": 21.07281613749816,
      "learning_rate": 9.642665579512835e-06,
      "loss": 3.3731,
      "step": 1841
    },
    {
      "epoch": 0.6278118609406953,
      "grad_norm": 26.417153229617224,
      "learning_rate": 9.641928997383352e-06,
      "loss": 3.7541,
      "step": 1842
    },
    {
      "epoch": 0.6281526925698705,
      "grad_norm": 23.53196803364887,
      "learning_rate": 9.641191685059958e-06,
      "loss": 2.9649,
      "step": 1843
    },
    {
      "epoch": 0.6284935241990457,
      "grad_norm": 31.738404525989697,
      "learning_rate": 9.640453642658637e-06,
      "loss": 3.3387,
      "step": 1844
    },
    {
      "epoch": 0.6288343558282209,
      "grad_norm": 22.51145971746865,
      "learning_rate": 9.639714870295484e-06,
      "loss": 3.61,
      "step": 1845
    },
    {
      "epoch": 0.6291751874573961,
      "grad_norm": 15.879174328582993,
      "learning_rate": 9.638975368086714e-06,
      "loss": 3.0667,
      "step": 1846
    },
    {
      "epoch": 0.6295160190865713,
      "grad_norm": 20.43142398231541,
      "learning_rate": 9.638235136148652e-06,
      "loss": 3.6289,
      "step": 1847
    },
    {
      "epoch": 0.6298568507157464,
      "grad_norm": 19.285960679220977,
      "learning_rate": 9.637494174597738e-06,
      "loss": 3.3616,
      "step": 1848
    },
    {
      "epoch": 0.6301976823449216,
      "grad_norm": 18.051598506379808,
      "learning_rate": 9.63675248355053e-06,
      "loss": 3.3051,
      "step": 1849
    },
    {
      "epoch": 0.6305385139740968,
      "grad_norm": 15.676887315075097,
      "learning_rate": 9.636010063123698e-06,
      "loss": 2.9841,
      "step": 1850
    },
    {
      "epoch": 0.630879345603272,
      "grad_norm": 19.804638979770427,
      "learning_rate": 9.63526691343403e-06,
      "loss": 3.0773,
      "step": 1851
    },
    {
      "epoch": 0.6312201772324472,
      "grad_norm": 13.653545484388776,
      "learning_rate": 9.634523034598421e-06,
      "loss": 3.5447,
      "step": 1852
    },
    {
      "epoch": 0.6315610088616224,
      "grad_norm": 19.035564323226595,
      "learning_rate": 9.633778426733893e-06,
      "loss": 3.4079,
      "step": 1853
    },
    {
      "epoch": 0.6319018404907976,
      "grad_norm": 25.59027052932139,
      "learning_rate": 9.633033089957572e-06,
      "loss": 3.3798,
      "step": 1854
    },
    {
      "epoch": 0.6322426721199728,
      "grad_norm": 18.623654800458535,
      "learning_rate": 9.632287024386702e-06,
      "loss": 3.6531,
      "step": 1855
    },
    {
      "epoch": 0.632583503749148,
      "grad_norm": 17.91486067353337,
      "learning_rate": 9.631540230138644e-06,
      "loss": 3.3088,
      "step": 1856
    },
    {
      "epoch": 0.6329243353783232,
      "grad_norm": 22.436591647253632,
      "learning_rate": 9.630792707330869e-06,
      "loss": 3.322,
      "step": 1857
    },
    {
      "epoch": 0.6332651670074982,
      "grad_norm": 32.648809351022855,
      "learning_rate": 9.630044456080968e-06,
      "loss": 3.4247,
      "step": 1858
    },
    {
      "epoch": 0.6336059986366734,
      "grad_norm": 15.771281612998978,
      "learning_rate": 9.629295476506644e-06,
      "loss": 3.2927,
      "step": 1859
    },
    {
      "epoch": 0.6339468302658486,
      "grad_norm": 53.36057380969888,
      "learning_rate": 9.628545768725713e-06,
      "loss": 3.3199,
      "step": 1860
    },
    {
      "epoch": 0.6342876618950238,
      "grad_norm": 18.92978631518339,
      "learning_rate": 9.627795332856107e-06,
      "loss": 2.8132,
      "step": 1861
    },
    {
      "epoch": 0.634628493524199,
      "grad_norm": 15.97104818504492,
      "learning_rate": 9.627044169015872e-06,
      "loss": 3.1322,
      "step": 1862
    },
    {
      "epoch": 0.6349693251533742,
      "grad_norm": 22.451825560784897,
      "learning_rate": 9.62629227732317e-06,
      "loss": 3.9186,
      "step": 1863
    },
    {
      "epoch": 0.6353101567825494,
      "grad_norm": 18.77877373387588,
      "learning_rate": 9.625539657896278e-06,
      "loss": 3.3421,
      "step": 1864
    },
    {
      "epoch": 0.6356509884117246,
      "grad_norm": 12.298264447879522,
      "learning_rate": 9.624786310853585e-06,
      "loss": 2.9925,
      "step": 1865
    },
    {
      "epoch": 0.6359918200408998,
      "grad_norm": 12.011692444925417,
      "learning_rate": 9.624032236313594e-06,
      "loss": 3.3325,
      "step": 1866
    },
    {
      "epoch": 0.636332651670075,
      "grad_norm": 18.262910516244414,
      "learning_rate": 9.623277434394925e-06,
      "loss": 3.0256,
      "step": 1867
    },
    {
      "epoch": 0.6366734832992502,
      "grad_norm": 19.748418919555885,
      "learning_rate": 9.622521905216312e-06,
      "loss": 3.6342,
      "step": 1868
    },
    {
      "epoch": 0.6370143149284253,
      "grad_norm": 15.754092164611281,
      "learning_rate": 9.6217656488966e-06,
      "loss": 3.3024,
      "step": 1869
    },
    {
      "epoch": 0.6373551465576005,
      "grad_norm": 15.970137893647482,
      "learning_rate": 9.621008665554756e-06,
      "loss": 3.3433,
      "step": 1870
    },
    {
      "epoch": 0.6376959781867757,
      "grad_norm": 18.526916135036977,
      "learning_rate": 9.620250955309852e-06,
      "loss": 3.4288,
      "step": 1871
    },
    {
      "epoch": 0.6380368098159509,
      "grad_norm": 14.551428276763954,
      "learning_rate": 9.619492518281081e-06,
      "loss": 3.3101,
      "step": 1872
    },
    {
      "epoch": 0.6383776414451261,
      "grad_norm": 15.74653833954483,
      "learning_rate": 9.618733354587748e-06,
      "loss": 3.4635,
      "step": 1873
    },
    {
      "epoch": 0.6387184730743013,
      "grad_norm": 14.75290322407769,
      "learning_rate": 9.617973464349272e-06,
      "loss": 2.9136,
      "step": 1874
    },
    {
      "epoch": 0.6390593047034765,
      "grad_norm": 21.96957910863123,
      "learning_rate": 9.617212847685188e-06,
      "loss": 3.3328,
      "step": 1875
    },
    {
      "epoch": 0.6394001363326517,
      "grad_norm": 18.264924863733192,
      "learning_rate": 9.61645150471514e-06,
      "loss": 3.3395,
      "step": 1876
    },
    {
      "epoch": 0.6397409679618269,
      "grad_norm": 15.346664389066275,
      "learning_rate": 9.615689435558897e-06,
      "loss": 3.4789,
      "step": 1877
    },
    {
      "epoch": 0.6400817995910021,
      "grad_norm": 23.667003584791548,
      "learning_rate": 9.61492664033633e-06,
      "loss": 3.5531,
      "step": 1878
    },
    {
      "epoch": 0.6404226312201773,
      "grad_norm": 47.26762825356943,
      "learning_rate": 9.614163119167431e-06,
      "loss": 3.9377,
      "step": 1879
    },
    {
      "epoch": 0.6407634628493524,
      "grad_norm": 21.21045084280319,
      "learning_rate": 9.613398872172305e-06,
      "loss": 3.2274,
      "step": 1880
    },
    {
      "epoch": 0.6411042944785276,
      "grad_norm": 23.88903123114681,
      "learning_rate": 9.612633899471173e-06,
      "loss": 3.1888,
      "step": 1881
    },
    {
      "epoch": 0.6414451261077028,
      "grad_norm": 15.04820980973465,
      "learning_rate": 9.611868201184366e-06,
      "loss": 3.2992,
      "step": 1882
    },
    {
      "epoch": 0.641785957736878,
      "grad_norm": 18.458041362633583,
      "learning_rate": 9.611101777432333e-06,
      "loss": 2.9019,
      "step": 1883
    },
    {
      "epoch": 0.6421267893660532,
      "grad_norm": 16.93714283146721,
      "learning_rate": 9.610334628335635e-06,
      "loss": 3.7668,
      "step": 1884
    },
    {
      "epoch": 0.6424676209952284,
      "grad_norm": 18.5555549969013,
      "learning_rate": 9.609566754014946e-06,
      "loss": 3.0663,
      "step": 1885
    },
    {
      "epoch": 0.6428084526244036,
      "grad_norm": 22.477578878578978,
      "learning_rate": 9.608798154591058e-06,
      "loss": 4.1019,
      "step": 1886
    },
    {
      "epoch": 0.6431492842535788,
      "grad_norm": 25.17848803550175,
      "learning_rate": 9.608028830184872e-06,
      "loss": 3.1324,
      "step": 1887
    },
    {
      "epoch": 0.643490115882754,
      "grad_norm": 17.313547263013028,
      "learning_rate": 9.60725878091741e-06,
      "loss": 3.1112,
      "step": 1888
    },
    {
      "epoch": 0.6438309475119292,
      "grad_norm": 16.3536722105224,
      "learning_rate": 9.606488006909799e-06,
      "loss": 2.8149,
      "step": 1889
    },
    {
      "epoch": 0.6441717791411042,
      "grad_norm": 19.700609418781053,
      "learning_rate": 9.60571650828329e-06,
      "loss": 3.8028,
      "step": 1890
    },
    {
      "epoch": 0.6445126107702794,
      "grad_norm": 17.783987025409406,
      "learning_rate": 9.604944285159236e-06,
      "loss": 3.4296,
      "step": 1891
    },
    {
      "epoch": 0.6448534423994546,
      "grad_norm": 22.636514256158335,
      "learning_rate": 9.604171337659118e-06,
      "loss": 3.2642,
      "step": 1892
    },
    {
      "epoch": 0.6451942740286298,
      "grad_norm": 14.140884043703355,
      "learning_rate": 9.60339766590452e-06,
      "loss": 2.9438,
      "step": 1893
    },
    {
      "epoch": 0.645535105657805,
      "grad_norm": 17.11698372260806,
      "learning_rate": 9.602623270017142e-06,
      "loss": 3.3118,
      "step": 1894
    },
    {
      "epoch": 0.6458759372869802,
      "grad_norm": 18.73075649203496,
      "learning_rate": 9.601848150118804e-06,
      "loss": 3.504,
      "step": 1895
    },
    {
      "epoch": 0.6462167689161554,
      "grad_norm": 17.500809016637444,
      "learning_rate": 9.601072306331431e-06,
      "loss": 3.6992,
      "step": 1896
    },
    {
      "epoch": 0.6465576005453306,
      "grad_norm": 11.73472881915135,
      "learning_rate": 9.60029573877707e-06,
      "loss": 2.6807,
      "step": 1897
    },
    {
      "epoch": 0.6468984321745058,
      "grad_norm": 16.040480551356406,
      "learning_rate": 9.599518447577875e-06,
      "loss": 3.3502,
      "step": 1898
    },
    {
      "epoch": 0.647239263803681,
      "grad_norm": 17.70617568612734,
      "learning_rate": 9.598740432856121e-06,
      "loss": 3.2048,
      "step": 1899
    },
    {
      "epoch": 0.6475800954328562,
      "grad_norm": 18.35507808954349,
      "learning_rate": 9.597961694734187e-06,
      "loss": 2.7568,
      "step": 1900
    },
    {
      "epoch": 0.6479209270620313,
      "grad_norm": 13.270561119274776,
      "learning_rate": 9.59718223333458e-06,
      "loss": 3.4024,
      "step": 1901
    },
    {
      "epoch": 0.6482617586912065,
      "grad_norm": 21.335269259904248,
      "learning_rate": 9.596402048779903e-06,
      "loss": 3.4578,
      "step": 1902
    },
    {
      "epoch": 0.6486025903203817,
      "grad_norm": 23.36209989585706,
      "learning_rate": 9.595621141192888e-06,
      "loss": 3.5526,
      "step": 1903
    },
    {
      "epoch": 0.6489434219495569,
      "grad_norm": 18.670957652435145,
      "learning_rate": 9.594839510696372e-06,
      "loss": 3.6742,
      "step": 1904
    },
    {
      "epoch": 0.6492842535787321,
      "grad_norm": 19.5686217959044,
      "learning_rate": 9.594057157413312e-06,
      "loss": 3.2888,
      "step": 1905
    },
    {
      "epoch": 0.6496250852079073,
      "grad_norm": 15.00830071391283,
      "learning_rate": 9.593274081466772e-06,
      "loss": 3.2568,
      "step": 1906
    },
    {
      "epoch": 0.6499659168370825,
      "grad_norm": 16.54793292260827,
      "learning_rate": 9.592490282979936e-06,
      "loss": 3.366,
      "step": 1907
    },
    {
      "epoch": 0.6503067484662577,
      "grad_norm": 17.764767429075814,
      "learning_rate": 9.591705762076095e-06,
      "loss": 3.6577,
      "step": 1908
    },
    {
      "epoch": 0.6506475800954329,
      "grad_norm": 13.529107916612718,
      "learning_rate": 9.59092051887866e-06,
      "loss": 3.2463,
      "step": 1909
    },
    {
      "epoch": 0.6509884117246081,
      "grad_norm": 16.26038191996233,
      "learning_rate": 9.590134553511152e-06,
      "loss": 3.309,
      "step": 1910
    },
    {
      "epoch": 0.6513292433537833,
      "grad_norm": 18.73827926920426,
      "learning_rate": 9.589347866097204e-06,
      "loss": 3.3594,
      "step": 1911
    },
    {
      "epoch": 0.6516700749829584,
      "grad_norm": 26.976021599516173,
      "learning_rate": 9.588560456760571e-06,
      "loss": 3.4447,
      "step": 1912
    },
    {
      "epoch": 0.6520109066121336,
      "grad_norm": 13.151316619228554,
      "learning_rate": 9.587772325625111e-06,
      "loss": 3.1514,
      "step": 1913
    },
    {
      "epoch": 0.6523517382413088,
      "grad_norm": 18.87965847820063,
      "learning_rate": 9.5869834728148e-06,
      "loss": 3.1424,
      "step": 1914
    },
    {
      "epoch": 0.652692569870484,
      "grad_norm": 11.904140114532062,
      "learning_rate": 9.586193898453731e-06,
      "loss": 2.9558,
      "step": 1915
    },
    {
      "epoch": 0.6530334014996592,
      "grad_norm": 14.223770608299395,
      "learning_rate": 9.585403602666105e-06,
      "loss": 3.0521,
      "step": 1916
    },
    {
      "epoch": 0.6533742331288344,
      "grad_norm": 18.08975802862374,
      "learning_rate": 9.584612585576239e-06,
      "loss": 3.0353,
      "step": 1917
    },
    {
      "epoch": 0.6537150647580096,
      "grad_norm": 21.310963754251947,
      "learning_rate": 9.583820847308561e-06,
      "loss": 3.9582,
      "step": 1918
    },
    {
      "epoch": 0.6540558963871848,
      "grad_norm": 16.260009858218904,
      "learning_rate": 9.583028387987619e-06,
      "loss": 3.3373,
      "step": 1919
    },
    {
      "epoch": 0.65439672801636,
      "grad_norm": 16.61412906976423,
      "learning_rate": 9.582235207738065e-06,
      "loss": 3.0913,
      "step": 1920
    },
    {
      "epoch": 0.6547375596455351,
      "grad_norm": 16.7022067110724,
      "learning_rate": 9.581441306684674e-06,
      "loss": 3.5353,
      "step": 1921
    },
    {
      "epoch": 0.6550783912747103,
      "grad_norm": 12.27511483728523,
      "learning_rate": 9.580646684952326e-06,
      "loss": 2.989,
      "step": 1922
    },
    {
      "epoch": 0.6554192229038854,
      "grad_norm": 43.68046807376579,
      "learning_rate": 9.579851342666022e-06,
      "loss": 2.4732,
      "step": 1923
    },
    {
      "epoch": 0.6557600545330606,
      "grad_norm": 19.9437374392077,
      "learning_rate": 9.579055279950869e-06,
      "loss": 3.1407,
      "step": 1924
    },
    {
      "epoch": 0.6561008861622358,
      "grad_norm": 18.39188423281928,
      "learning_rate": 9.578258496932093e-06,
      "loss": 3.9077,
      "step": 1925
    },
    {
      "epoch": 0.656441717791411,
      "grad_norm": 17.043915977312523,
      "learning_rate": 9.57746099373503e-06,
      "loss": 3.9808,
      "step": 1926
    },
    {
      "epoch": 0.6567825494205862,
      "grad_norm": 15.86645245485466,
      "learning_rate": 9.576662770485132e-06,
      "loss": 3.0901,
      "step": 1927
    },
    {
      "epoch": 0.6571233810497614,
      "grad_norm": 12.033290203276225,
      "learning_rate": 9.57586382730796e-06,
      "loss": 2.2891,
      "step": 1928
    },
    {
      "epoch": 0.6574642126789366,
      "grad_norm": 15.604763197265202,
      "learning_rate": 9.575064164329193e-06,
      "loss": 3.1663,
      "step": 1929
    },
    {
      "epoch": 0.6578050443081118,
      "grad_norm": 17.70071445940186,
      "learning_rate": 9.574263781674623e-06,
      "loss": 3.694,
      "step": 1930
    },
    {
      "epoch": 0.658145875937287,
      "grad_norm": 15.716572900608769,
      "learning_rate": 9.573462679470147e-06,
      "loss": 3.1166,
      "step": 1931
    },
    {
      "epoch": 0.6584867075664622,
      "grad_norm": 15.004924006794022,
      "learning_rate": 9.572660857841789e-06,
      "loss": 3.0325,
      "step": 1932
    },
    {
      "epoch": 0.6588275391956373,
      "grad_norm": 25.699409550847736,
      "learning_rate": 9.571858316915674e-06,
      "loss": 2.957,
      "step": 1933
    },
    {
      "epoch": 0.6591683708248125,
      "grad_norm": 32.67159332154362,
      "learning_rate": 9.571055056818044e-06,
      "loss": 3.4626,
      "step": 1934
    },
    {
      "epoch": 0.6595092024539877,
      "grad_norm": 14.598237438208814,
      "learning_rate": 9.57025107767526e-06,
      "loss": 3.1902,
      "step": 1935
    },
    {
      "epoch": 0.6598500340831629,
      "grad_norm": 16.25388583422308,
      "learning_rate": 9.569446379613788e-06,
      "loss": 3.2615,
      "step": 1936
    },
    {
      "epoch": 0.6601908657123381,
      "grad_norm": 17.66730506762041,
      "learning_rate": 9.56864096276021e-06,
      "loss": 3.2894,
      "step": 1937
    },
    {
      "epoch": 0.6605316973415133,
      "grad_norm": 18.102763026683714,
      "learning_rate": 9.567834827241223e-06,
      "loss": 2.8759,
      "step": 1938
    },
    {
      "epoch": 0.6608725289706885,
      "grad_norm": 16.658696811542644,
      "learning_rate": 9.567027973183633e-06,
      "loss": 2.788,
      "step": 1939
    },
    {
      "epoch": 0.6612133605998637,
      "grad_norm": 16.98936384727511,
      "learning_rate": 9.566220400714363e-06,
      "loss": 4.0868,
      "step": 1940
    },
    {
      "epoch": 0.6615541922290389,
      "grad_norm": 26.3458388830922,
      "learning_rate": 9.565412109960446e-06,
      "loss": 3.1901,
      "step": 1941
    },
    {
      "epoch": 0.6618950238582141,
      "grad_norm": 15.021459857865608,
      "learning_rate": 9.564603101049033e-06,
      "loss": 3.2874,
      "step": 1942
    },
    {
      "epoch": 0.6622358554873893,
      "grad_norm": 12.759852816393614,
      "learning_rate": 9.563793374107379e-06,
      "loss": 3.0907,
      "step": 1943
    },
    {
      "epoch": 0.6625766871165644,
      "grad_norm": 23.79871727152637,
      "learning_rate": 9.56298292926286e-06,
      "loss": 2.6903,
      "step": 1944
    },
    {
      "epoch": 0.6629175187457396,
      "grad_norm": 28.191367496337623,
      "learning_rate": 9.562171766642963e-06,
      "loss": 3.3377,
      "step": 1945
    },
    {
      "epoch": 0.6632583503749148,
      "grad_norm": 13.866116485727343,
      "learning_rate": 9.561359886375286e-06,
      "loss": 3.0477,
      "step": 1946
    },
    {
      "epoch": 0.66359918200409,
      "grad_norm": 19.243519963378194,
      "learning_rate": 9.56054728858754e-06,
      "loss": 3.4226,
      "step": 1947
    },
    {
      "epoch": 0.6639400136332652,
      "grad_norm": 15.929398638255925,
      "learning_rate": 9.559733973407552e-06,
      "loss": 2.9338,
      "step": 1948
    },
    {
      "epoch": 0.6642808452624404,
      "grad_norm": 11.56942208140459,
      "learning_rate": 9.558919940963258e-06,
      "loss": 2.8939,
      "step": 1949
    },
    {
      "epoch": 0.6646216768916156,
      "grad_norm": 19.16066336187289,
      "learning_rate": 9.55810519138271e-06,
      "loss": 3.7111,
      "step": 1950
    },
    {
      "epoch": 0.6649625085207908,
      "grad_norm": 19.049840050713676,
      "learning_rate": 9.55728972479407e-06,
      "loss": 2.9145,
      "step": 1951
    },
    {
      "epoch": 0.665303340149966,
      "grad_norm": 22.239054737509395,
      "learning_rate": 9.556473541325616e-06,
      "loss": 3.986,
      "step": 1952
    },
    {
      "epoch": 0.6656441717791411,
      "grad_norm": 27.066263272024667,
      "learning_rate": 9.555656641105735e-06,
      "loss": 3.7447,
      "step": 1953
    },
    {
      "epoch": 0.6659850034083163,
      "grad_norm": 16.99177305926694,
      "learning_rate": 9.554839024262929e-06,
      "loss": 3.4471,
      "step": 1954
    },
    {
      "epoch": 0.6663258350374914,
      "grad_norm": 25.00570873731563,
      "learning_rate": 9.554020690925812e-06,
      "loss": 3.8342,
      "step": 1955
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 14.789899790011257,
      "learning_rate": 9.553201641223112e-06,
      "loss": 3.6569,
      "step": 1956
    },
    {
      "epoch": 0.6670074982958418,
      "grad_norm": 18.89636273545075,
      "learning_rate": 9.552381875283671e-06,
      "loss": 2.9707,
      "step": 1957
    },
    {
      "epoch": 0.667348329925017,
      "grad_norm": 22.09130500278689,
      "learning_rate": 9.551561393236437e-06,
      "loss": 3.4975,
      "step": 1958
    },
    {
      "epoch": 0.6676891615541922,
      "grad_norm": 18.294943971563203,
      "learning_rate": 9.550740195210477e-06,
      "loss": 3.7537,
      "step": 1959
    },
    {
      "epoch": 0.6680299931833674,
      "grad_norm": 15.164806742011505,
      "learning_rate": 9.54991828133497e-06,
      "loss": 3.2224,
      "step": 1960
    },
    {
      "epoch": 0.6683708248125426,
      "grad_norm": 18.721670109188437,
      "learning_rate": 9.549095651739204e-06,
      "loss": 3.9872,
      "step": 1961
    },
    {
      "epoch": 0.6687116564417178,
      "grad_norm": 17.45966569474267,
      "learning_rate": 9.548272306552584e-06,
      "loss": 3.0352,
      "step": 1962
    },
    {
      "epoch": 0.669052488070893,
      "grad_norm": 17.85052185427996,
      "learning_rate": 9.547448245904624e-06,
      "loss": 3.1362,
      "step": 1963
    },
    {
      "epoch": 0.6693933197000682,
      "grad_norm": 24.286225086708743,
      "learning_rate": 9.546623469924954e-06,
      "loss": 3.3246,
      "step": 1964
    },
    {
      "epoch": 0.6697341513292433,
      "grad_norm": 14.403746426916642,
      "learning_rate": 9.545797978743312e-06,
      "loss": 3.3748,
      "step": 1965
    },
    {
      "epoch": 0.6700749829584185,
      "grad_norm": 19.134102992422886,
      "learning_rate": 9.54497177248955e-06,
      "loss": 4.0349,
      "step": 1966
    },
    {
      "epoch": 0.6704158145875937,
      "grad_norm": 37.96120278096082,
      "learning_rate": 9.54414485129364e-06,
      "loss": 3.3466,
      "step": 1967
    },
    {
      "epoch": 0.6707566462167689,
      "grad_norm": 20.098975124479157,
      "learning_rate": 9.543317215285652e-06,
      "loss": 3.0479,
      "step": 1968
    },
    {
      "epoch": 0.6710974778459441,
      "grad_norm": 18.712315631767304,
      "learning_rate": 9.542488864595782e-06,
      "loss": 3.378,
      "step": 1969
    },
    {
      "epoch": 0.6714383094751193,
      "grad_norm": 15.05536913786104,
      "learning_rate": 9.54165979935433e-06,
      "loss": 3.044,
      "step": 1970
    },
    {
      "epoch": 0.6717791411042945,
      "grad_norm": 31.09397876575775,
      "learning_rate": 9.540830019691714e-06,
      "loss": 3.3005,
      "step": 1971
    },
    {
      "epoch": 0.6721199727334697,
      "grad_norm": 12.36476215528246,
      "learning_rate": 9.539999525738459e-06,
      "loss": 2.9023,
      "step": 1972
    },
    {
      "epoch": 0.6724608043626449,
      "grad_norm": 15.809428659118469,
      "learning_rate": 9.539168317625205e-06,
      "loss": 3.0042,
      "step": 1973
    },
    {
      "epoch": 0.6728016359918201,
      "grad_norm": 19.490097196577153,
      "learning_rate": 9.538336395482705e-06,
      "loss": 3.5293,
      "step": 1974
    },
    {
      "epoch": 0.6731424676209953,
      "grad_norm": 25.19323498374576,
      "learning_rate": 9.537503759441825e-06,
      "loss": 3.9339,
      "step": 1975
    },
    {
      "epoch": 0.6734832992501704,
      "grad_norm": 15.062160769808289,
      "learning_rate": 9.536670409633542e-06,
      "loss": 3.1639,
      "step": 1976
    },
    {
      "epoch": 0.6738241308793456,
      "grad_norm": 19.9092094388048,
      "learning_rate": 9.535836346188943e-06,
      "loss": 2.8818,
      "step": 1977
    },
    {
      "epoch": 0.6741649625085208,
      "grad_norm": 16.669856716141275,
      "learning_rate": 9.53500156923923e-06,
      "loss": 3.3019,
      "step": 1978
    },
    {
      "epoch": 0.674505794137696,
      "grad_norm": 11.682213839884271,
      "learning_rate": 9.534166078915716e-06,
      "loss": 2.6365,
      "step": 1979
    },
    {
      "epoch": 0.6748466257668712,
      "grad_norm": 20.334246464188695,
      "learning_rate": 9.533329875349831e-06,
      "loss": 3.2939,
      "step": 1980
    },
    {
      "epoch": 0.6751874573960464,
      "grad_norm": 15.549475218468288,
      "learning_rate": 9.53249295867311e-06,
      "loss": 3.4871,
      "step": 1981
    },
    {
      "epoch": 0.6755282890252216,
      "grad_norm": 13.64488090397031,
      "learning_rate": 9.531655329017203e-06,
      "loss": 3.0567,
      "step": 1982
    },
    {
      "epoch": 0.6758691206543967,
      "grad_norm": 14.805701269289413,
      "learning_rate": 9.530816986513874e-06,
      "loss": 3.067,
      "step": 1983
    },
    {
      "epoch": 0.676209952283572,
      "grad_norm": 15.277722534191799,
      "learning_rate": 9.529977931294996e-06,
      "loss": 3.3365,
      "step": 1984
    },
    {
      "epoch": 0.6765507839127471,
      "grad_norm": 28.54031385126413,
      "learning_rate": 9.529138163492557e-06,
      "loss": 2.9569,
      "step": 1985
    },
    {
      "epoch": 0.6768916155419223,
      "grad_norm": 22.706198928572856,
      "learning_rate": 9.528297683238656e-06,
      "loss": 3.2074,
      "step": 1986
    },
    {
      "epoch": 0.6772324471710974,
      "grad_norm": 25.845670187432944,
      "learning_rate": 9.527456490665503e-06,
      "loss": 3.8666,
      "step": 1987
    },
    {
      "epoch": 0.6775732788002726,
      "grad_norm": 15.485237987012194,
      "learning_rate": 9.52661458590542e-06,
      "loss": 3.2799,
      "step": 1988
    },
    {
      "epoch": 0.6779141104294478,
      "grad_norm": 15.0153036051137,
      "learning_rate": 9.525771969090844e-06,
      "loss": 2.4864,
      "step": 1989
    },
    {
      "epoch": 0.678254942058623,
      "grad_norm": 24.39427774344475,
      "learning_rate": 9.524928640354322e-06,
      "loss": 3.6532,
      "step": 1990
    },
    {
      "epoch": 0.6785957736877982,
      "grad_norm": 14.881036940036829,
      "learning_rate": 9.524084599828513e-06,
      "loss": 3.2092,
      "step": 1991
    },
    {
      "epoch": 0.6789366053169734,
      "grad_norm": 20.068688339098646,
      "learning_rate": 9.523239847646186e-06,
      "loss": 3.7028,
      "step": 1992
    },
    {
      "epoch": 0.6792774369461486,
      "grad_norm": 26.19946953227523,
      "learning_rate": 9.522394383940224e-06,
      "loss": 3.5656,
      "step": 1993
    },
    {
      "epoch": 0.6796182685753238,
      "grad_norm": 16.342682339367393,
      "learning_rate": 9.521548208843623e-06,
      "loss": 3.5323,
      "step": 1994
    },
    {
      "epoch": 0.679959100204499,
      "grad_norm": 16.464620805398315,
      "learning_rate": 9.520701322489492e-06,
      "loss": 3.0097,
      "step": 1995
    },
    {
      "epoch": 0.6802999318336742,
      "grad_norm": 18.069072951410323,
      "learning_rate": 9.519853725011043e-06,
      "loss": 3.7281,
      "step": 1996
    },
    {
      "epoch": 0.6806407634628494,
      "grad_norm": 11.019877625539294,
      "learning_rate": 9.519005416541613e-06,
      "loss": 2.6042,
      "step": 1997
    },
    {
      "epoch": 0.6809815950920245,
      "grad_norm": 13.574420581578858,
      "learning_rate": 9.518156397214643e-06,
      "loss": 2.6014,
      "step": 1998
    },
    {
      "epoch": 0.6813224267211997,
      "grad_norm": 14.725261245536851,
      "learning_rate": 9.517306667163684e-06,
      "loss": 2.9867,
      "step": 1999
    },
    {
      "epoch": 0.6816632583503749,
      "grad_norm": 19.87163919316535,
      "learning_rate": 9.516456226522405e-06,
      "loss": 3.8756,
      "step": 2000
    },
    {
      "epoch": 0.6820040899795501,
      "grad_norm": 38.296241816809705,
      "learning_rate": 9.515605075424582e-06,
      "loss": 3.1252,
      "step": 2001
    },
    {
      "epoch": 0.6823449216087253,
      "grad_norm": 21.714693392757273,
      "learning_rate": 9.514753214004106e-06,
      "loss": 3.7938,
      "step": 2002
    },
    {
      "epoch": 0.6826857532379005,
      "grad_norm": 22.01524703004069,
      "learning_rate": 9.513900642394975e-06,
      "loss": 2.9821,
      "step": 2003
    },
    {
      "epoch": 0.6830265848670757,
      "grad_norm": 19.897512714745915,
      "learning_rate": 9.513047360731307e-06,
      "loss": 3.2912,
      "step": 2004
    },
    {
      "epoch": 0.6833674164962509,
      "grad_norm": 15.700482501062833,
      "learning_rate": 9.512193369147322e-06,
      "loss": 3.551,
      "step": 2005
    },
    {
      "epoch": 0.6837082481254261,
      "grad_norm": 15.248470032526114,
      "learning_rate": 9.511338667777357e-06,
      "loss": 2.8459,
      "step": 2006
    },
    {
      "epoch": 0.6840490797546013,
      "grad_norm": 17.794658509112953,
      "learning_rate": 9.510483256755863e-06,
      "loss": 3.65,
      "step": 2007
    },
    {
      "epoch": 0.6843899113837764,
      "grad_norm": 14.23043694221653,
      "learning_rate": 9.509627136217396e-06,
      "loss": 2.8497,
      "step": 2008
    },
    {
      "epoch": 0.6847307430129516,
      "grad_norm": 55.315257846910114,
      "learning_rate": 9.50877030629663e-06,
      "loss": 3.594,
      "step": 2009
    },
    {
      "epoch": 0.6850715746421268,
      "grad_norm": 19.991325447621154,
      "learning_rate": 9.507912767128347e-06,
      "loss": 3.5262,
      "step": 2010
    },
    {
      "epoch": 0.685412406271302,
      "grad_norm": 15.684575762939904,
      "learning_rate": 9.50705451884744e-06,
      "loss": 3.1289,
      "step": 2011
    },
    {
      "epoch": 0.6857532379004772,
      "grad_norm": 16.0804914661727,
      "learning_rate": 9.506195561588914e-06,
      "loss": 3.4769,
      "step": 2012
    },
    {
      "epoch": 0.6860940695296524,
      "grad_norm": 17.381004583815827,
      "learning_rate": 9.50533589548789e-06,
      "loss": 3.6077,
      "step": 2013
    },
    {
      "epoch": 0.6864349011588275,
      "grad_norm": 21.176472953167355,
      "learning_rate": 9.504475520679594e-06,
      "loss": 3.1482,
      "step": 2014
    },
    {
      "epoch": 0.6867757327880027,
      "grad_norm": 24.455751375714787,
      "learning_rate": 9.503614437299368e-06,
      "loss": 3.044,
      "step": 2015
    },
    {
      "epoch": 0.6871165644171779,
      "grad_norm": 27.525874253927697,
      "learning_rate": 9.502752645482662e-06,
      "loss": 2.938,
      "step": 2016
    },
    {
      "epoch": 0.6874573960463531,
      "grad_norm": 21.088439896738464,
      "learning_rate": 9.501890145365042e-06,
      "loss": 3.218,
      "step": 2017
    },
    {
      "epoch": 0.6877982276755283,
      "grad_norm": 12.957538250224953,
      "learning_rate": 9.50102693708218e-06,
      "loss": 2.9292,
      "step": 2018
    },
    {
      "epoch": 0.6881390593047034,
      "grad_norm": 12.066691008047316,
      "learning_rate": 9.500163020769864e-06,
      "loss": 3.2258,
      "step": 2019
    },
    {
      "epoch": 0.6884798909338786,
      "grad_norm": 15.473928274866065,
      "learning_rate": 9.49929839656399e-06,
      "loss": 3.0083,
      "step": 2020
    },
    {
      "epoch": 0.6888207225630538,
      "grad_norm": 14.863772285117992,
      "learning_rate": 9.49843306460057e-06,
      "loss": 3.1327,
      "step": 2021
    },
    {
      "epoch": 0.689161554192229,
      "grad_norm": 16.153843672496812,
      "learning_rate": 9.497567025015719e-06,
      "loss": 3.0741,
      "step": 2022
    },
    {
      "epoch": 0.6895023858214042,
      "grad_norm": 37.86599087583578,
      "learning_rate": 9.496700277945671e-06,
      "loss": 3.4046,
      "step": 2023
    },
    {
      "epoch": 0.6898432174505794,
      "grad_norm": 15.813383182354107,
      "learning_rate": 9.495832823526769e-06,
      "loss": 3.1282,
      "step": 2024
    },
    {
      "epoch": 0.6901840490797546,
      "grad_norm": 14.837438996710222,
      "learning_rate": 9.494964661895468e-06,
      "loss": 3.6571,
      "step": 2025
    },
    {
      "epoch": 0.6905248807089298,
      "grad_norm": 16.226959638496016,
      "learning_rate": 9.494095793188332e-06,
      "loss": 3.1592,
      "step": 2026
    },
    {
      "epoch": 0.690865712338105,
      "grad_norm": 21.39002557820156,
      "learning_rate": 9.493226217542037e-06,
      "loss": 3.7845,
      "step": 2027
    },
    {
      "epoch": 0.6912065439672802,
      "grad_norm": 13.73495251411619,
      "learning_rate": 9.492355935093373e-06,
      "loss": 3.0983,
      "step": 2028
    },
    {
      "epoch": 0.6915473755964554,
      "grad_norm": 23.653164891281357,
      "learning_rate": 9.491484945979237e-06,
      "loss": 3.507,
      "step": 2029
    },
    {
      "epoch": 0.6918882072256305,
      "grad_norm": 20.713499765264057,
      "learning_rate": 9.49061325033664e-06,
      "loss": 3.5846,
      "step": 2030
    },
    {
      "epoch": 0.6922290388548057,
      "grad_norm": 14.889074261047142,
      "learning_rate": 9.489740848302702e-06,
      "loss": 3.2062,
      "step": 2031
    },
    {
      "epoch": 0.6925698704839809,
      "grad_norm": 24.970506449710747,
      "learning_rate": 9.488867740014654e-06,
      "loss": 2.7953,
      "step": 2032
    },
    {
      "epoch": 0.6929107021131561,
      "grad_norm": 11.579003408088393,
      "learning_rate": 9.487993925609845e-06,
      "loss": 2.9836,
      "step": 2033
    },
    {
      "epoch": 0.6932515337423313,
      "grad_norm": 14.30339206451075,
      "learning_rate": 9.487119405225722e-06,
      "loss": 3.4967,
      "step": 2034
    },
    {
      "epoch": 0.6935923653715065,
      "grad_norm": 21.164812573692313,
      "learning_rate": 9.486244178999857e-06,
      "loss": 2.789,
      "step": 2035
    },
    {
      "epoch": 0.6939331970006817,
      "grad_norm": 15.530937088809129,
      "learning_rate": 9.485368247069922e-06,
      "loss": 3.2896,
      "step": 2036
    },
    {
      "epoch": 0.6942740286298569,
      "grad_norm": 21.434645487326826,
      "learning_rate": 9.484491609573709e-06,
      "loss": 4.0006,
      "step": 2037
    },
    {
      "epoch": 0.6946148602590321,
      "grad_norm": 27.88718392251804,
      "learning_rate": 9.483614266649113e-06,
      "loss": 3.2253,
      "step": 2038
    },
    {
      "epoch": 0.6949556918882073,
      "grad_norm": 23.76352991016491,
      "learning_rate": 9.482736218434144e-06,
      "loss": 3.7524,
      "step": 2039
    },
    {
      "epoch": 0.6952965235173824,
      "grad_norm": 20.62580595668517,
      "learning_rate": 9.481857465066923e-06,
      "loss": 3.8301,
      "step": 2040
    },
    {
      "epoch": 0.6956373551465576,
      "grad_norm": 15.922671726495734,
      "learning_rate": 9.480978006685681e-06,
      "loss": 2.7674,
      "step": 2041
    },
    {
      "epoch": 0.6959781867757328,
      "grad_norm": 15.586046888614947,
      "learning_rate": 9.480097843428762e-06,
      "loss": 3.1495,
      "step": 2042
    },
    {
      "epoch": 0.696319018404908,
      "grad_norm": 17.998063666559656,
      "learning_rate": 9.479216975434619e-06,
      "loss": 2.6914,
      "step": 2043
    },
    {
      "epoch": 0.6966598500340832,
      "grad_norm": 15.003571196411162,
      "learning_rate": 9.478335402841813e-06,
      "loss": 3.193,
      "step": 2044
    },
    {
      "epoch": 0.6970006816632583,
      "grad_norm": 17.479786551972047,
      "learning_rate": 9.477453125789023e-06,
      "loss": 3.4111,
      "step": 2045
    },
    {
      "epoch": 0.6973415132924335,
      "grad_norm": 14.542936529992526,
      "learning_rate": 9.476570144415031e-06,
      "loss": 3.2874,
      "step": 2046
    },
    {
      "epoch": 0.6976823449216087,
      "grad_norm": 20.954687418567097,
      "learning_rate": 9.475686458858735e-06,
      "loss": 3.4455,
      "step": 2047
    },
    {
      "epoch": 0.6980231765507839,
      "grad_norm": 16.384454023612744,
      "learning_rate": 9.474802069259143e-06,
      "loss": 3.7153,
      "step": 2048
    },
    {
      "epoch": 0.6983640081799591,
      "grad_norm": 15.239032377171956,
      "learning_rate": 9.47391697575537e-06,
      "loss": 3.5753,
      "step": 2049
    },
    {
      "epoch": 0.6987048398091343,
      "grad_norm": 17.779814979591002,
      "learning_rate": 9.473031178486651e-06,
      "loss": 3.1986,
      "step": 2050
    },
    {
      "epoch": 0.6990456714383094,
      "grad_norm": 13.291188816696545,
      "learning_rate": 9.472144677592318e-06,
      "loss": 2.8296,
      "step": 2051
    },
    {
      "epoch": 0.6993865030674846,
      "grad_norm": 19.592809071272427,
      "learning_rate": 9.471257473211828e-06,
      "loss": 3.5864,
      "step": 2052
    },
    {
      "epoch": 0.6997273346966598,
      "grad_norm": 35.99881977867805,
      "learning_rate": 9.470369565484735e-06,
      "loss": 4.0762,
      "step": 2053
    },
    {
      "epoch": 0.700068166325835,
      "grad_norm": 21.474521619323784,
      "learning_rate": 9.469480954550715e-06,
      "loss": 3.3391,
      "step": 2054
    },
    {
      "epoch": 0.7004089979550102,
      "grad_norm": 15.742164821540676,
      "learning_rate": 9.468591640549552e-06,
      "loss": 3.2388,
      "step": 2055
    },
    {
      "epoch": 0.7007498295841854,
      "grad_norm": 15.691178093887359,
      "learning_rate": 9.467701623621133e-06,
      "loss": 2.8942,
      "step": 2056
    },
    {
      "epoch": 0.7010906612133606,
      "grad_norm": 17.38995279912389,
      "learning_rate": 9.466810903905463e-06,
      "loss": 3.2667,
      "step": 2057
    },
    {
      "epoch": 0.7014314928425358,
      "grad_norm": 15.68282722800537,
      "learning_rate": 9.465919481542658e-06,
      "loss": 3.3339,
      "step": 2058
    },
    {
      "epoch": 0.701772324471711,
      "grad_norm": 11.103329742117824,
      "learning_rate": 9.465027356672939e-06,
      "loss": 2.5684,
      "step": 2059
    },
    {
      "epoch": 0.7021131561008862,
      "grad_norm": 16.384328869947428,
      "learning_rate": 9.464134529436644e-06,
      "loss": 2.812,
      "step": 2060
    },
    {
      "epoch": 0.7024539877300614,
      "grad_norm": 16.950746080588612,
      "learning_rate": 9.463240999974216e-06,
      "loss": 2.8176,
      "step": 2061
    },
    {
      "epoch": 0.7027948193592365,
      "grad_norm": 20.389368117168754,
      "learning_rate": 9.462346768426211e-06,
      "loss": 3.7322,
      "step": 2062
    },
    {
      "epoch": 0.7031356509884117,
      "grad_norm": 24.55072741774868,
      "learning_rate": 9.461451834933297e-06,
      "loss": 3.6454,
      "step": 2063
    },
    {
      "epoch": 0.7034764826175869,
      "grad_norm": 13.8184814815036,
      "learning_rate": 9.460556199636249e-06,
      "loss": 3.0659,
      "step": 2064
    },
    {
      "epoch": 0.7038173142467621,
      "grad_norm": 14.674628169910356,
      "learning_rate": 9.459659862675953e-06,
      "loss": 3.0557,
      "step": 2065
    },
    {
      "epoch": 0.7041581458759373,
      "grad_norm": 23.094601615012294,
      "learning_rate": 9.45876282419341e-06,
      "loss": 3.2231,
      "step": 2066
    },
    {
      "epoch": 0.7044989775051125,
      "grad_norm": 12.421104573764032,
      "learning_rate": 9.457865084329725e-06,
      "loss": 3.1167,
      "step": 2067
    },
    {
      "epoch": 0.7048398091342877,
      "grad_norm": 13.500578859018319,
      "learning_rate": 9.456966643226115e-06,
      "loss": 2.9939,
      "step": 2068
    },
    {
      "epoch": 0.7051806407634629,
      "grad_norm": 15.545337927020801,
      "learning_rate": 9.45606750102391e-06,
      "loss": 3.318,
      "step": 2069
    },
    {
      "epoch": 0.7055214723926381,
      "grad_norm": 27.60818941510369,
      "learning_rate": 9.455167657864551e-06,
      "loss": 3.4039,
      "step": 2070
    },
    {
      "epoch": 0.7058623040218133,
      "grad_norm": 18.942082086841143,
      "learning_rate": 9.454267113889584e-06,
      "loss": 3.4197,
      "step": 2071
    },
    {
      "epoch": 0.7062031356509885,
      "grad_norm": 17.033486535274417,
      "learning_rate": 9.453365869240666e-06,
      "loss": 3.2668,
      "step": 2072
    },
    {
      "epoch": 0.7065439672801636,
      "grad_norm": 34.93472275254403,
      "learning_rate": 9.45246392405957e-06,
      "loss": 3.7864,
      "step": 2073
    },
    {
      "epoch": 0.7068847989093388,
      "grad_norm": 22.216679073537343,
      "learning_rate": 9.451561278488175e-06,
      "loss": 3.5603,
      "step": 2074
    },
    {
      "epoch": 0.707225630538514,
      "grad_norm": 16.80746069633037,
      "learning_rate": 9.45065793266847e-06,
      "loss": 3.5328,
      "step": 2075
    },
    {
      "epoch": 0.7075664621676891,
      "grad_norm": 18.102792793165516,
      "learning_rate": 9.449753886742555e-06,
      "loss": 3.992,
      "step": 2076
    },
    {
      "epoch": 0.7079072937968643,
      "grad_norm": 20.898992756954325,
      "learning_rate": 9.44884914085264e-06,
      "loss": 3.0073,
      "step": 2077
    },
    {
      "epoch": 0.7082481254260395,
      "grad_norm": 24.514984834994927,
      "learning_rate": 9.447943695141047e-06,
      "loss": 3.6462,
      "step": 2078
    },
    {
      "epoch": 0.7085889570552147,
      "grad_norm": 18.497925421788572,
      "learning_rate": 9.447037549750203e-06,
      "loss": 3.3344,
      "step": 2079
    },
    {
      "epoch": 0.7089297886843899,
      "grad_norm": 25.20486231040838,
      "learning_rate": 9.446130704822649e-06,
      "loss": 3.8382,
      "step": 2080
    },
    {
      "epoch": 0.7092706203135651,
      "grad_norm": 14.465797807478049,
      "learning_rate": 9.445223160501038e-06,
      "loss": 2.9517,
      "step": 2081
    },
    {
      "epoch": 0.7096114519427403,
      "grad_norm": 22.067345121809517,
      "learning_rate": 9.444314916928128e-06,
      "loss": 3.0826,
      "step": 2082
    },
    {
      "epoch": 0.7099522835719154,
      "grad_norm": 16.106941238533324,
      "learning_rate": 9.443405974246788e-06,
      "loss": 2.9081,
      "step": 2083
    },
    {
      "epoch": 0.7102931152010906,
      "grad_norm": 12.03865445507256,
      "learning_rate": 9.442496332600001e-06,
      "loss": 2.8307,
      "step": 2084
    },
    {
      "epoch": 0.7106339468302658,
      "grad_norm": 14.76250585182107,
      "learning_rate": 9.441585992130857e-06,
      "loss": 3.301,
      "step": 2085
    },
    {
      "epoch": 0.710974778459441,
      "grad_norm": 16.030253900005775,
      "learning_rate": 9.440674952982556e-06,
      "loss": 3.4222,
      "step": 2086
    },
    {
      "epoch": 0.7113156100886162,
      "grad_norm": 35.08646623456906,
      "learning_rate": 9.439763215298407e-06,
      "loss": 3.248,
      "step": 2087
    },
    {
      "epoch": 0.7116564417177914,
      "grad_norm": 25.16838158433103,
      "learning_rate": 9.43885077922183e-06,
      "loss": 3.4755,
      "step": 2088
    },
    {
      "epoch": 0.7119972733469666,
      "grad_norm": 12.953209130445995,
      "learning_rate": 9.437937644896355e-06,
      "loss": 3.3373,
      "step": 2089
    },
    {
      "epoch": 0.7123381049761418,
      "grad_norm": 16.211260306196767,
      "learning_rate": 9.437023812465623e-06,
      "loss": 3.1976,
      "step": 2090
    },
    {
      "epoch": 0.712678936605317,
      "grad_norm": 18.180820567999817,
      "learning_rate": 9.436109282073382e-06,
      "loss": 3.3859,
      "step": 2091
    },
    {
      "epoch": 0.7130197682344922,
      "grad_norm": 15.358429843890118,
      "learning_rate": 9.435194053863492e-06,
      "loss": 3.5194,
      "step": 2092
    },
    {
      "epoch": 0.7133605998636674,
      "grad_norm": 19.412466957159683,
      "learning_rate": 9.434278127979923e-06,
      "loss": 3.7945,
      "step": 2093
    },
    {
      "epoch": 0.7137014314928425,
      "grad_norm": 14.194751426133202,
      "learning_rate": 9.433361504566751e-06,
      "loss": 3.0173,
      "step": 2094
    },
    {
      "epoch": 0.7140422631220177,
      "grad_norm": 23.21945102886546,
      "learning_rate": 9.432444183768167e-06,
      "loss": 3.4271,
      "step": 2095
    },
    {
      "epoch": 0.7143830947511929,
      "grad_norm": 16.140404586063685,
      "learning_rate": 9.431526165728468e-06,
      "loss": 3.6293,
      "step": 2096
    },
    {
      "epoch": 0.7147239263803681,
      "grad_norm": 20.025224319496953,
      "learning_rate": 9.430607450592061e-06,
      "loss": 2.9187,
      "step": 2097
    },
    {
      "epoch": 0.7150647580095433,
      "grad_norm": 24.1230327551842,
      "learning_rate": 9.429688038503468e-06,
      "loss": 3.4119,
      "step": 2098
    },
    {
      "epoch": 0.7154055896387185,
      "grad_norm": 21.860560373167758,
      "learning_rate": 9.428767929607311e-06,
      "loss": 3.2569,
      "step": 2099
    },
    {
      "epoch": 0.7157464212678937,
      "grad_norm": 16.190933343041902,
      "learning_rate": 9.427847124048327e-06,
      "loss": 3.5765,
      "step": 2100
    },
    {
      "epoch": 0.7160872528970689,
      "grad_norm": 19.205727110643647,
      "learning_rate": 9.426925621971368e-06,
      "loss": 3.6002,
      "step": 2101
    },
    {
      "epoch": 0.7164280845262441,
      "grad_norm": 22.122931669002806,
      "learning_rate": 9.426003423521384e-06,
      "loss": 3.5051,
      "step": 2102
    },
    {
      "epoch": 0.7167689161554193,
      "grad_norm": 17.123984370704385,
      "learning_rate": 9.425080528843443e-06,
      "loss": 3.524,
      "step": 2103
    },
    {
      "epoch": 0.7171097477845945,
      "grad_norm": 16.745855273147797,
      "learning_rate": 9.424156938082718e-06,
      "loss": 3.5285,
      "step": 2104
    },
    {
      "epoch": 0.7174505794137696,
      "grad_norm": 14.516706419087384,
      "learning_rate": 9.423232651384497e-06,
      "loss": 3.2747,
      "step": 2105
    },
    {
      "epoch": 0.7177914110429447,
      "grad_norm": 14.246778416977962,
      "learning_rate": 9.42230766889417e-06,
      "loss": 3.4641,
      "step": 2106
    },
    {
      "epoch": 0.71813224267212,
      "grad_norm": 16.58095857306711,
      "learning_rate": 9.421381990757246e-06,
      "loss": 3.4627,
      "step": 2107
    },
    {
      "epoch": 0.7184730743012951,
      "grad_norm": 22.663404026901187,
      "learning_rate": 9.420455617119331e-06,
      "loss": 3.2897,
      "step": 2108
    },
    {
      "epoch": 0.7188139059304703,
      "grad_norm": 14.044393794286663,
      "learning_rate": 9.419528548126152e-06,
      "loss": 3.2347,
      "step": 2109
    },
    {
      "epoch": 0.7191547375596455,
      "grad_norm": 15.128951272227432,
      "learning_rate": 9.41860078392354e-06,
      "loss": 2.6867,
      "step": 2110
    },
    {
      "epoch": 0.7194955691888207,
      "grad_norm": 16.0410928385781,
      "learning_rate": 9.417672324657434e-06,
      "loss": 3.2987,
      "step": 2111
    },
    {
      "epoch": 0.7198364008179959,
      "grad_norm": 33.89639004371312,
      "learning_rate": 9.416743170473885e-06,
      "loss": 3.1659,
      "step": 2112
    },
    {
      "epoch": 0.7201772324471711,
      "grad_norm": 18.634352909516263,
      "learning_rate": 9.415813321519054e-06,
      "loss": 3.6293,
      "step": 2113
    },
    {
      "epoch": 0.7205180640763463,
      "grad_norm": 15.035383910742356,
      "learning_rate": 9.41488277793921e-06,
      "loss": 3.1384,
      "step": 2114
    },
    {
      "epoch": 0.7208588957055214,
      "grad_norm": 20.508804092048123,
      "learning_rate": 9.413951539880731e-06,
      "loss": 3.5859,
      "step": 2115
    },
    {
      "epoch": 0.7211997273346966,
      "grad_norm": 15.80505551681569,
      "learning_rate": 9.413019607490102e-06,
      "loss": 3.6468,
      "step": 2116
    },
    {
      "epoch": 0.7215405589638718,
      "grad_norm": 29.067964705869628,
      "learning_rate": 9.412086980913922e-06,
      "loss": 3.2086,
      "step": 2117
    },
    {
      "epoch": 0.721881390593047,
      "grad_norm": 11.0571044334001,
      "learning_rate": 9.411153660298897e-06,
      "loss": 2.7291,
      "step": 2118
    },
    {
      "epoch": 0.7222222222222222,
      "grad_norm": 14.143067431664138,
      "learning_rate": 9.410219645791841e-06,
      "loss": 2.5099,
      "step": 2119
    },
    {
      "epoch": 0.7225630538513974,
      "grad_norm": 13.553828825464832,
      "learning_rate": 9.40928493753968e-06,
      "loss": 2.8003,
      "step": 2120
    },
    {
      "epoch": 0.7229038854805726,
      "grad_norm": 22.798779517546993,
      "learning_rate": 9.408349535689443e-06,
      "loss": 4.0223,
      "step": 2121
    },
    {
      "epoch": 0.7232447171097478,
      "grad_norm": 29.04272144834151,
      "learning_rate": 9.407413440388277e-06,
      "loss": 4.0865,
      "step": 2122
    },
    {
      "epoch": 0.723585548738923,
      "grad_norm": 21.708388757850162,
      "learning_rate": 9.406476651783433e-06,
      "loss": 3.5193,
      "step": 2123
    },
    {
      "epoch": 0.7239263803680982,
      "grad_norm": 16.543317884920462,
      "learning_rate": 9.405539170022269e-06,
      "loss": 3.3338,
      "step": 2124
    },
    {
      "epoch": 0.7242672119972734,
      "grad_norm": 11.630971341563763,
      "learning_rate": 9.404600995252256e-06,
      "loss": 3.0225,
      "step": 2125
    },
    {
      "epoch": 0.7246080436264485,
      "grad_norm": 58.67817623220733,
      "learning_rate": 9.403662127620974e-06,
      "loss": 2.3736,
      "step": 2126
    },
    {
      "epoch": 0.7249488752556237,
      "grad_norm": 20.675348292327357,
      "learning_rate": 9.402722567276107e-06,
      "loss": 3.2509,
      "step": 2127
    },
    {
      "epoch": 0.7252897068847989,
      "grad_norm": 20.1739415533614,
      "learning_rate": 9.401782314365458e-06,
      "loss": 3.3504,
      "step": 2128
    },
    {
      "epoch": 0.7256305385139741,
      "grad_norm": 12.62005522688663,
      "learning_rate": 9.400841369036926e-06,
      "loss": 2.8552,
      "step": 2129
    },
    {
      "epoch": 0.7259713701431493,
      "grad_norm": 17.015086628492835,
      "learning_rate": 9.399899731438527e-06,
      "loss": 2.9039,
      "step": 2130
    },
    {
      "epoch": 0.7263122017723245,
      "grad_norm": 26.28232266523384,
      "learning_rate": 9.398957401718387e-06,
      "loss": 3.2005,
      "step": 2131
    },
    {
      "epoch": 0.7266530334014997,
      "grad_norm": 19.00622615223013,
      "learning_rate": 9.398014380024736e-06,
      "loss": 3.3896,
      "step": 2132
    },
    {
      "epoch": 0.7269938650306749,
      "grad_norm": 33.76365152296656,
      "learning_rate": 9.397070666505914e-06,
      "loss": 3.1952,
      "step": 2133
    },
    {
      "epoch": 0.7273346966598501,
      "grad_norm": 17.865703123998077,
      "learning_rate": 9.396126261310375e-06,
      "loss": 3.2917,
      "step": 2134
    },
    {
      "epoch": 0.7276755282890253,
      "grad_norm": 16.878853737888385,
      "learning_rate": 9.395181164586675e-06,
      "loss": 3.4298,
      "step": 2135
    },
    {
      "epoch": 0.7280163599182005,
      "grad_norm": 19.229647258773202,
      "learning_rate": 9.39423537648348e-06,
      "loss": 3.714,
      "step": 2136
    },
    {
      "epoch": 0.7283571915473755,
      "grad_norm": 11.629717725532092,
      "learning_rate": 9.393288897149568e-06,
      "loss": 2.9651,
      "step": 2137
    },
    {
      "epoch": 0.7286980231765507,
      "grad_norm": 14.842689281654332,
      "learning_rate": 9.392341726733825e-06,
      "loss": 3.2475,
      "step": 2138
    },
    {
      "epoch": 0.7290388548057259,
      "grad_norm": 11.597251862638945,
      "learning_rate": 9.391393865385241e-06,
      "loss": 2.4716,
      "step": 2139
    },
    {
      "epoch": 0.7293796864349011,
      "grad_norm": 14.773770711380942,
      "learning_rate": 9.390445313252923e-06,
      "loss": 2.6929,
      "step": 2140
    },
    {
      "epoch": 0.7297205180640763,
      "grad_norm": 13.070233071968197,
      "learning_rate": 9.38949607048608e-06,
      "loss": 2.8606,
      "step": 2141
    },
    {
      "epoch": 0.7300613496932515,
      "grad_norm": 18.715139905950707,
      "learning_rate": 9.388546137234031e-06,
      "loss": 3.9204,
      "step": 2142
    },
    {
      "epoch": 0.7304021813224267,
      "grad_norm": 30.91185338467825,
      "learning_rate": 9.387595513646203e-06,
      "loss": 3.5411,
      "step": 2143
    },
    {
      "epoch": 0.7307430129516019,
      "grad_norm": 22.35001476303035,
      "learning_rate": 9.386644199872136e-06,
      "loss": 3.7739,
      "step": 2144
    },
    {
      "epoch": 0.7310838445807771,
      "grad_norm": 11.360516617175431,
      "learning_rate": 9.385692196061475e-06,
      "loss": 2.8393,
      "step": 2145
    },
    {
      "epoch": 0.7314246762099523,
      "grad_norm": 18.10681642479518,
      "learning_rate": 9.384739502363974e-06,
      "loss": 3.9415,
      "step": 2146
    },
    {
      "epoch": 0.7317655078391274,
      "grad_norm": 19.724109530396593,
      "learning_rate": 9.383786118929494e-06,
      "loss": 2.909,
      "step": 2147
    },
    {
      "epoch": 0.7321063394683026,
      "grad_norm": 15.763733893898985,
      "learning_rate": 9.382832045908005e-06,
      "loss": 3.5376,
      "step": 2148
    },
    {
      "epoch": 0.7324471710974778,
      "grad_norm": 16.484036594375144,
      "learning_rate": 9.38187728344959e-06,
      "loss": 3.1311,
      "step": 2149
    },
    {
      "epoch": 0.732788002726653,
      "grad_norm": 13.664964196419726,
      "learning_rate": 9.380921831704432e-06,
      "loss": 2.818,
      "step": 2150
    },
    {
      "epoch": 0.7331288343558282,
      "grad_norm": 21.79716829037097,
      "learning_rate": 9.379965690822834e-06,
      "loss": 2.9985,
      "step": 2151
    },
    {
      "epoch": 0.7334696659850034,
      "grad_norm": 16.84405114087349,
      "learning_rate": 9.379008860955197e-06,
      "loss": 3.4575,
      "step": 2152
    },
    {
      "epoch": 0.7338104976141786,
      "grad_norm": 20.297002063945683,
      "learning_rate": 9.378051342252033e-06,
      "loss": 3.0723,
      "step": 2153
    },
    {
      "epoch": 0.7341513292433538,
      "grad_norm": 15.229877054371403,
      "learning_rate": 9.377093134863964e-06,
      "loss": 3.5116,
      "step": 2154
    },
    {
      "epoch": 0.734492160872529,
      "grad_norm": 20.102329987105268,
      "learning_rate": 9.376134238941721e-06,
      "loss": 3.7979,
      "step": 2155
    },
    {
      "epoch": 0.7348329925017042,
      "grad_norm": 18.069566736179308,
      "learning_rate": 9.375174654636144e-06,
      "loss": 3.0663,
      "step": 2156
    },
    {
      "epoch": 0.7351738241308794,
      "grad_norm": 14.29597169236135,
      "learning_rate": 9.374214382098174e-06,
      "loss": 2.9961,
      "step": 2157
    },
    {
      "epoch": 0.7355146557600545,
      "grad_norm": 14.728883341795942,
      "learning_rate": 9.373253421478872e-06,
      "loss": 3.4193,
      "step": 2158
    },
    {
      "epoch": 0.7358554873892297,
      "grad_norm": 17.730710005851144,
      "learning_rate": 9.372291772929396e-06,
      "loss": 3.4843,
      "step": 2159
    },
    {
      "epoch": 0.7361963190184049,
      "grad_norm": 18.437335202751587,
      "learning_rate": 9.37132943660102e-06,
      "loss": 3.595,
      "step": 2160
    },
    {
      "epoch": 0.7365371506475801,
      "grad_norm": 19.340814698089993,
      "learning_rate": 9.370366412645122e-06,
      "loss": 3.5238,
      "step": 2161
    },
    {
      "epoch": 0.7368779822767553,
      "grad_norm": 12.794623594348831,
      "learning_rate": 9.36940270121319e-06,
      "loss": 2.9988,
      "step": 2162
    },
    {
      "epoch": 0.7372188139059305,
      "grad_norm": 42.404861330438266,
      "learning_rate": 9.36843830245682e-06,
      "loss": 3.2262,
      "step": 2163
    },
    {
      "epoch": 0.7375596455351057,
      "grad_norm": 13.372116795211642,
      "learning_rate": 9.367473216527714e-06,
      "loss": 3.1717,
      "step": 2164
    },
    {
      "epoch": 0.7379004771642809,
      "grad_norm": 18.502127879020858,
      "learning_rate": 9.366507443577687e-06,
      "loss": 3.515,
      "step": 2165
    },
    {
      "epoch": 0.7382413087934561,
      "grad_norm": 14.354791922216167,
      "learning_rate": 9.365540983758654e-06,
      "loss": 3.5344,
      "step": 2166
    },
    {
      "epoch": 0.7385821404226313,
      "grad_norm": 21.054251619549888,
      "learning_rate": 9.364573837222648e-06,
      "loss": 3.5727,
      "step": 2167
    },
    {
      "epoch": 0.7389229720518065,
      "grad_norm": 16.538471482234247,
      "learning_rate": 9.363606004121804e-06,
      "loss": 3.5327,
      "step": 2168
    },
    {
      "epoch": 0.7392638036809815,
      "grad_norm": 10.214764307764867,
      "learning_rate": 9.362637484608363e-06,
      "loss": 2.9283,
      "step": 2169
    },
    {
      "epoch": 0.7396046353101567,
      "grad_norm": 18.343435111081916,
      "learning_rate": 9.36166827883468e-06,
      "loss": 3.6453,
      "step": 2170
    },
    {
      "epoch": 0.7399454669393319,
      "grad_norm": 19.740084996092282,
      "learning_rate": 9.360698386953215e-06,
      "loss": 3.9665,
      "step": 2171
    },
    {
      "epoch": 0.7402862985685071,
      "grad_norm": 18.9907173073652,
      "learning_rate": 9.359727809116534e-06,
      "loss": 3.9118,
      "step": 2172
    },
    {
      "epoch": 0.7406271301976823,
      "grad_norm": 20.877752836602724,
      "learning_rate": 9.358756545477313e-06,
      "loss": 3.0459,
      "step": 2173
    },
    {
      "epoch": 0.7409679618268575,
      "grad_norm": 12.788947158766721,
      "learning_rate": 9.357784596188336e-06,
      "loss": 3.0421,
      "step": 2174
    },
    {
      "epoch": 0.7413087934560327,
      "grad_norm": 15.545949472715442,
      "learning_rate": 9.356811961402495e-06,
      "loss": 3.0669,
      "step": 2175
    },
    {
      "epoch": 0.7416496250852079,
      "grad_norm": 17.753827684355173,
      "learning_rate": 9.35583864127279e-06,
      "loss": 3.555,
      "step": 2176
    },
    {
      "epoch": 0.7419904567143831,
      "grad_norm": 18.625258748317222,
      "learning_rate": 9.354864635952326e-06,
      "loss": 3.8347,
      "step": 2177
    },
    {
      "epoch": 0.7423312883435583,
      "grad_norm": 17.975737396510112,
      "learning_rate": 9.35388994559432e-06,
      "loss": 3.7061,
      "step": 2178
    },
    {
      "epoch": 0.7426721199727335,
      "grad_norm": 12.675949702293165,
      "learning_rate": 9.352914570352091e-06,
      "loss": 2.6705,
      "step": 2179
    },
    {
      "epoch": 0.7430129516019086,
      "grad_norm": 19.815875738318564,
      "learning_rate": 9.351938510379076e-06,
      "loss": 3.6813,
      "step": 2180
    },
    {
      "epoch": 0.7433537832310838,
      "grad_norm": 16.588717999824787,
      "learning_rate": 9.350961765828807e-06,
      "loss": 3.6362,
      "step": 2181
    },
    {
      "epoch": 0.743694614860259,
      "grad_norm": 26.16639037155661,
      "learning_rate": 9.349984336854933e-06,
      "loss": 3.2156,
      "step": 2182
    },
    {
      "epoch": 0.7440354464894342,
      "grad_norm": 15.851113494129976,
      "learning_rate": 9.349006223611205e-06,
      "loss": 3.4348,
      "step": 2183
    },
    {
      "epoch": 0.7443762781186094,
      "grad_norm": 20.597915214000203,
      "learning_rate": 9.348027426251487e-06,
      "loss": 3.2739,
      "step": 2184
    },
    {
      "epoch": 0.7447171097477846,
      "grad_norm": 13.405937403467084,
      "learning_rate": 9.347047944929746e-06,
      "loss": 3.2374,
      "step": 2185
    },
    {
      "epoch": 0.7450579413769598,
      "grad_norm": 22.159732733387383,
      "learning_rate": 9.34606777980006e-06,
      "loss": 3.0936,
      "step": 2186
    },
    {
      "epoch": 0.745398773006135,
      "grad_norm": 16.58468274811247,
      "learning_rate": 9.34508693101661e-06,
      "loss": 3.3175,
      "step": 2187
    },
    {
      "epoch": 0.7457396046353102,
      "grad_norm": 17.542675982933208,
      "learning_rate": 9.34410539873369e-06,
      "loss": 2.8963,
      "step": 2188
    },
    {
      "epoch": 0.7460804362644854,
      "grad_norm": 34.37201658514218,
      "learning_rate": 9.343123183105696e-06,
      "loss": 3.0524,
      "step": 2189
    },
    {
      "epoch": 0.7464212678936605,
      "grad_norm": 17.002253441097135,
      "learning_rate": 9.342140284287138e-06,
      "loss": 3.3892,
      "step": 2190
    },
    {
      "epoch": 0.7467620995228357,
      "grad_norm": 17.641226622567473,
      "learning_rate": 9.341156702432628e-06,
      "loss": 3.2909,
      "step": 2191
    },
    {
      "epoch": 0.7471029311520109,
      "grad_norm": 12.917285103621877,
      "learning_rate": 9.340172437696887e-06,
      "loss": 3.1064,
      "step": 2192
    },
    {
      "epoch": 0.7474437627811861,
      "grad_norm": 18.73676190115936,
      "learning_rate": 9.339187490234745e-06,
      "loss": 3.3704,
      "step": 2193
    },
    {
      "epoch": 0.7477845944103613,
      "grad_norm": 17.494518310140794,
      "learning_rate": 9.338201860201136e-06,
      "loss": 4.096,
      "step": 2194
    },
    {
      "epoch": 0.7481254260395365,
      "grad_norm": 19.215726206872105,
      "learning_rate": 9.337215547751106e-06,
      "loss": 3.9718,
      "step": 2195
    },
    {
      "epoch": 0.7484662576687117,
      "grad_norm": 17.81620884597207,
      "learning_rate": 9.336228553039806e-06,
      "loss": 3.4317,
      "step": 2196
    },
    {
      "epoch": 0.7488070892978869,
      "grad_norm": 22.984282987381416,
      "learning_rate": 9.335240876222491e-06,
      "loss": 3.8808,
      "step": 2197
    },
    {
      "epoch": 0.7491479209270621,
      "grad_norm": 17.258399298271794,
      "learning_rate": 9.33425251745453e-06,
      "loss": 3.6919,
      "step": 2198
    },
    {
      "epoch": 0.7494887525562373,
      "grad_norm": 20.928123680354293,
      "learning_rate": 9.333263476891394e-06,
      "loss": 3.5518,
      "step": 2199
    },
    {
      "epoch": 0.7498295841854125,
      "grad_norm": 11.587359875322278,
      "learning_rate": 9.332273754688663e-06,
      "loss": 2.8778,
      "step": 2200
    },
    {
      "epoch": 0.7501704158145875,
      "grad_norm": 15.197307259098533,
      "learning_rate": 9.331283351002025e-06,
      "loss": 3.2932,
      "step": 2201
    },
    {
      "epoch": 0.7505112474437627,
      "grad_norm": 19.404660170563698,
      "learning_rate": 9.330292265987272e-06,
      "loss": 3.4958,
      "step": 2202
    },
    {
      "epoch": 0.7508520790729379,
      "grad_norm": 17.634293740189626,
      "learning_rate": 9.32930049980031e-06,
      "loss": 3.647,
      "step": 2203
    },
    {
      "epoch": 0.7511929107021131,
      "grad_norm": 13.922820779239487,
      "learning_rate": 9.328308052597146e-06,
      "loss": 2.8803,
      "step": 2204
    },
    {
      "epoch": 0.7515337423312883,
      "grad_norm": 17.900230465099092,
      "learning_rate": 9.327314924533895e-06,
      "loss": 3.6653,
      "step": 2205
    },
    {
      "epoch": 0.7518745739604635,
      "grad_norm": 19.71394320222073,
      "learning_rate": 9.32632111576678e-06,
      "loss": 3.3984,
      "step": 2206
    },
    {
      "epoch": 0.7522154055896387,
      "grad_norm": 15.041103905115955,
      "learning_rate": 9.325326626452132e-06,
      "loss": 3.8555,
      "step": 2207
    },
    {
      "epoch": 0.7525562372188139,
      "grad_norm": 22.596771660900377,
      "learning_rate": 9.324331456746386e-06,
      "loss": 4.0586,
      "step": 2208
    },
    {
      "epoch": 0.7528970688479891,
      "grad_norm": 12.474683116156665,
      "learning_rate": 9.323335606806087e-06,
      "loss": 3.2313,
      "step": 2209
    },
    {
      "epoch": 0.7532379004771643,
      "grad_norm": 12.748036023873059,
      "learning_rate": 9.32233907678789e-06,
      "loss": 3.0374,
      "step": 2210
    },
    {
      "epoch": 0.7535787321063395,
      "grad_norm": 20.779819095959084,
      "learning_rate": 9.321341866848548e-06,
      "loss": 3.5759,
      "step": 2211
    },
    {
      "epoch": 0.7539195637355146,
      "grad_norm": 16.760112147520704,
      "learning_rate": 9.320343977144928e-06,
      "loss": 3.6889,
      "step": 2212
    },
    {
      "epoch": 0.7542603953646898,
      "grad_norm": 19.769407717086892,
      "learning_rate": 9.319345407834001e-06,
      "loss": 3.4033,
      "step": 2213
    },
    {
      "epoch": 0.754601226993865,
      "grad_norm": 19.05678859880588,
      "learning_rate": 9.318346159072848e-06,
      "loss": 3.6113,
      "step": 2214
    },
    {
      "epoch": 0.7549420586230402,
      "grad_norm": 15.115377524683467,
      "learning_rate": 9.317346231018655e-06,
      "loss": 3.6396,
      "step": 2215
    },
    {
      "epoch": 0.7552828902522154,
      "grad_norm": 24.795231138669546,
      "learning_rate": 9.316345623828713e-06,
      "loss": 3.5372,
      "step": 2216
    },
    {
      "epoch": 0.7556237218813906,
      "grad_norm": 15.181193246515999,
      "learning_rate": 9.315344337660422e-06,
      "loss": 3.4035,
      "step": 2217
    },
    {
      "epoch": 0.7559645535105658,
      "grad_norm": 14.485168184672116,
      "learning_rate": 9.314342372671288e-06,
      "loss": 2.999,
      "step": 2218
    },
    {
      "epoch": 0.756305385139741,
      "grad_norm": 12.444307070806353,
      "learning_rate": 9.313339729018924e-06,
      "loss": 3.0643,
      "step": 2219
    },
    {
      "epoch": 0.7566462167689162,
      "grad_norm": 15.109182172426845,
      "learning_rate": 9.31233640686105e-06,
      "loss": 3.2432,
      "step": 2220
    },
    {
      "epoch": 0.7569870483980914,
      "grad_norm": 17.028611250437553,
      "learning_rate": 9.311332406355494e-06,
      "loss": 3.0913,
      "step": 2221
    },
    {
      "epoch": 0.7573278800272665,
      "grad_norm": 23.970153837870257,
      "learning_rate": 9.310327727660186e-06,
      "loss": 3.0553,
      "step": 2222
    },
    {
      "epoch": 0.7576687116564417,
      "grad_norm": 17.589589454302235,
      "learning_rate": 9.30932237093317e-06,
      "loss": 3.7805,
      "step": 2223
    },
    {
      "epoch": 0.7580095432856169,
      "grad_norm": 18.2585801475019,
      "learning_rate": 9.30831633633259e-06,
      "loss": 3.5395,
      "step": 2224
    },
    {
      "epoch": 0.7583503749147921,
      "grad_norm": 15.04097606023296,
      "learning_rate": 9.3073096240167e-06,
      "loss": 3.2287,
      "step": 2225
    },
    {
      "epoch": 0.7586912065439673,
      "grad_norm": 18.18052016031695,
      "learning_rate": 9.30630223414386e-06,
      "loss": 3.7671,
      "step": 2226
    },
    {
      "epoch": 0.7590320381731425,
      "grad_norm": 24.405177831804576,
      "learning_rate": 9.305294166872535e-06,
      "loss": 3.2137,
      "step": 2227
    },
    {
      "epoch": 0.7593728698023177,
      "grad_norm": 19.793330052440826,
      "learning_rate": 9.304285422361301e-06,
      "loss": 3.4576,
      "step": 2228
    },
    {
      "epoch": 0.7597137014314929,
      "grad_norm": 32.18943112293496,
      "learning_rate": 9.303276000768835e-06,
      "loss": 2.5832,
      "step": 2229
    },
    {
      "epoch": 0.7600545330606681,
      "grad_norm": 17.002338113604342,
      "learning_rate": 9.302265902253924e-06,
      "loss": 3.4641,
      "step": 2230
    },
    {
      "epoch": 0.7603953646898433,
      "grad_norm": 16.389134179340285,
      "learning_rate": 9.301255126975459e-06,
      "loss": 2.9225,
      "step": 2231
    },
    {
      "epoch": 0.7607361963190185,
      "grad_norm": 9.962315650258061,
      "learning_rate": 9.30024367509244e-06,
      "loss": 2.5274,
      "step": 2232
    },
    {
      "epoch": 0.7610770279481935,
      "grad_norm": 20.34876926769682,
      "learning_rate": 9.299231546763974e-06,
      "loss": 3.3728,
      "step": 2233
    },
    {
      "epoch": 0.7614178595773687,
      "grad_norm": 18.919413738017173,
      "learning_rate": 9.298218742149272e-06,
      "loss": 3.5811,
      "step": 2234
    },
    {
      "epoch": 0.7617586912065439,
      "grad_norm": 13.273083493935724,
      "learning_rate": 9.29720526140765e-06,
      "loss": 3.3602,
      "step": 2235
    },
    {
      "epoch": 0.7620995228357191,
      "grad_norm": 20.80465445159649,
      "learning_rate": 9.296191104698535e-06,
      "loss": 3.0763,
      "step": 2236
    },
    {
      "epoch": 0.7624403544648943,
      "grad_norm": 12.382199728235753,
      "learning_rate": 9.295176272181457e-06,
      "loss": 2.779,
      "step": 2237
    },
    {
      "epoch": 0.7627811860940695,
      "grad_norm": 33.94568656681278,
      "learning_rate": 9.294160764016053e-06,
      "loss": 3.6467,
      "step": 2238
    },
    {
      "epoch": 0.7631220177232447,
      "grad_norm": 14.433892941231413,
      "learning_rate": 9.293144580362067e-06,
      "loss": 2.5348,
      "step": 2239
    },
    {
      "epoch": 0.7634628493524199,
      "grad_norm": 18.577871939260245,
      "learning_rate": 9.292127721379348e-06,
      "loss": 3.5378,
      "step": 2240
    },
    {
      "epoch": 0.7638036809815951,
      "grad_norm": 19.160629364415474,
      "learning_rate": 9.291110187227853e-06,
      "loss": 3.8237,
      "step": 2241
    },
    {
      "epoch": 0.7641445126107703,
      "grad_norm": 17.76211823134256,
      "learning_rate": 9.290091978067643e-06,
      "loss": 3.7033,
      "step": 2242
    },
    {
      "epoch": 0.7644853442399455,
      "grad_norm": 20.019362488023077,
      "learning_rate": 9.289073094058889e-06,
      "loss": 3.0504,
      "step": 2243
    },
    {
      "epoch": 0.7648261758691206,
      "grad_norm": 16.212936357246484,
      "learning_rate": 9.28805353536186e-06,
      "loss": 3.5597,
      "step": 2244
    },
    {
      "epoch": 0.7651670074982958,
      "grad_norm": 19.19924710019374,
      "learning_rate": 9.287033302136943e-06,
      "loss": 3.4762,
      "step": 2245
    },
    {
      "epoch": 0.765507839127471,
      "grad_norm": 17.79562431377502,
      "learning_rate": 9.286012394544621e-06,
      "loss": 3.2457,
      "step": 2246
    },
    {
      "epoch": 0.7658486707566462,
      "grad_norm": 13.969697135843267,
      "learning_rate": 9.284990812745488e-06,
      "loss": 3.4271,
      "step": 2247
    },
    {
      "epoch": 0.7661895023858214,
      "grad_norm": 26.857724555096212,
      "learning_rate": 9.283968556900243e-06,
      "loss": 4.2087,
      "step": 2248
    },
    {
      "epoch": 0.7665303340149966,
      "grad_norm": 19.95592626278951,
      "learning_rate": 9.282945627169688e-06,
      "loss": 3.3688,
      "step": 2249
    },
    {
      "epoch": 0.7668711656441718,
      "grad_norm": 19.80020052981005,
      "learning_rate": 9.281922023714739e-06,
      "loss": 3.4484,
      "step": 2250
    },
    {
      "epoch": 0.767211997273347,
      "grad_norm": 13.9565209775081,
      "learning_rate": 9.280897746696409e-06,
      "loss": 2.9656,
      "step": 2251
    },
    {
      "epoch": 0.7675528289025222,
      "grad_norm": 15.959063701628548,
      "learning_rate": 9.279872796275823e-06,
      "loss": 3.312,
      "step": 2252
    },
    {
      "epoch": 0.7678936605316974,
      "grad_norm": 16.45381165903346,
      "learning_rate": 9.278847172614209e-06,
      "loss": 2.5325,
      "step": 2253
    },
    {
      "epoch": 0.7682344921608726,
      "grad_norm": 17.079219372645227,
      "learning_rate": 9.277820875872902e-06,
      "loss": 3.3,
      "step": 2254
    },
    {
      "epoch": 0.7685753237900477,
      "grad_norm": 20.111078753324165,
      "learning_rate": 9.276793906213342e-06,
      "loss": 3.2207,
      "step": 2255
    },
    {
      "epoch": 0.7689161554192229,
      "grad_norm": 20.44805689414811,
      "learning_rate": 9.275766263797075e-06,
      "loss": 2.8747,
      "step": 2256
    },
    {
      "epoch": 0.7692569870483981,
      "grad_norm": 16.534426584559167,
      "learning_rate": 9.274737948785752e-06,
      "loss": 3.3387,
      "step": 2257
    },
    {
      "epoch": 0.7695978186775733,
      "grad_norm": 11.455323245489279,
      "learning_rate": 9.273708961341137e-06,
      "loss": 2.8507,
      "step": 2258
    },
    {
      "epoch": 0.7699386503067485,
      "grad_norm": 14.9687767387331,
      "learning_rate": 9.272679301625088e-06,
      "loss": 2.9025,
      "step": 2259
    },
    {
      "epoch": 0.7702794819359237,
      "grad_norm": 17.259326533473462,
      "learning_rate": 9.271648969799574e-06,
      "loss": 2.9936,
      "step": 2260
    },
    {
      "epoch": 0.7706203135650989,
      "grad_norm": 20.58388894259241,
      "learning_rate": 9.270617966026676e-06,
      "loss": 3.06,
      "step": 2261
    },
    {
      "epoch": 0.7709611451942741,
      "grad_norm": 33.62407169145876,
      "learning_rate": 9.26958629046857e-06,
      "loss": 3.6448,
      "step": 2262
    },
    {
      "epoch": 0.7713019768234493,
      "grad_norm": 14.647001225837155,
      "learning_rate": 9.268553943287543e-06,
      "loss": 3.7968,
      "step": 2263
    },
    {
      "epoch": 0.7716428084526245,
      "grad_norm": 13.280144277351141,
      "learning_rate": 9.26752092464599e-06,
      "loss": 2.9437,
      "step": 2264
    },
    {
      "epoch": 0.7719836400817995,
      "grad_norm": 32.682413360006294,
      "learning_rate": 9.266487234706409e-06,
      "loss": 3.3698,
      "step": 2265
    },
    {
      "epoch": 0.7723244717109747,
      "grad_norm": 19.177515467381305,
      "learning_rate": 9.2654528736314e-06,
      "loss": 2.7436,
      "step": 2266
    },
    {
      "epoch": 0.7726653033401499,
      "grad_norm": 18.988050292988977,
      "learning_rate": 9.264417841583675e-06,
      "loss": 3.0641,
      "step": 2267
    },
    {
      "epoch": 0.7730061349693251,
      "grad_norm": 18.88222654804129,
      "learning_rate": 9.263382138726047e-06,
      "loss": 3.5783,
      "step": 2268
    },
    {
      "epoch": 0.7733469665985003,
      "grad_norm": 13.992875934243239,
      "learning_rate": 9.262345765221439e-06,
      "loss": 2.7541,
      "step": 2269
    },
    {
      "epoch": 0.7736877982276755,
      "grad_norm": 18.625825705576826,
      "learning_rate": 9.261308721232872e-06,
      "loss": 2.7369,
      "step": 2270
    },
    {
      "epoch": 0.7740286298568507,
      "grad_norm": 13.071364974931148,
      "learning_rate": 9.260271006923482e-06,
      "loss": 3.6027,
      "step": 2271
    },
    {
      "epoch": 0.7743694614860259,
      "grad_norm": 13.557491615555827,
      "learning_rate": 9.259232622456502e-06,
      "loss": 3.0037,
      "step": 2272
    },
    {
      "epoch": 0.7747102931152011,
      "grad_norm": 12.792879055578162,
      "learning_rate": 9.258193567995278e-06,
      "loss": 2.8715,
      "step": 2273
    },
    {
      "epoch": 0.7750511247443763,
      "grad_norm": 16.121132129924348,
      "learning_rate": 9.257153843703252e-06,
      "loss": 3.534,
      "step": 2274
    },
    {
      "epoch": 0.7753919563735515,
      "grad_norm": 29.461863476616383,
      "learning_rate": 9.256113449743981e-06,
      "loss": 3.0347,
      "step": 2275
    },
    {
      "epoch": 0.7757327880027266,
      "grad_norm": 20.16592517956135,
      "learning_rate": 9.255072386281121e-06,
      "loss": 3.4961,
      "step": 2276
    },
    {
      "epoch": 0.7760736196319018,
      "grad_norm": 22.014447477498507,
      "learning_rate": 9.254030653478436e-06,
      "loss": 3.1273,
      "step": 2277
    },
    {
      "epoch": 0.776414451261077,
      "grad_norm": 11.548066637272779,
      "learning_rate": 9.252988251499796e-06,
      "loss": 3.2111,
      "step": 2278
    },
    {
      "epoch": 0.7767552828902522,
      "grad_norm": 19.331526231968937,
      "learning_rate": 9.251945180509174e-06,
      "loss": 3.5842,
      "step": 2279
    },
    {
      "epoch": 0.7770961145194274,
      "grad_norm": 20.734471308717293,
      "learning_rate": 9.250901440670648e-06,
      "loss": 3.4387,
      "step": 2280
    },
    {
      "epoch": 0.7774369461486026,
      "grad_norm": 17.316072615428816,
      "learning_rate": 9.249857032148404e-06,
      "loss": 3.248,
      "step": 2281
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 13.855899764125656,
      "learning_rate": 9.248811955106732e-06,
      "loss": 3.5959,
      "step": 2282
    },
    {
      "epoch": 0.778118609406953,
      "grad_norm": 13.943943066000202,
      "learning_rate": 9.247766209710025e-06,
      "loss": 3.2347,
      "step": 2283
    },
    {
      "epoch": 0.7784594410361282,
      "grad_norm": 17.094030818433183,
      "learning_rate": 9.246719796122784e-06,
      "loss": 2.584,
      "step": 2284
    },
    {
      "epoch": 0.7788002726653034,
      "grad_norm": 22.135702305649247,
      "learning_rate": 9.245672714509613e-06,
      "loss": 3.5853,
      "step": 2285
    },
    {
      "epoch": 0.7791411042944786,
      "grad_norm": 13.344222546600374,
      "learning_rate": 9.244624965035223e-06,
      "loss": 3.4417,
      "step": 2286
    },
    {
      "epoch": 0.7794819359236537,
      "grad_norm": 17.669070849785857,
      "learning_rate": 9.24357654786443e-06,
      "loss": 3.3253,
      "step": 2287
    },
    {
      "epoch": 0.7798227675528289,
      "grad_norm": 14.474884755513854,
      "learning_rate": 9.242527463162155e-06,
      "loss": 3.34,
      "step": 2288
    },
    {
      "epoch": 0.7801635991820041,
      "grad_norm": 15.975704191024015,
      "learning_rate": 9.241477711093419e-06,
      "loss": 3.0057,
      "step": 2289
    },
    {
      "epoch": 0.7805044308111793,
      "grad_norm": 16.407186555053222,
      "learning_rate": 9.240427291823355e-06,
      "loss": 3.1972,
      "step": 2290
    },
    {
      "epoch": 0.7808452624403545,
      "grad_norm": 21.231020638901168,
      "learning_rate": 9.239376205517197e-06,
      "loss": 2.9554,
      "step": 2291
    },
    {
      "epoch": 0.7811860940695297,
      "grad_norm": 36.65695877619637,
      "learning_rate": 9.238324452340288e-06,
      "loss": 3.3557,
      "step": 2292
    },
    {
      "epoch": 0.7815269256987049,
      "grad_norm": 13.171241082042586,
      "learning_rate": 9.23727203245807e-06,
      "loss": 3.4168,
      "step": 2293
    },
    {
      "epoch": 0.78186775732788,
      "grad_norm": 13.201834507029362,
      "learning_rate": 9.236218946036093e-06,
      "loss": 2.9598,
      "step": 2294
    },
    {
      "epoch": 0.7822085889570553,
      "grad_norm": 27.226213326908056,
      "learning_rate": 9.235165193240015e-06,
      "loss": 3.6304,
      "step": 2295
    },
    {
      "epoch": 0.7825494205862304,
      "grad_norm": 17.431740106405698,
      "learning_rate": 9.23411077423559e-06,
      "loss": 3.5265,
      "step": 2296
    },
    {
      "epoch": 0.7828902522154055,
      "grad_norm": 16.44150270163108,
      "learning_rate": 9.233055689188686e-06,
      "loss": 3.423,
      "step": 2297
    },
    {
      "epoch": 0.7832310838445807,
      "grad_norm": 13.368735408602225,
      "learning_rate": 9.23199993826527e-06,
      "loss": 2.9745,
      "step": 2298
    },
    {
      "epoch": 0.7835719154737559,
      "grad_norm": 18.170126862758586,
      "learning_rate": 9.23094352163142e-06,
      "loss": 3.5841,
      "step": 2299
    },
    {
      "epoch": 0.7839127471029311,
      "grad_norm": 15.156829936723492,
      "learning_rate": 9.22988643945331e-06,
      "loss": 3.5537,
      "step": 2300
    },
    {
      "epoch": 0.7842535787321063,
      "grad_norm": 13.324848579705975,
      "learning_rate": 9.228828691897225e-06,
      "loss": 3.2456,
      "step": 2301
    },
    {
      "epoch": 0.7845944103612815,
      "grad_norm": 18.963853485920126,
      "learning_rate": 9.227770279129553e-06,
      "loss": 2.8456,
      "step": 2302
    },
    {
      "epoch": 0.7849352419904567,
      "grad_norm": 17.291228983034287,
      "learning_rate": 9.226711201316785e-06,
      "loss": 3.1616,
      "step": 2303
    },
    {
      "epoch": 0.7852760736196319,
      "grad_norm": 15.466287070573062,
      "learning_rate": 9.225651458625521e-06,
      "loss": 3.4873,
      "step": 2304
    },
    {
      "epoch": 0.7856169052488071,
      "grad_norm": 22.590751333665924,
      "learning_rate": 9.224591051222461e-06,
      "loss": 3.6783,
      "step": 2305
    },
    {
      "epoch": 0.7859577368779823,
      "grad_norm": 18.223262321042696,
      "learning_rate": 9.223529979274411e-06,
      "loss": 3.6829,
      "step": 2306
    },
    {
      "epoch": 0.7862985685071575,
      "grad_norm": 15.83825828874748,
      "learning_rate": 9.222468242948283e-06,
      "loss": 3.5601,
      "step": 2307
    },
    {
      "epoch": 0.7866394001363326,
      "grad_norm": 29.841818000501487,
      "learning_rate": 9.221405842411091e-06,
      "loss": 3.3108,
      "step": 2308
    },
    {
      "epoch": 0.7869802317655078,
      "grad_norm": 17.35144049008841,
      "learning_rate": 9.220342777829956e-06,
      "loss": 3.6368,
      "step": 2309
    },
    {
      "epoch": 0.787321063394683,
      "grad_norm": 27.032709484313855,
      "learning_rate": 9.219279049372101e-06,
      "loss": 3.2408,
      "step": 2310
    },
    {
      "epoch": 0.7876618950238582,
      "grad_norm": 15.38084381091063,
      "learning_rate": 9.218214657204857e-06,
      "loss": 2.9746,
      "step": 2311
    },
    {
      "epoch": 0.7880027266530334,
      "grad_norm": 16.244742832934143,
      "learning_rate": 9.217149601495654e-06,
      "loss": 3.0045,
      "step": 2312
    },
    {
      "epoch": 0.7883435582822086,
      "grad_norm": 16.832014690142568,
      "learning_rate": 9.216083882412033e-06,
      "loss": 3.2433,
      "step": 2313
    },
    {
      "epoch": 0.7886843899113838,
      "grad_norm": 14.8472917092808,
      "learning_rate": 9.215017500121632e-06,
      "loss": 3.5616,
      "step": 2314
    },
    {
      "epoch": 0.789025221540559,
      "grad_norm": 17.118380770971644,
      "learning_rate": 9.213950454792201e-06,
      "loss": 3.2948,
      "step": 2315
    },
    {
      "epoch": 0.7893660531697342,
      "grad_norm": 17.530113356950764,
      "learning_rate": 9.212882746591588e-06,
      "loss": 3.4175,
      "step": 2316
    },
    {
      "epoch": 0.7897068847989094,
      "grad_norm": 14.021223425700859,
      "learning_rate": 9.211814375687749e-06,
      "loss": 3.6983,
      "step": 2317
    },
    {
      "epoch": 0.7900477164280846,
      "grad_norm": 16.93241062662418,
      "learning_rate": 9.21074534224874e-06,
      "loss": 2.6702,
      "step": 2318
    },
    {
      "epoch": 0.7903885480572597,
      "grad_norm": 22.78694448352502,
      "learning_rate": 9.20967564644273e-06,
      "loss": 3.2993,
      "step": 2319
    },
    {
      "epoch": 0.7907293796864349,
      "grad_norm": 18.374178686965557,
      "learning_rate": 9.208605288437982e-06,
      "loss": 2.9033,
      "step": 2320
    },
    {
      "epoch": 0.7910702113156101,
      "grad_norm": 25.398005580063025,
      "learning_rate": 9.20753426840287e-06,
      "loss": 3.1823,
      "step": 2321
    },
    {
      "epoch": 0.7914110429447853,
      "grad_norm": 16.978090493951253,
      "learning_rate": 9.206462586505865e-06,
      "loss": 2.8169,
      "step": 2322
    },
    {
      "epoch": 0.7917518745739605,
      "grad_norm": 13.89558506091015,
      "learning_rate": 9.205390242915554e-06,
      "loss": 3.4932,
      "step": 2323
    },
    {
      "epoch": 0.7920927062031357,
      "grad_norm": 21.222777654744895,
      "learning_rate": 9.204317237800613e-06,
      "loss": 2.5054,
      "step": 2324
    },
    {
      "epoch": 0.7924335378323109,
      "grad_norm": 16.416770309343,
      "learning_rate": 9.203243571329836e-06,
      "loss": 4.0085,
      "step": 2325
    },
    {
      "epoch": 0.792774369461486,
      "grad_norm": 21.618829778549372,
      "learning_rate": 9.202169243672114e-06,
      "loss": 3.9931,
      "step": 2326
    },
    {
      "epoch": 0.7931152010906612,
      "grad_norm": 20.20824907964005,
      "learning_rate": 9.20109425499644e-06,
      "loss": 3.1761,
      "step": 2327
    },
    {
      "epoch": 0.7934560327198364,
      "grad_norm": 15.232954643706833,
      "learning_rate": 9.200018605471918e-06,
      "loss": 3.3236,
      "step": 2328
    },
    {
      "epoch": 0.7937968643490115,
      "grad_norm": 18.00184780641661,
      "learning_rate": 9.19894229526775e-06,
      "loss": 3.3332,
      "step": 2329
    },
    {
      "epoch": 0.7941376959781867,
      "grad_norm": 17.81055633299913,
      "learning_rate": 9.197865324553243e-06,
      "loss": 3.7239,
      "step": 2330
    },
    {
      "epoch": 0.7944785276073619,
      "grad_norm": 12.970887142862471,
      "learning_rate": 9.19678769349781e-06,
      "loss": 3.076,
      "step": 2331
    },
    {
      "epoch": 0.7948193592365371,
      "grad_norm": 32.018652840320236,
      "learning_rate": 9.195709402270967e-06,
      "loss": 2.8213,
      "step": 2332
    },
    {
      "epoch": 0.7951601908657123,
      "grad_norm": 12.105424253229685,
      "learning_rate": 9.194630451042331e-06,
      "loss": 3.1369,
      "step": 2333
    },
    {
      "epoch": 0.7955010224948875,
      "grad_norm": 15.309590307790572,
      "learning_rate": 9.19355083998163e-06,
      "loss": 3.6113,
      "step": 2334
    },
    {
      "epoch": 0.7958418541240627,
      "grad_norm": 18.470453927369803,
      "learning_rate": 9.192470569258684e-06,
      "loss": 3.7743,
      "step": 2335
    },
    {
      "epoch": 0.7961826857532379,
      "grad_norm": 34.254852433650235,
      "learning_rate": 9.191389639043432e-06,
      "loss": 3.6329,
      "step": 2336
    },
    {
      "epoch": 0.7965235173824131,
      "grad_norm": 15.13078254657666,
      "learning_rate": 9.190308049505905e-06,
      "loss": 3.0793,
      "step": 2337
    },
    {
      "epoch": 0.7968643490115883,
      "grad_norm": 16.778827493546654,
      "learning_rate": 9.18922580081624e-06,
      "loss": 3.257,
      "step": 2338
    },
    {
      "epoch": 0.7972051806407635,
      "grad_norm": 14.983664704644697,
      "learning_rate": 9.18814289314468e-06,
      "loss": 3.1499,
      "step": 2339
    },
    {
      "epoch": 0.7975460122699386,
      "grad_norm": 15.44211616455542,
      "learning_rate": 9.18705932666157e-06,
      "loss": 3.555,
      "step": 2340
    },
    {
      "epoch": 0.7978868438991138,
      "grad_norm": 14.747541869841813,
      "learning_rate": 9.185975101537361e-06,
      "loss": 3.5558,
      "step": 2341
    },
    {
      "epoch": 0.798227675528289,
      "grad_norm": 15.28420468166412,
      "learning_rate": 9.184890217942606e-06,
      "loss": 2.8295,
      "step": 2342
    },
    {
      "epoch": 0.7985685071574642,
      "grad_norm": 24.004605688689164,
      "learning_rate": 9.18380467604796e-06,
      "loss": 3.6598,
      "step": 2343
    },
    {
      "epoch": 0.7989093387866394,
      "grad_norm": 17.500257941226682,
      "learning_rate": 9.182718476024184e-06,
      "loss": 3.4082,
      "step": 2344
    },
    {
      "epoch": 0.7992501704158146,
      "grad_norm": 22.903309079374456,
      "learning_rate": 9.18163161804214e-06,
      "loss": 2.8414,
      "step": 2345
    },
    {
      "epoch": 0.7995910020449898,
      "grad_norm": 33.21018941675157,
      "learning_rate": 9.180544102272798e-06,
      "loss": 3.4036,
      "step": 2346
    },
    {
      "epoch": 0.799931833674165,
      "grad_norm": 20.71974394900947,
      "learning_rate": 9.179455928887226e-06,
      "loss": 3.6448,
      "step": 2347
    },
    {
      "epoch": 0.8002726653033402,
      "grad_norm": 40.499421515147944,
      "learning_rate": 9.1783670980566e-06,
      "loss": 3.4455,
      "step": 2348
    },
    {
      "epoch": 0.8006134969325154,
      "grad_norm": 14.967676188811344,
      "learning_rate": 9.177277609952196e-06,
      "loss": 2.917,
      "step": 2349
    },
    {
      "epoch": 0.8009543285616906,
      "grad_norm": 25.388501724176354,
      "learning_rate": 9.176187464745394e-06,
      "loss": 3.4399,
      "step": 2350
    },
    {
      "epoch": 0.8012951601908657,
      "grad_norm": 20.323614861389242,
      "learning_rate": 9.17509666260768e-06,
      "loss": 3.5136,
      "step": 2351
    },
    {
      "epoch": 0.8016359918200409,
      "grad_norm": 20.50972572332781,
      "learning_rate": 9.17400520371064e-06,
      "loss": 3.8936,
      "step": 2352
    },
    {
      "epoch": 0.8019768234492161,
      "grad_norm": 17.535716451912997,
      "learning_rate": 9.172913088225968e-06,
      "loss": 3.0149,
      "step": 2353
    },
    {
      "epoch": 0.8023176550783913,
      "grad_norm": 13.78351743644992,
      "learning_rate": 9.171820316325453e-06,
      "loss": 3.3189,
      "step": 2354
    },
    {
      "epoch": 0.8026584867075665,
      "grad_norm": 15.608262935134798,
      "learning_rate": 9.170726888180996e-06,
      "loss": 3.9005,
      "step": 2355
    },
    {
      "epoch": 0.8029993183367417,
      "grad_norm": 18.30825964047596,
      "learning_rate": 9.169632803964599e-06,
      "loss": 3.4273,
      "step": 2356
    },
    {
      "epoch": 0.8033401499659168,
      "grad_norm": 22.94558890042387,
      "learning_rate": 9.168538063848362e-06,
      "loss": 3.6191,
      "step": 2357
    },
    {
      "epoch": 0.803680981595092,
      "grad_norm": 14.26076532090428,
      "learning_rate": 9.167442668004493e-06,
      "loss": 2.7997,
      "step": 2358
    },
    {
      "epoch": 0.8040218132242672,
      "grad_norm": 22.582267832754667,
      "learning_rate": 9.166346616605304e-06,
      "loss": 3.5801,
      "step": 2359
    },
    {
      "epoch": 0.8043626448534424,
      "grad_norm": 19.14479656871101,
      "learning_rate": 9.165249909823206e-06,
      "loss": 3.2874,
      "step": 2360
    },
    {
      "epoch": 0.8047034764826176,
      "grad_norm": 18.825916450723792,
      "learning_rate": 9.164152547830715e-06,
      "loss": 3.5138,
      "step": 2361
    },
    {
      "epoch": 0.8050443081117927,
      "grad_norm": 23.888060596199878,
      "learning_rate": 9.163054530800455e-06,
      "loss": 3.4394,
      "step": 2362
    },
    {
      "epoch": 0.8053851397409679,
      "grad_norm": 16.73340068848952,
      "learning_rate": 9.161955858905142e-06,
      "loss": 3.8995,
      "step": 2363
    },
    {
      "epoch": 0.8057259713701431,
      "grad_norm": 22.950317693289552,
      "learning_rate": 9.160856532317607e-06,
      "loss": 3.6838,
      "step": 2364
    },
    {
      "epoch": 0.8060668029993183,
      "grad_norm": 19.803337146358142,
      "learning_rate": 9.159756551210774e-06,
      "loss": 3.4253,
      "step": 2365
    },
    {
      "epoch": 0.8064076346284935,
      "grad_norm": 17.603503635671508,
      "learning_rate": 9.158655915757678e-06,
      "loss": 3.0491,
      "step": 2366
    },
    {
      "epoch": 0.8067484662576687,
      "grad_norm": 22.762761041897896,
      "learning_rate": 9.15755462613145e-06,
      "loss": 3.1037,
      "step": 2367
    },
    {
      "epoch": 0.8070892978868439,
      "grad_norm": 16.939096282527,
      "learning_rate": 9.156452682505332e-06,
      "loss": 3.5573,
      "step": 2368
    },
    {
      "epoch": 0.8074301295160191,
      "grad_norm": 20.357862036535888,
      "learning_rate": 9.155350085052659e-06,
      "loss": 3.6476,
      "step": 2369
    },
    {
      "epoch": 0.8077709611451943,
      "grad_norm": 17.866436859867207,
      "learning_rate": 9.154246833946876e-06,
      "loss": 3.3504,
      "step": 2370
    },
    {
      "epoch": 0.8081117927743695,
      "grad_norm": 16.863999125204725,
      "learning_rate": 9.15314292936153e-06,
      "loss": 3.7055,
      "step": 2371
    },
    {
      "epoch": 0.8084526244035446,
      "grad_norm": 16.778370483826215,
      "learning_rate": 9.152038371470269e-06,
      "loss": 3.5891,
      "step": 2372
    },
    {
      "epoch": 0.8087934560327198,
      "grad_norm": 19.142149206087602,
      "learning_rate": 9.150933160446842e-06,
      "loss": 3.0667,
      "step": 2373
    },
    {
      "epoch": 0.809134287661895,
      "grad_norm": 13.203414553655334,
      "learning_rate": 9.149827296465107e-06,
      "loss": 3.4842,
      "step": 2374
    },
    {
      "epoch": 0.8094751192910702,
      "grad_norm": 12.894850604240622,
      "learning_rate": 9.148720779699016e-06,
      "loss": 3.145,
      "step": 2375
    },
    {
      "epoch": 0.8098159509202454,
      "grad_norm": 13.71550831433336,
      "learning_rate": 9.147613610322633e-06,
      "loss": 2.9144,
      "step": 2376
    },
    {
      "epoch": 0.8101567825494206,
      "grad_norm": 23.291364445482547,
      "learning_rate": 9.146505788510118e-06,
      "loss": 3.4701,
      "step": 2377
    },
    {
      "epoch": 0.8104976141785958,
      "grad_norm": 18.776741621628137,
      "learning_rate": 9.145397314435737e-06,
      "loss": 3.4881,
      "step": 2378
    },
    {
      "epoch": 0.810838445807771,
      "grad_norm": 18.634415765252427,
      "learning_rate": 9.144288188273856e-06,
      "loss": 2.9467,
      "step": 2379
    },
    {
      "epoch": 0.8111792774369462,
      "grad_norm": 17.09355088889809,
      "learning_rate": 9.143178410198945e-06,
      "loss": 3.1055,
      "step": 2380
    },
    {
      "epoch": 0.8115201090661214,
      "grad_norm": 13.722790094409966,
      "learning_rate": 9.142067980385579e-06,
      "loss": 2.8237,
      "step": 2381
    },
    {
      "epoch": 0.8118609406952966,
      "grad_norm": 27.042214737548687,
      "learning_rate": 9.140956899008428e-06,
      "loss": 3.2906,
      "step": 2382
    },
    {
      "epoch": 0.8122017723244717,
      "grad_norm": 17.997008144298885,
      "learning_rate": 9.139845166242274e-06,
      "loss": 2.8277,
      "step": 2383
    },
    {
      "epoch": 0.8125426039536469,
      "grad_norm": 13.97319413324819,
      "learning_rate": 9.138732782261997e-06,
      "loss": 3.6137,
      "step": 2384
    },
    {
      "epoch": 0.8128834355828221,
      "grad_norm": 18.020254213177534,
      "learning_rate": 9.137619747242577e-06,
      "loss": 3.3837,
      "step": 2385
    },
    {
      "epoch": 0.8132242672119973,
      "grad_norm": 13.357446076596586,
      "learning_rate": 9.136506061359098e-06,
      "loss": 3.1552,
      "step": 2386
    },
    {
      "epoch": 0.8135650988411725,
      "grad_norm": 15.06226979787106,
      "learning_rate": 9.135391724786751e-06,
      "loss": 3.5435,
      "step": 2387
    },
    {
      "epoch": 0.8139059304703476,
      "grad_norm": 26.298947481837843,
      "learning_rate": 9.134276737700824e-06,
      "loss": 2.4501,
      "step": 2388
    },
    {
      "epoch": 0.8142467620995228,
      "grad_norm": 21.213045751393793,
      "learning_rate": 9.133161100276708e-06,
      "loss": 3.0864,
      "step": 2389
    },
    {
      "epoch": 0.814587593728698,
      "grad_norm": 18.628118743982064,
      "learning_rate": 9.132044812689897e-06,
      "loss": 3.5066,
      "step": 2390
    },
    {
      "epoch": 0.8149284253578732,
      "grad_norm": 16.01425335415117,
      "learning_rate": 9.13092787511599e-06,
      "loss": 3.1766,
      "step": 2391
    },
    {
      "epoch": 0.8152692569870484,
      "grad_norm": 23.446313718431334,
      "learning_rate": 9.129810287730683e-06,
      "loss": 3.0271,
      "step": 2392
    },
    {
      "epoch": 0.8156100886162236,
      "grad_norm": 19.887136005355508,
      "learning_rate": 9.128692050709778e-06,
      "loss": 3.0478,
      "step": 2393
    },
    {
      "epoch": 0.8159509202453987,
      "grad_norm": 25.76636030563254,
      "learning_rate": 9.127573164229178e-06,
      "loss": 3.5309,
      "step": 2394
    },
    {
      "epoch": 0.8162917518745739,
      "grad_norm": 34.76967508042324,
      "learning_rate": 9.126453628464889e-06,
      "loss": 3.5035,
      "step": 2395
    },
    {
      "epoch": 0.8166325835037491,
      "grad_norm": 17.88164012731075,
      "learning_rate": 9.125333443593017e-06,
      "loss": 3.428,
      "step": 2396
    },
    {
      "epoch": 0.8169734151329243,
      "grad_norm": 16.423614585048547,
      "learning_rate": 9.12421260978977e-06,
      "loss": 2.7889,
      "step": 2397
    },
    {
      "epoch": 0.8173142467620995,
      "grad_norm": 12.427463435852337,
      "learning_rate": 9.123091127231465e-06,
      "loss": 3.0426,
      "step": 2398
    },
    {
      "epoch": 0.8176550783912747,
      "grad_norm": 22.3407799054145,
      "learning_rate": 9.121968996094513e-06,
      "loss": 3.5196,
      "step": 2399
    },
    {
      "epoch": 0.8179959100204499,
      "grad_norm": 15.373162885008144,
      "learning_rate": 9.120846216555428e-06,
      "loss": 3.486,
      "step": 2400
    },
    {
      "epoch": 0.8183367416496251,
      "grad_norm": 13.575897651562975,
      "learning_rate": 9.119722788790828e-06,
      "loss": 3.5517,
      "step": 2401
    },
    {
      "epoch": 0.8186775732788003,
      "grad_norm": 15.043518184401005,
      "learning_rate": 9.118598712977436e-06,
      "loss": 3.4028,
      "step": 2402
    },
    {
      "epoch": 0.8190184049079755,
      "grad_norm": 12.561947053777114,
      "learning_rate": 9.11747398929207e-06,
      "loss": 2.7213,
      "step": 2403
    },
    {
      "epoch": 0.8193592365371506,
      "grad_norm": 26.910260138370653,
      "learning_rate": 9.116348617911654e-06,
      "loss": 3.3318,
      "step": 2404
    },
    {
      "epoch": 0.8197000681663258,
      "grad_norm": 20.406348847036732,
      "learning_rate": 9.115222599013215e-06,
      "loss": 2.9833,
      "step": 2405
    },
    {
      "epoch": 0.820040899795501,
      "grad_norm": 20.077582629467837,
      "learning_rate": 9.11409593277388e-06,
      "loss": 3.6674,
      "step": 2406
    },
    {
      "epoch": 0.8203817314246762,
      "grad_norm": 20.5501928601293,
      "learning_rate": 9.112968619370877e-06,
      "loss": 3.3529,
      "step": 2407
    },
    {
      "epoch": 0.8207225630538514,
      "grad_norm": 15.496002874117803,
      "learning_rate": 9.111840658981537e-06,
      "loss": 3.1515,
      "step": 2408
    },
    {
      "epoch": 0.8210633946830266,
      "grad_norm": 46.059140692449326,
      "learning_rate": 9.110712051783293e-06,
      "loss": 3.3705,
      "step": 2409
    },
    {
      "epoch": 0.8214042263122018,
      "grad_norm": 13.272982148816922,
      "learning_rate": 9.10958279795368e-06,
      "loss": 2.8387,
      "step": 2410
    },
    {
      "epoch": 0.821745057941377,
      "grad_norm": 15.247866325710303,
      "learning_rate": 9.108452897670333e-06,
      "loss": 3.2503,
      "step": 2411
    },
    {
      "epoch": 0.8220858895705522,
      "grad_norm": 14.579323698663247,
      "learning_rate": 9.107322351110991e-06,
      "loss": 3.3064,
      "step": 2412
    },
    {
      "epoch": 0.8224267211997274,
      "grad_norm": 13.37530579257486,
      "learning_rate": 9.106191158453493e-06,
      "loss": 3.2966,
      "step": 2413
    },
    {
      "epoch": 0.8227675528289026,
      "grad_norm": 22.508503506103317,
      "learning_rate": 9.10505931987578e-06,
      "loss": 3.4239,
      "step": 2414
    },
    {
      "epoch": 0.8231083844580777,
      "grad_norm": 17.309053332200968,
      "learning_rate": 9.103926835555894e-06,
      "loss": 3.3768,
      "step": 2415
    },
    {
      "epoch": 0.8234492160872529,
      "grad_norm": 13.756427376603552,
      "learning_rate": 9.10279370567198e-06,
      "loss": 3.3378,
      "step": 2416
    },
    {
      "epoch": 0.823790047716428,
      "grad_norm": 25.795961643707024,
      "learning_rate": 9.101659930402284e-06,
      "loss": 2.4557,
      "step": 2417
    },
    {
      "epoch": 0.8241308793456033,
      "grad_norm": 20.225826048503585,
      "learning_rate": 9.100525509925154e-06,
      "loss": 3.4185,
      "step": 2418
    },
    {
      "epoch": 0.8244717109747784,
      "grad_norm": 16.17049597792254,
      "learning_rate": 9.099390444419038e-06,
      "loss": 2.7882,
      "step": 2419
    },
    {
      "epoch": 0.8248125426039536,
      "grad_norm": 19.218170968942456,
      "learning_rate": 9.098254734062486e-06,
      "loss": 2.9817,
      "step": 2420
    },
    {
      "epoch": 0.8251533742331288,
      "grad_norm": 17.045483323229746,
      "learning_rate": 9.09711837903415e-06,
      "loss": 3.2429,
      "step": 2421
    },
    {
      "epoch": 0.825494205862304,
      "grad_norm": 16.286686798040616,
      "learning_rate": 9.095981379512782e-06,
      "loss": 3.4329,
      "step": 2422
    },
    {
      "epoch": 0.8258350374914792,
      "grad_norm": 25.816461458544147,
      "learning_rate": 9.09484373567724e-06,
      "loss": 3.5333,
      "step": 2423
    },
    {
      "epoch": 0.8261758691206544,
      "grad_norm": 14.241033207995622,
      "learning_rate": 9.093705447706475e-06,
      "loss": 2.9512,
      "step": 2424
    },
    {
      "epoch": 0.8265167007498296,
      "grad_norm": 12.772043947182974,
      "learning_rate": 9.092566515779551e-06,
      "loss": 2.8783,
      "step": 2425
    },
    {
      "epoch": 0.8268575323790047,
      "grad_norm": 23.760269167717404,
      "learning_rate": 9.091426940075619e-06,
      "loss": 2.9539,
      "step": 2426
    },
    {
      "epoch": 0.8271983640081799,
      "grad_norm": 20.847602265662346,
      "learning_rate": 9.090286720773943e-06,
      "loss": 3.578,
      "step": 2427
    },
    {
      "epoch": 0.8275391956373551,
      "grad_norm": 24.873861100895716,
      "learning_rate": 9.089145858053885e-06,
      "loss": 3.3606,
      "step": 2428
    },
    {
      "epoch": 0.8278800272665303,
      "grad_norm": 12.696497252710875,
      "learning_rate": 9.088004352094904e-06,
      "loss": 3.1954,
      "step": 2429
    },
    {
      "epoch": 0.8282208588957055,
      "grad_norm": 15.355184871125521,
      "learning_rate": 9.086862203076565e-06,
      "loss": 2.8093,
      "step": 2430
    },
    {
      "epoch": 0.8285616905248807,
      "grad_norm": 19.41158390389094,
      "learning_rate": 9.085719411178533e-06,
      "loss": 3.4925,
      "step": 2431
    },
    {
      "epoch": 0.8289025221540559,
      "grad_norm": 17.812016141980532,
      "learning_rate": 9.084575976580574e-06,
      "loss": 3.1239,
      "step": 2432
    },
    {
      "epoch": 0.8292433537832311,
      "grad_norm": 18.1961037022868,
      "learning_rate": 9.083431899462554e-06,
      "loss": 3.2766,
      "step": 2433
    },
    {
      "epoch": 0.8295841854124063,
      "grad_norm": 20.94441451290299,
      "learning_rate": 9.08228718000444e-06,
      "loss": 2.9311,
      "step": 2434
    },
    {
      "epoch": 0.8299250170415815,
      "grad_norm": 13.459440113524185,
      "learning_rate": 9.081141818386302e-06,
      "loss": 3.1426,
      "step": 2435
    },
    {
      "epoch": 0.8302658486707567,
      "grad_norm": 20.99525134659867,
      "learning_rate": 9.079995814788311e-06,
      "loss": 3.0768,
      "step": 2436
    },
    {
      "epoch": 0.8306066802999318,
      "grad_norm": 16.907365971082864,
      "learning_rate": 9.078849169390736e-06,
      "loss": 3.114,
      "step": 2437
    },
    {
      "epoch": 0.830947511929107,
      "grad_norm": 15.335619913444427,
      "learning_rate": 9.077701882373949e-06,
      "loss": 3.4113,
      "step": 2438
    },
    {
      "epoch": 0.8312883435582822,
      "grad_norm": 19.82277632244142,
      "learning_rate": 9.076553953918426e-06,
      "loss": 2.9033,
      "step": 2439
    },
    {
      "epoch": 0.8316291751874574,
      "grad_norm": 18.6016498122997,
      "learning_rate": 9.075405384204736e-06,
      "loss": 2.9175,
      "step": 2440
    },
    {
      "epoch": 0.8319700068166326,
      "grad_norm": 14.73140702180242,
      "learning_rate": 9.074256173413555e-06,
      "loss": 3.2264,
      "step": 2441
    },
    {
      "epoch": 0.8323108384458078,
      "grad_norm": 25.396576792435877,
      "learning_rate": 9.073106321725662e-06,
      "loss": 3.9497,
      "step": 2442
    },
    {
      "epoch": 0.832651670074983,
      "grad_norm": 29.98033404228255,
      "learning_rate": 9.071955829321928e-06,
      "loss": 2.9884,
      "step": 2443
    },
    {
      "epoch": 0.8329925017041582,
      "grad_norm": 19.24445993020744,
      "learning_rate": 9.070804696383334e-06,
      "loss": 3.6053,
      "step": 2444
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 23.19408337857866,
      "learning_rate": 9.069652923090955e-06,
      "loss": 3.4011,
      "step": 2445
    },
    {
      "epoch": 0.8336741649625086,
      "grad_norm": 14.846727861619131,
      "learning_rate": 9.068500509625973e-06,
      "loss": 3.0189,
      "step": 2446
    },
    {
      "epoch": 0.8340149965916837,
      "grad_norm": 13.637859909094301,
      "learning_rate": 9.067347456169664e-06,
      "loss": 3.4156,
      "step": 2447
    },
    {
      "epoch": 0.8343558282208589,
      "grad_norm": 24.76251836849556,
      "learning_rate": 9.066193762903408e-06,
      "loss": 3.6889,
      "step": 2448
    },
    {
      "epoch": 0.834696659850034,
      "grad_norm": 10.799710767518677,
      "learning_rate": 9.065039430008689e-06,
      "loss": 2.9284,
      "step": 2449
    },
    {
      "epoch": 0.8350374914792092,
      "grad_norm": 19.117839046082647,
      "learning_rate": 9.063884457667083e-06,
      "loss": 3.5847,
      "step": 2450
    },
    {
      "epoch": 0.8353783231083844,
      "grad_norm": 20.512761502020748,
      "learning_rate": 9.062728846060276e-06,
      "loss": 3.854,
      "step": 2451
    },
    {
      "epoch": 0.8357191547375596,
      "grad_norm": 15.923363440071734,
      "learning_rate": 9.061572595370048e-06,
      "loss": 3.473,
      "step": 2452
    },
    {
      "epoch": 0.8360599863667348,
      "grad_norm": 19.425293891256654,
      "learning_rate": 9.060415705778282e-06,
      "loss": 3.6185,
      "step": 2453
    },
    {
      "epoch": 0.83640081799591,
      "grad_norm": 12.909506185386316,
      "learning_rate": 9.059258177466963e-06,
      "loss": 3.3467,
      "step": 2454
    },
    {
      "epoch": 0.8367416496250852,
      "grad_norm": 19.047711588685917,
      "learning_rate": 9.058100010618174e-06,
      "loss": 3.5846,
      "step": 2455
    },
    {
      "epoch": 0.8370824812542604,
      "grad_norm": 17.84880232164407,
      "learning_rate": 9.056941205414097e-06,
      "loss": 3.8702,
      "step": 2456
    },
    {
      "epoch": 0.8374233128834356,
      "grad_norm": 14.893759959810534,
      "learning_rate": 9.055781762037022e-06,
      "loss": 3.0417,
      "step": 2457
    },
    {
      "epoch": 0.8377641445126107,
      "grad_norm": 13.54748927652795,
      "learning_rate": 9.054621680669328e-06,
      "loss": 3.1686,
      "step": 2458
    },
    {
      "epoch": 0.8381049761417859,
      "grad_norm": 22.85508075538054,
      "learning_rate": 9.053460961493503e-06,
      "loss": 3.0726,
      "step": 2459
    },
    {
      "epoch": 0.8384458077709611,
      "grad_norm": 14.65925903924751,
      "learning_rate": 9.052299604692134e-06,
      "loss": 3.1139,
      "step": 2460
    },
    {
      "epoch": 0.8387866394001363,
      "grad_norm": 17.48110893045032,
      "learning_rate": 9.051137610447907e-06,
      "loss": 3.6512,
      "step": 2461
    },
    {
      "epoch": 0.8391274710293115,
      "grad_norm": 22.574932644727326,
      "learning_rate": 9.049974978943604e-06,
      "loss": 3.9933,
      "step": 2462
    },
    {
      "epoch": 0.8394683026584867,
      "grad_norm": 17.4262887398651,
      "learning_rate": 9.048811710362117e-06,
      "loss": 3.6163,
      "step": 2463
    },
    {
      "epoch": 0.8398091342876619,
      "grad_norm": 26.208599743132172,
      "learning_rate": 9.04764780488643e-06,
      "loss": 3.2557,
      "step": 2464
    },
    {
      "epoch": 0.8401499659168371,
      "grad_norm": 16.119061390475427,
      "learning_rate": 9.04648326269963e-06,
      "loss": 3.4775,
      "step": 2465
    },
    {
      "epoch": 0.8404907975460123,
      "grad_norm": 19.435518692497183,
      "learning_rate": 9.045318083984906e-06,
      "loss": 3.5049,
      "step": 2466
    },
    {
      "epoch": 0.8408316291751875,
      "grad_norm": 18.90957280580403,
      "learning_rate": 9.044152268925543e-06,
      "loss": 3.3982,
      "step": 2467
    },
    {
      "epoch": 0.8411724608043627,
      "grad_norm": 29.880578125610764,
      "learning_rate": 9.04298581770493e-06,
      "loss": 3.1759,
      "step": 2468
    },
    {
      "epoch": 0.8415132924335378,
      "grad_norm": 15.513053999997519,
      "learning_rate": 9.041818730506553e-06,
      "loss": 3.11,
      "step": 2469
    },
    {
      "epoch": 0.841854124062713,
      "grad_norm": 18.36905387435256,
      "learning_rate": 9.040651007514001e-06,
      "loss": 3.4063,
      "step": 2470
    },
    {
      "epoch": 0.8421949556918882,
      "grad_norm": 16.907538978448553,
      "learning_rate": 9.03948264891096e-06,
      "loss": 3.18,
      "step": 2471
    },
    {
      "epoch": 0.8425357873210634,
      "grad_norm": 13.4953091467825,
      "learning_rate": 9.038313654881218e-06,
      "loss": 3.4582,
      "step": 2472
    },
    {
      "epoch": 0.8428766189502386,
      "grad_norm": 16.874819246488784,
      "learning_rate": 9.037144025608661e-06,
      "loss": 3.5823,
      "step": 2473
    },
    {
      "epoch": 0.8432174505794138,
      "grad_norm": 16.821846157628837,
      "learning_rate": 9.035973761277281e-06,
      "loss": 3.2487,
      "step": 2474
    },
    {
      "epoch": 0.843558282208589,
      "grad_norm": 13.140188865934732,
      "learning_rate": 9.034802862071159e-06,
      "loss": 3.147,
      "step": 2475
    },
    {
      "epoch": 0.8438991138377642,
      "grad_norm": 14.556898203645924,
      "learning_rate": 9.033631328174488e-06,
      "loss": 3.5957,
      "step": 2476
    },
    {
      "epoch": 0.8442399454669394,
      "grad_norm": 20.063303111708247,
      "learning_rate": 9.03245915977155e-06,
      "loss": 2.8996,
      "step": 2477
    },
    {
      "epoch": 0.8445807770961146,
      "grad_norm": 18.330753179077945,
      "learning_rate": 9.031286357046737e-06,
      "loss": 4.0325,
      "step": 2478
    },
    {
      "epoch": 0.8449216087252897,
      "grad_norm": 19.045012816527162,
      "learning_rate": 9.030112920184528e-06,
      "loss": 3.4495,
      "step": 2479
    },
    {
      "epoch": 0.8452624403544649,
      "grad_norm": 22.706467739233304,
      "learning_rate": 9.028938849369516e-06,
      "loss": 3.6744,
      "step": 2480
    },
    {
      "epoch": 0.84560327198364,
      "grad_norm": 24.975215971900436,
      "learning_rate": 9.027764144786384e-06,
      "loss": 3.3368,
      "step": 2481
    },
    {
      "epoch": 0.8459441036128152,
      "grad_norm": 24.77526581804242,
      "learning_rate": 9.02658880661992e-06,
      "loss": 4.0264,
      "step": 2482
    },
    {
      "epoch": 0.8462849352419904,
      "grad_norm": 21.3977822655649,
      "learning_rate": 9.025412835055005e-06,
      "loss": 3.2643,
      "step": 2483
    },
    {
      "epoch": 0.8466257668711656,
      "grad_norm": 21.232365099631405,
      "learning_rate": 9.02423623027663e-06,
      "loss": 3.1365,
      "step": 2484
    },
    {
      "epoch": 0.8469665985003408,
      "grad_norm": 13.079303062069254,
      "learning_rate": 9.023058992469875e-06,
      "loss": 3.1411,
      "step": 2485
    },
    {
      "epoch": 0.847307430129516,
      "grad_norm": 15.993365839076109,
      "learning_rate": 9.021881121819925e-06,
      "loss": 3.1909,
      "step": 2486
    },
    {
      "epoch": 0.8476482617586912,
      "grad_norm": 21.11432695815161,
      "learning_rate": 9.020702618512064e-06,
      "loss": 4.1149,
      "step": 2487
    },
    {
      "epoch": 0.8479890933878664,
      "grad_norm": 17.577805747600387,
      "learning_rate": 9.019523482731678e-06,
      "loss": 3.7953,
      "step": 2488
    },
    {
      "epoch": 0.8483299250170416,
      "grad_norm": 19.426164864440853,
      "learning_rate": 9.018343714664245e-06,
      "loss": 3.4335,
      "step": 2489
    },
    {
      "epoch": 0.8486707566462167,
      "grad_norm": 23.54572524456234,
      "learning_rate": 9.01716331449535e-06,
      "loss": 2.9619,
      "step": 2490
    },
    {
      "epoch": 0.8490115882753919,
      "grad_norm": 17.943598071111115,
      "learning_rate": 9.015982282410675e-06,
      "loss": 3.6034,
      "step": 2491
    },
    {
      "epoch": 0.8493524199045671,
      "grad_norm": 19.280876985047392,
      "learning_rate": 9.014800618596e-06,
      "loss": 3.2417,
      "step": 2492
    },
    {
      "epoch": 0.8496932515337423,
      "grad_norm": 13.679680002212928,
      "learning_rate": 9.013618323237206e-06,
      "loss": 3.3964,
      "step": 2493
    },
    {
      "epoch": 0.8500340831629175,
      "grad_norm": 15.412282125205346,
      "learning_rate": 9.012435396520271e-06,
      "loss": 3.2454,
      "step": 2494
    },
    {
      "epoch": 0.8503749147920927,
      "grad_norm": 13.54520234144716,
      "learning_rate": 9.011251838631277e-06,
      "loss": 2.7674,
      "step": 2495
    },
    {
      "epoch": 0.8507157464212679,
      "grad_norm": 18.245441284771918,
      "learning_rate": 9.0100676497564e-06,
      "loss": 2.9394,
      "step": 2496
    },
    {
      "epoch": 0.8510565780504431,
      "grad_norm": 26.805163147213598,
      "learning_rate": 9.008882830081919e-06,
      "loss": 3.3999,
      "step": 2497
    },
    {
      "epoch": 0.8513974096796183,
      "grad_norm": 15.935504749274058,
      "learning_rate": 9.007697379794212e-06,
      "loss": 3.3175,
      "step": 2498
    },
    {
      "epoch": 0.8517382413087935,
      "grad_norm": 25.039339599793102,
      "learning_rate": 9.006511299079752e-06,
      "loss": 3.3837,
      "step": 2499
    },
    {
      "epoch": 0.8520790729379687,
      "grad_norm": 17.175925734247897,
      "learning_rate": 9.005324588125115e-06,
      "loss": 3.3645,
      "step": 2500
    },
    {
      "epoch": 0.8524199045671438,
      "grad_norm": 16.20280138368815,
      "learning_rate": 9.004137247116977e-06,
      "loss": 3.2318,
      "step": 2501
    },
    {
      "epoch": 0.852760736196319,
      "grad_norm": 16.67129545535022,
      "learning_rate": 9.00294927624211e-06,
      "loss": 3.4075,
      "step": 2502
    },
    {
      "epoch": 0.8531015678254942,
      "grad_norm": 18.902281290478683,
      "learning_rate": 9.001760675687387e-06,
      "loss": 3.1946,
      "step": 2503
    },
    {
      "epoch": 0.8534423994546694,
      "grad_norm": 18.585585563316393,
      "learning_rate": 9.000571445639779e-06,
      "loss": 2.9643,
      "step": 2504
    },
    {
      "epoch": 0.8537832310838446,
      "grad_norm": 17.959435765312417,
      "learning_rate": 8.999381586286359e-06,
      "loss": 3.1123,
      "step": 2505
    },
    {
      "epoch": 0.8541240627130198,
      "grad_norm": 13.102107563255688,
      "learning_rate": 8.998191097814292e-06,
      "loss": 3.2032,
      "step": 2506
    },
    {
      "epoch": 0.854464894342195,
      "grad_norm": 18.91979014082419,
      "learning_rate": 8.996999980410852e-06,
      "loss": 3.625,
      "step": 2507
    },
    {
      "epoch": 0.8548057259713702,
      "grad_norm": 14.285052271847208,
      "learning_rate": 8.995808234263404e-06,
      "loss": 3.155,
      "step": 2508
    },
    {
      "epoch": 0.8551465576005454,
      "grad_norm": 14.643966573414696,
      "learning_rate": 8.994615859559412e-06,
      "loss": 3.427,
      "step": 2509
    },
    {
      "epoch": 0.8554873892297206,
      "grad_norm": 13.061251265095386,
      "learning_rate": 8.993422856486446e-06,
      "loss": 3.1418,
      "step": 2510
    },
    {
      "epoch": 0.8558282208588958,
      "grad_norm": 17.155698451420008,
      "learning_rate": 8.992229225232167e-06,
      "loss": 3.3005,
      "step": 2511
    },
    {
      "epoch": 0.8561690524880708,
      "grad_norm": 19.064644379240814,
      "learning_rate": 8.99103496598434e-06,
      "loss": 3.8094,
      "step": 2512
    },
    {
      "epoch": 0.856509884117246,
      "grad_norm": 14.751398462071323,
      "learning_rate": 8.989840078930822e-06,
      "loss": 3.4364,
      "step": 2513
    },
    {
      "epoch": 0.8568507157464212,
      "grad_norm": 16.08250782149595,
      "learning_rate": 8.988644564259581e-06,
      "loss": 3.1978,
      "step": 2514
    },
    {
      "epoch": 0.8571915473755964,
      "grad_norm": 21.42620765697391,
      "learning_rate": 8.98744842215867e-06,
      "loss": 3.5852,
      "step": 2515
    },
    {
      "epoch": 0.8575323790047716,
      "grad_norm": 12.933862433069027,
      "learning_rate": 8.98625165281625e-06,
      "loss": 2.8531,
      "step": 2516
    },
    {
      "epoch": 0.8578732106339468,
      "grad_norm": 20.844742727043727,
      "learning_rate": 8.985054256420575e-06,
      "loss": 3.908,
      "step": 2517
    },
    {
      "epoch": 0.858214042263122,
      "grad_norm": 12.937282955917563,
      "learning_rate": 8.983856233160003e-06,
      "loss": 3.1311,
      "step": 2518
    },
    {
      "epoch": 0.8585548738922972,
      "grad_norm": 18.44922679132292,
      "learning_rate": 8.982657583222986e-06,
      "loss": 3.4946,
      "step": 2519
    },
    {
      "epoch": 0.8588957055214724,
      "grad_norm": 18.977044500937282,
      "learning_rate": 8.981458306798078e-06,
      "loss": 3.8432,
      "step": 2520
    },
    {
      "epoch": 0.8592365371506476,
      "grad_norm": 13.996163129910174,
      "learning_rate": 8.98025840407393e-06,
      "loss": 3.1883,
      "step": 2521
    },
    {
      "epoch": 0.8595773687798227,
      "grad_norm": 20.27690789780118,
      "learning_rate": 8.97905787523929e-06,
      "loss": 3.2158,
      "step": 2522
    },
    {
      "epoch": 0.8599182004089979,
      "grad_norm": 13.099504452489223,
      "learning_rate": 8.977856720483005e-06,
      "loss": 2.9615,
      "step": 2523
    },
    {
      "epoch": 0.8602590320381731,
      "grad_norm": 17.173977075701263,
      "learning_rate": 8.976654939994023e-06,
      "loss": 3.3923,
      "step": 2524
    },
    {
      "epoch": 0.8605998636673483,
      "grad_norm": 16.589256787511843,
      "learning_rate": 8.97545253396139e-06,
      "loss": 3.4206,
      "step": 2525
    },
    {
      "epoch": 0.8609406952965235,
      "grad_norm": 26.911864285849596,
      "learning_rate": 8.974249502574248e-06,
      "loss": 2.8757,
      "step": 2526
    },
    {
      "epoch": 0.8612815269256987,
      "grad_norm": 13.76769376218054,
      "learning_rate": 8.973045846021837e-06,
      "loss": 3.459,
      "step": 2527
    },
    {
      "epoch": 0.8616223585548739,
      "grad_norm": 17.32405517640472,
      "learning_rate": 8.971841564493502e-06,
      "loss": 3.6824,
      "step": 2528
    },
    {
      "epoch": 0.8619631901840491,
      "grad_norm": 12.6890369430944,
      "learning_rate": 8.970636658178675e-06,
      "loss": 3.2362,
      "step": 2529
    },
    {
      "epoch": 0.8623040218132243,
      "grad_norm": 16.261626020665148,
      "learning_rate": 8.969431127266896e-06,
      "loss": 3.6278,
      "step": 2530
    },
    {
      "epoch": 0.8626448534423995,
      "grad_norm": 22.831061455606793,
      "learning_rate": 8.9682249719478e-06,
      "loss": 3.6986,
      "step": 2531
    },
    {
      "epoch": 0.8629856850715747,
      "grad_norm": 21.416361626774027,
      "learning_rate": 8.967018192411118e-06,
      "loss": 3.4809,
      "step": 2532
    },
    {
      "epoch": 0.8633265167007498,
      "grad_norm": 15.244843351145281,
      "learning_rate": 8.965810788846685e-06,
      "loss": 3.1427,
      "step": 2533
    },
    {
      "epoch": 0.863667348329925,
      "grad_norm": 17.64426522358967,
      "learning_rate": 8.964602761444424e-06,
      "loss": 3.5285,
      "step": 2534
    },
    {
      "epoch": 0.8640081799591002,
      "grad_norm": 18.509463248903714,
      "learning_rate": 8.963394110394371e-06,
      "loss": 3.0782,
      "step": 2535
    },
    {
      "epoch": 0.8643490115882754,
      "grad_norm": 17.740652728035393,
      "learning_rate": 8.962184835886644e-06,
      "loss": 3.1384,
      "step": 2536
    },
    {
      "epoch": 0.8646898432174506,
      "grad_norm": 18.261553423135222,
      "learning_rate": 8.960974938111471e-06,
      "loss": 3.544,
      "step": 2537
    },
    {
      "epoch": 0.8650306748466258,
      "grad_norm": 19.23172705867055,
      "learning_rate": 8.959764417259172e-06,
      "loss": 3.6919,
      "step": 2538
    },
    {
      "epoch": 0.865371506475801,
      "grad_norm": 16.573000635196315,
      "learning_rate": 8.958553273520168e-06,
      "loss": 3.0245,
      "step": 2539
    },
    {
      "epoch": 0.8657123381049762,
      "grad_norm": 15.598739860282748,
      "learning_rate": 8.957341507084975e-06,
      "loss": 3.016,
      "step": 2540
    },
    {
      "epoch": 0.8660531697341514,
      "grad_norm": 23.669345053206843,
      "learning_rate": 8.956129118144211e-06,
      "loss": 3.9527,
      "step": 2541
    },
    {
      "epoch": 0.8663940013633266,
      "grad_norm": 16.45766325185923,
      "learning_rate": 8.954916106888587e-06,
      "loss": 3.3011,
      "step": 2542
    },
    {
      "epoch": 0.8667348329925018,
      "grad_norm": 13.616589793924241,
      "learning_rate": 8.953702473508916e-06,
      "loss": 3.5519,
      "step": 2543
    },
    {
      "epoch": 0.8670756646216768,
      "grad_norm": 10.94836284850705,
      "learning_rate": 8.952488218196107e-06,
      "loss": 3.3588,
      "step": 2544
    },
    {
      "epoch": 0.867416496250852,
      "grad_norm": 16.15433096412706,
      "learning_rate": 8.951273341141169e-06,
      "loss": 3.2061,
      "step": 2545
    },
    {
      "epoch": 0.8677573278800272,
      "grad_norm": 17.644814216403923,
      "learning_rate": 8.950057842535204e-06,
      "loss": 3.1495,
      "step": 2546
    },
    {
      "epoch": 0.8680981595092024,
      "grad_norm": 14.956051300410229,
      "learning_rate": 8.948841722569416e-06,
      "loss": 2.8823,
      "step": 2547
    },
    {
      "epoch": 0.8684389911383776,
      "grad_norm": 12.032389184245892,
      "learning_rate": 8.947624981435105e-06,
      "loss": 3.1144,
      "step": 2548
    },
    {
      "epoch": 0.8687798227675528,
      "grad_norm": 17.907122893700418,
      "learning_rate": 8.946407619323671e-06,
      "loss": 3.1172,
      "step": 2549
    },
    {
      "epoch": 0.869120654396728,
      "grad_norm": 14.804162630475096,
      "learning_rate": 8.945189636426608e-06,
      "loss": 3.2271,
      "step": 2550
    },
    {
      "epoch": 0.8694614860259032,
      "grad_norm": 15.914750924454655,
      "learning_rate": 8.94397103293551e-06,
      "loss": 3.2645,
      "step": 2551
    },
    {
      "epoch": 0.8698023176550784,
      "grad_norm": 18.380186634115482,
      "learning_rate": 8.94275180904207e-06,
      "loss": 2.9581,
      "step": 2552
    },
    {
      "epoch": 0.8701431492842536,
      "grad_norm": 25.636631508478025,
      "learning_rate": 8.941531964938076e-06,
      "loss": 3.7498,
      "step": 2553
    },
    {
      "epoch": 0.8704839809134287,
      "grad_norm": 19.45822524927454,
      "learning_rate": 8.940311500815411e-06,
      "loss": 3.4741,
      "step": 2554
    },
    {
      "epoch": 0.8708248125426039,
      "grad_norm": 26.137589895865297,
      "learning_rate": 8.939090416866062e-06,
      "loss": 3.4975,
      "step": 2555
    },
    {
      "epoch": 0.8711656441717791,
      "grad_norm": 15.744131524949397,
      "learning_rate": 8.93786871328211e-06,
      "loss": 3.1802,
      "step": 2556
    },
    {
      "epoch": 0.8715064758009543,
      "grad_norm": 15.206287328748104,
      "learning_rate": 8.936646390255734e-06,
      "loss": 3.3128,
      "step": 2557
    },
    {
      "epoch": 0.8718473074301295,
      "grad_norm": 16.97659502609929,
      "learning_rate": 8.93542344797921e-06,
      "loss": 3.3851,
      "step": 2558
    },
    {
      "epoch": 0.8721881390593047,
      "grad_norm": 13.400484151900626,
      "learning_rate": 8.934199886644911e-06,
      "loss": 3.0789,
      "step": 2559
    },
    {
      "epoch": 0.8725289706884799,
      "grad_norm": 42.66915334720432,
      "learning_rate": 8.93297570644531e-06,
      "loss": 3.5284,
      "step": 2560
    },
    {
      "epoch": 0.8728698023176551,
      "grad_norm": 15.135592834450856,
      "learning_rate": 8.931750907572972e-06,
      "loss": 3.3116,
      "step": 2561
    },
    {
      "epoch": 0.8732106339468303,
      "grad_norm": 18.72139133061072,
      "learning_rate": 8.930525490220566e-06,
      "loss": 3.1384,
      "step": 2562
    },
    {
      "epoch": 0.8735514655760055,
      "grad_norm": 14.627331797921567,
      "learning_rate": 8.929299454580855e-06,
      "loss": 3.1913,
      "step": 2563
    },
    {
      "epoch": 0.8738922972051807,
      "grad_norm": 14.619976276925701,
      "learning_rate": 8.928072800846697e-06,
      "loss": 3.4051,
      "step": 2564
    },
    {
      "epoch": 0.8742331288343558,
      "grad_norm": 16.836077326456703,
      "learning_rate": 8.92684552921105e-06,
      "loss": 3.2225,
      "step": 2565
    },
    {
      "epoch": 0.874573960463531,
      "grad_norm": 22.970586464106376,
      "learning_rate": 8.92561763986697e-06,
      "loss": 3.4569,
      "step": 2566
    },
    {
      "epoch": 0.8749147920927062,
      "grad_norm": 11.549589888000659,
      "learning_rate": 8.924389133007609e-06,
      "loss": 3.2314,
      "step": 2567
    },
    {
      "epoch": 0.8752556237218814,
      "grad_norm": 12.639525829463885,
      "learning_rate": 8.923160008826216e-06,
      "loss": 3.0867,
      "step": 2568
    },
    {
      "epoch": 0.8755964553510566,
      "grad_norm": 35.08922376351978,
      "learning_rate": 8.921930267516133e-06,
      "loss": 3.899,
      "step": 2569
    },
    {
      "epoch": 0.8759372869802318,
      "grad_norm": 16.791122756801638,
      "learning_rate": 8.920699909270809e-06,
      "loss": 3.533,
      "step": 2570
    },
    {
      "epoch": 0.876278118609407,
      "grad_norm": 11.905418741636423,
      "learning_rate": 8.91946893428378e-06,
      "loss": 3.1115,
      "step": 2571
    },
    {
      "epoch": 0.8766189502385822,
      "grad_norm": 16.963960207749334,
      "learning_rate": 8.918237342748687e-06,
      "loss": 2.9679,
      "step": 2572
    },
    {
      "epoch": 0.8769597818677574,
      "grad_norm": 23.23187854404608,
      "learning_rate": 8.917005134859263e-06,
      "loss": 3.8656,
      "step": 2573
    },
    {
      "epoch": 0.8773006134969326,
      "grad_norm": 29.571167720906537,
      "learning_rate": 8.915772310809338e-06,
      "loss": 3.6199,
      "step": 2574
    },
    {
      "epoch": 0.8776414451261078,
      "grad_norm": 19.864558980311482,
      "learning_rate": 8.914538870792841e-06,
      "loss": 3.6041,
      "step": 2575
    },
    {
      "epoch": 0.8779822767552828,
      "grad_norm": 32.40597445636004,
      "learning_rate": 8.913304815003796e-06,
      "loss": 4.233,
      "step": 2576
    },
    {
      "epoch": 0.878323108384458,
      "grad_norm": 16.756252922968606,
      "learning_rate": 8.912070143636325e-06,
      "loss": 2.9932,
      "step": 2577
    },
    {
      "epoch": 0.8786639400136332,
      "grad_norm": 16.444427012291722,
      "learning_rate": 8.910834856884648e-06,
      "loss": 3.0674,
      "step": 2578
    },
    {
      "epoch": 0.8790047716428084,
      "grad_norm": 20.24359367469236,
      "learning_rate": 8.909598954943081e-06,
      "loss": 3.2323,
      "step": 2579
    },
    {
      "epoch": 0.8793456032719836,
      "grad_norm": 16.84198541480077,
      "learning_rate": 8.908362438006033e-06,
      "loss": 3.3313,
      "step": 2580
    },
    {
      "epoch": 0.8796864349011588,
      "grad_norm": 18.65773262037242,
      "learning_rate": 8.907125306268016e-06,
      "loss": 3.8243,
      "step": 2581
    },
    {
      "epoch": 0.880027266530334,
      "grad_norm": 13.605233998524524,
      "learning_rate": 8.905887559923635e-06,
      "loss": 3.0313,
      "step": 2582
    },
    {
      "epoch": 0.8803680981595092,
      "grad_norm": 21.57099224760928,
      "learning_rate": 8.904649199167594e-06,
      "loss": 3.5689,
      "step": 2583
    },
    {
      "epoch": 0.8807089297886844,
      "grad_norm": 17.094000848151907,
      "learning_rate": 8.903410224194689e-06,
      "loss": 3.4659,
      "step": 2584
    },
    {
      "epoch": 0.8810497614178596,
      "grad_norm": 15.82122564627449,
      "learning_rate": 8.902170635199818e-06,
      "loss": 2.8154,
      "step": 2585
    },
    {
      "epoch": 0.8813905930470347,
      "grad_norm": 18.012259037032894,
      "learning_rate": 8.900930432377972e-06,
      "loss": 2.9192,
      "step": 2586
    },
    {
      "epoch": 0.8817314246762099,
      "grad_norm": 17.303825443668035,
      "learning_rate": 8.899689615924242e-06,
      "loss": 3.0612,
      "step": 2587
    },
    {
      "epoch": 0.8820722563053851,
      "grad_norm": 15.522802778342063,
      "learning_rate": 8.898448186033808e-06,
      "loss": 2.5114,
      "step": 2588
    },
    {
      "epoch": 0.8824130879345603,
      "grad_norm": 34.81177648216177,
      "learning_rate": 8.897206142901958e-06,
      "loss": 3.3171,
      "step": 2589
    },
    {
      "epoch": 0.8827539195637355,
      "grad_norm": 14.443844361280465,
      "learning_rate": 8.895963486724066e-06,
      "loss": 3.016,
      "step": 2590
    },
    {
      "epoch": 0.8830947511929107,
      "grad_norm": 16.608527475103152,
      "learning_rate": 8.894720217695609e-06,
      "loss": 3.2254,
      "step": 2591
    },
    {
      "epoch": 0.8834355828220859,
      "grad_norm": 18.07300450932185,
      "learning_rate": 8.89347633601216e-06,
      "loss": 3.314,
      "step": 2592
    },
    {
      "epoch": 0.8837764144512611,
      "grad_norm": 17.99991018618848,
      "learning_rate": 8.892231841869383e-06,
      "loss": 3.2984,
      "step": 2593
    },
    {
      "epoch": 0.8841172460804363,
      "grad_norm": 23.68684902248102,
      "learning_rate": 8.890986735463042e-06,
      "loss": 3.2368,
      "step": 2594
    },
    {
      "epoch": 0.8844580777096115,
      "grad_norm": 12.377640533811284,
      "learning_rate": 8.889741016989e-06,
      "loss": 2.7937,
      "step": 2595
    },
    {
      "epoch": 0.8847989093387867,
      "grad_norm": 18.93954919053635,
      "learning_rate": 8.88849468664321e-06,
      "loss": 3.2709,
      "step": 2596
    },
    {
      "epoch": 0.8851397409679618,
      "grad_norm": 20.97760510726073,
      "learning_rate": 8.887247744621728e-06,
      "loss": 3.4956,
      "step": 2597
    },
    {
      "epoch": 0.885480572597137,
      "grad_norm": 10.243451637219906,
      "learning_rate": 8.886000191120702e-06,
      "loss": 3.1416,
      "step": 2598
    },
    {
      "epoch": 0.8858214042263122,
      "grad_norm": 17.135549004785315,
      "learning_rate": 8.884752026336376e-06,
      "loss": 3.8823,
      "step": 2599
    },
    {
      "epoch": 0.8861622358554874,
      "grad_norm": 13.179906951645833,
      "learning_rate": 8.883503250465092e-06,
      "loss": 3.063,
      "step": 2600
    },
    {
      "epoch": 0.8865030674846626,
      "grad_norm": 13.801727936293917,
      "learning_rate": 8.882253863703287e-06,
      "loss": 3.1578,
      "step": 2601
    },
    {
      "epoch": 0.8868438991138378,
      "grad_norm": 24.02401969243966,
      "learning_rate": 8.881003866247495e-06,
      "loss": 3.0646,
      "step": 2602
    },
    {
      "epoch": 0.887184730743013,
      "grad_norm": 14.187545116786932,
      "learning_rate": 8.879753258294346e-06,
      "loss": 2.9487,
      "step": 2603
    },
    {
      "epoch": 0.8875255623721882,
      "grad_norm": 14.544883275338014,
      "learning_rate": 8.878502040040564e-06,
      "loss": 3.3355,
      "step": 2604
    },
    {
      "epoch": 0.8878663940013634,
      "grad_norm": 21.339704329949885,
      "learning_rate": 8.877250211682973e-06,
      "loss": 3.3834,
      "step": 2605
    },
    {
      "epoch": 0.8882072256305386,
      "grad_norm": 15.434572608069374,
      "learning_rate": 8.875997773418489e-06,
      "loss": 3.2591,
      "step": 2606
    },
    {
      "epoch": 0.8885480572597138,
      "grad_norm": 19.429338636395887,
      "learning_rate": 8.874744725444124e-06,
      "loss": 3.2709,
      "step": 2607
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 17.940726478218682,
      "learning_rate": 8.873491067956992e-06,
      "loss": 3.8305,
      "step": 2608
    },
    {
      "epoch": 0.889229720518064,
      "grad_norm": 15.15179529357824,
      "learning_rate": 8.872236801154295e-06,
      "loss": 3.0941,
      "step": 2609
    },
    {
      "epoch": 0.8895705521472392,
      "grad_norm": 11.341647373505868,
      "learning_rate": 8.870981925233333e-06,
      "loss": 2.9114,
      "step": 2610
    },
    {
      "epoch": 0.8899113837764144,
      "grad_norm": 19.85778552703168,
      "learning_rate": 8.869726440391506e-06,
      "loss": 3.5337,
      "step": 2611
    },
    {
      "epoch": 0.8902522154055896,
      "grad_norm": 19.261066530346714,
      "learning_rate": 8.868470346826305e-06,
      "loss": 3.0625,
      "step": 2612
    },
    {
      "epoch": 0.8905930470347648,
      "grad_norm": 14.48092293300439,
      "learning_rate": 8.867213644735319e-06,
      "loss": 3.5505,
      "step": 2613
    },
    {
      "epoch": 0.89093387866394,
      "grad_norm": 17.052113760321582,
      "learning_rate": 8.865956334316233e-06,
      "loss": 2.8712,
      "step": 2614
    },
    {
      "epoch": 0.8912747102931152,
      "grad_norm": 17.143128505549743,
      "learning_rate": 8.864698415766825e-06,
      "loss": 3.5641,
      "step": 2615
    },
    {
      "epoch": 0.8916155419222904,
      "grad_norm": 14.717468469098675,
      "learning_rate": 8.86343988928497e-06,
      "loss": 3.0236,
      "step": 2616
    },
    {
      "epoch": 0.8919563735514656,
      "grad_norm": 13.978465000620862,
      "learning_rate": 8.862180755068644e-06,
      "loss": 2.9009,
      "step": 2617
    },
    {
      "epoch": 0.8922972051806408,
      "grad_norm": 14.779266877690182,
      "learning_rate": 8.860921013315911e-06,
      "loss": 3.4647,
      "step": 2618
    },
    {
      "epoch": 0.8926380368098159,
      "grad_norm": 16.323610915739796,
      "learning_rate": 8.859660664224931e-06,
      "loss": 3.069,
      "step": 2619
    },
    {
      "epoch": 0.8929788684389911,
      "grad_norm": 30.01579058630744,
      "learning_rate": 8.858399707993965e-06,
      "loss": 3.3625,
      "step": 2620
    },
    {
      "epoch": 0.8933197000681663,
      "grad_norm": 12.51917313153003,
      "learning_rate": 8.857138144821366e-06,
      "loss": 2.6183,
      "step": 2621
    },
    {
      "epoch": 0.8936605316973415,
      "grad_norm": 18.223577426403516,
      "learning_rate": 8.85587597490558e-06,
      "loss": 2.6415,
      "step": 2622
    },
    {
      "epoch": 0.8940013633265167,
      "grad_norm": 20.171186724358783,
      "learning_rate": 8.854613198445156e-06,
      "loss": 3.1098,
      "step": 2623
    },
    {
      "epoch": 0.8943421949556919,
      "grad_norm": 30.65630707879049,
      "learning_rate": 8.853349815638733e-06,
      "loss": 3.1062,
      "step": 2624
    },
    {
      "epoch": 0.8946830265848671,
      "grad_norm": 22.752388972794545,
      "learning_rate": 8.852085826685043e-06,
      "loss": 3.1068,
      "step": 2625
    },
    {
      "epoch": 0.8950238582140423,
      "grad_norm": 20.93027634972307,
      "learning_rate": 8.850821231782918e-06,
      "loss": 2.9937,
      "step": 2626
    },
    {
      "epoch": 0.8953646898432175,
      "grad_norm": 22.977787477439108,
      "learning_rate": 8.849556031131285e-06,
      "loss": 3.6026,
      "step": 2627
    },
    {
      "epoch": 0.8957055214723927,
      "grad_norm": 17.35702790109682,
      "learning_rate": 8.848290224929163e-06,
      "loss": 2.8426,
      "step": 2628
    },
    {
      "epoch": 0.8960463531015678,
      "grad_norm": 31.52000216206936,
      "learning_rate": 8.84702381337567e-06,
      "loss": 3.3156,
      "step": 2629
    },
    {
      "epoch": 0.896387184730743,
      "grad_norm": 14.874042831185387,
      "learning_rate": 8.845756796670017e-06,
      "loss": 3.6809,
      "step": 2630
    },
    {
      "epoch": 0.8967280163599182,
      "grad_norm": 15.418573738933045,
      "learning_rate": 8.84448917501151e-06,
      "loss": 3.1221,
      "step": 2631
    },
    {
      "epoch": 0.8970688479890934,
      "grad_norm": 18.949302839500017,
      "learning_rate": 8.843220948599553e-06,
      "loss": 3.3833,
      "step": 2632
    },
    {
      "epoch": 0.8974096796182686,
      "grad_norm": 16.447862832899936,
      "learning_rate": 8.841952117633642e-06,
      "loss": 3.2716,
      "step": 2633
    },
    {
      "epoch": 0.8977505112474438,
      "grad_norm": 22.129351682372214,
      "learning_rate": 8.840682682313368e-06,
      "loss": 3.3612,
      "step": 2634
    },
    {
      "epoch": 0.898091342876619,
      "grad_norm": 21.19170689422571,
      "learning_rate": 8.839412642838418e-06,
      "loss": 3.6422,
      "step": 2635
    },
    {
      "epoch": 0.8984321745057942,
      "grad_norm": 17.49137166910942,
      "learning_rate": 8.838141999408577e-06,
      "loss": 3.6466,
      "step": 2636
    },
    {
      "epoch": 0.8987730061349694,
      "grad_norm": 20.8124007650098,
      "learning_rate": 8.83687075222372e-06,
      "loss": 2.9688,
      "step": 2637
    },
    {
      "epoch": 0.8991138377641446,
      "grad_norm": 26.594509608147657,
      "learning_rate": 8.835598901483822e-06,
      "loss": 4.4425,
      "step": 2638
    },
    {
      "epoch": 0.8994546693933198,
      "grad_norm": 17.26414529764415,
      "learning_rate": 8.834326447388946e-06,
      "loss": 3.2178,
      "step": 2639
    },
    {
      "epoch": 0.8997955010224948,
      "grad_norm": 17.110122102804514,
      "learning_rate": 8.833053390139257e-06,
      "loss": 3.6444,
      "step": 2640
    },
    {
      "epoch": 0.90013633265167,
      "grad_norm": 16.658017762330513,
      "learning_rate": 8.831779729935011e-06,
      "loss": 3.2089,
      "step": 2641
    },
    {
      "epoch": 0.9004771642808452,
      "grad_norm": 17.608557432701666,
      "learning_rate": 8.830505466976562e-06,
      "loss": 3.4995,
      "step": 2642
    },
    {
      "epoch": 0.9008179959100204,
      "grad_norm": 13.384039141678082,
      "learning_rate": 8.82923060146435e-06,
      "loss": 2.4669,
      "step": 2643
    },
    {
      "epoch": 0.9011588275391956,
      "grad_norm": 26.086203816286904,
      "learning_rate": 8.827955133598925e-06,
      "loss": 3.2553,
      "step": 2644
    },
    {
      "epoch": 0.9014996591683708,
      "grad_norm": 13.971221993817517,
      "learning_rate": 8.826679063580918e-06,
      "loss": 3.2402,
      "step": 2645
    },
    {
      "epoch": 0.901840490797546,
      "grad_norm": 16.752079543640747,
      "learning_rate": 8.825402391611062e-06,
      "loss": 3.0951,
      "step": 2646
    },
    {
      "epoch": 0.9021813224267212,
      "grad_norm": 17.22384166909549,
      "learning_rate": 8.824125117890182e-06,
      "loss": 3.0001,
      "step": 2647
    },
    {
      "epoch": 0.9025221540558964,
      "grad_norm": 17.316426255556912,
      "learning_rate": 8.822847242619198e-06,
      "loss": 3.2489,
      "step": 2648
    },
    {
      "epoch": 0.9028629856850716,
      "grad_norm": 18.68017830199737,
      "learning_rate": 8.821568765999124e-06,
      "loss": 2.8603,
      "step": 2649
    },
    {
      "epoch": 0.9032038173142468,
      "grad_norm": 20.210483931266573,
      "learning_rate": 8.820289688231069e-06,
      "loss": 3.6506,
      "step": 2650
    },
    {
      "epoch": 0.9035446489434219,
      "grad_norm": 21.567645590345684,
      "learning_rate": 8.81901000951624e-06,
      "loss": 3.3283,
      "step": 2651
    },
    {
      "epoch": 0.9038854805725971,
      "grad_norm": 16.419713208746742,
      "learning_rate": 8.817729730055935e-06,
      "loss": 3.2494,
      "step": 2652
    },
    {
      "epoch": 0.9042263122017723,
      "grad_norm": 14.687542786488185,
      "learning_rate": 8.816448850051544e-06,
      "loss": 3.5863,
      "step": 2653
    },
    {
      "epoch": 0.9045671438309475,
      "grad_norm": 14.057386467627571,
      "learning_rate": 8.815167369704557e-06,
      "loss": 2.8908,
      "step": 2654
    },
    {
      "epoch": 0.9049079754601227,
      "grad_norm": 14.726088569920323,
      "learning_rate": 8.813885289216555e-06,
      "loss": 2.9723,
      "step": 2655
    },
    {
      "epoch": 0.9052488070892979,
      "grad_norm": 14.117743389014294,
      "learning_rate": 8.812602608789216e-06,
      "loss": 2.9577,
      "step": 2656
    },
    {
      "epoch": 0.9055896387184731,
      "grad_norm": 17.904001631391075,
      "learning_rate": 8.811319328624308e-06,
      "loss": 3.6025,
      "step": 2657
    },
    {
      "epoch": 0.9059304703476483,
      "grad_norm": 15.645893785376682,
      "learning_rate": 8.810035448923699e-06,
      "loss": 3.475,
      "step": 2658
    },
    {
      "epoch": 0.9062713019768235,
      "grad_norm": 15.99775212059134,
      "learning_rate": 8.808750969889346e-06,
      "loss": 3.5906,
      "step": 2659
    },
    {
      "epoch": 0.9066121336059987,
      "grad_norm": 11.943592456920813,
      "learning_rate": 8.807465891723306e-06,
      "loss": 2.7103,
      "step": 2660
    },
    {
      "epoch": 0.9069529652351738,
      "grad_norm": 13.171650670181881,
      "learning_rate": 8.806180214627723e-06,
      "loss": 2.9712,
      "step": 2661
    },
    {
      "epoch": 0.907293796864349,
      "grad_norm": 14.68175101835619,
      "learning_rate": 8.804893938804839e-06,
      "loss": 2.9511,
      "step": 2662
    },
    {
      "epoch": 0.9076346284935242,
      "grad_norm": 20.49888645783011,
      "learning_rate": 8.803607064456994e-06,
      "loss": 3.5643,
      "step": 2663
    },
    {
      "epoch": 0.9079754601226994,
      "grad_norm": 13.692582967666993,
      "learning_rate": 8.802319591786617e-06,
      "loss": 2.4134,
      "step": 2664
    },
    {
      "epoch": 0.9083162917518746,
      "grad_norm": 15.720765656577715,
      "learning_rate": 8.801031520996231e-06,
      "loss": 3.4724,
      "step": 2665
    },
    {
      "epoch": 0.9086571233810498,
      "grad_norm": 22.311878984212996,
      "learning_rate": 8.799742852288456e-06,
      "loss": 3.5045,
      "step": 2666
    },
    {
      "epoch": 0.908997955010225,
      "grad_norm": 20.103471092021575,
      "learning_rate": 8.798453585866003e-06,
      "loss": 2.5639,
      "step": 2667
    },
    {
      "epoch": 0.9093387866394002,
      "grad_norm": 18.274332855727753,
      "learning_rate": 8.797163721931681e-06,
      "loss": 3.5239,
      "step": 2668
    },
    {
      "epoch": 0.9096796182685754,
      "grad_norm": 18.27311623503724,
      "learning_rate": 8.79587326068839e-06,
      "loss": 3.7754,
      "step": 2669
    },
    {
      "epoch": 0.9100204498977505,
      "grad_norm": 32.16892554943389,
      "learning_rate": 8.794582202339127e-06,
      "loss": 3.3482,
      "step": 2670
    },
    {
      "epoch": 0.9103612815269257,
      "grad_norm": 18.19777500467198,
      "learning_rate": 8.793290547086975e-06,
      "loss": 3.3267,
      "step": 2671
    },
    {
      "epoch": 0.9107021131561008,
      "grad_norm": 12.965224396513731,
      "learning_rate": 8.791998295135121e-06,
      "loss": 3.0989,
      "step": 2672
    },
    {
      "epoch": 0.911042944785276,
      "grad_norm": 16.202709702746365,
      "learning_rate": 8.79070544668684e-06,
      "loss": 3.5136,
      "step": 2673
    },
    {
      "epoch": 0.9113837764144512,
      "grad_norm": 14.108593067115311,
      "learning_rate": 8.789412001945501e-06,
      "loss": 3.0737,
      "step": 2674
    },
    {
      "epoch": 0.9117246080436264,
      "grad_norm": 18.884089862378104,
      "learning_rate": 8.788117961114571e-06,
      "loss": 3.5057,
      "step": 2675
    },
    {
      "epoch": 0.9120654396728016,
      "grad_norm": 13.962178225779653,
      "learning_rate": 8.786823324397606e-06,
      "loss": 3.0433,
      "step": 2676
    },
    {
      "epoch": 0.9124062713019768,
      "grad_norm": 15.059101769256207,
      "learning_rate": 8.785528091998258e-06,
      "loss": 3.2006,
      "step": 2677
    },
    {
      "epoch": 0.912747102931152,
      "grad_norm": 17.43805567503729,
      "learning_rate": 8.784232264120271e-06,
      "loss": 2.9184,
      "step": 2678
    },
    {
      "epoch": 0.9130879345603272,
      "grad_norm": 15.466016547319532,
      "learning_rate": 8.782935840967484e-06,
      "loss": 3.2058,
      "step": 2679
    },
    {
      "epoch": 0.9134287661895024,
      "grad_norm": 16.73806302869103,
      "learning_rate": 8.781638822743832e-06,
      "loss": 3.6101,
      "step": 2680
    },
    {
      "epoch": 0.9137695978186776,
      "grad_norm": 28.569871862204458,
      "learning_rate": 8.780341209653338e-06,
      "loss": 4.2538,
      "step": 2681
    },
    {
      "epoch": 0.9141104294478528,
      "grad_norm": 23.354157147359675,
      "learning_rate": 8.779043001900123e-06,
      "loss": 3.6315,
      "step": 2682
    },
    {
      "epoch": 0.9144512610770279,
      "grad_norm": 16.928938502856933,
      "learning_rate": 8.777744199688401e-06,
      "loss": 3.3808,
      "step": 2683
    },
    {
      "epoch": 0.9147920927062031,
      "grad_norm": 21.212963380276406,
      "learning_rate": 8.776444803222479e-06,
      "loss": 2.7492,
      "step": 2684
    },
    {
      "epoch": 0.9151329243353783,
      "grad_norm": 14.931545355197645,
      "learning_rate": 8.775144812706755e-06,
      "loss": 3.3506,
      "step": 2685
    },
    {
      "epoch": 0.9154737559645535,
      "grad_norm": 17.45626273219767,
      "learning_rate": 8.773844228345723e-06,
      "loss": 3.2479,
      "step": 2686
    },
    {
      "epoch": 0.9158145875937287,
      "grad_norm": 16.604700182268154,
      "learning_rate": 8.772543050343973e-06,
      "loss": 3.0862,
      "step": 2687
    },
    {
      "epoch": 0.9161554192229039,
      "grad_norm": 19.794854445518318,
      "learning_rate": 8.771241278906181e-06,
      "loss": 2.976,
      "step": 2688
    },
    {
      "epoch": 0.9164962508520791,
      "grad_norm": 11.841613925882248,
      "learning_rate": 8.769938914237127e-06,
      "loss": 2.255,
      "step": 2689
    },
    {
      "epoch": 0.9168370824812543,
      "grad_norm": 15.127237969462035,
      "learning_rate": 8.768635956541671e-06,
      "loss": 3.2708,
      "step": 2690
    },
    {
      "epoch": 0.9171779141104295,
      "grad_norm": 12.39247711522195,
      "learning_rate": 8.767332406024778e-06,
      "loss": 3.3598,
      "step": 2691
    },
    {
      "epoch": 0.9175187457396047,
      "grad_norm": 11.438859153788387,
      "learning_rate": 8.766028262891502e-06,
      "loss": 2.9445,
      "step": 2692
    },
    {
      "epoch": 0.9178595773687799,
      "grad_norm": 13.704188728985493,
      "learning_rate": 8.764723527346987e-06,
      "loss": 3.0133,
      "step": 2693
    },
    {
      "epoch": 0.918200408997955,
      "grad_norm": 27.923795480439864,
      "learning_rate": 8.763418199596476e-06,
      "loss": 3.0981,
      "step": 2694
    },
    {
      "epoch": 0.9185412406271302,
      "grad_norm": 27.89222571287306,
      "learning_rate": 8.762112279845301e-06,
      "loss": 3.0411,
      "step": 2695
    },
    {
      "epoch": 0.9188820722563054,
      "grad_norm": 17.081381969570064,
      "learning_rate": 8.760805768298888e-06,
      "loss": 3.0645,
      "step": 2696
    },
    {
      "epoch": 0.9192229038854806,
      "grad_norm": 13.462935900755797,
      "learning_rate": 8.759498665162755e-06,
      "loss": 3.0763,
      "step": 2697
    },
    {
      "epoch": 0.9195637355146558,
      "grad_norm": 15.352862861408548,
      "learning_rate": 8.758190970642518e-06,
      "loss": 3.1632,
      "step": 2698
    },
    {
      "epoch": 0.919904567143831,
      "grad_norm": 18.889489638635084,
      "learning_rate": 8.756882684943882e-06,
      "loss": 3.6544,
      "step": 2699
    },
    {
      "epoch": 0.9202453987730062,
      "grad_norm": 11.184898164431631,
      "learning_rate": 8.75557380827264e-06,
      "loss": 2.6107,
      "step": 2700
    },
    {
      "epoch": 0.9205862304021813,
      "grad_norm": 16.434723668721904,
      "learning_rate": 8.754264340834693e-06,
      "loss": 3.3371,
      "step": 2701
    },
    {
      "epoch": 0.9209270620313565,
      "grad_norm": 20.540947522442178,
      "learning_rate": 8.752954282836017e-06,
      "loss": 3.5615,
      "step": 2702
    },
    {
      "epoch": 0.9212678936605317,
      "grad_norm": 15.918671334533725,
      "learning_rate": 8.751643634482694e-06,
      "loss": 3.622,
      "step": 2703
    },
    {
      "epoch": 0.9216087252897068,
      "grad_norm": 18.41045783318474,
      "learning_rate": 8.750332395980894e-06,
      "loss": 3.6994,
      "step": 2704
    },
    {
      "epoch": 0.921949556918882,
      "grad_norm": 14.193659772694692,
      "learning_rate": 8.749020567536876e-06,
      "loss": 3.0117,
      "step": 2705
    },
    {
      "epoch": 0.9222903885480572,
      "grad_norm": 19.202775645818324,
      "learning_rate": 8.747708149357e-06,
      "loss": 3.6405,
      "step": 2706
    },
    {
      "epoch": 0.9226312201772324,
      "grad_norm": 19.771601071281292,
      "learning_rate": 8.746395141647715e-06,
      "loss": 3.7243,
      "step": 2707
    },
    {
      "epoch": 0.9229720518064076,
      "grad_norm": 18.844076701969367,
      "learning_rate": 8.745081544615558e-06,
      "loss": 3.4551,
      "step": 2708
    },
    {
      "epoch": 0.9233128834355828,
      "grad_norm": 15.583727724866202,
      "learning_rate": 8.743767358467166e-06,
      "loss": 3.5193,
      "step": 2709
    },
    {
      "epoch": 0.923653715064758,
      "grad_norm": 18.596674852453365,
      "learning_rate": 8.742452583409267e-06,
      "loss": 3.5285,
      "step": 2710
    },
    {
      "epoch": 0.9239945466939332,
      "grad_norm": 15.622796479760206,
      "learning_rate": 8.741137219648679e-06,
      "loss": 3.2276,
      "step": 2711
    },
    {
      "epoch": 0.9243353783231084,
      "grad_norm": 13.533598236824494,
      "learning_rate": 8.739821267392313e-06,
      "loss": 2.9683,
      "step": 2712
    },
    {
      "epoch": 0.9246762099522836,
      "grad_norm": 16.853232302027703,
      "learning_rate": 8.738504726847173e-06,
      "loss": 3.6098,
      "step": 2713
    },
    {
      "epoch": 0.9250170415814588,
      "grad_norm": 18.19295608536347,
      "learning_rate": 8.73718759822036e-06,
      "loss": 3.7979,
      "step": 2714
    },
    {
      "epoch": 0.9253578732106339,
      "grad_norm": 12.266374440986263,
      "learning_rate": 8.735869881719059e-06,
      "loss": 2.8422,
      "step": 2715
    },
    {
      "epoch": 0.9256987048398091,
      "grad_norm": 22.066630149072523,
      "learning_rate": 8.734551577550554e-06,
      "loss": 3.406,
      "step": 2716
    },
    {
      "epoch": 0.9260395364689843,
      "grad_norm": 19.703686450459372,
      "learning_rate": 8.73323268592222e-06,
      "loss": 3.3974,
      "step": 2717
    },
    {
      "epoch": 0.9263803680981595,
      "grad_norm": 13.97626936702887,
      "learning_rate": 8.731913207041523e-06,
      "loss": 3.2129,
      "step": 2718
    },
    {
      "epoch": 0.9267211997273347,
      "grad_norm": 14.229750218347712,
      "learning_rate": 8.730593141116023e-06,
      "loss": 3.1026,
      "step": 2719
    },
    {
      "epoch": 0.9270620313565099,
      "grad_norm": 13.293561825818738,
      "learning_rate": 8.729272488353371e-06,
      "loss": 3.1943,
      "step": 2720
    },
    {
      "epoch": 0.9274028629856851,
      "grad_norm": 13.035256160926515,
      "learning_rate": 8.727951248961312e-06,
      "loss": 2.7408,
      "step": 2721
    },
    {
      "epoch": 0.9277436946148603,
      "grad_norm": 14.915716848096986,
      "learning_rate": 8.72662942314768e-06,
      "loss": 3.275,
      "step": 2722
    },
    {
      "epoch": 0.9280845262440355,
      "grad_norm": 19.185895609534086,
      "learning_rate": 8.725307011120406e-06,
      "loss": 3.3366,
      "step": 2723
    },
    {
      "epoch": 0.9284253578732107,
      "grad_norm": 15.085483594969737,
      "learning_rate": 8.72398401308751e-06,
      "loss": 3.0332,
      "step": 2724
    },
    {
      "epoch": 0.9287661895023859,
      "grad_norm": 21.545592371847707,
      "learning_rate": 8.722660429257103e-06,
      "loss": 3.276,
      "step": 2725
    },
    {
      "epoch": 0.929107021131561,
      "grad_norm": 17.68774475685339,
      "learning_rate": 8.721336259837391e-06,
      "loss": 3.3784,
      "step": 2726
    },
    {
      "epoch": 0.9294478527607362,
      "grad_norm": 12.511006972962107,
      "learning_rate": 8.720011505036674e-06,
      "loss": 3.2995,
      "step": 2727
    },
    {
      "epoch": 0.9297886843899114,
      "grad_norm": 14.181119863045677,
      "learning_rate": 8.718686165063337e-06,
      "loss": 3.3101,
      "step": 2728
    },
    {
      "epoch": 0.9301295160190866,
      "grad_norm": 24.777411971917303,
      "learning_rate": 8.717360240125863e-06,
      "loss": 3.3187,
      "step": 2729
    },
    {
      "epoch": 0.9304703476482618,
      "grad_norm": 22.809261269472046,
      "learning_rate": 8.716033730432826e-06,
      "loss": 3.5693,
      "step": 2730
    },
    {
      "epoch": 0.930811179277437,
      "grad_norm": 36.59921946401887,
      "learning_rate": 8.714706636192892e-06,
      "loss": 3.401,
      "step": 2731
    },
    {
      "epoch": 0.9311520109066121,
      "grad_norm": 14.207790415825853,
      "learning_rate": 8.713378957614813e-06,
      "loss": 2.3107,
      "step": 2732
    },
    {
      "epoch": 0.9314928425357873,
      "grad_norm": 12.444214000057753,
      "learning_rate": 8.712050694907443e-06,
      "loss": 2.7456,
      "step": 2733
    },
    {
      "epoch": 0.9318336741649625,
      "grad_norm": 18.93552776140509,
      "learning_rate": 8.710721848279724e-06,
      "loss": 3.3428,
      "step": 2734
    },
    {
      "epoch": 0.9321745057941377,
      "grad_norm": 16.145907929162163,
      "learning_rate": 8.709392417940686e-06,
      "loss": 3.6313,
      "step": 2735
    },
    {
      "epoch": 0.9325153374233128,
      "grad_norm": 14.203083083293196,
      "learning_rate": 8.708062404099453e-06,
      "loss": 2.8628,
      "step": 2736
    },
    {
      "epoch": 0.932856169052488,
      "grad_norm": 12.18440033169147,
      "learning_rate": 8.706731806965244e-06,
      "loss": 2.965,
      "step": 2737
    },
    {
      "epoch": 0.9331970006816632,
      "grad_norm": 13.658952819883211,
      "learning_rate": 8.705400626747365e-06,
      "loss": 3.1434,
      "step": 2738
    },
    {
      "epoch": 0.9335378323108384,
      "grad_norm": 21.84118212201773,
      "learning_rate": 8.704068863655218e-06,
      "loss": 3.1146,
      "step": 2739
    },
    {
      "epoch": 0.9338786639400136,
      "grad_norm": 23.554320503840835,
      "learning_rate": 8.702736517898294e-06,
      "loss": 3.2486,
      "step": 2740
    },
    {
      "epoch": 0.9342194955691888,
      "grad_norm": 16.606611755516095,
      "learning_rate": 8.701403589686175e-06,
      "loss": 3.0872,
      "step": 2741
    },
    {
      "epoch": 0.934560327198364,
      "grad_norm": 18.685607040478704,
      "learning_rate": 8.700070079228537e-06,
      "loss": 3.4153,
      "step": 2742
    },
    {
      "epoch": 0.9349011588275392,
      "grad_norm": 35.15931435692519,
      "learning_rate": 8.698735986735148e-06,
      "loss": 4.0946,
      "step": 2743
    },
    {
      "epoch": 0.9352419904567144,
      "grad_norm": 14.370839468399605,
      "learning_rate": 8.697401312415863e-06,
      "loss": 3.1713,
      "step": 2744
    },
    {
      "epoch": 0.9355828220858896,
      "grad_norm": 17.09463103909829,
      "learning_rate": 8.696066056480633e-06,
      "loss": 3.6148,
      "step": 2745
    },
    {
      "epoch": 0.9359236537150648,
      "grad_norm": 22.230872435832456,
      "learning_rate": 8.6947302191395e-06,
      "loss": 3.6029,
      "step": 2746
    },
    {
      "epoch": 0.9362644853442399,
      "grad_norm": 16.04417335363527,
      "learning_rate": 8.693393800602594e-06,
      "loss": 3.4435,
      "step": 2747
    },
    {
      "epoch": 0.9366053169734151,
      "grad_norm": 16.284304007893816,
      "learning_rate": 8.692056801080143e-06,
      "loss": 2.6488,
      "step": 2748
    },
    {
      "epoch": 0.9369461486025903,
      "grad_norm": 11.445700258765418,
      "learning_rate": 8.690719220782458e-06,
      "loss": 2.3689,
      "step": 2749
    },
    {
      "epoch": 0.9372869802317655,
      "grad_norm": 17.7027424628809,
      "learning_rate": 8.689381059919949e-06,
      "loss": 3.3164,
      "step": 2750
    },
    {
      "epoch": 0.9376278118609407,
      "grad_norm": 15.125364443592483,
      "learning_rate": 8.688042318703111e-06,
      "loss": 3.4067,
      "step": 2751
    },
    {
      "epoch": 0.9379686434901159,
      "grad_norm": 15.873644132733526,
      "learning_rate": 8.686702997342538e-06,
      "loss": 3.415,
      "step": 2752
    },
    {
      "epoch": 0.9383094751192911,
      "grad_norm": 28.75980769495674,
      "learning_rate": 8.685363096048905e-06,
      "loss": 3.8048,
      "step": 2753
    },
    {
      "epoch": 0.9386503067484663,
      "grad_norm": 11.800571858379246,
      "learning_rate": 8.684022615032988e-06,
      "loss": 3.2646,
      "step": 2754
    },
    {
      "epoch": 0.9389911383776415,
      "grad_norm": 15.592548451780727,
      "learning_rate": 8.682681554505647e-06,
      "loss": 2.8499,
      "step": 2755
    },
    {
      "epoch": 0.9393319700068167,
      "grad_norm": 18.752358690905794,
      "learning_rate": 8.681339914677838e-06,
      "loss": 3.671,
      "step": 2756
    },
    {
      "epoch": 0.9396728016359919,
      "grad_norm": 18.08280658542198,
      "learning_rate": 8.679997695760606e-06,
      "loss": 3.1847,
      "step": 2757
    },
    {
      "epoch": 0.940013633265167,
      "grad_norm": 23.424175429876556,
      "learning_rate": 8.678654897965087e-06,
      "loss": 3.3421,
      "step": 2758
    },
    {
      "epoch": 0.9403544648943422,
      "grad_norm": 22.51120242730242,
      "learning_rate": 8.67731152150251e-06,
      "loss": 3.3113,
      "step": 2759
    },
    {
      "epoch": 0.9406952965235174,
      "grad_norm": 16.199334017975403,
      "learning_rate": 8.67596756658419e-06,
      "loss": 2.9616,
      "step": 2760
    },
    {
      "epoch": 0.9410361281526926,
      "grad_norm": 15.933501695689928,
      "learning_rate": 8.67462303342154e-06,
      "loss": 3.1875,
      "step": 2761
    },
    {
      "epoch": 0.9413769597818678,
      "grad_norm": 19.417927762783215,
      "learning_rate": 8.673277922226058e-06,
      "loss": 3.5381,
      "step": 2762
    },
    {
      "epoch": 0.941717791411043,
      "grad_norm": 18.158430632870704,
      "learning_rate": 8.671932233209338e-06,
      "loss": 3.1784,
      "step": 2763
    },
    {
      "epoch": 0.9420586230402181,
      "grad_norm": 18.198945425879312,
      "learning_rate": 8.670585966583059e-06,
      "loss": 3.4108,
      "step": 2764
    },
    {
      "epoch": 0.9423994546693933,
      "grad_norm": 19.265427772739574,
      "learning_rate": 8.669239122558996e-06,
      "loss": 3.1273,
      "step": 2765
    },
    {
      "epoch": 0.9427402862985685,
      "grad_norm": 16.406940735270705,
      "learning_rate": 8.667891701349014e-06,
      "loss": 3.1972,
      "step": 2766
    },
    {
      "epoch": 0.9430811179277437,
      "grad_norm": 16.183927657605565,
      "learning_rate": 8.666543703165065e-06,
      "loss": 3.0891,
      "step": 2767
    },
    {
      "epoch": 0.9434219495569189,
      "grad_norm": 14.948981902038565,
      "learning_rate": 8.665195128219197e-06,
      "loss": 3.4925,
      "step": 2768
    },
    {
      "epoch": 0.943762781186094,
      "grad_norm": 21.94712150785344,
      "learning_rate": 8.663845976723544e-06,
      "loss": 3.449,
      "step": 2769
    },
    {
      "epoch": 0.9441036128152692,
      "grad_norm": 16.179526533716203,
      "learning_rate": 8.662496248890333e-06,
      "loss": 3.3989,
      "step": 2770
    },
    {
      "epoch": 0.9444444444444444,
      "grad_norm": 14.886384930371841,
      "learning_rate": 8.661145944931884e-06,
      "loss": 3.4246,
      "step": 2771
    },
    {
      "epoch": 0.9447852760736196,
      "grad_norm": 13.871801816015386,
      "learning_rate": 8.659795065060603e-06,
      "loss": 3.016,
      "step": 2772
    },
    {
      "epoch": 0.9451261077027948,
      "grad_norm": 14.398798020441092,
      "learning_rate": 8.658443609488989e-06,
      "loss": 3.5073,
      "step": 2773
    },
    {
      "epoch": 0.94546693933197,
      "grad_norm": 12.433862849611405,
      "learning_rate": 8.657091578429632e-06,
      "loss": 3.0096,
      "step": 2774
    },
    {
      "epoch": 0.9458077709611452,
      "grad_norm": 16.096485777415598,
      "learning_rate": 8.655738972095214e-06,
      "loss": 3.4204,
      "step": 2775
    },
    {
      "epoch": 0.9461486025903204,
      "grad_norm": 15.210361184852108,
      "learning_rate": 8.6543857906985e-06,
      "loss": 3.0871,
      "step": 2776
    },
    {
      "epoch": 0.9464894342194956,
      "grad_norm": 17.54791898069647,
      "learning_rate": 8.653032034452356e-06,
      "loss": 3.8313,
      "step": 2777
    },
    {
      "epoch": 0.9468302658486708,
      "grad_norm": 16.327277131264548,
      "learning_rate": 8.65167770356973e-06,
      "loss": 3.4626,
      "step": 2778
    },
    {
      "epoch": 0.9471710974778459,
      "grad_norm": 16.533522416838554,
      "learning_rate": 8.650322798263664e-06,
      "loss": 3.166,
      "step": 2779
    },
    {
      "epoch": 0.9475119291070211,
      "grad_norm": 9.905318344692155,
      "learning_rate": 8.648967318747293e-06,
      "loss": 2.904,
      "step": 2780
    },
    {
      "epoch": 0.9478527607361963,
      "grad_norm": 16.51945061178983,
      "learning_rate": 8.647611265233836e-06,
      "loss": 3.5982,
      "step": 2781
    },
    {
      "epoch": 0.9481935923653715,
      "grad_norm": 18.722744250204418,
      "learning_rate": 8.646254637936606e-06,
      "loss": 3.6472,
      "step": 2782
    },
    {
      "epoch": 0.9485344239945467,
      "grad_norm": 17.55355286550268,
      "learning_rate": 8.644897437069007e-06,
      "loss": 3.5161,
      "step": 2783
    },
    {
      "epoch": 0.9488752556237219,
      "grad_norm": 16.187964154447286,
      "learning_rate": 8.643539662844531e-06,
      "loss": 3.1311,
      "step": 2784
    },
    {
      "epoch": 0.9492160872528971,
      "grad_norm": 16.21546403529476,
      "learning_rate": 8.642181315476764e-06,
      "loss": 3.0491,
      "step": 2785
    },
    {
      "epoch": 0.9495569188820723,
      "grad_norm": 17.23886591257753,
      "learning_rate": 8.640822395179377e-06,
      "loss": 3.5102,
      "step": 2786
    },
    {
      "epoch": 0.9498977505112475,
      "grad_norm": 18.12918007257363,
      "learning_rate": 8.63946290216613e-06,
      "loss": 3.6069,
      "step": 2787
    },
    {
      "epoch": 0.9502385821404227,
      "grad_norm": 20.17444667108366,
      "learning_rate": 8.638102836650885e-06,
      "loss": 3.6134,
      "step": 2788
    },
    {
      "epoch": 0.9505794137695979,
      "grad_norm": 31.810541330336193,
      "learning_rate": 8.636742198847579e-06,
      "loss": 3.2792,
      "step": 2789
    },
    {
      "epoch": 0.950920245398773,
      "grad_norm": 21.62079264536407,
      "learning_rate": 8.635380988970249e-06,
      "loss": 3.3417,
      "step": 2790
    },
    {
      "epoch": 0.9512610770279482,
      "grad_norm": 14.991020595826525,
      "learning_rate": 8.634019207233017e-06,
      "loss": 3.3229,
      "step": 2791
    },
    {
      "epoch": 0.9516019086571234,
      "grad_norm": 24.13921038669793,
      "learning_rate": 8.632656853850098e-06,
      "loss": 3.6314,
      "step": 2792
    },
    {
      "epoch": 0.9519427402862985,
      "grad_norm": 13.864205026343505,
      "learning_rate": 8.631293929035797e-06,
      "loss": 3.2958,
      "step": 2793
    },
    {
      "epoch": 0.9522835719154737,
      "grad_norm": 16.135151450139325,
      "learning_rate": 8.629930433004504e-06,
      "loss": 3.3525,
      "step": 2794
    },
    {
      "epoch": 0.952624403544649,
      "grad_norm": 13.530345918015703,
      "learning_rate": 8.628566365970703e-06,
      "loss": 3.1166,
      "step": 2795
    },
    {
      "epoch": 0.9529652351738241,
      "grad_norm": 13.395524803528065,
      "learning_rate": 8.62720172814897e-06,
      "loss": 3.031,
      "step": 2796
    },
    {
      "epoch": 0.9533060668029993,
      "grad_norm": 20.088889289219562,
      "learning_rate": 8.625836519753965e-06,
      "loss": 3.437,
      "step": 2797
    },
    {
      "epoch": 0.9536468984321745,
      "grad_norm": 16.8772019269885,
      "learning_rate": 8.624470741000443e-06,
      "loss": 3.3881,
      "step": 2798
    },
    {
      "epoch": 0.9539877300613497,
      "grad_norm": 12.476572434961101,
      "learning_rate": 8.623104392103243e-06,
      "loss": 2.7556,
      "step": 2799
    },
    {
      "epoch": 0.9543285616905249,
      "grad_norm": 28.22830494037833,
      "learning_rate": 8.621737473277301e-06,
      "loss": 3.1807,
      "step": 2800
    },
    {
      "epoch": 0.9546693933197,
      "grad_norm": 11.542727984623077,
      "learning_rate": 8.620369984737638e-06,
      "loss": 3.3033,
      "step": 2801
    },
    {
      "epoch": 0.9550102249488752,
      "grad_norm": 17.371922159090055,
      "learning_rate": 8.619001926699363e-06,
      "loss": 3.3585,
      "step": 2802
    },
    {
      "epoch": 0.9553510565780504,
      "grad_norm": 17.33519073412746,
      "learning_rate": 8.617633299377679e-06,
      "loss": 3.4994,
      "step": 2803
    },
    {
      "epoch": 0.9556918882072256,
      "grad_norm": 14.920739044936182,
      "learning_rate": 8.616264102987875e-06,
      "loss": 3.1437,
      "step": 2804
    },
    {
      "epoch": 0.9560327198364008,
      "grad_norm": 18.10337417746916,
      "learning_rate": 8.614894337745334e-06,
      "loss": 3.2551,
      "step": 2805
    },
    {
      "epoch": 0.956373551465576,
      "grad_norm": 18.483606071374336,
      "learning_rate": 8.613524003865523e-06,
      "loss": 3.4272,
      "step": 2806
    },
    {
      "epoch": 0.9567143830947512,
      "grad_norm": 19.349040179563065,
      "learning_rate": 8.612153101564e-06,
      "loss": 3.1107,
      "step": 2807
    },
    {
      "epoch": 0.9570552147239264,
      "grad_norm": 24.70065785818656,
      "learning_rate": 8.610781631056417e-06,
      "loss": 3.7859,
      "step": 2808
    },
    {
      "epoch": 0.9573960463531016,
      "grad_norm": 15.628566646374786,
      "learning_rate": 8.609409592558509e-06,
      "loss": 3.7556,
      "step": 2809
    },
    {
      "epoch": 0.9577368779822768,
      "grad_norm": 32.49965967039522,
      "learning_rate": 8.608036986286104e-06,
      "loss": 4.3735,
      "step": 2810
    },
    {
      "epoch": 0.9580777096114519,
      "grad_norm": 17.341015277489873,
      "learning_rate": 8.606663812455118e-06,
      "loss": 3.0412,
      "step": 2811
    },
    {
      "epoch": 0.9584185412406271,
      "grad_norm": 19.28534746699952,
      "learning_rate": 8.605290071281556e-06,
      "loss": 3.2379,
      "step": 2812
    },
    {
      "epoch": 0.9587593728698023,
      "grad_norm": 20.42147143675425,
      "learning_rate": 8.603915762981516e-06,
      "loss": 3.1202,
      "step": 2813
    },
    {
      "epoch": 0.9591002044989775,
      "grad_norm": 22.623211392126045,
      "learning_rate": 8.602540887771179e-06,
      "loss": 3.2372,
      "step": 2814
    },
    {
      "epoch": 0.9594410361281527,
      "grad_norm": 21.14254697777601,
      "learning_rate": 8.60116544586682e-06,
      "loss": 3.6696,
      "step": 2815
    },
    {
      "epoch": 0.9597818677573279,
      "grad_norm": 14.773857188665795,
      "learning_rate": 8.599789437484801e-06,
      "loss": 3.0701,
      "step": 2816
    },
    {
      "epoch": 0.9601226993865031,
      "grad_norm": 14.412174191576682,
      "learning_rate": 8.598412862841572e-06,
      "loss": 2.5487,
      "step": 2817
    },
    {
      "epoch": 0.9604635310156783,
      "grad_norm": 13.288714071078264,
      "learning_rate": 8.597035722153677e-06,
      "loss": 3.2021,
      "step": 2818
    },
    {
      "epoch": 0.9608043626448535,
      "grad_norm": 14.045592999443734,
      "learning_rate": 8.595658015637744e-06,
      "loss": 3.1294,
      "step": 2819
    },
    {
      "epoch": 0.9611451942740287,
      "grad_norm": 15.447504999895811,
      "learning_rate": 8.59427974351049e-06,
      "loss": 3.6895,
      "step": 2820
    },
    {
      "epoch": 0.9614860259032039,
      "grad_norm": 12.542480614008712,
      "learning_rate": 8.592900905988726e-06,
      "loss": 2.817,
      "step": 2821
    },
    {
      "epoch": 0.961826857532379,
      "grad_norm": 17.228522426485057,
      "learning_rate": 8.591521503289347e-06,
      "loss": 3.113,
      "step": 2822
    },
    {
      "epoch": 0.9621676891615542,
      "grad_norm": 25.144880355395923,
      "learning_rate": 8.590141535629338e-06,
      "loss": 3.6082,
      "step": 2823
    },
    {
      "epoch": 0.9625085207907293,
      "grad_norm": 17.27745823503467,
      "learning_rate": 8.588761003225772e-06,
      "loss": 3.5341,
      "step": 2824
    },
    {
      "epoch": 0.9628493524199045,
      "grad_norm": 18.39118197066289,
      "learning_rate": 8.587379906295817e-06,
      "loss": 3.5686,
      "step": 2825
    },
    {
      "epoch": 0.9631901840490797,
      "grad_norm": 13.146205143065188,
      "learning_rate": 8.585998245056721e-06,
      "loss": 3.2085,
      "step": 2826
    },
    {
      "epoch": 0.9635310156782549,
      "grad_norm": 25.776063887366817,
      "learning_rate": 8.584616019725826e-06,
      "loss": 3.1908,
      "step": 2827
    },
    {
      "epoch": 0.9638718473074301,
      "grad_norm": 23.965906389494332,
      "learning_rate": 8.583233230520562e-06,
      "loss": 3.1167,
      "step": 2828
    },
    {
      "epoch": 0.9642126789366053,
      "grad_norm": 16.25548217711139,
      "learning_rate": 8.581849877658445e-06,
      "loss": 3.2865,
      "step": 2829
    },
    {
      "epoch": 0.9645535105657805,
      "grad_norm": 14.881080416380206,
      "learning_rate": 8.580465961357087e-06,
      "loss": 3.2609,
      "step": 2830
    },
    {
      "epoch": 0.9648943421949557,
      "grad_norm": 43.1211620956575,
      "learning_rate": 8.579081481834177e-06,
      "loss": 3.4217,
      "step": 2831
    },
    {
      "epoch": 0.9652351738241309,
      "grad_norm": 16.068810100939025,
      "learning_rate": 8.577696439307506e-06,
      "loss": 3.2813,
      "step": 2832
    },
    {
      "epoch": 0.965576005453306,
      "grad_norm": 14.097840241761935,
      "learning_rate": 8.57631083399494e-06,
      "loss": 3.2195,
      "step": 2833
    },
    {
      "epoch": 0.9659168370824812,
      "grad_norm": 19.43650617921922,
      "learning_rate": 8.574924666114445e-06,
      "loss": 3.5986,
      "step": 2834
    },
    {
      "epoch": 0.9662576687116564,
      "grad_norm": 25.312086093731,
      "learning_rate": 8.573537935884067e-06,
      "loss": 3.8416,
      "step": 2835
    },
    {
      "epoch": 0.9665985003408316,
      "grad_norm": 24.261166727951263,
      "learning_rate": 8.572150643521949e-06,
      "loss": 3.6539,
      "step": 2836
    },
    {
      "epoch": 0.9669393319700068,
      "grad_norm": 13.29018363665877,
      "learning_rate": 8.570762789246314e-06,
      "loss": 3.0425,
      "step": 2837
    },
    {
      "epoch": 0.967280163599182,
      "grad_norm": 12.720455962080838,
      "learning_rate": 8.569374373275477e-06,
      "loss": 2.9033,
      "step": 2838
    },
    {
      "epoch": 0.9676209952283572,
      "grad_norm": 21.084955632282604,
      "learning_rate": 8.567985395827843e-06,
      "loss": 3.848,
      "step": 2839
    },
    {
      "epoch": 0.9679618268575324,
      "grad_norm": 19.519430392014677,
      "learning_rate": 8.566595857121902e-06,
      "loss": 2.5476,
      "step": 2840
    },
    {
      "epoch": 0.9683026584867076,
      "grad_norm": 16.45533444935172,
      "learning_rate": 8.565205757376235e-06,
      "loss": 3.6921,
      "step": 2841
    },
    {
      "epoch": 0.9686434901158828,
      "grad_norm": 17.016346178739973,
      "learning_rate": 8.563815096809512e-06,
      "loss": 3.1871,
      "step": 2842
    },
    {
      "epoch": 0.9689843217450579,
      "grad_norm": 12.110584699619704,
      "learning_rate": 8.562423875640485e-06,
      "loss": 3.3109,
      "step": 2843
    },
    {
      "epoch": 0.9693251533742331,
      "grad_norm": 19.45734810317632,
      "learning_rate": 8.561032094088002e-06,
      "loss": 3.1773,
      "step": 2844
    },
    {
      "epoch": 0.9696659850034083,
      "grad_norm": 17.766459954813246,
      "learning_rate": 8.559639752370997e-06,
      "loss": 3.6258,
      "step": 2845
    },
    {
      "epoch": 0.9700068166325835,
      "grad_norm": 16.937165269537452,
      "learning_rate": 8.558246850708485e-06,
      "loss": 3.152,
      "step": 2846
    },
    {
      "epoch": 0.9703476482617587,
      "grad_norm": 20.992291599324776,
      "learning_rate": 8.556853389319581e-06,
      "loss": 2.7195,
      "step": 2847
    },
    {
      "epoch": 0.9706884798909339,
      "grad_norm": 13.08818643863256,
      "learning_rate": 8.55545936842348e-06,
      "loss": 2.9693,
      "step": 2848
    },
    {
      "epoch": 0.9710293115201091,
      "grad_norm": 18.672059535501482,
      "learning_rate": 8.554064788239467e-06,
      "loss": 2.6755,
      "step": 2849
    },
    {
      "epoch": 0.9713701431492843,
      "grad_norm": 29.020310783601275,
      "learning_rate": 8.552669648986914e-06,
      "loss": 3.7968,
      "step": 2850
    },
    {
      "epoch": 0.9717109747784595,
      "grad_norm": 18.18303304751928,
      "learning_rate": 8.551273950885284e-06,
      "loss": 3.5419,
      "step": 2851
    },
    {
      "epoch": 0.9720518064076347,
      "grad_norm": 12.975828210593502,
      "learning_rate": 8.549877694154123e-06,
      "loss": 4.333,
      "step": 2852
    },
    {
      "epoch": 0.9723926380368099,
      "grad_norm": 29.32352474008414,
      "learning_rate": 8.54848087901307e-06,
      "loss": 3.1318,
      "step": 2853
    },
    {
      "epoch": 0.972733469665985,
      "grad_norm": 16.581712830546596,
      "learning_rate": 8.547083505681851e-06,
      "loss": 3.2858,
      "step": 2854
    },
    {
      "epoch": 0.9730743012951601,
      "grad_norm": 12.72710372072233,
      "learning_rate": 8.545685574380274e-06,
      "loss": 2.8104,
      "step": 2855
    },
    {
      "epoch": 0.9734151329243353,
      "grad_norm": 16.169013520474138,
      "learning_rate": 8.544287085328244e-06,
      "loss": 3.514,
      "step": 2856
    },
    {
      "epoch": 0.9737559645535105,
      "grad_norm": 17.275093009137397,
      "learning_rate": 8.542888038745745e-06,
      "loss": 3.4519,
      "step": 2857
    },
    {
      "epoch": 0.9740967961826857,
      "grad_norm": 14.065451497692326,
      "learning_rate": 8.541488434852854e-06,
      "loss": 2.6869,
      "step": 2858
    },
    {
      "epoch": 0.9744376278118609,
      "grad_norm": 23.742721073734234,
      "learning_rate": 8.540088273869735e-06,
      "loss": 3.7695,
      "step": 2859
    },
    {
      "epoch": 0.9747784594410361,
      "grad_norm": 13.588248602132353,
      "learning_rate": 8.538687556016638e-06,
      "loss": 2.9224,
      "step": 2860
    },
    {
      "epoch": 0.9751192910702113,
      "grad_norm": 14.900002884669592,
      "learning_rate": 8.537286281513901e-06,
      "loss": 3.5515,
      "step": 2861
    },
    {
      "epoch": 0.9754601226993865,
      "grad_norm": 16.134431053705402,
      "learning_rate": 8.53588445058195e-06,
      "loss": 3.5142,
      "step": 2862
    },
    {
      "epoch": 0.9758009543285617,
      "grad_norm": 10.883769333870887,
      "learning_rate": 8.534482063441303e-06,
      "loss": 3.0691,
      "step": 2863
    },
    {
      "epoch": 0.9761417859577369,
      "grad_norm": 13.600577558971208,
      "learning_rate": 8.533079120312554e-06,
      "loss": 2.7937,
      "step": 2864
    },
    {
      "epoch": 0.976482617586912,
      "grad_norm": 13.573475197972893,
      "learning_rate": 8.531675621416396e-06,
      "loss": 3.1008,
      "step": 2865
    },
    {
      "epoch": 0.9768234492160872,
      "grad_norm": 17.649423643282336,
      "learning_rate": 8.530271566973605e-06,
      "loss": 3.784,
      "step": 2866
    },
    {
      "epoch": 0.9771642808452624,
      "grad_norm": 14.960649376697887,
      "learning_rate": 8.528866957205041e-06,
      "loss": 2.6888,
      "step": 2867
    },
    {
      "epoch": 0.9775051124744376,
      "grad_norm": 16.23850326656891,
      "learning_rate": 8.527461792331659e-06,
      "loss": 3.4392,
      "step": 2868
    },
    {
      "epoch": 0.9778459441036128,
      "grad_norm": 17.200012591753573,
      "learning_rate": 8.526056072574495e-06,
      "loss": 3.2917,
      "step": 2869
    },
    {
      "epoch": 0.978186775732788,
      "grad_norm": 38.536473449098004,
      "learning_rate": 8.524649798154674e-06,
      "loss": 2.436,
      "step": 2870
    },
    {
      "epoch": 0.9785276073619632,
      "grad_norm": 20.5623174557345,
      "learning_rate": 8.523242969293408e-06,
      "loss": 2.8192,
      "step": 2871
    },
    {
      "epoch": 0.9788684389911384,
      "grad_norm": 24.979037942336948,
      "learning_rate": 8.521835586211998e-06,
      "loss": 3.7572,
      "step": 2872
    },
    {
      "epoch": 0.9792092706203136,
      "grad_norm": 9.606435350033108,
      "learning_rate": 8.520427649131833e-06,
      "loss": 2.8564,
      "step": 2873
    },
    {
      "epoch": 0.9795501022494888,
      "grad_norm": 19.990228436366976,
      "learning_rate": 8.519019158274382e-06,
      "loss": 2.9949,
      "step": 2874
    },
    {
      "epoch": 0.979890933878664,
      "grad_norm": 10.992664656057984,
      "learning_rate": 8.51761011386121e-06,
      "loss": 3.2158,
      "step": 2875
    },
    {
      "epoch": 0.9802317655078391,
      "grad_norm": 33.40465598857599,
      "learning_rate": 8.516200516113965e-06,
      "loss": 3.5498,
      "step": 2876
    },
    {
      "epoch": 0.9805725971370143,
      "grad_norm": 11.92556448221305,
      "learning_rate": 8.514790365254379e-06,
      "loss": 2.9747,
      "step": 2877
    },
    {
      "epoch": 0.9809134287661895,
      "grad_norm": 24.93437219490221,
      "learning_rate": 8.51337966150428e-06,
      "loss": 2.6377,
      "step": 2878
    },
    {
      "epoch": 0.9812542603953647,
      "grad_norm": 17.719848072683153,
      "learning_rate": 8.511968405085574e-06,
      "loss": 3.5006,
      "step": 2879
    },
    {
      "epoch": 0.9815950920245399,
      "grad_norm": 17.49859602308155,
      "learning_rate": 8.510556596220257e-06,
      "loss": 2.8587,
      "step": 2880
    },
    {
      "epoch": 0.9819359236537151,
      "grad_norm": 18.445338803475764,
      "learning_rate": 8.509144235130412e-06,
      "loss": 3.5915,
      "step": 2881
    },
    {
      "epoch": 0.9822767552828903,
      "grad_norm": 23.00355003264217,
      "learning_rate": 8.50773132203821e-06,
      "loss": 3.2127,
      "step": 2882
    },
    {
      "epoch": 0.9826175869120655,
      "grad_norm": 20.32650459532349,
      "learning_rate": 8.506317857165906e-06,
      "loss": 3.7823,
      "step": 2883
    },
    {
      "epoch": 0.9829584185412407,
      "grad_norm": 19.230098443090707,
      "learning_rate": 8.504903840735847e-06,
      "loss": 2.6148,
      "step": 2884
    },
    {
      "epoch": 0.9832992501704159,
      "grad_norm": 25.51961535534717,
      "learning_rate": 8.503489272970463e-06,
      "loss": 3.4922,
      "step": 2885
    },
    {
      "epoch": 0.983640081799591,
      "grad_norm": 28.534901733508026,
      "learning_rate": 8.502074154092268e-06,
      "loss": 3.3244,
      "step": 2886
    },
    {
      "epoch": 0.9839809134287661,
      "grad_norm": 12.550966436926648,
      "learning_rate": 8.500658484323868e-06,
      "loss": 3.0892,
      "step": 2887
    },
    {
      "epoch": 0.9843217450579413,
      "grad_norm": 16.253272217913988,
      "learning_rate": 8.499242263887953e-06,
      "loss": 3.7861,
      "step": 2888
    },
    {
      "epoch": 0.9846625766871165,
      "grad_norm": 16.6406372188896,
      "learning_rate": 8.4978254930073e-06,
      "loss": 3.6091,
      "step": 2889
    },
    {
      "epoch": 0.9850034083162917,
      "grad_norm": 11.43871572211583,
      "learning_rate": 8.496408171904773e-06,
      "loss": 3.1258,
      "step": 2890
    },
    {
      "epoch": 0.9853442399454669,
      "grad_norm": 13.619775771258272,
      "learning_rate": 8.494990300803323e-06,
      "loss": 3.0446,
      "step": 2891
    },
    {
      "epoch": 0.9856850715746421,
      "grad_norm": 30.91177081318645,
      "learning_rate": 8.493571879925986e-06,
      "loss": 3.3357,
      "step": 2892
    },
    {
      "epoch": 0.9860259032038173,
      "grad_norm": 11.886838084074983,
      "learning_rate": 8.492152909495883e-06,
      "loss": 3.1527,
      "step": 2893
    },
    {
      "epoch": 0.9863667348329925,
      "grad_norm": 21.1999653831076,
      "learning_rate": 8.490733389736227e-06,
      "loss": 3.5012,
      "step": 2894
    },
    {
      "epoch": 0.9867075664621677,
      "grad_norm": 18.61527971186803,
      "learning_rate": 8.489313320870313e-06,
      "loss": 3.7613,
      "step": 2895
    },
    {
      "epoch": 0.9870483980913429,
      "grad_norm": 24.84226702527376,
      "learning_rate": 8.487892703121525e-06,
      "loss": 3.642,
      "step": 2896
    },
    {
      "epoch": 0.987389229720518,
      "grad_norm": 12.852360192140019,
      "learning_rate": 8.486471536713328e-06,
      "loss": 2.6919,
      "step": 2897
    },
    {
      "epoch": 0.9877300613496932,
      "grad_norm": 12.885476803797896,
      "learning_rate": 8.485049821869282e-06,
      "loss": 3.1084,
      "step": 2898
    },
    {
      "epoch": 0.9880708929788684,
      "grad_norm": 17.997597498370826,
      "learning_rate": 8.483627558813023e-06,
      "loss": 3.4732,
      "step": 2899
    },
    {
      "epoch": 0.9884117246080436,
      "grad_norm": 14.358909573180082,
      "learning_rate": 8.482204747768283e-06,
      "loss": 3.5364,
      "step": 2900
    },
    {
      "epoch": 0.9887525562372188,
      "grad_norm": 15.392120553878687,
      "learning_rate": 8.480781388958874e-06,
      "loss": 3.1144,
      "step": 2901
    },
    {
      "epoch": 0.989093387866394,
      "grad_norm": 21.631958549697835,
      "learning_rate": 8.479357482608698e-06,
      "loss": 3.227,
      "step": 2902
    },
    {
      "epoch": 0.9894342194955692,
      "grad_norm": 14.790179902885969,
      "learning_rate": 8.477933028941738e-06,
      "loss": 2.9892,
      "step": 2903
    },
    {
      "epoch": 0.9897750511247444,
      "grad_norm": 16.929650054160785,
      "learning_rate": 8.476508028182068e-06,
      "loss": 3.6609,
      "step": 2904
    },
    {
      "epoch": 0.9901158827539196,
      "grad_norm": 13.301036879306281,
      "learning_rate": 8.475082480553845e-06,
      "loss": 2.2391,
      "step": 2905
    },
    {
      "epoch": 0.9904567143830948,
      "grad_norm": 20.154337392298867,
      "learning_rate": 8.473656386281315e-06,
      "loss": 4.0546,
      "step": 2906
    },
    {
      "epoch": 0.99079754601227,
      "grad_norm": 17.84756462943088,
      "learning_rate": 8.472229745588806e-06,
      "loss": 2.4679,
      "step": 2907
    },
    {
      "epoch": 0.9911383776414451,
      "grad_norm": 20.064756421277345,
      "learning_rate": 8.470802558700739e-06,
      "loss": 3.2025,
      "step": 2908
    },
    {
      "epoch": 0.9914792092706203,
      "grad_norm": 17.983850631707597,
      "learning_rate": 8.46937482584161e-06,
      "loss": 3.388,
      "step": 2909
    },
    {
      "epoch": 0.9918200408997955,
      "grad_norm": 20.03977673484492,
      "learning_rate": 8.467946547236013e-06,
      "loss": 3.3407,
      "step": 2910
    },
    {
      "epoch": 0.9921608725289707,
      "grad_norm": 14.867860162207226,
      "learning_rate": 8.466517723108618e-06,
      "loss": 3.2095,
      "step": 2911
    },
    {
      "epoch": 0.9925017041581459,
      "grad_norm": 27.1159081128393,
      "learning_rate": 8.465088353684186e-06,
      "loss": 3.2513,
      "step": 2912
    },
    {
      "epoch": 0.9928425357873211,
      "grad_norm": 19.577186027232766,
      "learning_rate": 8.46365843918756e-06,
      "loss": 2.9644,
      "step": 2913
    },
    {
      "epoch": 0.9931833674164963,
      "grad_norm": 17.43438646495004,
      "learning_rate": 8.462227979843675e-06,
      "loss": 3.2013,
      "step": 2914
    },
    {
      "epoch": 0.9935241990456715,
      "grad_norm": 19.519496729977845,
      "learning_rate": 8.460796975877547e-06,
      "loss": 3.8508,
      "step": 2915
    },
    {
      "epoch": 0.9938650306748467,
      "grad_norm": 14.94106355333602,
      "learning_rate": 8.459365427514278e-06,
      "loss": 3.4555,
      "step": 2916
    },
    {
      "epoch": 0.9942058623040219,
      "grad_norm": 18.060576882678305,
      "learning_rate": 8.457933334979057e-06,
      "loss": 3.2966,
      "step": 2917
    },
    {
      "epoch": 0.994546693933197,
      "grad_norm": 25.75220601130435,
      "learning_rate": 8.456500698497154e-06,
      "loss": 3.3007,
      "step": 2918
    },
    {
      "epoch": 0.9948875255623721,
      "grad_norm": 12.508177303606423,
      "learning_rate": 8.455067518293934e-06,
      "loss": 2.8538,
      "step": 2919
    },
    {
      "epoch": 0.9952283571915473,
      "grad_norm": 23.619607550320882,
      "learning_rate": 8.45363379459484e-06,
      "loss": 3.0823,
      "step": 2920
    },
    {
      "epoch": 0.9955691888207225,
      "grad_norm": 22.57181039332717,
      "learning_rate": 8.4521995276254e-06,
      "loss": 3.1306,
      "step": 2921
    },
    {
      "epoch": 0.9959100204498977,
      "grad_norm": 19.866340546346457,
      "learning_rate": 8.450764717611235e-06,
      "loss": 2.9677,
      "step": 2922
    },
    {
      "epoch": 0.9962508520790729,
      "grad_norm": 23.428682125495197,
      "learning_rate": 8.44932936477804e-06,
      "loss": 3.4723,
      "step": 2923
    },
    {
      "epoch": 0.9965916837082481,
      "grad_norm": 15.052286149566417,
      "learning_rate": 8.447893469351607e-06,
      "loss": 3.2926,
      "step": 2924
    },
    {
      "epoch": 0.9969325153374233,
      "grad_norm": 20.456393357248533,
      "learning_rate": 8.446457031557805e-06,
      "loss": 3.6024,
      "step": 2925
    },
    {
      "epoch": 0.9972733469665985,
      "grad_norm": 15.38630556060551,
      "learning_rate": 8.445020051622592e-06,
      "loss": 3.1217,
      "step": 2926
    },
    {
      "epoch": 0.9976141785957737,
      "grad_norm": 27.493943900141844,
      "learning_rate": 8.443582529772012e-06,
      "loss": 2.8535,
      "step": 2927
    },
    {
      "epoch": 0.9979550102249489,
      "grad_norm": 27.95479975603998,
      "learning_rate": 8.442144466232191e-06,
      "loss": 3.727,
      "step": 2928
    },
    {
      "epoch": 0.998295841854124,
      "grad_norm": 16.751768113758825,
      "learning_rate": 8.440705861229344e-06,
      "loss": 2.8237,
      "step": 2929
    },
    {
      "epoch": 0.9986366734832992,
      "grad_norm": 23.543302606910213,
      "learning_rate": 8.439266714989768e-06,
      "loss": 3.922,
      "step": 2930
    },
    {
      "epoch": 0.9989775051124744,
      "grad_norm": 15.002184047165716,
      "learning_rate": 8.437827027739848e-06,
      "loss": 2.9091,
      "step": 2931
    },
    {
      "epoch": 0.9993183367416496,
      "grad_norm": 14.93972205274078,
      "learning_rate": 8.43638679970605e-06,
      "loss": 3.03,
      "step": 2932
    },
    {
      "epoch": 0.9996591683708248,
      "grad_norm": 17.272998430086624,
      "learning_rate": 8.43494603111493e-06,
      "loss": 3.473,
      "step": 2933
    },
    {
      "epoch": 1.0,
      "grad_norm": 24.182192278620526,
      "learning_rate": 8.433504722193124e-06,
      "loss": 3.8168,
      "step": 2934
    },
    {
      "epoch": 1.000340831629175,
      "grad_norm": 16.870011694155398,
      "learning_rate": 8.432062873167357e-06,
      "loss": 2.3946,
      "step": 2935
    },
    {
      "epoch": 1.0006816632583504,
      "grad_norm": 15.45378259252126,
      "learning_rate": 8.43062048426444e-06,
      "loss": 3.0908,
      "step": 2936
    },
    {
      "epoch": 1.0010224948875255,
      "grad_norm": 11.436594610398576,
      "learning_rate": 8.429177555711263e-06,
      "loss": 2.1897,
      "step": 2937
    },
    {
      "epoch": 1.0013633265167008,
      "grad_norm": 16.925804087550564,
      "learning_rate": 8.427734087734806e-06,
      "loss": 2.5268,
      "step": 2938
    },
    {
      "epoch": 1.0017041581458759,
      "grad_norm": 10.936642764224866,
      "learning_rate": 8.426290080562132e-06,
      "loss": 2.4222,
      "step": 2939
    },
    {
      "epoch": 1.0020449897750512,
      "grad_norm": 27.38259886368823,
      "learning_rate": 8.424845534420389e-06,
      "loss": 2.0305,
      "step": 2940
    },
    {
      "epoch": 1.0023858214042263,
      "grad_norm": 11.719159609130703,
      "learning_rate": 8.423400449536809e-06,
      "loss": 1.9912,
      "step": 2941
    },
    {
      "epoch": 1.0027266530334016,
      "grad_norm": 10.675697988550219,
      "learning_rate": 8.421954826138712e-06,
      "loss": 1.9478,
      "step": 2942
    },
    {
      "epoch": 1.0030674846625767,
      "grad_norm": 19.456917261726804,
      "learning_rate": 8.420508664453497e-06,
      "loss": 1.9202,
      "step": 2943
    },
    {
      "epoch": 1.003408316291752,
      "grad_norm": 10.166978228025158,
      "learning_rate": 8.419061964708655e-06,
      "loss": 2.2339,
      "step": 2944
    },
    {
      "epoch": 1.003749147920927,
      "grad_norm": 13.900642311041366,
      "learning_rate": 8.417614727131753e-06,
      "loss": 2.7878,
      "step": 2945
    },
    {
      "epoch": 1.0040899795501022,
      "grad_norm": 16.068470277631796,
      "learning_rate": 8.416166951950452e-06,
      "loss": 2.5421,
      "step": 2946
    },
    {
      "epoch": 1.0044308111792775,
      "grad_norm": 11.435515601051637,
      "learning_rate": 8.41471863939249e-06,
      "loss": 2.5403,
      "step": 2947
    },
    {
      "epoch": 1.0047716428084525,
      "grad_norm": 18.08518657540011,
      "learning_rate": 8.413269789685693e-06,
      "loss": 2.2287,
      "step": 2948
    },
    {
      "epoch": 1.0051124744376279,
      "grad_norm": 15.212083433230257,
      "learning_rate": 8.41182040305797e-06,
      "loss": 2.2113,
      "step": 2949
    },
    {
      "epoch": 1.005453306066803,
      "grad_norm": 12.007250073913866,
      "learning_rate": 8.410370479737316e-06,
      "loss": 2.0771,
      "step": 2950
    },
    {
      "epoch": 1.0057941376959783,
      "grad_norm": 20.114843549987683,
      "learning_rate": 8.408920019951808e-06,
      "loss": 2.2451,
      "step": 2951
    },
    {
      "epoch": 1.0061349693251533,
      "grad_norm": 19.993448412036706,
      "learning_rate": 8.407469023929613e-06,
      "loss": 2.4454,
      "step": 2952
    },
    {
      "epoch": 1.0064758009543286,
      "grad_norm": 14.495211520181796,
      "learning_rate": 8.406017491898975e-06,
      "loss": 2.5565,
      "step": 2953
    },
    {
      "epoch": 1.0068166325835037,
      "grad_norm": 20.145187321678865,
      "learning_rate": 8.404565424088226e-06,
      "loss": 1.6723,
      "step": 2954
    },
    {
      "epoch": 1.007157464212679,
      "grad_norm": 15.202571997451127,
      "learning_rate": 8.403112820725787e-06,
      "loss": 2.0155,
      "step": 2955
    },
    {
      "epoch": 1.0074982958418541,
      "grad_norm": 16.25349625331195,
      "learning_rate": 8.40165968204015e-06,
      "loss": 1.8035,
      "step": 2956
    },
    {
      "epoch": 1.0078391274710292,
      "grad_norm": 16.56891070562071,
      "learning_rate": 8.400206008259906e-06,
      "loss": 2.495,
      "step": 2957
    },
    {
      "epoch": 1.0081799591002045,
      "grad_norm": 14.020564107765805,
      "learning_rate": 8.398751799613723e-06,
      "loss": 2.1595,
      "step": 2958
    },
    {
      "epoch": 1.0085207907293796,
      "grad_norm": 23.990834631503944,
      "learning_rate": 8.39729705633035e-06,
      "loss": 3.113,
      "step": 2959
    },
    {
      "epoch": 1.008861622358555,
      "grad_norm": 13.012646640483593,
      "learning_rate": 8.395841778638625e-06,
      "loss": 2.5958,
      "step": 2960
    },
    {
      "epoch": 1.00920245398773,
      "grad_norm": 16.149216876893217,
      "learning_rate": 8.394385966767473e-06,
      "loss": 2.0657,
      "step": 2961
    },
    {
      "epoch": 1.0095432856169053,
      "grad_norm": 17.8509695216834,
      "learning_rate": 8.392929620945893e-06,
      "loss": 2.2783,
      "step": 2962
    },
    {
      "epoch": 1.0098841172460804,
      "grad_norm": 19.17849910347601,
      "learning_rate": 8.39147274140298e-06,
      "loss": 2.4315,
      "step": 2963
    },
    {
      "epoch": 1.0102249488752557,
      "grad_norm": 15.403360051243343,
      "learning_rate": 8.390015328367903e-06,
      "loss": 2.4307,
      "step": 2964
    },
    {
      "epoch": 1.0105657805044308,
      "grad_norm": 15.282855319669597,
      "learning_rate": 8.38855738206992e-06,
      "loss": 2.5902,
      "step": 2965
    },
    {
      "epoch": 1.010906612133606,
      "grad_norm": 12.45547651457662,
      "learning_rate": 8.387098902738371e-06,
      "loss": 2.3073,
      "step": 2966
    },
    {
      "epoch": 1.0112474437627812,
      "grad_norm": 28.854770704634024,
      "learning_rate": 8.385639890602682e-06,
      "loss": 2.5377,
      "step": 2967
    },
    {
      "epoch": 1.0115882753919563,
      "grad_norm": 20.287647826450243,
      "learning_rate": 8.38418034589236e-06,
      "loss": 2.8899,
      "step": 2968
    },
    {
      "epoch": 1.0119291070211316,
      "grad_norm": 13.308280237858709,
      "learning_rate": 8.382720268837e-06,
      "loss": 2.8145,
      "step": 2969
    },
    {
      "epoch": 1.0122699386503067,
      "grad_norm": 19.532227245613758,
      "learning_rate": 8.381259659666274e-06,
      "loss": 2.7608,
      "step": 2970
    },
    {
      "epoch": 1.012610770279482,
      "grad_norm": 15.327175206797714,
      "learning_rate": 8.379798518609946e-06,
      "loss": 2.3319,
      "step": 2971
    },
    {
      "epoch": 1.012951601908657,
      "grad_norm": 17.9940879384752,
      "learning_rate": 8.378336845897855e-06,
      "loss": 2.3474,
      "step": 2972
    },
    {
      "epoch": 1.0132924335378324,
      "grad_norm": 13.345762996994136,
      "learning_rate": 8.376874641759928e-06,
      "loss": 2.4487,
      "step": 2973
    },
    {
      "epoch": 1.0136332651670075,
      "grad_norm": 14.429591322414812,
      "learning_rate": 8.375411906426179e-06,
      "loss": 2.6001,
      "step": 2974
    },
    {
      "epoch": 1.0139740967961828,
      "grad_norm": 15.791211626973098,
      "learning_rate": 8.3739486401267e-06,
      "loss": 2.4211,
      "step": 2975
    },
    {
      "epoch": 1.0143149284253579,
      "grad_norm": 14.140575515538307,
      "learning_rate": 8.37248484309167e-06,
      "loss": 2.0912,
      "step": 2976
    },
    {
      "epoch": 1.014655760054533,
      "grad_norm": 15.035755659838799,
      "learning_rate": 8.37102051555135e-06,
      "loss": 2.6132,
      "step": 2977
    },
    {
      "epoch": 1.0149965916837083,
      "grad_norm": 8.884969883377437,
      "learning_rate": 8.369555657736082e-06,
      "loss": 1.6373,
      "step": 2978
    },
    {
      "epoch": 1.0153374233128833,
      "grad_norm": 17.02520619141162,
      "learning_rate": 8.368090269876294e-06,
      "loss": 2.5518,
      "step": 2979
    },
    {
      "epoch": 1.0156782549420587,
      "grad_norm": 12.37932527232506,
      "learning_rate": 8.366624352202501e-06,
      "loss": 1.9395,
      "step": 2980
    },
    {
      "epoch": 1.0160190865712337,
      "grad_norm": 38.663894152801625,
      "learning_rate": 8.365157904945296e-06,
      "loss": 2.2597,
      "step": 2981
    },
    {
      "epoch": 1.016359918200409,
      "grad_norm": 13.781998441443758,
      "learning_rate": 8.363690928335356e-06,
      "loss": 1.9735,
      "step": 2982
    },
    {
      "epoch": 1.0167007498295841,
      "grad_norm": 23.563536883980035,
      "learning_rate": 8.362223422603442e-06,
      "loss": 1.8197,
      "step": 2983
    },
    {
      "epoch": 1.0170415814587594,
      "grad_norm": 18.37201266737557,
      "learning_rate": 8.360755387980402e-06,
      "loss": 2.2621,
      "step": 2984
    },
    {
      "epoch": 1.0173824130879345,
      "grad_norm": 22.031476829186673,
      "learning_rate": 8.35928682469716e-06,
      "loss": 2.174,
      "step": 2985
    },
    {
      "epoch": 1.0177232447171098,
      "grad_norm": 16.59961550867403,
      "learning_rate": 8.357817732984726e-06,
      "loss": 2.3534,
      "step": 2986
    },
    {
      "epoch": 1.018064076346285,
      "grad_norm": 14.419030983901294,
      "learning_rate": 8.356348113074197e-06,
      "loss": 2.8242,
      "step": 2987
    },
    {
      "epoch": 1.01840490797546,
      "grad_norm": 15.918666933985195,
      "learning_rate": 8.35487796519675e-06,
      "loss": 2.2908,
      "step": 2988
    },
    {
      "epoch": 1.0187457396046353,
      "grad_norm": 21.98539237098907,
      "learning_rate": 8.353407289583644e-06,
      "loss": 2.1171,
      "step": 2989
    },
    {
      "epoch": 1.0190865712338104,
      "grad_norm": 15.929630028381956,
      "learning_rate": 8.35193608646622e-06,
      "loss": 2.3836,
      "step": 2990
    },
    {
      "epoch": 1.0194274028629857,
      "grad_norm": 17.434592350149966,
      "learning_rate": 8.350464356075908e-06,
      "loss": 2.2862,
      "step": 2991
    },
    {
      "epoch": 1.0197682344921608,
      "grad_norm": 18.652329104261216,
      "learning_rate": 8.348992098644216e-06,
      "loss": 2.5439,
      "step": 2992
    },
    {
      "epoch": 1.0201090661213361,
      "grad_norm": 17.068499944435413,
      "learning_rate": 8.347519314402734e-06,
      "loss": 2.1013,
      "step": 2993
    },
    {
      "epoch": 1.0204498977505112,
      "grad_norm": 16.910992019311166,
      "learning_rate": 8.346046003583138e-06,
      "loss": 2.7617,
      "step": 2994
    },
    {
      "epoch": 1.0207907293796865,
      "grad_norm": 11.663874612756022,
      "learning_rate": 8.344572166417186e-06,
      "loss": 1.8726,
      "step": 2995
    },
    {
      "epoch": 1.0211315610088616,
      "grad_norm": 17.185465730930233,
      "learning_rate": 8.34309780313672e-06,
      "loss": 2.9399,
      "step": 2996
    },
    {
      "epoch": 1.021472392638037,
      "grad_norm": 23.51640520313903,
      "learning_rate": 8.34162291397366e-06,
      "loss": 2.3387,
      "step": 2997
    },
    {
      "epoch": 1.021813224267212,
      "grad_norm": 14.337619550104389,
      "learning_rate": 8.340147499160014e-06,
      "loss": 1.7297,
      "step": 2998
    },
    {
      "epoch": 1.022154055896387,
      "grad_norm": 12.20701893107469,
      "learning_rate": 8.33867155892787e-06,
      "loss": 2.1076,
      "step": 2999
    },
    {
      "epoch": 1.0224948875255624,
      "grad_norm": 19.945522742097726,
      "learning_rate": 8.337195093509397e-06,
      "loss": 1.1566,
      "step": 3000
    },
    {
      "epoch": 1.0228357191547375,
      "grad_norm": 19.078014494751812,
      "learning_rate": 8.335718103136854e-06,
      "loss": 2.5271,
      "step": 3001
    },
    {
      "epoch": 1.0231765507839128,
      "grad_norm": 14.209076050734804,
      "learning_rate": 8.334240588042573e-06,
      "loss": 2.9133,
      "step": 3002
    },
    {
      "epoch": 1.0235173824130879,
      "grad_norm": 14.115100055936848,
      "learning_rate": 8.332762548458975e-06,
      "loss": 2.1811,
      "step": 3003
    },
    {
      "epoch": 1.0238582140422632,
      "grad_norm": 22.729049435929607,
      "learning_rate": 8.331283984618562e-06,
      "loss": 3.0598,
      "step": 3004
    },
    {
      "epoch": 1.0241990456714383,
      "grad_norm": 12.312665406876421,
      "learning_rate": 8.329804896753917e-06,
      "loss": 1.7742,
      "step": 3005
    },
    {
      "epoch": 1.0245398773006136,
      "grad_norm": 14.63908031957038,
      "learning_rate": 8.328325285097707e-06,
      "loss": 2.1559,
      "step": 3006
    },
    {
      "epoch": 1.0248807089297887,
      "grad_norm": 13.028367967791633,
      "learning_rate": 8.32684514988268e-06,
      "loss": 2.0566,
      "step": 3007
    },
    {
      "epoch": 1.025221540558964,
      "grad_norm": 14.490406216923038,
      "learning_rate": 8.325364491341666e-06,
      "loss": 1.7624,
      "step": 3008
    },
    {
      "epoch": 1.025562372188139,
      "grad_norm": 21.799879219927558,
      "learning_rate": 8.32388330970758e-06,
      "loss": 1.8941,
      "step": 3009
    },
    {
      "epoch": 1.0259032038173141,
      "grad_norm": 15.598456944995535,
      "learning_rate": 8.32240160521342e-06,
      "loss": 2.3004,
      "step": 3010
    },
    {
      "epoch": 1.0262440354464895,
      "grad_norm": 14.566478606049442,
      "learning_rate": 8.32091937809226e-06,
      "loss": 2.0838,
      "step": 3011
    },
    {
      "epoch": 1.0265848670756645,
      "grad_norm": 14.074719088265235,
      "learning_rate": 8.319436628577262e-06,
      "loss": 2.4881,
      "step": 3012
    },
    {
      "epoch": 1.0269256987048399,
      "grad_norm": 15.744788862216913,
      "learning_rate": 8.31795335690167e-06,
      "loss": 2.1748,
      "step": 3013
    },
    {
      "epoch": 1.027266530334015,
      "grad_norm": 13.7999587978313,
      "learning_rate": 8.316469563298805e-06,
      "loss": 2.1522,
      "step": 3014
    },
    {
      "epoch": 1.0276073619631902,
      "grad_norm": 11.462483224095424,
      "learning_rate": 8.314985248002077e-06,
      "loss": 1.8285,
      "step": 3015
    },
    {
      "epoch": 1.0279481935923653,
      "grad_norm": 20.105956892459854,
      "learning_rate": 8.313500411244973e-06,
      "loss": 2.2915,
      "step": 3016
    },
    {
      "epoch": 1.0282890252215406,
      "grad_norm": 17.332264797104557,
      "learning_rate": 8.312015053261065e-06,
      "loss": 2.5517,
      "step": 3017
    },
    {
      "epoch": 1.0286298568507157,
      "grad_norm": 25.736804068350896,
      "learning_rate": 8.310529174284004e-06,
      "loss": 2.5159,
      "step": 3018
    },
    {
      "epoch": 1.028970688479891,
      "grad_norm": 18.15206055189928,
      "learning_rate": 8.309042774547525e-06,
      "loss": 2.4168,
      "step": 3019
    },
    {
      "epoch": 1.0293115201090661,
      "grad_norm": 19.642150752155256,
      "learning_rate": 8.307555854285445e-06,
      "loss": 2.4011,
      "step": 3020
    },
    {
      "epoch": 1.0296523517382412,
      "grad_norm": 18.954346281667657,
      "learning_rate": 8.306068413731662e-06,
      "loss": 2.653,
      "step": 3021
    },
    {
      "epoch": 1.0299931833674165,
      "grad_norm": 13.839428774791044,
      "learning_rate": 8.304580453120157e-06,
      "loss": 2.2454,
      "step": 3022
    },
    {
      "epoch": 1.0303340149965916,
      "grad_norm": 14.938233306906922,
      "learning_rate": 8.303091972684993e-06,
      "loss": 1.8844,
      "step": 3023
    },
    {
      "epoch": 1.030674846625767,
      "grad_norm": 20.15701905690113,
      "learning_rate": 8.301602972660313e-06,
      "loss": 2.4438,
      "step": 3024
    },
    {
      "epoch": 1.031015678254942,
      "grad_norm": 19.042476037802064,
      "learning_rate": 8.30011345328034e-06,
      "loss": 1.968,
      "step": 3025
    },
    {
      "epoch": 1.0313565098841173,
      "grad_norm": 15.821410892679895,
      "learning_rate": 8.298623414779386e-06,
      "loss": 2.3062,
      "step": 3026
    },
    {
      "epoch": 1.0316973415132924,
      "grad_norm": 27.05541356306815,
      "learning_rate": 8.297132857391836e-06,
      "loss": 2.509,
      "step": 3027
    },
    {
      "epoch": 1.0320381731424677,
      "grad_norm": 15.850551969880144,
      "learning_rate": 8.295641781352165e-06,
      "loss": 2.0777,
      "step": 3028
    },
    {
      "epoch": 1.0323790047716428,
      "grad_norm": 14.96001653023231,
      "learning_rate": 8.294150186894922e-06,
      "loss": 2.3675,
      "step": 3029
    },
    {
      "epoch": 1.032719836400818,
      "grad_norm": 27.89636177307034,
      "learning_rate": 8.29265807425474e-06,
      "loss": 2.4957,
      "step": 3030
    },
    {
      "epoch": 1.0330606680299932,
      "grad_norm": 16.92886848697703,
      "learning_rate": 8.291165443666336e-06,
      "loss": 2.215,
      "step": 3031
    },
    {
      "epoch": 1.0334014996591683,
      "grad_norm": 16.741689227119593,
      "learning_rate": 8.289672295364507e-06,
      "loss": 2.2491,
      "step": 3032
    },
    {
      "epoch": 1.0337423312883436,
      "grad_norm": 18.36470972359791,
      "learning_rate": 8.288178629584131e-06,
      "loss": 1.804,
      "step": 3033
    },
    {
      "epoch": 1.0340831629175187,
      "grad_norm": 16.696433116735296,
      "learning_rate": 8.286684446560165e-06,
      "loss": 2.0143,
      "step": 3034
    },
    {
      "epoch": 1.034423994546694,
      "grad_norm": 17.29207957921798,
      "learning_rate": 8.285189746527653e-06,
      "loss": 1.4633,
      "step": 3035
    },
    {
      "epoch": 1.034764826175869,
      "grad_norm": 29.185405326344267,
      "learning_rate": 8.283694529721717e-06,
      "loss": 2.0241,
      "step": 3036
    },
    {
      "epoch": 1.0351056578050444,
      "grad_norm": 19.333599457372,
      "learning_rate": 8.282198796377561e-06,
      "loss": 1.717,
      "step": 3037
    },
    {
      "epoch": 1.0354464894342195,
      "grad_norm": 17.57626224692412,
      "learning_rate": 8.280702546730466e-06,
      "loss": 2.3111,
      "step": 3038
    },
    {
      "epoch": 1.0357873210633948,
      "grad_norm": 15.01687174716507,
      "learning_rate": 8.279205781015801e-06,
      "loss": 2.4606,
      "step": 3039
    },
    {
      "epoch": 1.0361281526925699,
      "grad_norm": 39.19575455555872,
      "learning_rate": 8.277708499469016e-06,
      "loss": 2.0121,
      "step": 3040
    },
    {
      "epoch": 1.0364689843217452,
      "grad_norm": 14.732989123613407,
      "learning_rate": 8.276210702325631e-06,
      "loss": 2.0396,
      "step": 3041
    },
    {
      "epoch": 1.0368098159509203,
      "grad_norm": 21.350769311608744,
      "learning_rate": 8.274712389821264e-06,
      "loss": 2.4751,
      "step": 3042
    },
    {
      "epoch": 1.0371506475800953,
      "grad_norm": 19.892160191363168,
      "learning_rate": 8.273213562191603e-06,
      "loss": 2.8179,
      "step": 3043
    },
    {
      "epoch": 1.0374914792092707,
      "grad_norm": 14.921919109080436,
      "learning_rate": 8.271714219672417e-06,
      "loss": 2.5831,
      "step": 3044
    },
    {
      "epoch": 1.0378323108384457,
      "grad_norm": 13.297269986800188,
      "learning_rate": 8.27021436249956e-06,
      "loss": 2.2095,
      "step": 3045
    },
    {
      "epoch": 1.038173142467621,
      "grad_norm": 16.456075493483166,
      "learning_rate": 8.268713990908967e-06,
      "loss": 2.1204,
      "step": 3046
    },
    {
      "epoch": 1.0385139740967961,
      "grad_norm": 17.29213970067709,
      "learning_rate": 8.267213105136652e-06,
      "loss": 1.7629,
      "step": 3047
    },
    {
      "epoch": 1.0388548057259714,
      "grad_norm": 15.555417633395297,
      "learning_rate": 8.265711705418706e-06,
      "loss": 2.6193,
      "step": 3048
    },
    {
      "epoch": 1.0391956373551465,
      "grad_norm": 15.99762491491181,
      "learning_rate": 8.264209791991312e-06,
      "loss": 2.1822,
      "step": 3049
    },
    {
      "epoch": 1.0395364689843218,
      "grad_norm": 175.37585414932568,
      "learning_rate": 8.262707365090723e-06,
      "loss": 2.1971,
      "step": 3050
    },
    {
      "epoch": 1.039877300613497,
      "grad_norm": 24.310899222953914,
      "learning_rate": 8.261204424953278e-06,
      "loss": 2.3931,
      "step": 3051
    },
    {
      "epoch": 1.0402181322426722,
      "grad_norm": 32.42973777525591,
      "learning_rate": 8.259700971815393e-06,
      "loss": 1.8215,
      "step": 3052
    },
    {
      "epoch": 1.0405589638718473,
      "grad_norm": 17.883271855069403,
      "learning_rate": 8.258197005913569e-06,
      "loss": 1.8461,
      "step": 3053
    },
    {
      "epoch": 1.0408997955010224,
      "grad_norm": 22.660466018032217,
      "learning_rate": 8.256692527484387e-06,
      "loss": 2.0931,
      "step": 3054
    },
    {
      "epoch": 1.0412406271301977,
      "grad_norm": 19.92534481769709,
      "learning_rate": 8.255187536764506e-06,
      "loss": 2.7074,
      "step": 3055
    },
    {
      "epoch": 1.0415814587593728,
      "grad_norm": 16.47940260883423,
      "learning_rate": 8.25368203399067e-06,
      "loss": 2.5494,
      "step": 3056
    },
    {
      "epoch": 1.0419222903885481,
      "grad_norm": 11.898123259954165,
      "learning_rate": 8.252176019399694e-06,
      "loss": 1.8906,
      "step": 3057
    },
    {
      "epoch": 1.0422631220177232,
      "grad_norm": 19.460783811219137,
      "learning_rate": 8.250669493228485e-06,
      "loss": 2.3604,
      "step": 3058
    },
    {
      "epoch": 1.0426039536468985,
      "grad_norm": 16.434052869889303,
      "learning_rate": 8.249162455714024e-06,
      "loss": 2.2501,
      "step": 3059
    },
    {
      "epoch": 1.0429447852760736,
      "grad_norm": 15.397466721566715,
      "learning_rate": 8.247654907093377e-06,
      "loss": 1.9695,
      "step": 3060
    },
    {
      "epoch": 1.043285616905249,
      "grad_norm": 18.73175844908087,
      "learning_rate": 8.246146847603683e-06,
      "loss": 1.7781,
      "step": 3061
    },
    {
      "epoch": 1.043626448534424,
      "grad_norm": 12.85943596935143,
      "learning_rate": 8.244638277482167e-06,
      "loss": 2.245,
      "step": 3062
    },
    {
      "epoch": 1.043967280163599,
      "grad_norm": 12.659537510025315,
      "learning_rate": 8.243129196966135e-06,
      "loss": 1.8459,
      "step": 3063
    },
    {
      "epoch": 1.0443081117927744,
      "grad_norm": 25.71753211914107,
      "learning_rate": 8.24161960629297e-06,
      "loss": 2.8123,
      "step": 3064
    },
    {
      "epoch": 1.0446489434219495,
      "grad_norm": 25.319083245912555,
      "learning_rate": 8.240109505700137e-06,
      "loss": 2.384,
      "step": 3065
    },
    {
      "epoch": 1.0449897750511248,
      "grad_norm": 11.222100149877608,
      "learning_rate": 8.23859889542518e-06,
      "loss": 2.1389,
      "step": 3066
    },
    {
      "epoch": 1.0453306066802999,
      "grad_norm": 13.376979373234589,
      "learning_rate": 8.237087775705725e-06,
      "loss": 1.9939,
      "step": 3067
    },
    {
      "epoch": 1.0456714383094752,
      "grad_norm": 13.333347281002322,
      "learning_rate": 8.235576146779479e-06,
      "loss": 2.5065,
      "step": 3068
    },
    {
      "epoch": 1.0460122699386503,
      "grad_norm": 11.79330833163563,
      "learning_rate": 8.234064008884221e-06,
      "loss": 2.0235,
      "step": 3069
    },
    {
      "epoch": 1.0463531015678256,
      "grad_norm": 18.378096754268437,
      "learning_rate": 8.232551362257822e-06,
      "loss": 1.956,
      "step": 3070
    },
    {
      "epoch": 1.0466939331970007,
      "grad_norm": 12.615913099361963,
      "learning_rate": 8.231038207138226e-06,
      "loss": 1.8079,
      "step": 3071
    },
    {
      "epoch": 1.047034764826176,
      "grad_norm": 20.89480984488167,
      "learning_rate": 8.229524543763458e-06,
      "loss": 1.8613,
      "step": 3072
    },
    {
      "epoch": 1.047375596455351,
      "grad_norm": 17.762584047722363,
      "learning_rate": 8.228010372371624e-06,
      "loss": 2.5874,
      "step": 3073
    },
    {
      "epoch": 1.0477164280845261,
      "grad_norm": 18.70707792620371,
      "learning_rate": 8.226495693200906e-06,
      "loss": 2.2664,
      "step": 3074
    },
    {
      "epoch": 1.0480572597137015,
      "grad_norm": 13.9576596780897,
      "learning_rate": 8.224980506489573e-06,
      "loss": 1.885,
      "step": 3075
    },
    {
      "epoch": 1.0483980913428765,
      "grad_norm": 13.762414859366428,
      "learning_rate": 8.223464812475968e-06,
      "loss": 2.3124,
      "step": 3076
    },
    {
      "epoch": 1.0487389229720518,
      "grad_norm": 22.720594171765338,
      "learning_rate": 8.221948611398518e-06,
      "loss": 1.8868,
      "step": 3077
    },
    {
      "epoch": 1.049079754601227,
      "grad_norm": 12.771370687025618,
      "learning_rate": 8.220431903495725e-06,
      "loss": 1.971,
      "step": 3078
    },
    {
      "epoch": 1.0494205862304022,
      "grad_norm": 11.024773695045058,
      "learning_rate": 8.218914689006173e-06,
      "loss": 1.2333,
      "step": 3079
    },
    {
      "epoch": 1.0497614178595773,
      "grad_norm": 25.89196536520435,
      "learning_rate": 8.217396968168526e-06,
      "loss": 2.2958,
      "step": 3080
    },
    {
      "epoch": 1.0501022494887526,
      "grad_norm": 14.675102868463854,
      "learning_rate": 8.215878741221527e-06,
      "loss": 1.3539,
      "step": 3081
    },
    {
      "epoch": 1.0504430811179277,
      "grad_norm": 13.623436268438427,
      "learning_rate": 8.214360008404004e-06,
      "loss": 2.2616,
      "step": 3082
    },
    {
      "epoch": 1.050783912747103,
      "grad_norm": 20.58460107592449,
      "learning_rate": 8.212840769954855e-06,
      "loss": 2.0933,
      "step": 3083
    },
    {
      "epoch": 1.0511247443762781,
      "grad_norm": 15.375796267023995,
      "learning_rate": 8.211321026113061e-06,
      "loss": 2.6491,
      "step": 3084
    },
    {
      "epoch": 1.0514655760054532,
      "grad_norm": 21.247743340922828,
      "learning_rate": 8.20980077711769e-06,
      "loss": 3.327,
      "step": 3085
    },
    {
      "epoch": 1.0518064076346285,
      "grad_norm": 22.528173769699457,
      "learning_rate": 8.208280023207875e-06,
      "loss": 1.9887,
      "step": 3086
    },
    {
      "epoch": 1.0521472392638036,
      "grad_norm": 15.12423234545914,
      "learning_rate": 8.206758764622843e-06,
      "loss": 2.0256,
      "step": 3087
    },
    {
      "epoch": 1.052488070892979,
      "grad_norm": 14.04171805336934,
      "learning_rate": 8.205237001601892e-06,
      "loss": 2.3245,
      "step": 3088
    },
    {
      "epoch": 1.052828902522154,
      "grad_norm": 14.901915791061269,
      "learning_rate": 8.203714734384402e-06,
      "loss": 2.06,
      "step": 3089
    },
    {
      "epoch": 1.0531697341513293,
      "grad_norm": 13.384545382799981,
      "learning_rate": 8.20219196320983e-06,
      "loss": 2.4424,
      "step": 3090
    },
    {
      "epoch": 1.0535105657805044,
      "grad_norm": 12.349643604676519,
      "learning_rate": 8.200668688317714e-06,
      "loss": 1.8983,
      "step": 3091
    },
    {
      "epoch": 1.0538513974096797,
      "grad_norm": 20.40328846007424,
      "learning_rate": 8.199144909947674e-06,
      "loss": 2.5484,
      "step": 3092
    },
    {
      "epoch": 1.0541922290388548,
      "grad_norm": 15.328369508663503,
      "learning_rate": 8.197620628339404e-06,
      "loss": 2.3304,
      "step": 3093
    },
    {
      "epoch": 1.05453306066803,
      "grad_norm": 14.071576763447034,
      "learning_rate": 8.19609584373268e-06,
      "loss": 2.3661,
      "step": 3094
    },
    {
      "epoch": 1.0548738922972052,
      "grad_norm": 17.135689142809646,
      "learning_rate": 8.194570556367356e-06,
      "loss": 1.6676,
      "step": 3095
    },
    {
      "epoch": 1.0552147239263803,
      "grad_norm": 16.375276998764477,
      "learning_rate": 8.19304476648337e-06,
      "loss": 2.3436,
      "step": 3096
    },
    {
      "epoch": 1.0555555555555556,
      "grad_norm": 14.446102849696802,
      "learning_rate": 8.191518474320727e-06,
      "loss": 2.6259,
      "step": 3097
    },
    {
      "epoch": 1.0558963871847307,
      "grad_norm": 13.177257196662163,
      "learning_rate": 8.189991680119526e-06,
      "loss": 2.6169,
      "step": 3098
    },
    {
      "epoch": 1.056237218813906,
      "grad_norm": 16.105307990036593,
      "learning_rate": 8.188464384119936e-06,
      "loss": 2.3657,
      "step": 3099
    },
    {
      "epoch": 1.056578050443081,
      "grad_norm": 16.362586551654793,
      "learning_rate": 8.186936586562203e-06,
      "loss": 2.3062,
      "step": 3100
    },
    {
      "epoch": 1.0569188820722564,
      "grad_norm": 30.489239018594386,
      "learning_rate": 8.18540828768666e-06,
      "loss": 2.573,
      "step": 3101
    },
    {
      "epoch": 1.0572597137014315,
      "grad_norm": 14.538415121204771,
      "learning_rate": 8.183879487733713e-06,
      "loss": 2.9195,
      "step": 3102
    },
    {
      "epoch": 1.0576005453306068,
      "grad_norm": 19.212063214858837,
      "learning_rate": 8.182350186943849e-06,
      "loss": 2.9697,
      "step": 3103
    },
    {
      "epoch": 1.0579413769597819,
      "grad_norm": 14.010212454036367,
      "learning_rate": 8.18082038555763e-06,
      "loss": 2.4357,
      "step": 3104
    },
    {
      "epoch": 1.058282208588957,
      "grad_norm": 15.292007301305084,
      "learning_rate": 8.179290083815705e-06,
      "loss": 2.1053,
      "step": 3105
    },
    {
      "epoch": 1.0586230402181322,
      "grad_norm": 21.748373388893622,
      "learning_rate": 8.177759281958792e-06,
      "loss": 1.636,
      "step": 3106
    },
    {
      "epoch": 1.0589638718473073,
      "grad_norm": 22.018796826211727,
      "learning_rate": 8.176227980227693e-06,
      "loss": 2.8344,
      "step": 3107
    },
    {
      "epoch": 1.0593047034764826,
      "grad_norm": 19.024505779983993,
      "learning_rate": 8.174696178863291e-06,
      "loss": 2.421,
      "step": 3108
    },
    {
      "epoch": 1.0596455351056577,
      "grad_norm": 13.672222687232669,
      "learning_rate": 8.173163878106543e-06,
      "loss": 2.1616,
      "step": 3109
    },
    {
      "epoch": 1.059986366734833,
      "grad_norm": 15.25692059145895,
      "learning_rate": 8.171631078198483e-06,
      "loss": 1.9193,
      "step": 3110
    },
    {
      "epoch": 1.0603271983640081,
      "grad_norm": 22.784208982359797,
      "learning_rate": 8.170097779380231e-06,
      "loss": 2.2727,
      "step": 3111
    },
    {
      "epoch": 1.0606680299931834,
      "grad_norm": 22.044856795176475,
      "learning_rate": 8.168563981892977e-06,
      "loss": 2.5161,
      "step": 3112
    },
    {
      "epoch": 1.0610088616223585,
      "grad_norm": 16.659562410192517,
      "learning_rate": 8.167029685977995e-06,
      "loss": 2.5196,
      "step": 3113
    },
    {
      "epoch": 1.0613496932515338,
      "grad_norm": 27.850046323118775,
      "learning_rate": 8.165494891876637e-06,
      "loss": 2.2761,
      "step": 3114
    },
    {
      "epoch": 1.061690524880709,
      "grad_norm": 15.914888722676201,
      "learning_rate": 8.163959599830333e-06,
      "loss": 2.2639,
      "step": 3115
    },
    {
      "epoch": 1.062031356509884,
      "grad_norm": 16.791400288915664,
      "learning_rate": 8.162423810080587e-06,
      "loss": 1.9989,
      "step": 3116
    },
    {
      "epoch": 1.0623721881390593,
      "grad_norm": 16.48877382215794,
      "learning_rate": 8.160887522868988e-06,
      "loss": 1.7909,
      "step": 3117
    },
    {
      "epoch": 1.0627130197682344,
      "grad_norm": 22.309940422078828,
      "learning_rate": 8.159350738437198e-06,
      "loss": 2.1785,
      "step": 3118
    },
    {
      "epoch": 1.0630538513974097,
      "grad_norm": 21.833094537866675,
      "learning_rate": 8.157813457026961e-06,
      "loss": 2.5992,
      "step": 3119
    },
    {
      "epoch": 1.0633946830265848,
      "grad_norm": 15.263392397783726,
      "learning_rate": 8.156275678880095e-06,
      "loss": 2.9478,
      "step": 3120
    },
    {
      "epoch": 1.06373551465576,
      "grad_norm": 14.880288203539916,
      "learning_rate": 8.154737404238501e-06,
      "loss": 2.1008,
      "step": 3121
    },
    {
      "epoch": 1.0640763462849352,
      "grad_norm": 17.02765989582731,
      "learning_rate": 8.153198633344158e-06,
      "loss": 2.1832,
      "step": 3122
    },
    {
      "epoch": 1.0644171779141105,
      "grad_norm": 24.829570188693502,
      "learning_rate": 8.151659366439114e-06,
      "loss": 2.5098,
      "step": 3123
    },
    {
      "epoch": 1.0647580095432856,
      "grad_norm": 20.9863948508211,
      "learning_rate": 8.150119603765507e-06,
      "loss": 2.6556,
      "step": 3124
    },
    {
      "epoch": 1.065098841172461,
      "grad_norm": 15.840303815302837,
      "learning_rate": 8.148579345565545e-06,
      "loss": 2.2784,
      "step": 3125
    },
    {
      "epoch": 1.065439672801636,
      "grad_norm": 23.482187577863737,
      "learning_rate": 8.147038592081521e-06,
      "loss": 1.9724,
      "step": 3126
    },
    {
      "epoch": 1.065780504430811,
      "grad_norm": 11.319770090064555,
      "learning_rate": 8.145497343555796e-06,
      "loss": 2.0975,
      "step": 3127
    },
    {
      "epoch": 1.0661213360599864,
      "grad_norm": 18.983544105546937,
      "learning_rate": 8.14395560023082e-06,
      "loss": 2.447,
      "step": 3128
    },
    {
      "epoch": 1.0664621676891615,
      "grad_norm": 11.654782941688701,
      "learning_rate": 8.142413362349112e-06,
      "loss": 2.3275,
      "step": 3129
    },
    {
      "epoch": 1.0668029993183368,
      "grad_norm": 14.04893958573976,
      "learning_rate": 8.140870630153273e-06,
      "loss": 1.9325,
      "step": 3130
    },
    {
      "epoch": 1.0671438309475119,
      "grad_norm": 20.954136504741882,
      "learning_rate": 8.13932740388598e-06,
      "loss": 1.9308,
      "step": 3131
    },
    {
      "epoch": 1.0674846625766872,
      "grad_norm": 14.758378219582532,
      "learning_rate": 8.137783683789988e-06,
      "loss": 2.407,
      "step": 3132
    },
    {
      "epoch": 1.0678254942058623,
      "grad_norm": 19.200079817109145,
      "learning_rate": 8.136239470108134e-06,
      "loss": 2.1451,
      "step": 3133
    },
    {
      "epoch": 1.0681663258350376,
      "grad_norm": 15.804376397751135,
      "learning_rate": 8.134694763083326e-06,
      "loss": 2.079,
      "step": 3134
    },
    {
      "epoch": 1.0685071574642127,
      "grad_norm": 19.835563801842827,
      "learning_rate": 8.133149562958553e-06,
      "loss": 2.3474,
      "step": 3135
    },
    {
      "epoch": 1.068847989093388,
      "grad_norm": 35.89955463328483,
      "learning_rate": 8.131603869976883e-06,
      "loss": 2.1913,
      "step": 3136
    },
    {
      "epoch": 1.069188820722563,
      "grad_norm": 12.141430853569297,
      "learning_rate": 8.130057684381457e-06,
      "loss": 2.163,
      "step": 3137
    },
    {
      "epoch": 1.0695296523517381,
      "grad_norm": 17.015096675282965,
      "learning_rate": 8.128511006415496e-06,
      "loss": 2.4282,
      "step": 3138
    },
    {
      "epoch": 1.0698704839809134,
      "grad_norm": 17.38373503661309,
      "learning_rate": 8.126963836322301e-06,
      "loss": 2.3006,
      "step": 3139
    },
    {
      "epoch": 1.0702113156100885,
      "grad_norm": 12.707510282825181,
      "learning_rate": 8.125416174345246e-06,
      "loss": 2.0925,
      "step": 3140
    },
    {
      "epoch": 1.0705521472392638,
      "grad_norm": 20.923561258168093,
      "learning_rate": 8.123868020727787e-06,
      "loss": 2.4227,
      "step": 3141
    },
    {
      "epoch": 1.070892978868439,
      "grad_norm": 27.126265781637404,
      "learning_rate": 8.122319375713451e-06,
      "loss": 2.2754,
      "step": 3142
    },
    {
      "epoch": 1.0712338104976142,
      "grad_norm": 18.458042558699006,
      "learning_rate": 8.120770239545849e-06,
      "loss": 1.9235,
      "step": 3143
    },
    {
      "epoch": 1.0715746421267893,
      "grad_norm": 17.370463613680858,
      "learning_rate": 8.119220612468665e-06,
      "loss": 1.6792,
      "step": 3144
    },
    {
      "epoch": 1.0719154737559646,
      "grad_norm": 16.92558827196925,
      "learning_rate": 8.117670494725661e-06,
      "loss": 2.4309,
      "step": 3145
    },
    {
      "epoch": 1.0722563053851397,
      "grad_norm": 9.877990698772642,
      "learning_rate": 8.116119886560679e-06,
      "loss": 1.6414,
      "step": 3146
    },
    {
      "epoch": 1.072597137014315,
      "grad_norm": 16.614410430672475,
      "learning_rate": 8.114568788217633e-06,
      "loss": 1.997,
      "step": 3147
    },
    {
      "epoch": 1.0729379686434901,
      "grad_norm": 21.713234715385564,
      "learning_rate": 8.113017199940518e-06,
      "loss": 2.3511,
      "step": 3148
    },
    {
      "epoch": 1.0732788002726652,
      "grad_norm": 20.57344536376163,
      "learning_rate": 8.111465121973408e-06,
      "loss": 2.4847,
      "step": 3149
    },
    {
      "epoch": 1.0736196319018405,
      "grad_norm": 15.338302964548546,
      "learning_rate": 8.109912554560446e-06,
      "loss": 2.5154,
      "step": 3150
    },
    {
      "epoch": 1.0739604635310156,
      "grad_norm": 20.89790366932806,
      "learning_rate": 8.10835949794586e-06,
      "loss": 2.0382,
      "step": 3151
    },
    {
      "epoch": 1.074301295160191,
      "grad_norm": 13.169943436210849,
      "learning_rate": 8.10680595237395e-06,
      "loss": 1.8008,
      "step": 3152
    },
    {
      "epoch": 1.074642126789366,
      "grad_norm": 27.222254481362327,
      "learning_rate": 8.105251918089097e-06,
      "loss": 2.6793,
      "step": 3153
    },
    {
      "epoch": 1.0749829584185413,
      "grad_norm": 26.495432946335,
      "learning_rate": 8.103697395335755e-06,
      "loss": 2.3513,
      "step": 3154
    },
    {
      "epoch": 1.0753237900477164,
      "grad_norm": 18.598940071613598,
      "learning_rate": 8.10214238435846e-06,
      "loss": 2.2787,
      "step": 3155
    },
    {
      "epoch": 1.0756646216768917,
      "grad_norm": 25.750243343036917,
      "learning_rate": 8.100586885401817e-06,
      "loss": 2.7696,
      "step": 3156
    },
    {
      "epoch": 1.0760054533060668,
      "grad_norm": 15.517251216461291,
      "learning_rate": 8.099030898710516e-06,
      "loss": 2.1773,
      "step": 3157
    },
    {
      "epoch": 1.076346284935242,
      "grad_norm": 14.975727028667999,
      "learning_rate": 8.097474424529317e-06,
      "loss": 2.1912,
      "step": 3158
    },
    {
      "epoch": 1.0766871165644172,
      "grad_norm": 22.827158912443142,
      "learning_rate": 8.09591746310306e-06,
      "loss": 2.6409,
      "step": 3159
    },
    {
      "epoch": 1.0770279481935923,
      "grad_norm": 18.317163088871478,
      "learning_rate": 8.094360014676662e-06,
      "loss": 1.9059,
      "step": 3160
    },
    {
      "epoch": 1.0773687798227676,
      "grad_norm": 15.867549967970058,
      "learning_rate": 8.092802079495116e-06,
      "loss": 2.4134,
      "step": 3161
    },
    {
      "epoch": 1.0777096114519427,
      "grad_norm": 13.187792784716995,
      "learning_rate": 8.091243657803492e-06,
      "loss": 1.5458,
      "step": 3162
    },
    {
      "epoch": 1.078050443081118,
      "grad_norm": 53.62445693892496,
      "learning_rate": 8.089684749846936e-06,
      "loss": 1.9946,
      "step": 3163
    },
    {
      "epoch": 1.078391274710293,
      "grad_norm": 19.476277318605565,
      "learning_rate": 8.088125355870667e-06,
      "loss": 2.4941,
      "step": 3164
    },
    {
      "epoch": 1.0787321063394684,
      "grad_norm": 18.940823804116544,
      "learning_rate": 8.086565476119987e-06,
      "loss": 2.1838,
      "step": 3165
    },
    {
      "epoch": 1.0790729379686435,
      "grad_norm": 18.780629244048033,
      "learning_rate": 8.085005110840272e-06,
      "loss": 2.3449,
      "step": 3166
    },
    {
      "epoch": 1.0794137695978188,
      "grad_norm": 15.95047682505284,
      "learning_rate": 8.083444260276972e-06,
      "loss": 1.6398,
      "step": 3167
    },
    {
      "epoch": 1.0797546012269938,
      "grad_norm": 20.216216292697318,
      "learning_rate": 8.081882924675615e-06,
      "loss": 2.1688,
      "step": 3168
    },
    {
      "epoch": 1.0800954328561692,
      "grad_norm": 11.366547208295799,
      "learning_rate": 8.080321104281806e-06,
      "loss": 1.9828,
      "step": 3169
    },
    {
      "epoch": 1.0804362644853442,
      "grad_norm": 20.5170626760122,
      "learning_rate": 8.078758799341224e-06,
      "loss": 2.3839,
      "step": 3170
    },
    {
      "epoch": 1.0807770961145193,
      "grad_norm": 16.55415403907038,
      "learning_rate": 8.077196010099628e-06,
      "loss": 2.5871,
      "step": 3171
    },
    {
      "epoch": 1.0811179277436946,
      "grad_norm": 15.970253486454304,
      "learning_rate": 8.07563273680285e-06,
      "loss": 2.7828,
      "step": 3172
    },
    {
      "epoch": 1.0814587593728697,
      "grad_norm": 19.99269069641523,
      "learning_rate": 8.0740689796968e-06,
      "loss": 1.9075,
      "step": 3173
    },
    {
      "epoch": 1.081799591002045,
      "grad_norm": 17.696483931820286,
      "learning_rate": 8.072504739027461e-06,
      "loss": 2.3156,
      "step": 3174
    },
    {
      "epoch": 1.0821404226312201,
      "grad_norm": 36.2668693526958,
      "learning_rate": 8.070940015040895e-06,
      "loss": 2.4483,
      "step": 3175
    },
    {
      "epoch": 1.0824812542603954,
      "grad_norm": 17.135151907343378,
      "learning_rate": 8.069374807983243e-06,
      "loss": 2.0338,
      "step": 3176
    },
    {
      "epoch": 1.0828220858895705,
      "grad_norm": 20.436114926869138,
      "learning_rate": 8.067809118100712e-06,
      "loss": 2.3836,
      "step": 3177
    },
    {
      "epoch": 1.0831629175187458,
      "grad_norm": 17.590043716308852,
      "learning_rate": 8.066242945639596e-06,
      "loss": 2.3458,
      "step": 3178
    },
    {
      "epoch": 1.083503749147921,
      "grad_norm": 30.126377058694498,
      "learning_rate": 8.064676290846259e-06,
      "loss": 2.295,
      "step": 3179
    },
    {
      "epoch": 1.0838445807770962,
      "grad_norm": 13.084091317004107,
      "learning_rate": 8.063109153967141e-06,
      "loss": 2.1502,
      "step": 3180
    },
    {
      "epoch": 1.0841854124062713,
      "grad_norm": 11.181764811947094,
      "learning_rate": 8.061541535248759e-06,
      "loss": 2.1741,
      "step": 3181
    },
    {
      "epoch": 1.0845262440354464,
      "grad_norm": 13.436189627830931,
      "learning_rate": 8.059973434937707e-06,
      "loss": 2.3652,
      "step": 3182
    },
    {
      "epoch": 1.0848670756646217,
      "grad_norm": 16.622685731419896,
      "learning_rate": 8.058404853280652e-06,
      "loss": 1.891,
      "step": 3183
    },
    {
      "epoch": 1.0852079072937968,
      "grad_norm": 15.315723970787205,
      "learning_rate": 8.05683579052434e-06,
      "loss": 2.5497,
      "step": 3184
    },
    {
      "epoch": 1.085548738922972,
      "grad_norm": 14.416864373302854,
      "learning_rate": 8.055266246915589e-06,
      "loss": 1.9673,
      "step": 3185
    },
    {
      "epoch": 1.0858895705521472,
      "grad_norm": 12.671401902873622,
      "learning_rate": 8.053696222701294e-06,
      "loss": 1.8941,
      "step": 3186
    },
    {
      "epoch": 1.0862304021813225,
      "grad_norm": 15.50992439158444,
      "learning_rate": 8.052125718128428e-06,
      "loss": 2.6163,
      "step": 3187
    },
    {
      "epoch": 1.0865712338104976,
      "grad_norm": 13.587385570349172,
      "learning_rate": 8.050554733444036e-06,
      "loss": 2.0261,
      "step": 3188
    },
    {
      "epoch": 1.086912065439673,
      "grad_norm": 18.53938444912566,
      "learning_rate": 8.048983268895241e-06,
      "loss": 1.9923,
      "step": 3189
    },
    {
      "epoch": 1.087252897068848,
      "grad_norm": 21.9480655032452,
      "learning_rate": 8.047411324729242e-06,
      "loss": 1.8066,
      "step": 3190
    },
    {
      "epoch": 1.0875937286980233,
      "grad_norm": 22.938067352679735,
      "learning_rate": 8.04583890119331e-06,
      "loss": 2.207,
      "step": 3191
    },
    {
      "epoch": 1.0879345603271984,
      "grad_norm": 14.514017640209069,
      "learning_rate": 8.044265998534793e-06,
      "loss": 2.1644,
      "step": 3192
    },
    {
      "epoch": 1.0882753919563735,
      "grad_norm": 17.07840466492976,
      "learning_rate": 8.042692617001118e-06,
      "loss": 1.9882,
      "step": 3193
    },
    {
      "epoch": 1.0886162235855488,
      "grad_norm": 13.33423410121783,
      "learning_rate": 8.04111875683978e-06,
      "loss": 1.6925,
      "step": 3194
    },
    {
      "epoch": 1.0889570552147239,
      "grad_norm": 19.510102516520295,
      "learning_rate": 8.039544418298358e-06,
      "loss": 2.3903,
      "step": 3195
    },
    {
      "epoch": 1.0892978868438992,
      "grad_norm": 22.088352787750637,
      "learning_rate": 8.037969601624495e-06,
      "loss": 2.1696,
      "step": 3196
    },
    {
      "epoch": 1.0896387184730743,
      "grad_norm": 18.47955253328768,
      "learning_rate": 8.036394307065923e-06,
      "loss": 2.8124,
      "step": 3197
    },
    {
      "epoch": 1.0899795501022496,
      "grad_norm": 15.0813233130154,
      "learning_rate": 8.03481853487044e-06,
      "loss": 2.3649,
      "step": 3198
    },
    {
      "epoch": 1.0903203817314246,
      "grad_norm": 14.962493066940002,
      "learning_rate": 8.033242285285919e-06,
      "loss": 2.6459,
      "step": 3199
    },
    {
      "epoch": 1.0906612133606,
      "grad_norm": 16.415201521999474,
      "learning_rate": 8.031665558560313e-06,
      "loss": 2.711,
      "step": 3200
    },
    {
      "epoch": 1.091002044989775,
      "grad_norm": 14.39468983110036,
      "learning_rate": 8.030088354941645e-06,
      "loss": 1.9284,
      "step": 3201
    },
    {
      "epoch": 1.0913428766189504,
      "grad_norm": 16.84682654519699,
      "learning_rate": 8.028510674678018e-06,
      "loss": 1.8573,
      "step": 3202
    },
    {
      "epoch": 1.0916837082481254,
      "grad_norm": 16.049978740945726,
      "learning_rate": 8.026932518017605e-06,
      "loss": 2.7321,
      "step": 3203
    },
    {
      "epoch": 1.0920245398773005,
      "grad_norm": 14.954382030364723,
      "learning_rate": 8.025353885208657e-06,
      "loss": 1.7487,
      "step": 3204
    },
    {
      "epoch": 1.0923653715064758,
      "grad_norm": 16.13165540799585,
      "learning_rate": 8.023774776499502e-06,
      "loss": 2.2835,
      "step": 3205
    },
    {
      "epoch": 1.092706203135651,
      "grad_norm": 11.592206059275936,
      "learning_rate": 8.022195192138535e-06,
      "loss": 1.978,
      "step": 3206
    },
    {
      "epoch": 1.0930470347648262,
      "grad_norm": 14.525045083009802,
      "learning_rate": 8.020615132374236e-06,
      "loss": 2.642,
      "step": 3207
    },
    {
      "epoch": 1.0933878663940013,
      "grad_norm": 15.609060441606163,
      "learning_rate": 8.01903459745515e-06,
      "loss": 1.8288,
      "step": 3208
    },
    {
      "epoch": 1.0937286980231766,
      "grad_norm": 17.76783627876322,
      "learning_rate": 8.017453587629906e-06,
      "loss": 2.2987,
      "step": 3209
    },
    {
      "epoch": 1.0940695296523517,
      "grad_norm": 15.39128602718989,
      "learning_rate": 8.015872103147199e-06,
      "loss": 2.4194,
      "step": 3210
    },
    {
      "epoch": 1.094410361281527,
      "grad_norm": 14.804288407980847,
      "learning_rate": 8.014290144255806e-06,
      "loss": 2.4688,
      "step": 3211
    },
    {
      "epoch": 1.094751192910702,
      "grad_norm": 13.584996030905666,
      "learning_rate": 8.012707711204573e-06,
      "loss": 2.3963,
      "step": 3212
    },
    {
      "epoch": 1.0950920245398774,
      "grad_norm": 14.198138519229806,
      "learning_rate": 8.011124804242424e-06,
      "loss": 2.0981,
      "step": 3213
    },
    {
      "epoch": 1.0954328561690525,
      "grad_norm": 12.815754882040364,
      "learning_rate": 8.009541423618359e-06,
      "loss": 1.8702,
      "step": 3214
    },
    {
      "epoch": 1.0957736877982276,
      "grad_norm": 24.274237888834897,
      "learning_rate": 8.007957569581447e-06,
      "loss": 1.4268,
      "step": 3215
    },
    {
      "epoch": 1.096114519427403,
      "grad_norm": 15.883622539814077,
      "learning_rate": 8.006373242380834e-06,
      "loss": 2.2147,
      "step": 3216
    },
    {
      "epoch": 1.096455351056578,
      "grad_norm": 11.89931443888466,
      "learning_rate": 8.00478844226574e-06,
      "loss": 1.6344,
      "step": 3217
    },
    {
      "epoch": 1.0967961826857533,
      "grad_norm": 13.88622674461722,
      "learning_rate": 8.003203169485467e-06,
      "loss": 1.8146,
      "step": 3218
    },
    {
      "epoch": 1.0971370143149284,
      "grad_norm": 15.156307949321688,
      "learning_rate": 8.001617424289379e-06,
      "loss": 2.2974,
      "step": 3219
    },
    {
      "epoch": 1.0974778459441037,
      "grad_norm": 14.258564864768744,
      "learning_rate": 8.000031206926921e-06,
      "loss": 2.2082,
      "step": 3220
    },
    {
      "epoch": 1.0978186775732788,
      "grad_norm": 14.784228365521683,
      "learning_rate": 7.998444517647613e-06,
      "loss": 1.9239,
      "step": 3221
    },
    {
      "epoch": 1.098159509202454,
      "grad_norm": 17.266649501201943,
      "learning_rate": 7.996857356701045e-06,
      "loss": 2.2911,
      "step": 3222
    },
    {
      "epoch": 1.0985003408316292,
      "grad_norm": 14.980744064622227,
      "learning_rate": 7.995269724336886e-06,
      "loss": 2.3697,
      "step": 3223
    },
    {
      "epoch": 1.0988411724608043,
      "grad_norm": 14.999418377518635,
      "learning_rate": 7.993681620804874e-06,
      "loss": 1.7488,
      "step": 3224
    },
    {
      "epoch": 1.0991820040899796,
      "grad_norm": 15.92277697198263,
      "learning_rate": 7.992093046354827e-06,
      "loss": 2.509,
      "step": 3225
    },
    {
      "epoch": 1.0995228357191547,
      "grad_norm": 18.557292311315123,
      "learning_rate": 7.990504001236633e-06,
      "loss": 2.3265,
      "step": 3226
    },
    {
      "epoch": 1.09986366734833,
      "grad_norm": 18.608192194420543,
      "learning_rate": 7.988914485700255e-06,
      "loss": 1.9073,
      "step": 3227
    },
    {
      "epoch": 1.100204498977505,
      "grad_norm": 14.105403110058363,
      "learning_rate": 7.98732449999573e-06,
      "loss": 2.1848,
      "step": 3228
    },
    {
      "epoch": 1.1005453306066804,
      "grad_norm": 24.20272429202263,
      "learning_rate": 7.98573404437317e-06,
      "loss": 2.4388,
      "step": 3229
    },
    {
      "epoch": 1.1008861622358554,
      "grad_norm": 17.68374320712257,
      "learning_rate": 7.984143119082759e-06,
      "loss": 1.5977,
      "step": 3230
    },
    {
      "epoch": 1.1012269938650308,
      "grad_norm": 19.960332774244147,
      "learning_rate": 7.982551724374755e-06,
      "loss": 2.1678,
      "step": 3231
    },
    {
      "epoch": 1.1015678254942058,
      "grad_norm": 27.83798927372564,
      "learning_rate": 7.980959860499494e-06,
      "loss": 2.3072,
      "step": 3232
    },
    {
      "epoch": 1.1019086571233812,
      "grad_norm": 14.431823204692204,
      "learning_rate": 7.979367527707379e-06,
      "loss": 2.3311,
      "step": 3233
    },
    {
      "epoch": 1.1022494887525562,
      "grad_norm": 22.4567782738246,
      "learning_rate": 7.97777472624889e-06,
      "loss": 2.3087,
      "step": 3234
    },
    {
      "epoch": 1.1025903203817313,
      "grad_norm": 13.615487994904296,
      "learning_rate": 7.976181456374587e-06,
      "loss": 2.1664,
      "step": 3235
    },
    {
      "epoch": 1.1029311520109066,
      "grad_norm": 22.971670825941,
      "learning_rate": 7.974587718335091e-06,
      "loss": 2.8007,
      "step": 3236
    },
    {
      "epoch": 1.1032719836400817,
      "grad_norm": 17.044162581632712,
      "learning_rate": 7.972993512381105e-06,
      "loss": 2.074,
      "step": 3237
    },
    {
      "epoch": 1.103612815269257,
      "grad_norm": 16.396271642580803,
      "learning_rate": 7.971398838763406e-06,
      "loss": 1.9353,
      "step": 3238
    },
    {
      "epoch": 1.1039536468984321,
      "grad_norm": 17.516829745061383,
      "learning_rate": 7.96980369773284e-06,
      "loss": 2.3333,
      "step": 3239
    },
    {
      "epoch": 1.1042944785276074,
      "grad_norm": 16.752776902730993,
      "learning_rate": 7.968208089540332e-06,
      "loss": 2.5672,
      "step": 3240
    },
    {
      "epoch": 1.1046353101567825,
      "grad_norm": 14.80584551306591,
      "learning_rate": 7.966612014436876e-06,
      "loss": 2.2669,
      "step": 3241
    },
    {
      "epoch": 1.1049761417859578,
      "grad_norm": 17.883990157920547,
      "learning_rate": 7.965015472673538e-06,
      "loss": 2.1144,
      "step": 3242
    },
    {
      "epoch": 1.105316973415133,
      "grad_norm": 15.038055043054378,
      "learning_rate": 7.963418464501464e-06,
      "loss": 2.7557,
      "step": 3243
    },
    {
      "epoch": 1.105657805044308,
      "grad_norm": 18.394911270103695,
      "learning_rate": 7.961820990171871e-06,
      "loss": 2.9488,
      "step": 3244
    },
    {
      "epoch": 1.1059986366734833,
      "grad_norm": 14.051256310531572,
      "learning_rate": 7.960223049936045e-06,
      "loss": 1.8045,
      "step": 3245
    },
    {
      "epoch": 1.1063394683026584,
      "grad_norm": 15.271525743524034,
      "learning_rate": 7.958624644045348e-06,
      "loss": 2.4993,
      "step": 3246
    },
    {
      "epoch": 1.1066802999318337,
      "grad_norm": 15.82914739666783,
      "learning_rate": 7.95702577275122e-06,
      "loss": 1.9199,
      "step": 3247
    },
    {
      "epoch": 1.1070211315610088,
      "grad_norm": 20.853518637008605,
      "learning_rate": 7.955426436305163e-06,
      "loss": 2.4308,
      "step": 3248
    },
    {
      "epoch": 1.107361963190184,
      "grad_norm": 22.694153393964385,
      "learning_rate": 7.953826634958763e-06,
      "loss": 2.2159,
      "step": 3249
    },
    {
      "epoch": 1.1077027948193592,
      "grad_norm": 18.463564680008517,
      "learning_rate": 7.952226368963677e-06,
      "loss": 2.1331,
      "step": 3250
    },
    {
      "epoch": 1.1080436264485345,
      "grad_norm": 17.732599113691062,
      "learning_rate": 7.95062563857163e-06,
      "loss": 2.3031,
      "step": 3251
    },
    {
      "epoch": 1.1083844580777096,
      "grad_norm": 21.87268528872712,
      "learning_rate": 7.949024444034424e-06,
      "loss": 1.786,
      "step": 3252
    },
    {
      "epoch": 1.1087252897068849,
      "grad_norm": 23.539314601127803,
      "learning_rate": 7.947422785603934e-06,
      "loss": 2.7283,
      "step": 3253
    },
    {
      "epoch": 1.10906612133606,
      "grad_norm": 12.940591684480484,
      "learning_rate": 7.945820663532105e-06,
      "loss": 1.9,
      "step": 3254
    },
    {
      "epoch": 1.109406952965235,
      "grad_norm": 17.02730000764426,
      "learning_rate": 7.944218078070963e-06,
      "loss": 2.8901,
      "step": 3255
    },
    {
      "epoch": 1.1097477845944104,
      "grad_norm": 17.621491587086457,
      "learning_rate": 7.942615029472593e-06,
      "loss": 1.6036,
      "step": 3256
    },
    {
      "epoch": 1.1100886162235855,
      "grad_norm": 12.794252726550406,
      "learning_rate": 7.941011517989166e-06,
      "loss": 1.6849,
      "step": 3257
    },
    {
      "epoch": 1.1104294478527608,
      "grad_norm": 20.575298570531164,
      "learning_rate": 7.939407543872921e-06,
      "loss": 2.5317,
      "step": 3258
    },
    {
      "epoch": 1.1107702794819359,
      "grad_norm": 15.648678887510131,
      "learning_rate": 7.937803107376168e-06,
      "loss": 2.748,
      "step": 3259
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 13.95351087607938,
      "learning_rate": 7.936198208751292e-06,
      "loss": 1.994,
      "step": 3260
    },
    {
      "epoch": 1.1114519427402862,
      "grad_norm": 14.322990342698946,
      "learning_rate": 7.934592848250747e-06,
      "loss": 2.059,
      "step": 3261
    },
    {
      "epoch": 1.1117927743694616,
      "grad_norm": 10.203945081638588,
      "learning_rate": 7.932987026127066e-06,
      "loss": 1.7368,
      "step": 3262
    },
    {
      "epoch": 1.1121336059986366,
      "grad_norm": 21.683401523568225,
      "learning_rate": 7.931380742632852e-06,
      "loss": 2.1464,
      "step": 3263
    },
    {
      "epoch": 1.112474437627812,
      "grad_norm": 20.25418430358129,
      "learning_rate": 7.929773998020776e-06,
      "loss": 2.2328,
      "step": 3264
    },
    {
      "epoch": 1.112815269256987,
      "grad_norm": 16.26802081380709,
      "learning_rate": 7.928166792543589e-06,
      "loss": 2.6076,
      "step": 3265
    },
    {
      "epoch": 1.1131561008861621,
      "grad_norm": 18.504566926869845,
      "learning_rate": 7.926559126454109e-06,
      "loss": 2.988,
      "step": 3266
    },
    {
      "epoch": 1.1134969325153374,
      "grad_norm": 18.93975025957297,
      "learning_rate": 7.924951000005228e-06,
      "loss": 1.9905,
      "step": 3267
    },
    {
      "epoch": 1.1138377641445125,
      "grad_norm": 17.185824360338753,
      "learning_rate": 7.92334241344991e-06,
      "loss": 2.6204,
      "step": 3268
    },
    {
      "epoch": 1.1141785957736878,
      "grad_norm": 19.216871543627892,
      "learning_rate": 7.921733367041195e-06,
      "loss": 2.3781,
      "step": 3269
    },
    {
      "epoch": 1.114519427402863,
      "grad_norm": 18.667527287921267,
      "learning_rate": 7.92012386103219e-06,
      "loss": 2.2127,
      "step": 3270
    },
    {
      "epoch": 1.1148602590320382,
      "grad_norm": 21.872414346059074,
      "learning_rate": 7.918513895676077e-06,
      "loss": 2.4983,
      "step": 3271
    },
    {
      "epoch": 1.1152010906612133,
      "grad_norm": 18.13650247120749,
      "learning_rate": 7.91690347122611e-06,
      "loss": 2.1931,
      "step": 3272
    },
    {
      "epoch": 1.1155419222903886,
      "grad_norm": 18.133826628312377,
      "learning_rate": 7.915292587935615e-06,
      "loss": 2.0104,
      "step": 3273
    },
    {
      "epoch": 1.1158827539195637,
      "grad_norm": 24.25160585149576,
      "learning_rate": 7.913681246057992e-06,
      "loss": 2.2624,
      "step": 3274
    },
    {
      "epoch": 1.116223585548739,
      "grad_norm": 19.143270950434935,
      "learning_rate": 7.91206944584671e-06,
      "loss": 2.4288,
      "step": 3275
    },
    {
      "epoch": 1.116564417177914,
      "grad_norm": 23.63434836978218,
      "learning_rate": 7.910457187555311e-06,
      "loss": 2.78,
      "step": 3276
    },
    {
      "epoch": 1.1169052488070892,
      "grad_norm": 11.9739410906923,
      "learning_rate": 7.908844471437413e-06,
      "loss": 2.0488,
      "step": 3277
    },
    {
      "epoch": 1.1172460804362645,
      "grad_norm": 19.107530415917232,
      "learning_rate": 7.907231297746698e-06,
      "loss": 2.2045,
      "step": 3278
    },
    {
      "epoch": 1.1175869120654396,
      "grad_norm": 25.106956024708573,
      "learning_rate": 7.905617666736927e-06,
      "loss": 2.6024,
      "step": 3279
    },
    {
      "epoch": 1.117927743694615,
      "grad_norm": 21.84524489752337,
      "learning_rate": 7.904003578661933e-06,
      "loss": 2.3441,
      "step": 3280
    },
    {
      "epoch": 1.11826857532379,
      "grad_norm": 19.531750962038622,
      "learning_rate": 7.902389033775612e-06,
      "loss": 2.8603,
      "step": 3281
    },
    {
      "epoch": 1.1186094069529653,
      "grad_norm": 15.334834321326309,
      "learning_rate": 7.900774032331943e-06,
      "loss": 2.3338,
      "step": 3282
    },
    {
      "epoch": 1.1189502385821404,
      "grad_norm": 18.522699254017493,
      "learning_rate": 7.899158574584973e-06,
      "loss": 2.7134,
      "step": 3283
    },
    {
      "epoch": 1.1192910702113157,
      "grad_norm": 16.16630811759295,
      "learning_rate": 7.897542660788818e-06,
      "loss": 2.1765,
      "step": 3284
    },
    {
      "epoch": 1.1196319018404908,
      "grad_norm": 16.705871435697475,
      "learning_rate": 7.895926291197667e-06,
      "loss": 1.7722,
      "step": 3285
    },
    {
      "epoch": 1.119972733469666,
      "grad_norm": 16.645465159156743,
      "learning_rate": 7.894309466065784e-06,
      "loss": 1.4785,
      "step": 3286
    },
    {
      "epoch": 1.1203135650988412,
      "grad_norm": 15.593261771515197,
      "learning_rate": 7.892692185647499e-06,
      "loss": 2.4151,
      "step": 3287
    },
    {
      "epoch": 1.1206543967280163,
      "grad_norm": 10.427086529340585,
      "learning_rate": 7.891074450197219e-06,
      "loss": 1.9486,
      "step": 3288
    },
    {
      "epoch": 1.1209952283571916,
      "grad_norm": 12.052091473818175,
      "learning_rate": 7.889456259969418e-06,
      "loss": 1.8618,
      "step": 3289
    },
    {
      "epoch": 1.1213360599863667,
      "grad_norm": 15.107123953188882,
      "learning_rate": 7.887837615218648e-06,
      "loss": 2.6977,
      "step": 3290
    },
    {
      "epoch": 1.121676891615542,
      "grad_norm": 16.451701529990903,
      "learning_rate": 7.886218516199522e-06,
      "loss": 1.8724,
      "step": 3291
    },
    {
      "epoch": 1.122017723244717,
      "grad_norm": 21.820672552648134,
      "learning_rate": 7.884598963166736e-06,
      "loss": 2.0172,
      "step": 3292
    },
    {
      "epoch": 1.1223585548738924,
      "grad_norm": 18.33351244253047,
      "learning_rate": 7.882978956375048e-06,
      "loss": 2.9295,
      "step": 3293
    },
    {
      "epoch": 1.1226993865030674,
      "grad_norm": 15.533486967515652,
      "learning_rate": 7.881358496079297e-06,
      "loss": 2.7047,
      "step": 3294
    },
    {
      "epoch": 1.1230402181322428,
      "grad_norm": 18.96750096306392,
      "learning_rate": 7.879737582534382e-06,
      "loss": 2.2042,
      "step": 3295
    },
    {
      "epoch": 1.1233810497614178,
      "grad_norm": 19.035064047222846,
      "learning_rate": 7.878116215995282e-06,
      "loss": 2.3091,
      "step": 3296
    },
    {
      "epoch": 1.1237218813905931,
      "grad_norm": 20.420009180448417,
      "learning_rate": 7.876494396717044e-06,
      "loss": 2.5589,
      "step": 3297
    },
    {
      "epoch": 1.1240627130197682,
      "grad_norm": 16.209909886959682,
      "learning_rate": 7.874872124954788e-06,
      "loss": 2.6532,
      "step": 3298
    },
    {
      "epoch": 1.1244035446489433,
      "grad_norm": 16.285719033829782,
      "learning_rate": 7.873249400963701e-06,
      "loss": 2.0218,
      "step": 3299
    },
    {
      "epoch": 1.1247443762781186,
      "grad_norm": 18.799979813347477,
      "learning_rate": 7.871626224999048e-06,
      "loss": 1.7604,
      "step": 3300
    },
    {
      "epoch": 1.1250852079072937,
      "grad_norm": 14.377028544178179,
      "learning_rate": 7.870002597316155e-06,
      "loss": 2.3859,
      "step": 3301
    },
    {
      "epoch": 1.125426039536469,
      "grad_norm": 16.509835587620174,
      "learning_rate": 7.86837851817043e-06,
      "loss": 2.3378,
      "step": 3302
    },
    {
      "epoch": 1.1257668711656441,
      "grad_norm": 11.50260809591511,
      "learning_rate": 7.866753987817347e-06,
      "loss": 1.9984,
      "step": 3303
    },
    {
      "epoch": 1.1261077027948194,
      "grad_norm": 17.701129683771814,
      "learning_rate": 7.865129006512448e-06,
      "loss": 2.2657,
      "step": 3304
    },
    {
      "epoch": 1.1264485344239945,
      "grad_norm": 21.116108566263353,
      "learning_rate": 7.863503574511354e-06,
      "loss": 2.2838,
      "step": 3305
    },
    {
      "epoch": 1.1267893660531698,
      "grad_norm": 23.30332735577259,
      "learning_rate": 7.861877692069746e-06,
      "loss": 2.8452,
      "step": 3306
    },
    {
      "epoch": 1.127130197682345,
      "grad_norm": 17.81914103254937,
      "learning_rate": 7.860251359443386e-06,
      "loss": 2.0968,
      "step": 3307
    },
    {
      "epoch": 1.1274710293115202,
      "grad_norm": 25.252530302535245,
      "learning_rate": 7.8586245768881e-06,
      "loss": 2.1439,
      "step": 3308
    },
    {
      "epoch": 1.1278118609406953,
      "grad_norm": 15.226571702762017,
      "learning_rate": 7.856997344659788e-06,
      "loss": 2.6405,
      "step": 3309
    },
    {
      "epoch": 1.1281526925698704,
      "grad_norm": 21.776894887482452,
      "learning_rate": 7.85536966301442e-06,
      "loss": 2.1528,
      "step": 3310
    },
    {
      "epoch": 1.1284935241990457,
      "grad_norm": 19.194700210366815,
      "learning_rate": 7.85374153220804e-06,
      "loss": 2.4896,
      "step": 3311
    },
    {
      "epoch": 1.1288343558282208,
      "grad_norm": 12.180401821803889,
      "learning_rate": 7.852112952496755e-06,
      "loss": 2.1481,
      "step": 3312
    },
    {
      "epoch": 1.129175187457396,
      "grad_norm": 14.109818023776713,
      "learning_rate": 7.850483924136751e-06,
      "loss": 2.0975,
      "step": 3313
    },
    {
      "epoch": 1.1295160190865712,
      "grad_norm": 16.473100559633057,
      "learning_rate": 7.848854447384278e-06,
      "loss": 2.2334,
      "step": 3314
    },
    {
      "epoch": 1.1298568507157465,
      "grad_norm": 15.058589659764682,
      "learning_rate": 7.847224522495657e-06,
      "loss": 2.2939,
      "step": 3315
    },
    {
      "epoch": 1.1301976823449216,
      "grad_norm": 14.06300254435552,
      "learning_rate": 7.845594149727288e-06,
      "loss": 1.9568,
      "step": 3316
    },
    {
      "epoch": 1.1305385139740969,
      "grad_norm": 20.44809883669552,
      "learning_rate": 7.843963329335631e-06,
      "loss": 2.6436,
      "step": 3317
    },
    {
      "epoch": 1.130879345603272,
      "grad_norm": 19.11521334290607,
      "learning_rate": 7.84233206157722e-06,
      "loss": 2.1564,
      "step": 3318
    },
    {
      "epoch": 1.1312201772324473,
      "grad_norm": 14.530084262792466,
      "learning_rate": 7.84070034670866e-06,
      "loss": 1.8501,
      "step": 3319
    },
    {
      "epoch": 1.1315610088616224,
      "grad_norm": 14.187532402727902,
      "learning_rate": 7.839068184986629e-06,
      "loss": 1.9951,
      "step": 3320
    },
    {
      "epoch": 1.1319018404907975,
      "grad_norm": 13.612620243325834,
      "learning_rate": 7.837435576667872e-06,
      "loss": 2.0879,
      "step": 3321
    },
    {
      "epoch": 1.1322426721199728,
      "grad_norm": 18.229536225185527,
      "learning_rate": 7.835802522009201e-06,
      "loss": 2.2611,
      "step": 3322
    },
    {
      "epoch": 1.1325835037491478,
      "grad_norm": 16.122363331723726,
      "learning_rate": 7.834169021267505e-06,
      "loss": 1.9306,
      "step": 3323
    },
    {
      "epoch": 1.1329243353783232,
      "grad_norm": 21.18358213001543,
      "learning_rate": 7.832535074699741e-06,
      "loss": 2.6036,
      "step": 3324
    },
    {
      "epoch": 1.1332651670074982,
      "grad_norm": 17.554908609211026,
      "learning_rate": 7.830900682562933e-06,
      "loss": 2.2412,
      "step": 3325
    },
    {
      "epoch": 1.1336059986366736,
      "grad_norm": 13.123615864731933,
      "learning_rate": 7.82926584511418e-06,
      "loss": 1.7771,
      "step": 3326
    },
    {
      "epoch": 1.1339468302658486,
      "grad_norm": 17.11383097258368,
      "learning_rate": 7.827630562610647e-06,
      "loss": 2.7906,
      "step": 3327
    },
    {
      "epoch": 1.134287661895024,
      "grad_norm": 23.490094613840775,
      "learning_rate": 7.825994835309569e-06,
      "loss": 2.0116,
      "step": 3328
    },
    {
      "epoch": 1.134628493524199,
      "grad_norm": 14.725307100363581,
      "learning_rate": 7.824358663468254e-06,
      "loss": 1.6223,
      "step": 3329
    },
    {
      "epoch": 1.1349693251533743,
      "grad_norm": 24.104110069847007,
      "learning_rate": 7.822722047344077e-06,
      "loss": 1.8455,
      "step": 3330
    },
    {
      "epoch": 1.1353101567825494,
      "grad_norm": 29.83963352904978,
      "learning_rate": 7.821084987194487e-06,
      "loss": 2.1759,
      "step": 3331
    },
    {
      "epoch": 1.1356509884117245,
      "grad_norm": 35.9832726150975,
      "learning_rate": 7.819447483276998e-06,
      "loss": 2.2729,
      "step": 3332
    },
    {
      "epoch": 1.1359918200408998,
      "grad_norm": 15.07591925243444,
      "learning_rate": 7.817809535849197e-06,
      "loss": 2.2584,
      "step": 3333
    },
    {
      "epoch": 1.136332651670075,
      "grad_norm": 18.970961587440705,
      "learning_rate": 7.816171145168739e-06,
      "loss": 1.8775,
      "step": 3334
    },
    {
      "epoch": 1.1366734832992502,
      "grad_norm": 23.194811955429742,
      "learning_rate": 7.814532311493348e-06,
      "loss": 2.1792,
      "step": 3335
    },
    {
      "epoch": 1.1370143149284253,
      "grad_norm": 13.860033359785337,
      "learning_rate": 7.81289303508082e-06,
      "loss": 2.2617,
      "step": 3336
    },
    {
      "epoch": 1.1373551465576006,
      "grad_norm": 17.84631157666731,
      "learning_rate": 7.811253316189023e-06,
      "loss": 1.9161,
      "step": 3337
    },
    {
      "epoch": 1.1376959781867757,
      "grad_norm": 19.20181840671097,
      "learning_rate": 7.809613155075886e-06,
      "loss": 2.1374,
      "step": 3338
    },
    {
      "epoch": 1.138036809815951,
      "grad_norm": 27.240052937782078,
      "learning_rate": 7.807972551999414e-06,
      "loss": 2.7086,
      "step": 3339
    },
    {
      "epoch": 1.138377641445126,
      "grad_norm": 22.05717611502453,
      "learning_rate": 7.806331507217684e-06,
      "loss": 2.2785,
      "step": 3340
    },
    {
      "epoch": 1.1387184730743014,
      "grad_norm": 24.31840941539864,
      "learning_rate": 7.804690020988837e-06,
      "loss": 2.1519,
      "step": 3341
    },
    {
      "epoch": 1.1390593047034765,
      "grad_norm": 16.40075977500821,
      "learning_rate": 7.803048093571084e-06,
      "loss": 2.0425,
      "step": 3342
    },
    {
      "epoch": 1.1394001363326516,
      "grad_norm": 27.57541687635015,
      "learning_rate": 7.801405725222708e-06,
      "loss": 2.6323,
      "step": 3343
    },
    {
      "epoch": 1.139740967961827,
      "grad_norm": 22.142334980760744,
      "learning_rate": 7.799762916202058e-06,
      "loss": 2.8017,
      "step": 3344
    },
    {
      "epoch": 1.140081799591002,
      "grad_norm": 11.935185376071686,
      "learning_rate": 7.798119666767559e-06,
      "loss": 2.4363,
      "step": 3345
    },
    {
      "epoch": 1.1404226312201773,
      "grad_norm": 18.36382555861202,
      "learning_rate": 7.796475977177695e-06,
      "loss": 2.2847,
      "step": 3346
    },
    {
      "epoch": 1.1407634628493524,
      "grad_norm": 17.42969349311105,
      "learning_rate": 7.794831847691028e-06,
      "loss": 1.9215,
      "step": 3347
    },
    {
      "epoch": 1.1411042944785277,
      "grad_norm": 21.699400493683267,
      "learning_rate": 7.793187278566186e-06,
      "loss": 2.852,
      "step": 3348
    },
    {
      "epoch": 1.1414451261077028,
      "grad_norm": 17.75710302799252,
      "learning_rate": 7.791542270061867e-06,
      "loss": 2.4463,
      "step": 3349
    },
    {
      "epoch": 1.141785957736878,
      "grad_norm": 19.760825981433086,
      "learning_rate": 7.789896822436835e-06,
      "loss": 2.0645,
      "step": 3350
    },
    {
      "epoch": 1.1421267893660532,
      "grad_norm": 14.071506526439624,
      "learning_rate": 7.788250935949929e-06,
      "loss": 2.4669,
      "step": 3351
    },
    {
      "epoch": 1.1424676209952285,
      "grad_norm": 13.745633775600728,
      "learning_rate": 7.78660461086005e-06,
      "loss": 2.043,
      "step": 3352
    },
    {
      "epoch": 1.1428084526244036,
      "grad_norm": 14.615078090108394,
      "learning_rate": 7.784957847426174e-06,
      "loss": 2.7033,
      "step": 3353
    },
    {
      "epoch": 1.1431492842535786,
      "grad_norm": 17.545058052844286,
      "learning_rate": 7.78331064590734e-06,
      "loss": 1.9087,
      "step": 3354
    },
    {
      "epoch": 1.143490115882754,
      "grad_norm": 16.32023807855148,
      "learning_rate": 7.781663006562663e-06,
      "loss": 2.5155,
      "step": 3355
    },
    {
      "epoch": 1.143830947511929,
      "grad_norm": 18.114867547116596,
      "learning_rate": 7.780014929651323e-06,
      "loss": 2.267,
      "step": 3356
    },
    {
      "epoch": 1.1441717791411044,
      "grad_norm": 19.658994992005546,
      "learning_rate": 7.778366415432566e-06,
      "loss": 2.7726,
      "step": 3357
    },
    {
      "epoch": 1.1445126107702794,
      "grad_norm": 17.985881600089815,
      "learning_rate": 7.776717464165713e-06,
      "loss": 2.9494,
      "step": 3358
    },
    {
      "epoch": 1.1448534423994547,
      "grad_norm": 15.587798550104226,
      "learning_rate": 7.77506807611015e-06,
      "loss": 2.0557,
      "step": 3359
    },
    {
      "epoch": 1.1451942740286298,
      "grad_norm": 16.207464430475465,
      "learning_rate": 7.773418251525331e-06,
      "loss": 2.7022,
      "step": 3360
    },
    {
      "epoch": 1.145535105657805,
      "grad_norm": 16.389940244758403,
      "learning_rate": 7.77176799067078e-06,
      "loss": 1.8798,
      "step": 3361
    },
    {
      "epoch": 1.1458759372869802,
      "grad_norm": 20.90196986928099,
      "learning_rate": 7.77011729380609e-06,
      "loss": 2.5315,
      "step": 3362
    },
    {
      "epoch": 1.1462167689161555,
      "grad_norm": 14.491794805676072,
      "learning_rate": 7.768466161190924e-06,
      "loss": 2.3364,
      "step": 3363
    },
    {
      "epoch": 1.1465576005453306,
      "grad_norm": 13.75439902019629,
      "learning_rate": 7.76681459308501e-06,
      "loss": 2.1735,
      "step": 3364
    },
    {
      "epoch": 1.1468984321745057,
      "grad_norm": 23.875586320181622,
      "learning_rate": 7.765162589748146e-06,
      "loss": 2.4275,
      "step": 3365
    },
    {
      "epoch": 1.147239263803681,
      "grad_norm": 13.669817000461418,
      "learning_rate": 7.7635101514402e-06,
      "loss": 1.7657,
      "step": 3366
    },
    {
      "epoch": 1.147580095432856,
      "grad_norm": 18.458613062388867,
      "learning_rate": 7.761857278421104e-06,
      "loss": 2.4672,
      "step": 3367
    },
    {
      "epoch": 1.1479209270620314,
      "grad_norm": 19.486741022578194,
      "learning_rate": 7.760203970950866e-06,
      "loss": 2.3249,
      "step": 3368
    },
    {
      "epoch": 1.1482617586912065,
      "grad_norm": 11.250491843122916,
      "learning_rate": 7.758550229289554e-06,
      "loss": 2.0424,
      "step": 3369
    },
    {
      "epoch": 1.1486025903203818,
      "grad_norm": 15.186805206006467,
      "learning_rate": 7.75689605369731e-06,
      "loss": 2.3832,
      "step": 3370
    },
    {
      "epoch": 1.148943421949557,
      "grad_norm": 15.462041510349598,
      "learning_rate": 7.755241444434346e-06,
      "loss": 2.1058,
      "step": 3371
    },
    {
      "epoch": 1.149284253578732,
      "grad_norm": 12.484444103319479,
      "learning_rate": 7.75358640176093e-06,
      "loss": 2.3678,
      "step": 3372
    },
    {
      "epoch": 1.1496250852079073,
      "grad_norm": 14.955214170161877,
      "learning_rate": 7.751930925937414e-06,
      "loss": 2.095,
      "step": 3373
    },
    {
      "epoch": 1.1499659168370826,
      "grad_norm": 9.249919633147934,
      "learning_rate": 7.750275017224208e-06,
      "loss": 1.8462,
      "step": 3374
    },
    {
      "epoch": 1.1503067484662577,
      "grad_norm": 17.43559972989925,
      "learning_rate": 7.748618675881793e-06,
      "loss": 2.2142,
      "step": 3375
    },
    {
      "epoch": 1.1506475800954328,
      "grad_norm": 32.791435441444136,
      "learning_rate": 7.74696190217072e-06,
      "loss": 2.7314,
      "step": 3376
    },
    {
      "epoch": 1.150988411724608,
      "grad_norm": 18.809398036581218,
      "learning_rate": 7.745304696351603e-06,
      "loss": 2.4865,
      "step": 3377
    },
    {
      "epoch": 1.1513292433537832,
      "grad_norm": 13.661697032719012,
      "learning_rate": 7.74364705868513e-06,
      "loss": 1.8984,
      "step": 3378
    },
    {
      "epoch": 1.1516700749829585,
      "grad_norm": 23.15591430817826,
      "learning_rate": 7.741988989432052e-06,
      "loss": 2.4242,
      "step": 3379
    },
    {
      "epoch": 1.1520109066121336,
      "grad_norm": 15.32816807623184,
      "learning_rate": 7.740330488853192e-06,
      "loss": 2.8042,
      "step": 3380
    },
    {
      "epoch": 1.1523517382413089,
      "grad_norm": 15.753941385475837,
      "learning_rate": 7.738671557209437e-06,
      "loss": 1.7111,
      "step": 3381
    },
    {
      "epoch": 1.152692569870484,
      "grad_norm": 19.632660391574486,
      "learning_rate": 7.737012194761743e-06,
      "loss": 2.4625,
      "step": 3382
    },
    {
      "epoch": 1.153033401499659,
      "grad_norm": 28.10078405182242,
      "learning_rate": 7.735352401771136e-06,
      "loss": 1.9864,
      "step": 3383
    },
    {
      "epoch": 1.1533742331288344,
      "grad_norm": 15.665789735392659,
      "learning_rate": 7.733692178498705e-06,
      "loss": 2.2073,
      "step": 3384
    },
    {
      "epoch": 1.1537150647580097,
      "grad_norm": 14.621939992101218,
      "learning_rate": 7.732031525205616e-06,
      "loss": 2.7708,
      "step": 3385
    },
    {
      "epoch": 1.1540558963871848,
      "grad_norm": 13.009102069699392,
      "learning_rate": 7.730370442153092e-06,
      "loss": 1.8405,
      "step": 3386
    },
    {
      "epoch": 1.1543967280163598,
      "grad_norm": 12.992180995514437,
      "learning_rate": 7.728708929602428e-06,
      "loss": 1.8741,
      "step": 3387
    },
    {
      "epoch": 1.1547375596455351,
      "grad_norm": 12.947531557969745,
      "learning_rate": 7.727046987814986e-06,
      "loss": 2.2434,
      "step": 3388
    },
    {
      "epoch": 1.1550783912747102,
      "grad_norm": 11.832831115328096,
      "learning_rate": 7.725384617052198e-06,
      "loss": 2.3875,
      "step": 3389
    },
    {
      "epoch": 1.1554192229038855,
      "grad_norm": 14.865087761302346,
      "learning_rate": 7.72372181757556e-06,
      "loss": 1.9367,
      "step": 3390
    },
    {
      "epoch": 1.1557600545330606,
      "grad_norm": 17.9234545085971,
      "learning_rate": 7.722058589646638e-06,
      "loss": 1.92,
      "step": 3391
    },
    {
      "epoch": 1.156100886162236,
      "grad_norm": 12.283253763437534,
      "learning_rate": 7.720394933527065e-06,
      "loss": 2.1613,
      "step": 3392
    },
    {
      "epoch": 1.156441717791411,
      "grad_norm": 15.005772546890071,
      "learning_rate": 7.718730849478538e-06,
      "loss": 2.1218,
      "step": 3393
    },
    {
      "epoch": 1.1567825494205861,
      "grad_norm": 29.68738866772878,
      "learning_rate": 7.717066337762826e-06,
      "loss": 2.1741,
      "step": 3394
    },
    {
      "epoch": 1.1571233810497614,
      "grad_norm": 10.25110259695917,
      "learning_rate": 7.715401398641764e-06,
      "loss": 1.8942,
      "step": 3395
    },
    {
      "epoch": 1.1574642126789365,
      "grad_norm": 14.954398557298907,
      "learning_rate": 7.713736032377252e-06,
      "loss": 2.263,
      "step": 3396
    },
    {
      "epoch": 1.1578050443081118,
      "grad_norm": 29.226301537696163,
      "learning_rate": 7.712070239231259e-06,
      "loss": 2.9588,
      "step": 3397
    },
    {
      "epoch": 1.158145875937287,
      "grad_norm": 19.49017209369269,
      "learning_rate": 7.710404019465821e-06,
      "loss": 1.9383,
      "step": 3398
    },
    {
      "epoch": 1.1584867075664622,
      "grad_norm": 14.755534164061041,
      "learning_rate": 7.708737373343042e-06,
      "loss": 2.5223,
      "step": 3399
    },
    {
      "epoch": 1.1588275391956373,
      "grad_norm": 18.325033933927585,
      "learning_rate": 7.70707030112509e-06,
      "loss": 2.0835,
      "step": 3400
    },
    {
      "epoch": 1.1591683708248126,
      "grad_norm": 13.855776675171917,
      "learning_rate": 7.705402803074205e-06,
      "loss": 2.2114,
      "step": 3401
    },
    {
      "epoch": 1.1595092024539877,
      "grad_norm": 16.98276213234163,
      "learning_rate": 7.70373487945269e-06,
      "loss": 1.9083,
      "step": 3402
    },
    {
      "epoch": 1.159850034083163,
      "grad_norm": 20.28503275423625,
      "learning_rate": 7.702066530522914e-06,
      "loss": 2.3952,
      "step": 3403
    },
    {
      "epoch": 1.160190865712338,
      "grad_norm": 18.43051255013586,
      "learning_rate": 7.700397756547316e-06,
      "loss": 2.4471,
      "step": 3404
    },
    {
      "epoch": 1.1605316973415132,
      "grad_norm": 14.837971430236047,
      "learning_rate": 7.698728557788404e-06,
      "loss": 1.57,
      "step": 3405
    },
    {
      "epoch": 1.1608725289706885,
      "grad_norm": 15.377154919196006,
      "learning_rate": 7.697058934508743e-06,
      "loss": 1.8119,
      "step": 3406
    },
    {
      "epoch": 1.1612133605998636,
      "grad_norm": 17.607149835689807,
      "learning_rate": 7.695388886970979e-06,
      "loss": 2.2873,
      "step": 3407
    },
    {
      "epoch": 1.1615541922290389,
      "grad_norm": 14.985937145461904,
      "learning_rate": 7.69371841543781e-06,
      "loss": 2.5539,
      "step": 3408
    },
    {
      "epoch": 1.161895023858214,
      "grad_norm": 19.72346597645711,
      "learning_rate": 7.692047520172014e-06,
      "loss": 2.1347,
      "step": 3409
    },
    {
      "epoch": 1.1622358554873893,
      "grad_norm": 16.392937516017952,
      "learning_rate": 7.690376201436427e-06,
      "loss": 1.8872,
      "step": 3410
    },
    {
      "epoch": 1.1625766871165644,
      "grad_norm": 18.411805358276542,
      "learning_rate": 7.688704459493951e-06,
      "loss": 2.3606,
      "step": 3411
    },
    {
      "epoch": 1.1629175187457397,
      "grad_norm": 15.575856750297906,
      "learning_rate": 7.687032294607561e-06,
      "loss": 1.7731,
      "step": 3412
    },
    {
      "epoch": 1.1632583503749148,
      "grad_norm": 17.095945003572407,
      "learning_rate": 7.685359707040296e-06,
      "loss": 2.252,
      "step": 3413
    },
    {
      "epoch": 1.16359918200409,
      "grad_norm": 17.244386775196546,
      "learning_rate": 7.683686697055256e-06,
      "loss": 2.1421,
      "step": 3414
    },
    {
      "epoch": 1.1639400136332652,
      "grad_norm": 20.829577947213316,
      "learning_rate": 7.68201326491562e-06,
      "loss": 1.7814,
      "step": 3415
    },
    {
      "epoch": 1.1642808452624402,
      "grad_norm": 18.960435020596222,
      "learning_rate": 7.680339410884619e-06,
      "loss": 2.1998,
      "step": 3416
    },
    {
      "epoch": 1.1646216768916156,
      "grad_norm": 13.268958142318443,
      "learning_rate": 7.678665135225558e-06,
      "loss": 2.5912,
      "step": 3417
    },
    {
      "epoch": 1.1649625085207906,
      "grad_norm": 17.094918093522853,
      "learning_rate": 7.676990438201808e-06,
      "loss": 2.2172,
      "step": 3418
    },
    {
      "epoch": 1.165303340149966,
      "grad_norm": 26.658287165110018,
      "learning_rate": 7.675315320076809e-06,
      "loss": 2.4682,
      "step": 3419
    },
    {
      "epoch": 1.165644171779141,
      "grad_norm": 15.486209747800526,
      "learning_rate": 7.673639781114055e-06,
      "loss": 2.6389,
      "step": 3420
    },
    {
      "epoch": 1.1659850034083163,
      "grad_norm": 20.69204977197211,
      "learning_rate": 7.671963821577123e-06,
      "loss": 2.3443,
      "step": 3421
    },
    {
      "epoch": 1.1663258350374914,
      "grad_norm": 16.01857032250532,
      "learning_rate": 7.670287441729647e-06,
      "loss": 2.1366,
      "step": 3422
    },
    {
      "epoch": 1.1666666666666667,
      "grad_norm": 18.80879519637838,
      "learning_rate": 7.668610641835325e-06,
      "loss": 1.8906,
      "step": 3423
    },
    {
      "epoch": 1.1670074982958418,
      "grad_norm": 16.27658153006072,
      "learning_rate": 7.666933422157926e-06,
      "loss": 2.2458,
      "step": 3424
    },
    {
      "epoch": 1.1673483299250171,
      "grad_norm": 24.01513884687653,
      "learning_rate": 7.665255782961282e-06,
      "loss": 2.1456,
      "step": 3425
    },
    {
      "epoch": 1.1676891615541922,
      "grad_norm": 13.952236201548029,
      "learning_rate": 7.663577724509295e-06,
      "loss": 2.3332,
      "step": 3426
    },
    {
      "epoch": 1.1680299931833673,
      "grad_norm": 16.895645287690993,
      "learning_rate": 7.661899247065929e-06,
      "loss": 2.5937,
      "step": 3427
    },
    {
      "epoch": 1.1683708248125426,
      "grad_norm": 11.108179074757164,
      "learning_rate": 7.660220350895216e-06,
      "loss": 1.634,
      "step": 3428
    },
    {
      "epoch": 1.1687116564417177,
      "grad_norm": 12.698037047350054,
      "learning_rate": 7.658541036261248e-06,
      "loss": 1.8058,
      "step": 3429
    },
    {
      "epoch": 1.169052488070893,
      "grad_norm": 14.551575941801643,
      "learning_rate": 7.656861303428196e-06,
      "loss": 2.3621,
      "step": 3430
    },
    {
      "epoch": 1.169393319700068,
      "grad_norm": 14.521570394262133,
      "learning_rate": 7.655181152660283e-06,
      "loss": 2.091,
      "step": 3431
    },
    {
      "epoch": 1.1697341513292434,
      "grad_norm": 18.76161708598552,
      "learning_rate": 7.653500584221804e-06,
      "loss": 2.2979,
      "step": 3432
    },
    {
      "epoch": 1.1700749829584185,
      "grad_norm": 16.98365152451555,
      "learning_rate": 7.651819598377122e-06,
      "loss": 3.2343,
      "step": 3433
    },
    {
      "epoch": 1.1704158145875938,
      "grad_norm": 31.006627607798507,
      "learning_rate": 7.65013819539066e-06,
      "loss": 2.4609,
      "step": 3434
    },
    {
      "epoch": 1.170756646216769,
      "grad_norm": 40.630471870113595,
      "learning_rate": 7.648456375526912e-06,
      "loss": 2.3612,
      "step": 3435
    },
    {
      "epoch": 1.1710974778459442,
      "grad_norm": 16.226210703905597,
      "learning_rate": 7.646774139050433e-06,
      "loss": 1.9699,
      "step": 3436
    },
    {
      "epoch": 1.1714383094751193,
      "grad_norm": 20.227294055544572,
      "learning_rate": 7.645091486225844e-06,
      "loss": 2.5663,
      "step": 3437
    },
    {
      "epoch": 1.1717791411042944,
      "grad_norm": 16.20322603923074,
      "learning_rate": 7.643408417317833e-06,
      "loss": 2.2977,
      "step": 3438
    },
    {
      "epoch": 1.1721199727334697,
      "grad_norm": 14.854843781940165,
      "learning_rate": 7.64172493259116e-06,
      "loss": 2.5916,
      "step": 3439
    },
    {
      "epoch": 1.1724608043626448,
      "grad_norm": 16.29702578753333,
      "learning_rate": 7.640041032310634e-06,
      "loss": 1.6729,
      "step": 3440
    },
    {
      "epoch": 1.17280163599182,
      "grad_norm": 19.534156049314966,
      "learning_rate": 7.63835671674115e-06,
      "loss": 1.9279,
      "step": 3441
    },
    {
      "epoch": 1.1731424676209952,
      "grad_norm": 18.968088419834693,
      "learning_rate": 7.636671986147646e-06,
      "loss": 2.3098,
      "step": 3442
    },
    {
      "epoch": 1.1734832992501705,
      "grad_norm": 16.630196106036905,
      "learning_rate": 7.634986840795145e-06,
      "loss": 2.2328,
      "step": 3443
    },
    {
      "epoch": 1.1738241308793456,
      "grad_norm": 14.9746838328074,
      "learning_rate": 7.633301280948726e-06,
      "loss": 2.209,
      "step": 3444
    },
    {
      "epoch": 1.1741649625085209,
      "grad_norm": 16.314422963903528,
      "learning_rate": 7.63161530687353e-06,
      "loss": 2.653,
      "step": 3445
    },
    {
      "epoch": 1.174505794137696,
      "grad_norm": 13.613653704986504,
      "learning_rate": 7.629928918834773e-06,
      "loss": 2.1704,
      "step": 3446
    },
    {
      "epoch": 1.1748466257668713,
      "grad_norm": 14.778198186893983,
      "learning_rate": 7.628242117097726e-06,
      "loss": 2.4202,
      "step": 3447
    },
    {
      "epoch": 1.1751874573960464,
      "grad_norm": 19.452591408056715,
      "learning_rate": 7.626554901927733e-06,
      "loss": 2.0892,
      "step": 3448
    },
    {
      "epoch": 1.1755282890252214,
      "grad_norm": 15.84077361299141,
      "learning_rate": 7.624867273590197e-06,
      "loss": 2.3837,
      "step": 3449
    },
    {
      "epoch": 1.1758691206543967,
      "grad_norm": 19.112587468243554,
      "learning_rate": 7.6231792323505905e-06,
      "loss": 2.3467,
      "step": 3450
    },
    {
      "epoch": 1.1762099522835718,
      "grad_norm": 19.397293979358363,
      "learning_rate": 7.621490778474449e-06,
      "loss": 2.611,
      "step": 3451
    },
    {
      "epoch": 1.1765507839127471,
      "grad_norm": 19.610323334801798,
      "learning_rate": 7.619801912227371e-06,
      "loss": 2.1823,
      "step": 3452
    },
    {
      "epoch": 1.1768916155419222,
      "grad_norm": 18.76558405734609,
      "learning_rate": 7.618112633875025e-06,
      "loss": 2.0959,
      "step": 3453
    },
    {
      "epoch": 1.1772324471710975,
      "grad_norm": 14.953699537050474,
      "learning_rate": 7.616422943683139e-06,
      "loss": 2.6041,
      "step": 3454
    },
    {
      "epoch": 1.1775732788002726,
      "grad_norm": 15.151714738165927,
      "learning_rate": 7.61473284191751e-06,
      "loss": 2.2419,
      "step": 3455
    },
    {
      "epoch": 1.177914110429448,
      "grad_norm": 22.50036755961169,
      "learning_rate": 7.613042328843994e-06,
      "loss": 2.8255,
      "step": 3456
    },
    {
      "epoch": 1.178254942058623,
      "grad_norm": 18.6934667462161,
      "learning_rate": 7.611351404728519e-06,
      "loss": 2.2603,
      "step": 3457
    },
    {
      "epoch": 1.1785957736877983,
      "grad_norm": 14.762762997775798,
      "learning_rate": 7.609660069837073e-06,
      "loss": 1.7921,
      "step": 3458
    },
    {
      "epoch": 1.1789366053169734,
      "grad_norm": 17.097164622593485,
      "learning_rate": 7.607968324435708e-06,
      "loss": 2.6404,
      "step": 3459
    },
    {
      "epoch": 1.1792774369461485,
      "grad_norm": 26.073557694704736,
      "learning_rate": 7.6062761687905455e-06,
      "loss": 2.2309,
      "step": 3460
    },
    {
      "epoch": 1.1796182685753238,
      "grad_norm": 16.157653947715477,
      "learning_rate": 7.604583603167766e-06,
      "loss": 2.1912,
      "step": 3461
    },
    {
      "epoch": 1.179959100204499,
      "grad_norm": 13.294126846034418,
      "learning_rate": 7.602890627833618e-06,
      "loss": 2.408,
      "step": 3462
    },
    {
      "epoch": 1.1802999318336742,
      "grad_norm": 12.93306572507711,
      "learning_rate": 7.601197243054411e-06,
      "loss": 2.4086,
      "step": 3463
    },
    {
      "epoch": 1.1806407634628493,
      "grad_norm": 13.442976883451449,
      "learning_rate": 7.599503449096526e-06,
      "loss": 2.4956,
      "step": 3464
    },
    {
      "epoch": 1.1809815950920246,
      "grad_norm": 14.37042108451778,
      "learning_rate": 7.597809246226398e-06,
      "loss": 2.5281,
      "step": 3465
    },
    {
      "epoch": 1.1813224267211997,
      "grad_norm": 15.29923013691024,
      "learning_rate": 7.596114634710534e-06,
      "loss": 1.8084,
      "step": 3466
    },
    {
      "epoch": 1.181663258350375,
      "grad_norm": 16.542785688982672,
      "learning_rate": 7.5944196148155046e-06,
      "loss": 2.3021,
      "step": 3467
    },
    {
      "epoch": 1.18200408997955,
      "grad_norm": 29.492018360272354,
      "learning_rate": 7.592724186807941e-06,
      "loss": 1.8462,
      "step": 3468
    },
    {
      "epoch": 1.1823449216087254,
      "grad_norm": 16.49800962249054,
      "learning_rate": 7.591028350954544e-06,
      "loss": 1.3855,
      "step": 3469
    },
    {
      "epoch": 1.1826857532379005,
      "grad_norm": 18.50151909277116,
      "learning_rate": 7.589332107522071e-06,
      "loss": 2.2691,
      "step": 3470
    },
    {
      "epoch": 1.1830265848670756,
      "grad_norm": 16.582138227934575,
      "learning_rate": 7.587635456777352e-06,
      "loss": 2.6979,
      "step": 3471
    },
    {
      "epoch": 1.1833674164962509,
      "grad_norm": 10.106485964794352,
      "learning_rate": 7.585938398987271e-06,
      "loss": 1.6698,
      "step": 3472
    },
    {
      "epoch": 1.183708248125426,
      "grad_norm": 19.7502523266852,
      "learning_rate": 7.58424093441879e-06,
      "loss": 2.718,
      "step": 3473
    },
    {
      "epoch": 1.1840490797546013,
      "grad_norm": 21.747719715611176,
      "learning_rate": 7.582543063338921e-06,
      "loss": 2.4786,
      "step": 3474
    },
    {
      "epoch": 1.1843899113837764,
      "grad_norm": 15.2404388449691,
      "learning_rate": 7.580844786014747e-06,
      "loss": 2.5842,
      "step": 3475
    },
    {
      "epoch": 1.1847307430129517,
      "grad_norm": 15.633919224149883,
      "learning_rate": 7.579146102713415e-06,
      "loss": 2.5067,
      "step": 3476
    },
    {
      "epoch": 1.1850715746421268,
      "grad_norm": 18.539120473212122,
      "learning_rate": 7.577447013702134e-06,
      "loss": 2.2343,
      "step": 3477
    },
    {
      "epoch": 1.185412406271302,
      "grad_norm": 17.447620857011277,
      "learning_rate": 7.5757475192481775e-06,
      "loss": 2.5094,
      "step": 3478
    },
    {
      "epoch": 1.1857532379004772,
      "grad_norm": 15.063753855306148,
      "learning_rate": 7.574047619618883e-06,
      "loss": 2.7539,
      "step": 3479
    },
    {
      "epoch": 1.1860940695296525,
      "grad_norm": 20.61940711644142,
      "learning_rate": 7.572347315081651e-06,
      "loss": 2.2919,
      "step": 3480
    },
    {
      "epoch": 1.1864349011588275,
      "grad_norm": 13.07162816423925,
      "learning_rate": 7.570646605903946e-06,
      "loss": 2.5554,
      "step": 3481
    },
    {
      "epoch": 1.1867757327880026,
      "grad_norm": 13.773736762052865,
      "learning_rate": 7.568945492353298e-06,
      "loss": 1.9924,
      "step": 3482
    },
    {
      "epoch": 1.187116564417178,
      "grad_norm": 21.899165872731015,
      "learning_rate": 7.567243974697297e-06,
      "loss": 2.272,
      "step": 3483
    },
    {
      "epoch": 1.187457396046353,
      "grad_norm": 17.739357909718006,
      "learning_rate": 7.5655420532035986e-06,
      "loss": 2.469,
      "step": 3484
    },
    {
      "epoch": 1.1877982276755283,
      "grad_norm": 14.158885780186182,
      "learning_rate": 7.5638397281399235e-06,
      "loss": 2.2376,
      "step": 3485
    },
    {
      "epoch": 1.1881390593047034,
      "grad_norm": 19.128590764027564,
      "learning_rate": 7.562136999774053e-06,
      "loss": 1.658,
      "step": 3486
    },
    {
      "epoch": 1.1884798909338787,
      "grad_norm": 14.202242255225126,
      "learning_rate": 7.560433868373833e-06,
      "loss": 2.9048,
      "step": 3487
    },
    {
      "epoch": 1.1888207225630538,
      "grad_norm": 15.492488452833802,
      "learning_rate": 7.558730334207172e-06,
      "loss": 2.22,
      "step": 3488
    },
    {
      "epoch": 1.1891615541922291,
      "grad_norm": 17.351415090181845,
      "learning_rate": 7.557026397542044e-06,
      "loss": 2.6243,
      "step": 3489
    },
    {
      "epoch": 1.1895023858214042,
      "grad_norm": 17.71521766015105,
      "learning_rate": 7.555322058646486e-06,
      "loss": 2.2206,
      "step": 3490
    },
    {
      "epoch": 1.1898432174505795,
      "grad_norm": 14.930696113850889,
      "learning_rate": 7.553617317788595e-06,
      "loss": 2.0301,
      "step": 3491
    },
    {
      "epoch": 1.1901840490797546,
      "grad_norm": 16.163062023195884,
      "learning_rate": 7.551912175236533e-06,
      "loss": 2.4345,
      "step": 3492
    },
    {
      "epoch": 1.1905248807089297,
      "grad_norm": 19.363770558920987,
      "learning_rate": 7.55020663125853e-06,
      "loss": 1.8443,
      "step": 3493
    },
    {
      "epoch": 1.190865712338105,
      "grad_norm": 18.827202837121124,
      "learning_rate": 7.5485006861228706e-06,
      "loss": 2.4356,
      "step": 3494
    },
    {
      "epoch": 1.19120654396728,
      "grad_norm": 13.827659993959035,
      "learning_rate": 7.546794340097909e-06,
      "loss": 2.4537,
      "step": 3495
    },
    {
      "epoch": 1.1915473755964554,
      "grad_norm": 21.1689276759338,
      "learning_rate": 7.545087593452061e-06,
      "loss": 1.9085,
      "step": 3496
    },
    {
      "epoch": 1.1918882072256305,
      "grad_norm": 18.274134830929608,
      "learning_rate": 7.5433804464538e-06,
      "loss": 2.735,
      "step": 3497
    },
    {
      "epoch": 1.1922290388548058,
      "grad_norm": 512.2358164496825,
      "learning_rate": 7.541672899371672e-06,
      "loss": 2.1946,
      "step": 3498
    },
    {
      "epoch": 1.1925698704839809,
      "grad_norm": 13.868302182733672,
      "learning_rate": 7.539964952474279e-06,
      "loss": 2.1872,
      "step": 3499
    },
    {
      "epoch": 1.1929107021131562,
      "grad_norm": 15.590274893294607,
      "learning_rate": 7.538256606030287e-06,
      "loss": 2.104,
      "step": 3500
    },
    {
      "epoch": 1.1932515337423313,
      "grad_norm": 17.775128656643314,
      "learning_rate": 7.536547860308428e-06,
      "loss": 2.1162,
      "step": 3501
    },
    {
      "epoch": 1.1935923653715066,
      "grad_norm": 14.211993104134171,
      "learning_rate": 7.534838715577491e-06,
      "loss": 1.9346,
      "step": 3502
    },
    {
      "epoch": 1.1939331970006817,
      "grad_norm": 14.916928468703803,
      "learning_rate": 7.533129172106335e-06,
      "loss": 2.4239,
      "step": 3503
    },
    {
      "epoch": 1.1942740286298568,
      "grad_norm": 16.080622769918858,
      "learning_rate": 7.531419230163876e-06,
      "loss": 2.3989,
      "step": 3504
    },
    {
      "epoch": 1.194614860259032,
      "grad_norm": 16.338626271468147,
      "learning_rate": 7.529708890019096e-06,
      "loss": 2.0289,
      "step": 3505
    },
    {
      "epoch": 1.1949556918882072,
      "grad_norm": 14.813461801271133,
      "learning_rate": 7.527998151941037e-06,
      "loss": 2.511,
      "step": 3506
    },
    {
      "epoch": 1.1952965235173825,
      "grad_norm": 12.28832208796034,
      "learning_rate": 7.526287016198805e-06,
      "loss": 1.8742,
      "step": 3507
    },
    {
      "epoch": 1.1956373551465576,
      "grad_norm": 21.62545993358652,
      "learning_rate": 7.524575483061569e-06,
      "loss": 2.4573,
      "step": 3508
    },
    {
      "epoch": 1.1959781867757329,
      "grad_norm": 14.123824505739906,
      "learning_rate": 7.52286355279856e-06,
      "loss": 2.0966,
      "step": 3509
    },
    {
      "epoch": 1.196319018404908,
      "grad_norm": 14.862403428297199,
      "learning_rate": 7.521151225679071e-06,
      "loss": 2.3123,
      "step": 3510
    },
    {
      "epoch": 1.196659850034083,
      "grad_norm": 22.112961896284595,
      "learning_rate": 7.519438501972458e-06,
      "loss": 2.3274,
      "step": 3511
    },
    {
      "epoch": 1.1970006816632583,
      "grad_norm": 15.942780992008942,
      "learning_rate": 7.5177253819481396e-06,
      "loss": 2.4161,
      "step": 3512
    },
    {
      "epoch": 1.1973415132924337,
      "grad_norm": 18.485970328902294,
      "learning_rate": 7.5160118658755964e-06,
      "loss": 2.0957,
      "step": 3513
    },
    {
      "epoch": 1.1976823449216087,
      "grad_norm": 17.268528589898256,
      "learning_rate": 7.5142979540243685e-06,
      "loss": 2.0666,
      "step": 3514
    },
    {
      "epoch": 1.1980231765507838,
      "grad_norm": 15.542683061783723,
      "learning_rate": 7.512583646664067e-06,
      "loss": 2.2258,
      "step": 3515
    },
    {
      "epoch": 1.1983640081799591,
      "grad_norm": 15.588339621167648,
      "learning_rate": 7.510868944064354e-06,
      "loss": 2.1227,
      "step": 3516
    },
    {
      "epoch": 1.1987048398091342,
      "grad_norm": 41.67270728093458,
      "learning_rate": 7.509153846494961e-06,
      "loss": 2.3,
      "step": 3517
    },
    {
      "epoch": 1.1990456714383095,
      "grad_norm": 17.54488124921181,
      "learning_rate": 7.50743835422568e-06,
      "loss": 2.373,
      "step": 3518
    },
    {
      "epoch": 1.1993865030674846,
      "grad_norm": 16.313485615169153,
      "learning_rate": 7.505722467526364e-06,
      "loss": 2.1935,
      "step": 3519
    },
    {
      "epoch": 1.19972733469666,
      "grad_norm": 13.0819249225591,
      "learning_rate": 7.50400618666693e-06,
      "loss": 1.9396,
      "step": 3520
    },
    {
      "epoch": 1.200068166325835,
      "grad_norm": 18.89921947599337,
      "learning_rate": 7.502289511917355e-06,
      "loss": 2.2365,
      "step": 3521
    },
    {
      "epoch": 1.20040899795501,
      "grad_norm": 16.189927551102976,
      "learning_rate": 7.500572443547677e-06,
      "loss": 1.8924,
      "step": 3522
    },
    {
      "epoch": 1.2007498295841854,
      "grad_norm": 14.195569715103849,
      "learning_rate": 7.498854981828001e-06,
      "loss": 2.3932,
      "step": 3523
    },
    {
      "epoch": 1.2010906612133607,
      "grad_norm": 17.267910079188276,
      "learning_rate": 7.4971371270284885e-06,
      "loss": 2.4188,
      "step": 3524
    },
    {
      "epoch": 1.2014314928425358,
      "grad_norm": 94.83606544723433,
      "learning_rate": 7.495418879419365e-06,
      "loss": 2.5374,
      "step": 3525
    },
    {
      "epoch": 1.201772324471711,
      "grad_norm": 18.84020120230579,
      "learning_rate": 7.493700239270919e-06,
      "loss": 2.4035,
      "step": 3526
    },
    {
      "epoch": 1.2021131561008862,
      "grad_norm": 24.47806679942933,
      "learning_rate": 7.491981206853498e-06,
      "loss": 2.1002,
      "step": 3527
    },
    {
      "epoch": 1.2024539877300613,
      "grad_norm": 14.287761051079594,
      "learning_rate": 7.4902617824375134e-06,
      "loss": 2.2792,
      "step": 3528
    },
    {
      "epoch": 1.2027948193592366,
      "grad_norm": 20.46669042358702,
      "learning_rate": 7.488541966293438e-06,
      "loss": 2.1284,
      "step": 3529
    },
    {
      "epoch": 1.2031356509884117,
      "grad_norm": 13.302351299017708,
      "learning_rate": 7.4868217586918036e-06,
      "loss": 2.4003,
      "step": 3530
    },
    {
      "epoch": 1.203476482617587,
      "grad_norm": 24.480064446251045,
      "learning_rate": 7.485101159903207e-06,
      "loss": 2.0672,
      "step": 3531
    },
    {
      "epoch": 1.203817314246762,
      "grad_norm": 34.68355908976286,
      "learning_rate": 7.4833801701983066e-06,
      "loss": 2.2491,
      "step": 3532
    },
    {
      "epoch": 1.2041581458759372,
      "grad_norm": 18.179813543348498,
      "learning_rate": 7.481658789847819e-06,
      "loss": 2.0346,
      "step": 3533
    },
    {
      "epoch": 1.2044989775051125,
      "grad_norm": 15.267720903621887,
      "learning_rate": 7.479937019122524e-06,
      "loss": 2.4428,
      "step": 3534
    },
    {
      "epoch": 1.2048398091342876,
      "grad_norm": 17.243204896623645,
      "learning_rate": 7.478214858293264e-06,
      "loss": 2.0495,
      "step": 3535
    },
    {
      "epoch": 1.2051806407634629,
      "grad_norm": 13.933753906527432,
      "learning_rate": 7.476492307630943e-06,
      "loss": 2.0939,
      "step": 3536
    },
    {
      "epoch": 1.205521472392638,
      "grad_norm": 19.04749311815318,
      "learning_rate": 7.474769367406524e-06,
      "loss": 2.2846,
      "step": 3537
    },
    {
      "epoch": 1.2058623040218133,
      "grad_norm": 20.264200700466226,
      "learning_rate": 7.47304603789103e-06,
      "loss": 2.1143,
      "step": 3538
    },
    {
      "epoch": 1.2062031356509884,
      "grad_norm": 28.577205650221806,
      "learning_rate": 7.47132231935555e-06,
      "loss": 2.7169,
      "step": 3539
    },
    {
      "epoch": 1.2065439672801637,
      "grad_norm": 16.47604754668848,
      "learning_rate": 7.4695982120712315e-06,
      "loss": 1.8882,
      "step": 3540
    },
    {
      "epoch": 1.2068847989093388,
      "grad_norm": 14.210003805229501,
      "learning_rate": 7.467873716309284e-06,
      "loss": 2.1854,
      "step": 3541
    },
    {
      "epoch": 1.207225630538514,
      "grad_norm": 16.27891176840281,
      "learning_rate": 7.466148832340977e-06,
      "loss": 2.5056,
      "step": 3542
    },
    {
      "epoch": 1.2075664621676891,
      "grad_norm": 17.821009301867054,
      "learning_rate": 7.4644235604376416e-06,
      "loss": 2.2501,
      "step": 3543
    },
    {
      "epoch": 1.2079072937968642,
      "grad_norm": 20.500257986162847,
      "learning_rate": 7.462697900870668e-06,
      "loss": 2.7542,
      "step": 3544
    },
    {
      "epoch": 1.2082481254260395,
      "grad_norm": 13.582886360699128,
      "learning_rate": 7.460971853911513e-06,
      "loss": 2.2556,
      "step": 3545
    },
    {
      "epoch": 1.2085889570552146,
      "grad_norm": 15.78654101357696,
      "learning_rate": 7.459245419831689e-06,
      "loss": 2.5435,
      "step": 3546
    },
    {
      "epoch": 1.20892978868439,
      "grad_norm": 23.048750222946015,
      "learning_rate": 7.4575185989027695e-06,
      "loss": 2.1116,
      "step": 3547
    },
    {
      "epoch": 1.209270620313565,
      "grad_norm": 15.515721714143368,
      "learning_rate": 7.4557913913963916e-06,
      "loss": 2.6657,
      "step": 3548
    },
    {
      "epoch": 1.2096114519427403,
      "grad_norm": 10.832170245095888,
      "learning_rate": 7.454063797584253e-06,
      "loss": 2.1862,
      "step": 3549
    },
    {
      "epoch": 1.2099522835719154,
      "grad_norm": 18.582673382198987,
      "learning_rate": 7.45233581773811e-06,
      "loss": 3.2494,
      "step": 3550
    },
    {
      "epoch": 1.2102931152010907,
      "grad_norm": 10.861338513620241,
      "learning_rate": 7.45060745212978e-06,
      "loss": 2.0246,
      "step": 3551
    },
    {
      "epoch": 1.2106339468302658,
      "grad_norm": 17.498058943391214,
      "learning_rate": 7.4488787010311425e-06,
      "loss": 2.3863,
      "step": 3552
    },
    {
      "epoch": 1.2109747784594411,
      "grad_norm": 28.044141153900842,
      "learning_rate": 7.447149564714137e-06,
      "loss": 2.0357,
      "step": 3553
    },
    {
      "epoch": 1.2113156100886162,
      "grad_norm": 13.903379749251021,
      "learning_rate": 7.445420043450764e-06,
      "loss": 2.819,
      "step": 3554
    },
    {
      "epoch": 1.2116564417177913,
      "grad_norm": 17.04732605546652,
      "learning_rate": 7.443690137513083e-06,
      "loss": 2.414,
      "step": 3555
    },
    {
      "epoch": 1.2119972733469666,
      "grad_norm": 17.554506437243358,
      "learning_rate": 7.441959847173215e-06,
      "loss": 2.2809,
      "step": 3556
    },
    {
      "epoch": 1.2123381049761417,
      "grad_norm": 19.633217608871956,
      "learning_rate": 7.440229172703344e-06,
      "loss": 2.4076,
      "step": 3557
    },
    {
      "epoch": 1.212678936605317,
      "grad_norm": 18.132705488540648,
      "learning_rate": 7.438498114375707e-06,
      "loss": 2.4662,
      "step": 3558
    },
    {
      "epoch": 1.213019768234492,
      "grad_norm": 14.304748730703931,
      "learning_rate": 7.436766672462612e-06,
      "loss": 1.9257,
      "step": 3559
    },
    {
      "epoch": 1.2133605998636674,
      "grad_norm": 22.10002847812441,
      "learning_rate": 7.435034847236419e-06,
      "loss": 2.5424,
      "step": 3560
    },
    {
      "epoch": 1.2137014314928425,
      "grad_norm": 16.60651224361225,
      "learning_rate": 7.43330263896955e-06,
      "loss": 2.518,
      "step": 3561
    },
    {
      "epoch": 1.2140422631220178,
      "grad_norm": 15.743519571927683,
      "learning_rate": 7.431570047934491e-06,
      "loss": 1.7291,
      "step": 3562
    },
    {
      "epoch": 1.2143830947511929,
      "grad_norm": 15.163739235112855,
      "learning_rate": 7.429837074403784e-06,
      "loss": 1.9598,
      "step": 3563
    },
    {
      "epoch": 1.2147239263803682,
      "grad_norm": 10.663582874639896,
      "learning_rate": 7.42810371865003e-06,
      "loss": 2.0449,
      "step": 3564
    },
    {
      "epoch": 1.2150647580095433,
      "grad_norm": 12.569432341045202,
      "learning_rate": 7.4263699809458975e-06,
      "loss": 1.4582,
      "step": 3565
    },
    {
      "epoch": 1.2154055896387184,
      "grad_norm": 14.032354167235962,
      "learning_rate": 7.424635861564109e-06,
      "loss": 2.3405,
      "step": 3566
    },
    {
      "epoch": 1.2157464212678937,
      "grad_norm": 15.933161410948284,
      "learning_rate": 7.422901360777446e-06,
      "loss": 2.526,
      "step": 3567
    },
    {
      "epoch": 1.2160872528970688,
      "grad_norm": 18.96289055351101,
      "learning_rate": 7.421166478858754e-06,
      "loss": 2.383,
      "step": 3568
    },
    {
      "epoch": 1.216428084526244,
      "grad_norm": 13.634793147707958,
      "learning_rate": 7.419431216080937e-06,
      "loss": 2.2559,
      "step": 3569
    },
    {
      "epoch": 1.2167689161554192,
      "grad_norm": 19.176283877738257,
      "learning_rate": 7.417695572716959e-06,
      "loss": 2.4939,
      "step": 3570
    },
    {
      "epoch": 1.2171097477845945,
      "grad_norm": 14.145313757268626,
      "learning_rate": 7.415959549039843e-06,
      "loss": 2.2575,
      "step": 3571
    },
    {
      "epoch": 1.2174505794137696,
      "grad_norm": 13.506269090606859,
      "learning_rate": 7.414223145322672e-06,
      "loss": 2.4066,
      "step": 3572
    },
    {
      "epoch": 1.2177914110429449,
      "grad_norm": 16.208544855207588,
      "learning_rate": 7.412486361838589e-06,
      "loss": 2.2495,
      "step": 3573
    },
    {
      "epoch": 1.21813224267212,
      "grad_norm": 17.00047808621785,
      "learning_rate": 7.410749198860801e-06,
      "loss": 2.7603,
      "step": 3574
    },
    {
      "epoch": 1.2184730743012953,
      "grad_norm": 16.77865054907014,
      "learning_rate": 7.409011656662563e-06,
      "loss": 2.4175,
      "step": 3575
    },
    {
      "epoch": 1.2188139059304703,
      "grad_norm": 18.522825891218556,
      "learning_rate": 7.407273735517205e-06,
      "loss": 2.6185,
      "step": 3576
    },
    {
      "epoch": 1.2191547375596454,
      "grad_norm": 25.719382641282785,
      "learning_rate": 7.405535435698104e-06,
      "loss": 2.8405,
      "step": 3577
    },
    {
      "epoch": 1.2194955691888207,
      "grad_norm": 33.16608760979264,
      "learning_rate": 7.4037967574787025e-06,
      "loss": 2.2128,
      "step": 3578
    },
    {
      "epoch": 1.2198364008179958,
      "grad_norm": 11.372800386115012,
      "learning_rate": 7.4020577011325035e-06,
      "loss": 2.0253,
      "step": 3579
    },
    {
      "epoch": 1.2201772324471711,
      "grad_norm": 22.366444820514957,
      "learning_rate": 7.400318266933066e-06,
      "loss": 1.5555,
      "step": 3580
    },
    {
      "epoch": 1.2205180640763462,
      "grad_norm": 29.399124022101848,
      "learning_rate": 7.398578455154009e-06,
      "loss": 2.4454,
      "step": 3581
    },
    {
      "epoch": 1.2208588957055215,
      "grad_norm": 27.562563264581122,
      "learning_rate": 7.396838266069014e-06,
      "loss": 2.5242,
      "step": 3582
    },
    {
      "epoch": 1.2211997273346966,
      "grad_norm": 13.891928276533982,
      "learning_rate": 7.3950976999518185e-06,
      "loss": 2.4913,
      "step": 3583
    },
    {
      "epoch": 1.221540558963872,
      "grad_norm": 16.024094939996104,
      "learning_rate": 7.39335675707622e-06,
      "loss": 2.317,
      "step": 3584
    },
    {
      "epoch": 1.221881390593047,
      "grad_norm": 17.505714302371455,
      "learning_rate": 7.391615437716076e-06,
      "loss": 2.4136,
      "step": 3585
    },
    {
      "epoch": 1.2222222222222223,
      "grad_norm": 16.343953020760413,
      "learning_rate": 7.3898737421453034e-06,
      "loss": 1.8229,
      "step": 3586
    },
    {
      "epoch": 1.2225630538513974,
      "grad_norm": 16.97690128784145,
      "learning_rate": 7.388131670637878e-06,
      "loss": 2.1067,
      "step": 3587
    },
    {
      "epoch": 1.2229038854805725,
      "grad_norm": 14.864053574854012,
      "learning_rate": 7.386389223467835e-06,
      "loss": 2.4956,
      "step": 3588
    },
    {
      "epoch": 1.2232447171097478,
      "grad_norm": 14.119579036689775,
      "learning_rate": 7.3846464009092655e-06,
      "loss": 1.7231,
      "step": 3589
    },
    {
      "epoch": 1.223585548738923,
      "grad_norm": 16.314088045715625,
      "learning_rate": 7.382903203236325e-06,
      "loss": 2.5451,
      "step": 3590
    },
    {
      "epoch": 1.2239263803680982,
      "grad_norm": 19.38342666966621,
      "learning_rate": 7.381159630723225e-06,
      "loss": 2.8963,
      "step": 3591
    },
    {
      "epoch": 1.2242672119972733,
      "grad_norm": 15.487674623972373,
      "learning_rate": 7.379415683644236e-06,
      "loss": 2.1022,
      "step": 3592
    },
    {
      "epoch": 1.2246080436264486,
      "grad_norm": 17.92195627423995,
      "learning_rate": 7.377671362273689e-06,
      "loss": 2.1033,
      "step": 3593
    },
    {
      "epoch": 1.2249488752556237,
      "grad_norm": 16.262260137133545,
      "learning_rate": 7.375926666885971e-06,
      "loss": 1.546,
      "step": 3594
    },
    {
      "epoch": 1.225289706884799,
      "grad_norm": 15.89408210170526,
      "learning_rate": 7.374181597755529e-06,
      "loss": 2.183,
      "step": 3595
    },
    {
      "epoch": 1.225630538513974,
      "grad_norm": 18.469562931029316,
      "learning_rate": 7.372436155156872e-06,
      "loss": 2.0773,
      "step": 3596
    },
    {
      "epoch": 1.2259713701431494,
      "grad_norm": 17.12863860000587,
      "learning_rate": 7.370690339364565e-06,
      "loss": 2.6311,
      "step": 3597
    },
    {
      "epoch": 1.2263122017723245,
      "grad_norm": 15.775235916227796,
      "learning_rate": 7.368944150653228e-06,
      "loss": 2.8052,
      "step": 3598
    },
    {
      "epoch": 1.2266530334014996,
      "grad_norm": 13.794288200412472,
      "learning_rate": 7.367197589297547e-06,
      "loss": 2.398,
      "step": 3599
    },
    {
      "epoch": 1.2269938650306749,
      "grad_norm": 19.147697773223854,
      "learning_rate": 7.365450655572262e-06,
      "loss": 2.8047,
      "step": 3600
    },
    {
      "epoch": 1.22733469665985,
      "grad_norm": 15.751517694515401,
      "learning_rate": 7.3637033497521725e-06,
      "loss": 2.2046,
      "step": 3601
    },
    {
      "epoch": 1.2276755282890253,
      "grad_norm": 24.154804313763915,
      "learning_rate": 7.361955672112136e-06,
      "loss": 2.0774,
      "step": 3602
    },
    {
      "epoch": 1.2280163599182004,
      "grad_norm": 15.546371046546769,
      "learning_rate": 7.360207622927071e-06,
      "loss": 2.0272,
      "step": 3603
    },
    {
      "epoch": 1.2283571915473757,
      "grad_norm": 25.426513864822095,
      "learning_rate": 7.358459202471951e-06,
      "loss": 2.7677,
      "step": 3604
    },
    {
      "epoch": 1.2286980231765507,
      "grad_norm": 17.418597950357185,
      "learning_rate": 7.35671041102181e-06,
      "loss": 1.7659,
      "step": 3605
    },
    {
      "epoch": 1.229038854805726,
      "grad_norm": 18.729767833124114,
      "learning_rate": 7.35496124885174e-06,
      "loss": 2.4308,
      "step": 3606
    },
    {
      "epoch": 1.2293796864349011,
      "grad_norm": 18.305064868506342,
      "learning_rate": 7.35321171623689e-06,
      "loss": 2.0991,
      "step": 3607
    },
    {
      "epoch": 1.2297205180640765,
      "grad_norm": 15.373779513110566,
      "learning_rate": 7.3514618134524716e-06,
      "loss": 2.54,
      "step": 3608
    },
    {
      "epoch": 1.2300613496932515,
      "grad_norm": 19.259790957542098,
      "learning_rate": 7.349711540773748e-06,
      "loss": 2.1264,
      "step": 3609
    },
    {
      "epoch": 1.2304021813224266,
      "grad_norm": 21.47896800489412,
      "learning_rate": 7.347960898476045e-06,
      "loss": 2.781,
      "step": 3610
    },
    {
      "epoch": 1.230743012951602,
      "grad_norm": 15.161500767014147,
      "learning_rate": 7.346209886834746e-06,
      "loss": 2.5495,
      "step": 3611
    },
    {
      "epoch": 1.231083844580777,
      "grad_norm": 13.889952071557646,
      "learning_rate": 7.344458506125293e-06,
      "loss": 1.8115,
      "step": 3612
    },
    {
      "epoch": 1.2314246762099523,
      "grad_norm": 24.16475273179867,
      "learning_rate": 7.342706756623183e-06,
      "loss": 2.4175,
      "step": 3613
    },
    {
      "epoch": 1.2317655078391274,
      "grad_norm": 34.21255014185103,
      "learning_rate": 7.340954638603976e-06,
      "loss": 2.4033,
      "step": 3614
    },
    {
      "epoch": 1.2321063394683027,
      "grad_norm": 19.938915763709094,
      "learning_rate": 7.339202152343283e-06,
      "loss": 1.8758,
      "step": 3615
    },
    {
      "epoch": 1.2324471710974778,
      "grad_norm": 26.12016810456602,
      "learning_rate": 7.3374492981167814e-06,
      "loss": 2.5257,
      "step": 3616
    },
    {
      "epoch": 1.2327880027266531,
      "grad_norm": 19.1147483744932,
      "learning_rate": 7.3356960762002e-06,
      "loss": 2.1305,
      "step": 3617
    },
    {
      "epoch": 1.2331288343558282,
      "grad_norm": 14.251846406810511,
      "learning_rate": 7.333942486869329e-06,
      "loss": 2.1164,
      "step": 3618
    },
    {
      "epoch": 1.2334696659850035,
      "grad_norm": 22.259691582331016,
      "learning_rate": 7.3321885304000126e-06,
      "loss": 2.3459,
      "step": 3619
    },
    {
      "epoch": 1.2338104976141786,
      "grad_norm": 17.013063049768054,
      "learning_rate": 7.330434207068157e-06,
      "loss": 2.4156,
      "step": 3620
    },
    {
      "epoch": 1.2341513292433537,
      "grad_norm": 14.26328016968089,
      "learning_rate": 7.328679517149724e-06,
      "loss": 2.1003,
      "step": 3621
    },
    {
      "epoch": 1.234492160872529,
      "grad_norm": 22.956007543878307,
      "learning_rate": 7.326924460920733e-06,
      "loss": 2.0122,
      "step": 3622
    },
    {
      "epoch": 1.234832992501704,
      "grad_norm": 12.334238676172612,
      "learning_rate": 7.325169038657261e-06,
      "loss": 2.1148,
      "step": 3623
    },
    {
      "epoch": 1.2351738241308794,
      "grad_norm": 15.49345597405138,
      "learning_rate": 7.323413250635441e-06,
      "loss": 2.6483,
      "step": 3624
    },
    {
      "epoch": 1.2355146557600545,
      "grad_norm": 17.65315424438424,
      "learning_rate": 7.321657097131472e-06,
      "loss": 2.1299,
      "step": 3625
    },
    {
      "epoch": 1.2358554873892298,
      "grad_norm": 17.51455354817194,
      "learning_rate": 7.319900578421597e-06,
      "loss": 2.6466,
      "step": 3626
    },
    {
      "epoch": 1.2361963190184049,
      "grad_norm": 19.88197306189632,
      "learning_rate": 7.318143694782127e-06,
      "loss": 1.8308,
      "step": 3627
    },
    {
      "epoch": 1.2365371506475802,
      "grad_norm": 16.856828636480945,
      "learning_rate": 7.316386446489425e-06,
      "loss": 2.04,
      "step": 3628
    },
    {
      "epoch": 1.2368779822767553,
      "grad_norm": 10.56997087841179,
      "learning_rate": 7.3146288338199145e-06,
      "loss": 2.2035,
      "step": 3629
    },
    {
      "epoch": 1.2372188139059306,
      "grad_norm": 16.899634874085223,
      "learning_rate": 7.312870857050074e-06,
      "loss": 2.4581,
      "step": 3630
    },
    {
      "epoch": 1.2375596455351057,
      "grad_norm": 13.313648804076688,
      "learning_rate": 7.311112516456443e-06,
      "loss": 1.9576,
      "step": 3631
    },
    {
      "epoch": 1.2379004771642808,
      "grad_norm": 17.75099127621394,
      "learning_rate": 7.309353812315609e-06,
      "loss": 2.0217,
      "step": 3632
    },
    {
      "epoch": 1.238241308793456,
      "grad_norm": 14.656985683641402,
      "learning_rate": 7.307594744904229e-06,
      "loss": 2.2107,
      "step": 3633
    },
    {
      "epoch": 1.2385821404226312,
      "grad_norm": 13.06548540717677,
      "learning_rate": 7.305835314499011e-06,
      "loss": 2.2674,
      "step": 3634
    },
    {
      "epoch": 1.2389229720518065,
      "grad_norm": 21.397860634460624,
      "learning_rate": 7.304075521376718e-06,
      "loss": 2.08,
      "step": 3635
    },
    {
      "epoch": 1.2392638036809815,
      "grad_norm": 12.566087373723544,
      "learning_rate": 7.302315365814173e-06,
      "loss": 1.7831,
      "step": 3636
    },
    {
      "epoch": 1.2396046353101569,
      "grad_norm": 33.1896651854774,
      "learning_rate": 7.300554848088257e-06,
      "loss": 2.5044,
      "step": 3637
    },
    {
      "epoch": 1.239945466939332,
      "grad_norm": 23.74768981833149,
      "learning_rate": 7.298793968475904e-06,
      "loss": 2.312,
      "step": 3638
    },
    {
      "epoch": 1.2402862985685073,
      "grad_norm": 15.188595188232458,
      "learning_rate": 7.297032727254112e-06,
      "loss": 2.6253,
      "step": 3639
    },
    {
      "epoch": 1.2406271301976823,
      "grad_norm": 14.39876554462012,
      "learning_rate": 7.295271124699924e-06,
      "loss": 2.0787,
      "step": 3640
    },
    {
      "epoch": 1.2409679618268576,
      "grad_norm": 16.287596516751098,
      "learning_rate": 7.293509161090453e-06,
      "loss": 2.0401,
      "step": 3641
    },
    {
      "epoch": 1.2413087934560327,
      "grad_norm": 19.446975964759353,
      "learning_rate": 7.291746836702862e-06,
      "loss": 2.4434,
      "step": 3642
    },
    {
      "epoch": 1.2416496250852078,
      "grad_norm": 16.77775285898734,
      "learning_rate": 7.289984151814369e-06,
      "loss": 2.8362,
      "step": 3643
    },
    {
      "epoch": 1.2419904567143831,
      "grad_norm": 16.62440181717258,
      "learning_rate": 7.288221106702253e-06,
      "loss": 2.6221,
      "step": 3644
    },
    {
      "epoch": 1.2423312883435582,
      "grad_norm": 17.34875144230253,
      "learning_rate": 7.2864577016438494e-06,
      "loss": 1.6827,
      "step": 3645
    },
    {
      "epoch": 1.2426721199727335,
      "grad_norm": 16.193107764629545,
      "learning_rate": 7.284693936916547e-06,
      "loss": 2.0995,
      "step": 3646
    },
    {
      "epoch": 1.2430129516019086,
      "grad_norm": 17.238087072530522,
      "learning_rate": 7.2829298127977936e-06,
      "loss": 1.7588,
      "step": 3647
    },
    {
      "epoch": 1.243353783231084,
      "grad_norm": 11.847236587834873,
      "learning_rate": 7.281165329565093e-06,
      "loss": 2.1484,
      "step": 3648
    },
    {
      "epoch": 1.243694614860259,
      "grad_norm": 22.27607964369925,
      "learning_rate": 7.279400487496004e-06,
      "loss": 2.504,
      "step": 3649
    },
    {
      "epoch": 1.2440354464894343,
      "grad_norm": 24.64762202265775,
      "learning_rate": 7.277635286868145e-06,
      "loss": 2.4988,
      "step": 3650
    },
    {
      "epoch": 1.2443762781186094,
      "grad_norm": 18.37783823701687,
      "learning_rate": 7.2758697279591895e-06,
      "loss": 2.0012,
      "step": 3651
    },
    {
      "epoch": 1.2447171097477847,
      "grad_norm": 13.239947199227654,
      "learning_rate": 7.274103811046865e-06,
      "loss": 2.3486,
      "step": 3652
    },
    {
      "epoch": 1.2450579413769598,
      "grad_norm": 22.847529348998265,
      "learning_rate": 7.272337536408959e-06,
      "loss": 2.4851,
      "step": 3653
    },
    {
      "epoch": 1.2453987730061349,
      "grad_norm": 13.942043654602312,
      "learning_rate": 7.270570904323313e-06,
      "loss": 2.564,
      "step": 3654
    },
    {
      "epoch": 1.2457396046353102,
      "grad_norm": 20.820386011217337,
      "learning_rate": 7.268803915067825e-06,
      "loss": 2.6842,
      "step": 3655
    },
    {
      "epoch": 1.2460804362644853,
      "grad_norm": 24.097777647772805,
      "learning_rate": 7.267036568920449e-06,
      "loss": 1.983,
      "step": 3656
    },
    {
      "epoch": 1.2464212678936606,
      "grad_norm": 13.68577090247775,
      "learning_rate": 7.2652688661591965e-06,
      "loss": 1.5271,
      "step": 3657
    },
    {
      "epoch": 1.2467620995228357,
      "grad_norm": 15.951467681372325,
      "learning_rate": 7.263500807062134e-06,
      "loss": 2.3209,
      "step": 3658
    },
    {
      "epoch": 1.247102931152011,
      "grad_norm": 20.389797762689586,
      "learning_rate": 7.261732391907385e-06,
      "loss": 2.1197,
      "step": 3659
    },
    {
      "epoch": 1.247443762781186,
      "grad_norm": 18.989525721505032,
      "learning_rate": 7.259963620973125e-06,
      "loss": 2.4304,
      "step": 3660
    },
    {
      "epoch": 1.2477845944103612,
      "grad_norm": 21.74870228476153,
      "learning_rate": 7.258194494537592e-06,
      "loss": 2.0293,
      "step": 3661
    },
    {
      "epoch": 1.2481254260395365,
      "grad_norm": 21.422464289628643,
      "learning_rate": 7.256425012879077e-06,
      "loss": 2.0886,
      "step": 3662
    },
    {
      "epoch": 1.2484662576687118,
      "grad_norm": 20.153990719533265,
      "learning_rate": 7.254655176275925e-06,
      "loss": 2.0151,
      "step": 3663
    },
    {
      "epoch": 1.2488070892978869,
      "grad_norm": 19.83330937573755,
      "learning_rate": 7.252884985006536e-06,
      "loss": 2.0695,
      "step": 3664
    },
    {
      "epoch": 1.249147920927062,
      "grad_norm": 16.964472573661798,
      "learning_rate": 7.251114439349373e-06,
      "loss": 2.3528,
      "step": 3665
    },
    {
      "epoch": 1.2494887525562373,
      "grad_norm": 23.070960293333485,
      "learning_rate": 7.249343539582946e-06,
      "loss": 2.3885,
      "step": 3666
    },
    {
      "epoch": 1.2498295841854123,
      "grad_norm": 24.82480837973142,
      "learning_rate": 7.247572285985828e-06,
      "loss": 2.2924,
      "step": 3667
    },
    {
      "epoch": 1.2501704158145877,
      "grad_norm": 17.172035190634162,
      "learning_rate": 7.245800678836641e-06,
      "loss": 2.1755,
      "step": 3668
    },
    {
      "epoch": 1.2505112474437627,
      "grad_norm": 16.098259533528285,
      "learning_rate": 7.244028718414068e-06,
      "loss": 2.1169,
      "step": 3669
    },
    {
      "epoch": 1.250852079072938,
      "grad_norm": 15.1495515853279,
      "learning_rate": 7.242256404996843e-06,
      "loss": 2.8136,
      "step": 3670
    },
    {
      "epoch": 1.2511929107021131,
      "grad_norm": 15.40782236546987,
      "learning_rate": 7.2404837388637615e-06,
      "loss": 2.0665,
      "step": 3671
    },
    {
      "epoch": 1.2515337423312882,
      "grad_norm": 18.14346338438745,
      "learning_rate": 7.238710720293668e-06,
      "loss": 2.0689,
      "step": 3672
    },
    {
      "epoch": 1.2518745739604635,
      "grad_norm": 18.323794991317495,
      "learning_rate": 7.236937349565467e-06,
      "loss": 2.0104,
      "step": 3673
    },
    {
      "epoch": 1.2522154055896388,
      "grad_norm": 15.692228195390742,
      "learning_rate": 7.235163626958115e-06,
      "loss": 2.211,
      "step": 3674
    },
    {
      "epoch": 1.252556237218814,
      "grad_norm": 17.252851080542197,
      "learning_rate": 7.233389552750627e-06,
      "loss": 2.4453,
      "step": 3675
    },
    {
      "epoch": 1.252897068847989,
      "grad_norm": 14.00637277882569,
      "learning_rate": 7.231615127222073e-06,
      "loss": 2.206,
      "step": 3676
    },
    {
      "epoch": 1.2532379004771643,
      "grad_norm": 16.014570837461253,
      "learning_rate": 7.229840350651576e-06,
      "loss": 2.0001,
      "step": 3677
    },
    {
      "epoch": 1.2535787321063394,
      "grad_norm": 14.258182110252342,
      "learning_rate": 7.228065223318315e-06,
      "loss": 2.4256,
      "step": 3678
    },
    {
      "epoch": 1.2539195637355147,
      "grad_norm": 13.687875728895541,
      "learning_rate": 7.226289745501525e-06,
      "loss": 2.4321,
      "step": 3679
    },
    {
      "epoch": 1.2542603953646898,
      "grad_norm": 16.403374397464667,
      "learning_rate": 7.224513917480494e-06,
      "loss": 2.0139,
      "step": 3680
    },
    {
      "epoch": 1.2546012269938651,
      "grad_norm": 14.720103557377202,
      "learning_rate": 7.22273773953457e-06,
      "loss": 2.3864,
      "step": 3681
    },
    {
      "epoch": 1.2549420586230402,
      "grad_norm": 38.54492462279358,
      "learning_rate": 7.220961211943152e-06,
      "loss": 2.2495,
      "step": 3682
    },
    {
      "epoch": 1.2552828902522153,
      "grad_norm": 23.97868581794325,
      "learning_rate": 7.219184334985693e-06,
      "loss": 2.5508,
      "step": 3683
    },
    {
      "epoch": 1.2556237218813906,
      "grad_norm": 22.859679771810203,
      "learning_rate": 7.2174071089417055e-06,
      "loss": 1.872,
      "step": 3684
    },
    {
      "epoch": 1.255964553510566,
      "grad_norm": 20.794676308730185,
      "learning_rate": 7.215629534090751e-06,
      "loss": 2.7151,
      "step": 3685
    },
    {
      "epoch": 1.256305385139741,
      "grad_norm": 20.86996989047925,
      "learning_rate": 7.213851610712451e-06,
      "loss": 2.4518,
      "step": 3686
    },
    {
      "epoch": 1.256646216768916,
      "grad_norm": 16.44249677514919,
      "learning_rate": 7.21207333908648e-06,
      "loss": 2.6101,
      "step": 3687
    },
    {
      "epoch": 1.2569870483980914,
      "grad_norm": 14.76804563925919,
      "learning_rate": 7.210294719492568e-06,
      "loss": 1.9824,
      "step": 3688
    },
    {
      "epoch": 1.2573278800272665,
      "grad_norm": 17.93763763390976,
      "learning_rate": 7.208515752210498e-06,
      "loss": 2.5755,
      "step": 3689
    },
    {
      "epoch": 1.2576687116564418,
      "grad_norm": 11.328468724357327,
      "learning_rate": 7.206736437520108e-06,
      "loss": 2.2547,
      "step": 3690
    },
    {
      "epoch": 1.2580095432856169,
      "grad_norm": 15.539394648004325,
      "learning_rate": 7.204956775701291e-06,
      "loss": 2.3193,
      "step": 3691
    },
    {
      "epoch": 1.2583503749147922,
      "grad_norm": 14.197250277951442,
      "learning_rate": 7.2031767670339965e-06,
      "loss": 2.8538,
      "step": 3692
    },
    {
      "epoch": 1.2586912065439673,
      "grad_norm": 22.848376891239475,
      "learning_rate": 7.201396411798227e-06,
      "loss": 1.9927,
      "step": 3693
    },
    {
      "epoch": 1.2590320381731424,
      "grad_norm": 13.51183605747328,
      "learning_rate": 7.199615710274038e-06,
      "loss": 2.237,
      "step": 3694
    },
    {
      "epoch": 1.2593728698023177,
      "grad_norm": 16.20305595763514,
      "learning_rate": 7.197834662741541e-06,
      "loss": 2.4897,
      "step": 3695
    },
    {
      "epoch": 1.259713701431493,
      "grad_norm": 11.322045556683854,
      "learning_rate": 7.196053269480904e-06,
      "loss": 1.6763,
      "step": 3696
    },
    {
      "epoch": 1.260054533060668,
      "grad_norm": 13.643094697521581,
      "learning_rate": 7.194271530772345e-06,
      "loss": 2.0263,
      "step": 3697
    },
    {
      "epoch": 1.2603953646898431,
      "grad_norm": 17.3611405173959,
      "learning_rate": 7.192489446896138e-06,
      "loss": 2.0476,
      "step": 3698
    },
    {
      "epoch": 1.2607361963190185,
      "grad_norm": 20.94899664277064,
      "learning_rate": 7.190707018132614e-06,
      "loss": 2.9461,
      "step": 3699
    },
    {
      "epoch": 1.2610770279481935,
      "grad_norm": 15.575273299437612,
      "learning_rate": 7.188924244762155e-06,
      "loss": 2.35,
      "step": 3700
    },
    {
      "epoch": 1.2614178595773688,
      "grad_norm": 25.72149759538044,
      "learning_rate": 7.187141127065201e-06,
      "loss": 1.7777,
      "step": 3701
    },
    {
      "epoch": 1.261758691206544,
      "grad_norm": 17.950575339975554,
      "learning_rate": 7.185357665322239e-06,
      "loss": 1.8057,
      "step": 3702
    },
    {
      "epoch": 1.2620995228357192,
      "grad_norm": 13.587031926458616,
      "learning_rate": 7.183573859813817e-06,
      "loss": 1.9299,
      "step": 3703
    },
    {
      "epoch": 1.2624403544648943,
      "grad_norm": 13.791604680695215,
      "learning_rate": 7.181789710820536e-06,
      "loss": 2.3635,
      "step": 3704
    },
    {
      "epoch": 1.2627811860940694,
      "grad_norm": 16.140014145535208,
      "learning_rate": 7.180005218623048e-06,
      "loss": 1.7355,
      "step": 3705
    },
    {
      "epoch": 1.2631220177232447,
      "grad_norm": 15.963108292827416,
      "learning_rate": 7.178220383502059e-06,
      "loss": 2.1832,
      "step": 3706
    },
    {
      "epoch": 1.26346284935242,
      "grad_norm": 20.994247530587163,
      "learning_rate": 7.176435205738335e-06,
      "loss": 2.4604,
      "step": 3707
    },
    {
      "epoch": 1.2638036809815951,
      "grad_norm": 15.946137899438389,
      "learning_rate": 7.1746496856126865e-06,
      "loss": 2.2899,
      "step": 3708
    },
    {
      "epoch": 1.2641445126107702,
      "grad_norm": 13.64137519680237,
      "learning_rate": 7.172863823405988e-06,
      "loss": 2.5356,
      "step": 3709
    },
    {
      "epoch": 1.2644853442399455,
      "grad_norm": 17.51475938214446,
      "learning_rate": 7.17107761939916e-06,
      "loss": 2.0512,
      "step": 3710
    },
    {
      "epoch": 1.2648261758691206,
      "grad_norm": 17.512868223116282,
      "learning_rate": 7.169291073873179e-06,
      "loss": 2.1393,
      "step": 3711
    },
    {
      "epoch": 1.265167007498296,
      "grad_norm": 17.060912144559335,
      "learning_rate": 7.167504187109076e-06,
      "loss": 1.6684,
      "step": 3712
    },
    {
      "epoch": 1.265507839127471,
      "grad_norm": 26.214947038258945,
      "learning_rate": 7.165716959387939e-06,
      "loss": 2.347,
      "step": 3713
    },
    {
      "epoch": 1.2658486707566463,
      "grad_norm": 13.966110211556815,
      "learning_rate": 7.163929390990902e-06,
      "loss": 2.0988,
      "step": 3714
    },
    {
      "epoch": 1.2661895023858214,
      "grad_norm": 18.518786001706943,
      "learning_rate": 7.162141482199158e-06,
      "loss": 2.253,
      "step": 3715
    },
    {
      "epoch": 1.2665303340149965,
      "grad_norm": 12.1668961272498,
      "learning_rate": 7.1603532332939485e-06,
      "loss": 2.1297,
      "step": 3716
    },
    {
      "epoch": 1.2668711656441718,
      "grad_norm": 14.788341995781185,
      "learning_rate": 7.158564644556578e-06,
      "loss": 2.145,
      "step": 3717
    },
    {
      "epoch": 1.267211997273347,
      "grad_norm": 32.02556275925098,
      "learning_rate": 7.156775716268398e-06,
      "loss": 2.4607,
      "step": 3718
    },
    {
      "epoch": 1.2675528289025222,
      "grad_norm": 16.166922398182674,
      "learning_rate": 7.15498644871081e-06,
      "loss": 2.4897,
      "step": 3719
    },
    {
      "epoch": 1.2678936605316973,
      "grad_norm": 21.445102445737547,
      "learning_rate": 7.153196842165275e-06,
      "loss": 2.2605,
      "step": 3720
    },
    {
      "epoch": 1.2682344921608726,
      "grad_norm": 16.164061186165192,
      "learning_rate": 7.151406896913305e-06,
      "loss": 1.9331,
      "step": 3721
    },
    {
      "epoch": 1.2685753237900477,
      "grad_norm": 31.12364610588024,
      "learning_rate": 7.149616613236467e-06,
      "loss": 2.1843,
      "step": 3722
    },
    {
      "epoch": 1.268916155419223,
      "grad_norm": 22.818928331955632,
      "learning_rate": 7.147825991416378e-06,
      "loss": 1.4311,
      "step": 3723
    },
    {
      "epoch": 1.269256987048398,
      "grad_norm": 17.508669148512013,
      "learning_rate": 7.146035031734712e-06,
      "loss": 2.0645,
      "step": 3724
    },
    {
      "epoch": 1.2695978186775734,
      "grad_norm": 19.474090759446494,
      "learning_rate": 7.1442437344731905e-06,
      "loss": 2.2306,
      "step": 3725
    },
    {
      "epoch": 1.2699386503067485,
      "grad_norm": 18.269335355242923,
      "learning_rate": 7.142452099913595e-06,
      "loss": 2.0307,
      "step": 3726
    },
    {
      "epoch": 1.2702794819359235,
      "grad_norm": 22.22154946992752,
      "learning_rate": 7.140660128337756e-06,
      "loss": 2.344,
      "step": 3727
    },
    {
      "epoch": 1.2706203135650989,
      "grad_norm": 15.99404764224144,
      "learning_rate": 7.138867820027557e-06,
      "loss": 2.1198,
      "step": 3728
    },
    {
      "epoch": 1.2709611451942742,
      "grad_norm": 22.541173080196703,
      "learning_rate": 7.137075175264933e-06,
      "loss": 2.2798,
      "step": 3729
    },
    {
      "epoch": 1.2713019768234493,
      "grad_norm": 13.392917296373598,
      "learning_rate": 7.135282194331881e-06,
      "loss": 2.4277,
      "step": 3730
    },
    {
      "epoch": 1.2716428084526243,
      "grad_norm": 15.705109751428056,
      "learning_rate": 7.133488877510437e-06,
      "loss": 1.9805,
      "step": 3731
    },
    {
      "epoch": 1.2719836400817996,
      "grad_norm": 14.690096405002484,
      "learning_rate": 7.131695225082702e-06,
      "loss": 2.4047,
      "step": 3732
    },
    {
      "epoch": 1.2723244717109747,
      "grad_norm": 13.724640227952932,
      "learning_rate": 7.129901237330818e-06,
      "loss": 2.164,
      "step": 3733
    },
    {
      "epoch": 1.27266530334015,
      "grad_norm": 10.34029065124664,
      "learning_rate": 7.128106914536994e-06,
      "loss": 1.9271,
      "step": 3734
    },
    {
      "epoch": 1.2730061349693251,
      "grad_norm": 18.06973657312537,
      "learning_rate": 7.126312256983482e-06,
      "loss": 2.1255,
      "step": 3735
    },
    {
      "epoch": 1.2733469665985004,
      "grad_norm": 13.892712278341005,
      "learning_rate": 7.124517264952585e-06,
      "loss": 2.1176,
      "step": 3736
    },
    {
      "epoch": 1.2736877982276755,
      "grad_norm": 44.355496836618784,
      "learning_rate": 7.122721938726666e-06,
      "loss": 2.5669,
      "step": 3737
    },
    {
      "epoch": 1.2740286298568506,
      "grad_norm": 10.753398221079161,
      "learning_rate": 7.120926278588136e-06,
      "loss": 2.2778,
      "step": 3738
    },
    {
      "epoch": 1.274369461486026,
      "grad_norm": 16.288167855662188,
      "learning_rate": 7.11913028481946e-06,
      "loss": 2.3772,
      "step": 3739
    },
    {
      "epoch": 1.274710293115201,
      "grad_norm": 16.34578825036036,
      "learning_rate": 7.117333957703153e-06,
      "loss": 2.3134,
      "step": 3740
    },
    {
      "epoch": 1.2750511247443763,
      "grad_norm": 18.478408938875653,
      "learning_rate": 7.115537297521786e-06,
      "loss": 2.0647,
      "step": 3741
    },
    {
      "epoch": 1.2753919563735514,
      "grad_norm": 17.586263583059125,
      "learning_rate": 7.113740304557981e-06,
      "loss": 2.1819,
      "step": 3742
    },
    {
      "epoch": 1.2757327880027267,
      "grad_norm": 16.996301232399176,
      "learning_rate": 7.111942979094413e-06,
      "loss": 1.7532,
      "step": 3743
    },
    {
      "epoch": 1.2760736196319018,
      "grad_norm": 13.833686982485709,
      "learning_rate": 7.110145321413806e-06,
      "loss": 2.2939,
      "step": 3744
    },
    {
      "epoch": 1.276414451261077,
      "grad_norm": 28.710953585619997,
      "learning_rate": 7.10834733179894e-06,
      "loss": 2.4974,
      "step": 3745
    },
    {
      "epoch": 1.2767552828902522,
      "grad_norm": 16.178509682054422,
      "learning_rate": 7.106549010532645e-06,
      "loss": 2.7058,
      "step": 3746
    },
    {
      "epoch": 1.2770961145194275,
      "grad_norm": 17.450003349016598,
      "learning_rate": 7.104750357897807e-06,
      "loss": 1.8309,
      "step": 3747
    },
    {
      "epoch": 1.2774369461486026,
      "grad_norm": 16.863796153632872,
      "learning_rate": 7.102951374177358e-06,
      "loss": 2.4437,
      "step": 3748
    },
    {
      "epoch": 1.2777777777777777,
      "grad_norm": 17.874556975013288,
      "learning_rate": 7.101152059654289e-06,
      "loss": 2.8323,
      "step": 3749
    },
    {
      "epoch": 1.278118609406953,
      "grad_norm": 17.267761995059164,
      "learning_rate": 7.099352414611632e-06,
      "loss": 2.3078,
      "step": 3750
    },
    {
      "epoch": 1.278459441036128,
      "grad_norm": 19.40396564310511,
      "learning_rate": 7.097552439332485e-06,
      "loss": 2.5031,
      "step": 3751
    },
    {
      "epoch": 1.2788002726653034,
      "grad_norm": 13.082954656403249,
      "learning_rate": 7.0957521340999915e-06,
      "loss": 2.6615,
      "step": 3752
    },
    {
      "epoch": 1.2791411042944785,
      "grad_norm": 14.336417135367721,
      "learning_rate": 7.093951499197343e-06,
      "loss": 2.147,
      "step": 3753
    },
    {
      "epoch": 1.2794819359236538,
      "grad_norm": 16.58989622862361,
      "learning_rate": 7.092150534907788e-06,
      "loss": 2.4459,
      "step": 3754
    },
    {
      "epoch": 1.2798227675528289,
      "grad_norm": 49.52086014494663,
      "learning_rate": 7.090349241514626e-06,
      "loss": 2.8512,
      "step": 3755
    },
    {
      "epoch": 1.280163599182004,
      "grad_norm": 14.29664636275805,
      "learning_rate": 7.088547619301207e-06,
      "loss": 1.4556,
      "step": 3756
    },
    {
      "epoch": 1.2805044308111793,
      "grad_norm": 23.579285432048326,
      "learning_rate": 7.086745668550933e-06,
      "loss": 2.2616,
      "step": 3757
    },
    {
      "epoch": 1.2808452624403546,
      "grad_norm": 20.192980823659056,
      "learning_rate": 7.084943389547259e-06,
      "loss": 2.3728,
      "step": 3758
    },
    {
      "epoch": 1.2811860940695297,
      "grad_norm": 17.07722230485793,
      "learning_rate": 7.083140782573689e-06,
      "loss": 2.7283,
      "step": 3759
    },
    {
      "epoch": 1.2815269256987047,
      "grad_norm": 14.544329762573188,
      "learning_rate": 7.0813378479137826e-06,
      "loss": 2.1605,
      "step": 3760
    },
    {
      "epoch": 1.28186775732788,
      "grad_norm": 18.985501763131865,
      "learning_rate": 7.079534585851148e-06,
      "loss": 2.299,
      "step": 3761
    },
    {
      "epoch": 1.2822085889570551,
      "grad_norm": 16.11022061984173,
      "learning_rate": 7.077730996669444e-06,
      "loss": 2.3366,
      "step": 3762
    },
    {
      "epoch": 1.2825494205862304,
      "grad_norm": 16.390645666488037,
      "learning_rate": 7.0759270806523825e-06,
      "loss": 2.6201,
      "step": 3763
    },
    {
      "epoch": 1.2828902522154055,
      "grad_norm": 16.988516140041288,
      "learning_rate": 7.074122838083729e-06,
      "loss": 2.2052,
      "step": 3764
    },
    {
      "epoch": 1.2832310838445808,
      "grad_norm": 24.14292466984292,
      "learning_rate": 7.072318269247296e-06,
      "loss": 1.7218,
      "step": 3765
    },
    {
      "epoch": 1.283571915473756,
      "grad_norm": 26.556960851123417,
      "learning_rate": 7.0705133744269486e-06,
      "loss": 2.1515,
      "step": 3766
    },
    {
      "epoch": 1.283912747102931,
      "grad_norm": 33.996840165558986,
      "learning_rate": 7.068708153906606e-06,
      "loss": 2.2266,
      "step": 3767
    },
    {
      "epoch": 1.2842535787321063,
      "grad_norm": 15.01834847954008,
      "learning_rate": 7.066902607970236e-06,
      "loss": 2.2618,
      "step": 3768
    },
    {
      "epoch": 1.2845944103612816,
      "grad_norm": 25.759240342123856,
      "learning_rate": 7.065096736901858e-06,
      "loss": 2.0937,
      "step": 3769
    },
    {
      "epoch": 1.2849352419904567,
      "grad_norm": 22.256072098184465,
      "learning_rate": 7.063290540985542e-06,
      "loss": 2.4018,
      "step": 3770
    },
    {
      "epoch": 1.2852760736196318,
      "grad_norm": 15.752480682852333,
      "learning_rate": 7.06148402050541e-06,
      "loss": 2.4036,
      "step": 3771
    },
    {
      "epoch": 1.2856169052488071,
      "grad_norm": 20.17090155966045,
      "learning_rate": 7.0596771757456374e-06,
      "loss": 2.2025,
      "step": 3772
    },
    {
      "epoch": 1.2859577368779822,
      "grad_norm": 15.399228724624912,
      "learning_rate": 7.057870006990444e-06,
      "loss": 2.2929,
      "step": 3773
    },
    {
      "epoch": 1.2862985685071575,
      "grad_norm": 15.270486766229137,
      "learning_rate": 7.056062514524107e-06,
      "loss": 2.4827,
      "step": 3774
    },
    {
      "epoch": 1.2866394001363326,
      "grad_norm": 21.082439953456948,
      "learning_rate": 7.0542546986309525e-06,
      "loss": 2.5273,
      "step": 3775
    },
    {
      "epoch": 1.286980231765508,
      "grad_norm": 16.0760950361454,
      "learning_rate": 7.052446559595355e-06,
      "loss": 2.4812,
      "step": 3776
    },
    {
      "epoch": 1.287321063394683,
      "grad_norm": 27.059540784712492,
      "learning_rate": 7.050638097701744e-06,
      "loss": 2.3431,
      "step": 3777
    },
    {
      "epoch": 1.287661895023858,
      "grad_norm": 15.321601262344549,
      "learning_rate": 7.048829313234599e-06,
      "loss": 2.1457,
      "step": 3778
    },
    {
      "epoch": 1.2880027266530334,
      "grad_norm": 22.260117436393212,
      "learning_rate": 7.047020206478445e-06,
      "loss": 2.1484,
      "step": 3779
    },
    {
      "epoch": 1.2883435582822087,
      "grad_norm": 14.19641294874661,
      "learning_rate": 7.045210777717863e-06,
      "loss": 2.0489,
      "step": 3780
    },
    {
      "epoch": 1.2886843899113838,
      "grad_norm": 23.004512751203315,
      "learning_rate": 7.043401027237487e-06,
      "loss": 2.8292,
      "step": 3781
    },
    {
      "epoch": 1.2890252215405589,
      "grad_norm": 13.607049189246933,
      "learning_rate": 7.041590955321993e-06,
      "loss": 2.8227,
      "step": 3782
    },
    {
      "epoch": 1.2893660531697342,
      "grad_norm": 19.5666032634259,
      "learning_rate": 7.039780562256114e-06,
      "loss": 2.7048,
      "step": 3783
    },
    {
      "epoch": 1.2897068847989093,
      "grad_norm": 17.145176813623035,
      "learning_rate": 7.037969848324634e-06,
      "loss": 2.028,
      "step": 3784
    },
    {
      "epoch": 1.2900477164280846,
      "grad_norm": 10.761088024799015,
      "learning_rate": 7.0361588138123834e-06,
      "loss": 1.662,
      "step": 3785
    },
    {
      "epoch": 1.2903885480572597,
      "grad_norm": 11.002414673730952,
      "learning_rate": 7.034347459004248e-06,
      "loss": 2.0564,
      "step": 3786
    },
    {
      "epoch": 1.290729379686435,
      "grad_norm": 12.746279826635107,
      "learning_rate": 7.032535784185157e-06,
      "loss": 2.2679,
      "step": 3787
    },
    {
      "epoch": 1.29107021131561,
      "grad_norm": 21.962513130618053,
      "learning_rate": 7.030723789640096e-06,
      "loss": 2.0115,
      "step": 3788
    },
    {
      "epoch": 1.2914110429447851,
      "grad_norm": 20.403204844049817,
      "learning_rate": 7.0289114756541e-06,
      "loss": 2.211,
      "step": 3789
    },
    {
      "epoch": 1.2917518745739605,
      "grad_norm": 19.27932942532655,
      "learning_rate": 7.027098842512252e-06,
      "loss": 1.8851,
      "step": 3790
    },
    {
      "epoch": 1.2920927062031358,
      "grad_norm": 13.483887844269695,
      "learning_rate": 7.0252858904996865e-06,
      "loss": 2.0915,
      "step": 3791
    },
    {
      "epoch": 1.2924335378323109,
      "grad_norm": 20.544548244301815,
      "learning_rate": 7.023472619901588e-06,
      "loss": 1.6905,
      "step": 3792
    },
    {
      "epoch": 1.292774369461486,
      "grad_norm": 15.79545120136861,
      "learning_rate": 7.021659031003192e-06,
      "loss": 1.7279,
      "step": 3793
    },
    {
      "epoch": 1.2931152010906612,
      "grad_norm": 11.666670389612552,
      "learning_rate": 7.019845124089781e-06,
      "loss": 2.1689,
      "step": 3794
    },
    {
      "epoch": 1.2934560327198363,
      "grad_norm": 15.20685481589324,
      "learning_rate": 7.018030899446695e-06,
      "loss": 2.6791,
      "step": 3795
    },
    {
      "epoch": 1.2937968643490116,
      "grad_norm": 20.214502943483648,
      "learning_rate": 7.016216357359312e-06,
      "loss": 2.5583,
      "step": 3796
    },
    {
      "epoch": 1.2941376959781867,
      "grad_norm": 16.28502856956678,
      "learning_rate": 7.014401498113071e-06,
      "loss": 1.9494,
      "step": 3797
    },
    {
      "epoch": 1.294478527607362,
      "grad_norm": 15.76952442977511,
      "learning_rate": 7.0125863219934555e-06,
      "loss": 1.9297,
      "step": 3798
    },
    {
      "epoch": 1.2948193592365371,
      "grad_norm": 14.133704175332756,
      "learning_rate": 7.010770829285999e-06,
      "loss": 2.3135,
      "step": 3799
    },
    {
      "epoch": 1.2951601908657122,
      "grad_norm": 18.41931850714591,
      "learning_rate": 7.008955020276287e-06,
      "loss": 2.2711,
      "step": 3800
    },
    {
      "epoch": 1.2955010224948875,
      "grad_norm": 23.706394855297354,
      "learning_rate": 7.007138895249954e-06,
      "loss": 1.6989,
      "step": 3801
    },
    {
      "epoch": 1.2958418541240628,
      "grad_norm": 13.719301313900244,
      "learning_rate": 7.005322454492683e-06,
      "loss": 2.5631,
      "step": 3802
    },
    {
      "epoch": 1.296182685753238,
      "grad_norm": 13.736088366879637,
      "learning_rate": 7.003505698290206e-06,
      "loss": 2.1676,
      "step": 3803
    },
    {
      "epoch": 1.296523517382413,
      "grad_norm": 16.18381550836006,
      "learning_rate": 7.001688626928308e-06,
      "loss": 2.7626,
      "step": 3804
    },
    {
      "epoch": 1.2968643490115883,
      "grad_norm": 17.88112989367456,
      "learning_rate": 6.999871240692819e-06,
      "loss": 2.2959,
      "step": 3805
    },
    {
      "epoch": 1.2972051806407634,
      "grad_norm": 17.943608471140614,
      "learning_rate": 6.998053539869625e-06,
      "loss": 2.6957,
      "step": 3806
    },
    {
      "epoch": 1.2975460122699387,
      "grad_norm": 15.176219953897414,
      "learning_rate": 6.996235524744654e-06,
      "loss": 2.0043,
      "step": 3807
    },
    {
      "epoch": 1.2978868438991138,
      "grad_norm": 17.00877462941904,
      "learning_rate": 6.99441719560389e-06,
      "loss": 2.4033,
      "step": 3808
    },
    {
      "epoch": 1.298227675528289,
      "grad_norm": 13.562789337247633,
      "learning_rate": 6.992598552733361e-06,
      "loss": 2.1669,
      "step": 3809
    },
    {
      "epoch": 1.2985685071574642,
      "grad_norm": 17.701036969394373,
      "learning_rate": 6.990779596419146e-06,
      "loss": 2.4451,
      "step": 3810
    },
    {
      "epoch": 1.2989093387866393,
      "grad_norm": 31.158446010449282,
      "learning_rate": 6.988960326947377e-06,
      "loss": 1.8016,
      "step": 3811
    },
    {
      "epoch": 1.2992501704158146,
      "grad_norm": 18.069948939811734,
      "learning_rate": 6.987140744604233e-06,
      "loss": 2.4876,
      "step": 3812
    },
    {
      "epoch": 1.29959100204499,
      "grad_norm": 13.331018302460079,
      "learning_rate": 6.985320849675936e-06,
      "loss": 2.2038,
      "step": 3813
    },
    {
      "epoch": 1.299931833674165,
      "grad_norm": 23.814632441704408,
      "learning_rate": 6.983500642448768e-06,
      "loss": 2.5977,
      "step": 3814
    },
    {
      "epoch": 1.30027266530334,
      "grad_norm": 14.327957420922402,
      "learning_rate": 6.981680123209053e-06,
      "loss": 2.3641,
      "step": 3815
    },
    {
      "epoch": 1.3006134969325154,
      "grad_norm": 21.71344146825744,
      "learning_rate": 6.979859292243166e-06,
      "loss": 2.5209,
      "step": 3816
    },
    {
      "epoch": 1.3009543285616905,
      "grad_norm": 11.235556770709811,
      "learning_rate": 6.97803814983753e-06,
      "loss": 2.1399,
      "step": 3817
    },
    {
      "epoch": 1.3012951601908658,
      "grad_norm": 15.765563185230238,
      "learning_rate": 6.97621669627862e-06,
      "loss": 2.1947,
      "step": 3818
    },
    {
      "epoch": 1.3016359918200409,
      "grad_norm": 19.802488915710846,
      "learning_rate": 6.974394931852957e-06,
      "loss": 2.031,
      "step": 3819
    },
    {
      "epoch": 1.3019768234492162,
      "grad_norm": 16.113011389938585,
      "learning_rate": 6.9725728568471115e-06,
      "loss": 2.2338,
      "step": 3820
    },
    {
      "epoch": 1.3023176550783913,
      "grad_norm": 21.525374451225744,
      "learning_rate": 6.970750471547702e-06,
      "loss": 2.7755,
      "step": 3821
    },
    {
      "epoch": 1.3026584867075663,
      "grad_norm": 17.266303511026507,
      "learning_rate": 6.968927776241399e-06,
      "loss": 1.5499,
      "step": 3822
    },
    {
      "epoch": 1.3029993183367417,
      "grad_norm": 22.38979318558162,
      "learning_rate": 6.96710477121492e-06,
      "loss": 2.2827,
      "step": 3823
    },
    {
      "epoch": 1.303340149965917,
      "grad_norm": 12.726958361894487,
      "learning_rate": 6.965281456755029e-06,
      "loss": 2.7568,
      "step": 3824
    },
    {
      "epoch": 1.303680981595092,
      "grad_norm": 14.238078979057878,
      "learning_rate": 6.963457833148542e-06,
      "loss": 2.2191,
      "step": 3825
    },
    {
      "epoch": 1.3040218132242671,
      "grad_norm": 14.853074182453383,
      "learning_rate": 6.961633900682321e-06,
      "loss": 2.6203,
      "step": 3826
    },
    {
      "epoch": 1.3043626448534424,
      "grad_norm": 19.79831389407027,
      "learning_rate": 6.959809659643279e-06,
      "loss": 2.452,
      "step": 3827
    },
    {
      "epoch": 1.3047034764826175,
      "grad_norm": 14.531131930603108,
      "learning_rate": 6.957985110318375e-06,
      "loss": 2.3774,
      "step": 3828
    },
    {
      "epoch": 1.3050443081117928,
      "grad_norm": 22.75988063470874,
      "learning_rate": 6.956160252994621e-06,
      "loss": 2.6234,
      "step": 3829
    },
    {
      "epoch": 1.305385139740968,
      "grad_norm": 14.721229115063348,
      "learning_rate": 6.9543350879590695e-06,
      "loss": 2.0447,
      "step": 3830
    },
    {
      "epoch": 1.3057259713701432,
      "grad_norm": 17.766640867718724,
      "learning_rate": 6.952509615498831e-06,
      "loss": 1.9995,
      "step": 3831
    },
    {
      "epoch": 1.3060668029993183,
      "grad_norm": 41.907392634743424,
      "learning_rate": 6.950683835901058e-06,
      "loss": 3.4796,
      "step": 3832
    },
    {
      "epoch": 1.3064076346284934,
      "grad_norm": 18.76698499741493,
      "learning_rate": 6.9488577494529505e-06,
      "loss": 1.9274,
      "step": 3833
    },
    {
      "epoch": 1.3067484662576687,
      "grad_norm": 15.920612965141467,
      "learning_rate": 6.947031356441761e-06,
      "loss": 2.348,
      "step": 3834
    },
    {
      "epoch": 1.307089297886844,
      "grad_norm": 11.264950480488507,
      "learning_rate": 6.94520465715479e-06,
      "loss": 1.8341,
      "step": 3835
    },
    {
      "epoch": 1.3074301295160191,
      "grad_norm": 21.89454174662031,
      "learning_rate": 6.943377651879382e-06,
      "loss": 2.454,
      "step": 3836
    },
    {
      "epoch": 1.3077709611451942,
      "grad_norm": 16.342115785360992,
      "learning_rate": 6.941550340902934e-06,
      "loss": 2.3671,
      "step": 3837
    },
    {
      "epoch": 1.3081117927743695,
      "grad_norm": 15.893470486127335,
      "learning_rate": 6.9397227245128874e-06,
      "loss": 1.8579,
      "step": 3838
    },
    {
      "epoch": 1.3084526244035446,
      "grad_norm": 26.6255229480798,
      "learning_rate": 6.937894802996734e-06,
      "loss": 2.1495,
      "step": 3839
    },
    {
      "epoch": 1.30879345603272,
      "grad_norm": 17.23120689851566,
      "learning_rate": 6.9360665766420165e-06,
      "loss": 2.4746,
      "step": 3840
    },
    {
      "epoch": 1.309134287661895,
      "grad_norm": 21.53173103278389,
      "learning_rate": 6.9342380457363176e-06,
      "loss": 1.8712,
      "step": 3841
    },
    {
      "epoch": 1.3094751192910703,
      "grad_norm": 20.698300110930877,
      "learning_rate": 6.932409210567275e-06,
      "loss": 2.3424,
      "step": 3842
    },
    {
      "epoch": 1.3098159509202454,
      "grad_norm": 16.892418105218574,
      "learning_rate": 6.93058007142257e-06,
      "loss": 2.3896,
      "step": 3843
    },
    {
      "epoch": 1.3101567825494205,
      "grad_norm": 16.50291823044476,
      "learning_rate": 6.928750628589935e-06,
      "loss": 2.2649,
      "step": 3844
    },
    {
      "epoch": 1.3104976141785958,
      "grad_norm": 23.887986586962686,
      "learning_rate": 6.92692088235715e-06,
      "loss": 2.1136,
      "step": 3845
    },
    {
      "epoch": 1.310838445807771,
      "grad_norm": 31.893850825636118,
      "learning_rate": 6.925090833012038e-06,
      "loss": 2.435,
      "step": 3846
    },
    {
      "epoch": 1.3111792774369462,
      "grad_norm": 17.038578214236743,
      "learning_rate": 6.9232604808424734e-06,
      "loss": 2.9094,
      "step": 3847
    },
    {
      "epoch": 1.3115201090661213,
      "grad_norm": 36.44132066024481,
      "learning_rate": 6.921429826136381e-06,
      "loss": 2.1493,
      "step": 3848
    },
    {
      "epoch": 1.3118609406952966,
      "grad_norm": 16.874113622746687,
      "learning_rate": 6.919598869181728e-06,
      "loss": 2.4017,
      "step": 3849
    },
    {
      "epoch": 1.3122017723244717,
      "grad_norm": 13.961713399703555,
      "learning_rate": 6.917767610266532e-06,
      "loss": 1.8282,
      "step": 3850
    },
    {
      "epoch": 1.312542603953647,
      "grad_norm": 14.153486178986439,
      "learning_rate": 6.915936049678855e-06,
      "loss": 1.6796,
      "step": 3851
    },
    {
      "epoch": 1.312883435582822,
      "grad_norm": 14.110829660547475,
      "learning_rate": 6.9141041877068105e-06,
      "loss": 2.2185,
      "step": 3852
    },
    {
      "epoch": 1.3132242672119974,
      "grad_norm": 11.232971768337123,
      "learning_rate": 6.912272024638559e-06,
      "loss": 2.3297,
      "step": 3853
    },
    {
      "epoch": 1.3135650988411725,
      "grad_norm": 19.58429388575918,
      "learning_rate": 6.910439560762305e-06,
      "loss": 2.1426,
      "step": 3854
    },
    {
      "epoch": 1.3139059304703475,
      "grad_norm": 19.029083995641212,
      "learning_rate": 6.9086067963663035e-06,
      "loss": 2.3334,
      "step": 3855
    },
    {
      "epoch": 1.3142467620995228,
      "grad_norm": 15.097944336104568,
      "learning_rate": 6.9067737317388535e-06,
      "loss": 2.431,
      "step": 3856
    },
    {
      "epoch": 1.3145875937286982,
      "grad_norm": 15.291953220666416,
      "learning_rate": 6.904940367168308e-06,
      "loss": 1.8898,
      "step": 3857
    },
    {
      "epoch": 1.3149284253578732,
      "grad_norm": 15.036577290849394,
      "learning_rate": 6.903106702943058e-06,
      "loss": 2.1919,
      "step": 3858
    },
    {
      "epoch": 1.3152692569870483,
      "grad_norm": 15.696085941190859,
      "learning_rate": 6.901272739351548e-06,
      "loss": 2.1346,
      "step": 3859
    },
    {
      "epoch": 1.3156100886162236,
      "grad_norm": 42.334378276211574,
      "learning_rate": 6.899438476682268e-06,
      "loss": 2.5064,
      "step": 3860
    },
    {
      "epoch": 1.3159509202453987,
      "grad_norm": 17.722484695755906,
      "learning_rate": 6.897603915223753e-06,
      "loss": 2.113,
      "step": 3861
    },
    {
      "epoch": 1.316291751874574,
      "grad_norm": 18.79507572915995,
      "learning_rate": 6.8957690552645896e-06,
      "loss": 2.5326,
      "step": 3862
    },
    {
      "epoch": 1.3166325835037491,
      "grad_norm": 18.1943425112976,
      "learning_rate": 6.893933897093406e-06,
      "loss": 1.8083,
      "step": 3863
    },
    {
      "epoch": 1.3169734151329244,
      "grad_norm": 11.799467558981958,
      "learning_rate": 6.892098440998881e-06,
      "loss": 1.75,
      "step": 3864
    },
    {
      "epoch": 1.3173142467620995,
      "grad_norm": 16.679822131383403,
      "learning_rate": 6.890262687269741e-06,
      "loss": 2.2385,
      "step": 3865
    },
    {
      "epoch": 1.3176550783912746,
      "grad_norm": 14.598476483988199,
      "learning_rate": 6.8884266361947545e-06,
      "loss": 2.393,
      "step": 3866
    },
    {
      "epoch": 1.31799591002045,
      "grad_norm": 17.89356812516282,
      "learning_rate": 6.886590288062741e-06,
      "loss": 2.2009,
      "step": 3867
    },
    {
      "epoch": 1.3183367416496252,
      "grad_norm": 18.590437797510663,
      "learning_rate": 6.884753643162565e-06,
      "loss": 2.722,
      "step": 3868
    },
    {
      "epoch": 1.3186775732788003,
      "grad_norm": 21.528736081202418,
      "learning_rate": 6.882916701783138e-06,
      "loss": 3.1708,
      "step": 3869
    },
    {
      "epoch": 1.3190184049079754,
      "grad_norm": 18.839244312680087,
      "learning_rate": 6.8810794642134185e-06,
      "loss": 2.2638,
      "step": 3870
    },
    {
      "epoch": 1.3193592365371507,
      "grad_norm": 14.608862100968034,
      "learning_rate": 6.87924193074241e-06,
      "loss": 2.5404,
      "step": 3871
    },
    {
      "epoch": 1.3197000681663258,
      "grad_norm": 18.427000878562087,
      "learning_rate": 6.877404101659166e-06,
      "loss": 2.3884,
      "step": 3872
    },
    {
      "epoch": 1.320040899795501,
      "grad_norm": 22.065773912202683,
      "learning_rate": 6.875565977252783e-06,
      "loss": 2.5673,
      "step": 3873
    },
    {
      "epoch": 1.3203817314246762,
      "grad_norm": 19.576269174829772,
      "learning_rate": 6.873727557812406e-06,
      "loss": 2.815,
      "step": 3874
    },
    {
      "epoch": 1.3207225630538515,
      "grad_norm": 17.10120052163096,
      "learning_rate": 6.871888843627226e-06,
      "loss": 2.3887,
      "step": 3875
    },
    {
      "epoch": 1.3210633946830266,
      "grad_norm": 24.16216454444984,
      "learning_rate": 6.870049834986479e-06,
      "loss": 2.5343,
      "step": 3876
    },
    {
      "epoch": 1.3214042263122017,
      "grad_norm": 15.806096200698017,
      "learning_rate": 6.868210532179447e-06,
      "loss": 1.9626,
      "step": 3877
    },
    {
      "epoch": 1.321745057941377,
      "grad_norm": 18.135263125124396,
      "learning_rate": 6.866370935495464e-06,
      "loss": 2.2331,
      "step": 3878
    },
    {
      "epoch": 1.322085889570552,
      "grad_norm": 14.323540270727536,
      "learning_rate": 6.864531045223901e-06,
      "loss": 2.0516,
      "step": 3879
    },
    {
      "epoch": 1.3224267211997274,
      "grad_norm": 37.71207770076337,
      "learning_rate": 6.862690861654184e-06,
      "loss": 1.8953,
      "step": 3880
    },
    {
      "epoch": 1.3227675528289025,
      "grad_norm": 13.633724129407634,
      "learning_rate": 6.860850385075777e-06,
      "loss": 2.3467,
      "step": 3881
    },
    {
      "epoch": 1.3231083844580778,
      "grad_norm": 14.90965598982734,
      "learning_rate": 6.859009615778199e-06,
      "loss": 2.5551,
      "step": 3882
    },
    {
      "epoch": 1.3234492160872529,
      "grad_norm": 21.7211468237817,
      "learning_rate": 6.857168554051008e-06,
      "loss": 2.7332,
      "step": 3883
    },
    {
      "epoch": 1.3237900477164282,
      "grad_norm": 18.31080426131572,
      "learning_rate": 6.85532720018381e-06,
      "loss": 2.1016,
      "step": 3884
    },
    {
      "epoch": 1.3241308793456033,
      "grad_norm": 23.606419565618964,
      "learning_rate": 6.853485554466258e-06,
      "loss": 2.3942,
      "step": 3885
    },
    {
      "epoch": 1.3244717109747786,
      "grad_norm": 14.84780720228075,
      "learning_rate": 6.85164361718805e-06,
      "loss": 2.5824,
      "step": 3886
    },
    {
      "epoch": 1.3248125426039536,
      "grad_norm": 16.01953339890038,
      "learning_rate": 6.849801388638929e-06,
      "loss": 2.3101,
      "step": 3887
    },
    {
      "epoch": 1.3251533742331287,
      "grad_norm": 13.969837033547517,
      "learning_rate": 6.847958869108689e-06,
      "loss": 1.9326,
      "step": 3888
    },
    {
      "epoch": 1.325494205862304,
      "grad_norm": 22.151160404559853,
      "learning_rate": 6.846116058887159e-06,
      "loss": 2.2138,
      "step": 3889
    },
    {
      "epoch": 1.3258350374914791,
      "grad_norm": 14.47859385097851,
      "learning_rate": 6.8442729582642245e-06,
      "loss": 2.4827,
      "step": 3890
    },
    {
      "epoch": 1.3261758691206544,
      "grad_norm": 18.01067543605298,
      "learning_rate": 6.842429567529815e-06,
      "loss": 2.3365,
      "step": 3891
    },
    {
      "epoch": 1.3265167007498295,
      "grad_norm": 14.596166241583779,
      "learning_rate": 6.840585886973897e-06,
      "loss": 2.7574,
      "step": 3892
    },
    {
      "epoch": 1.3268575323790048,
      "grad_norm": 16.90890573980185,
      "learning_rate": 6.838741916886493e-06,
      "loss": 2.1978,
      "step": 3893
    },
    {
      "epoch": 1.32719836400818,
      "grad_norm": 17.75469843037621,
      "learning_rate": 6.836897657557666e-06,
      "loss": 2.2504,
      "step": 3894
    },
    {
      "epoch": 1.3275391956373552,
      "grad_norm": 16.41835900070921,
      "learning_rate": 6.835053109277525e-06,
      "loss": 2.1525,
      "step": 3895
    },
    {
      "epoch": 1.3278800272665303,
      "grad_norm": 22.960236610631313,
      "learning_rate": 6.833208272336227e-06,
      "loss": 2.7451,
      "step": 3896
    },
    {
      "epoch": 1.3282208588957056,
      "grad_norm": 21.40550883112957,
      "learning_rate": 6.831363147023968e-06,
      "loss": 2.1508,
      "step": 3897
    },
    {
      "epoch": 1.3285616905248807,
      "grad_norm": 15.090253184796694,
      "learning_rate": 6.829517733630995e-06,
      "loss": 2.3375,
      "step": 3898
    },
    {
      "epoch": 1.3289025221540558,
      "grad_norm": 18.086659181070058,
      "learning_rate": 6.827672032447602e-06,
      "loss": 2.0773,
      "step": 3899
    },
    {
      "epoch": 1.329243353783231,
      "grad_norm": 16.060954192303036,
      "learning_rate": 6.825826043764121e-06,
      "loss": 2.6167,
      "step": 3900
    },
    {
      "epoch": 1.3295841854124062,
      "grad_norm": 21.79962843725842,
      "learning_rate": 6.823979767870936e-06,
      "loss": 2.1114,
      "step": 3901
    },
    {
      "epoch": 1.3299250170415815,
      "grad_norm": 19.094688767103232,
      "learning_rate": 6.822133205058472e-06,
      "loss": 2.2896,
      "step": 3902
    },
    {
      "epoch": 1.3302658486707566,
      "grad_norm": 13.715316023411846,
      "learning_rate": 6.820286355617202e-06,
      "loss": 2.0012,
      "step": 3903
    },
    {
      "epoch": 1.330606680299932,
      "grad_norm": 14.649565002096471,
      "learning_rate": 6.818439219837642e-06,
      "loss": 2.2898,
      "step": 3904
    },
    {
      "epoch": 1.330947511929107,
      "grad_norm": 15.79462281622963,
      "learning_rate": 6.816591798010354e-06,
      "loss": 1.9373,
      "step": 3905
    },
    {
      "epoch": 1.331288343558282,
      "grad_norm": 16.34796310845031,
      "learning_rate": 6.814744090425945e-06,
      "loss": 2.3773,
      "step": 3906
    },
    {
      "epoch": 1.3316291751874574,
      "grad_norm": 18.138759245882046,
      "learning_rate": 6.812896097375065e-06,
      "loss": 2.6431,
      "step": 3907
    },
    {
      "epoch": 1.3319700068166327,
      "grad_norm": 16.57133826933695,
      "learning_rate": 6.811047819148413e-06,
      "loss": 2.0701,
      "step": 3908
    },
    {
      "epoch": 1.3323108384458078,
      "grad_norm": 14.277452310457255,
      "learning_rate": 6.8091992560367305e-06,
      "loss": 2.0092,
      "step": 3909
    },
    {
      "epoch": 1.3326516700749829,
      "grad_norm": 14.713774267725489,
      "learning_rate": 6.807350408330802e-06,
      "loss": 1.935,
      "step": 3910
    },
    {
      "epoch": 1.3329925017041582,
      "grad_norm": 17.264725490728086,
      "learning_rate": 6.80550127632146e-06,
      "loss": 2.4897,
      "step": 3911
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 19.393539640646576,
      "learning_rate": 6.80365186029958e-06,
      "loss": 2.111,
      "step": 3912
    },
    {
      "epoch": 1.3336741649625086,
      "grad_norm": 20.135367266973134,
      "learning_rate": 6.801802160556084e-06,
      "loss": 1.4603,
      "step": 3913
    },
    {
      "epoch": 1.3340149965916837,
      "grad_norm": 12.417195162624207,
      "learning_rate": 6.799952177381933e-06,
      "loss": 1.8347,
      "step": 3914
    },
    {
      "epoch": 1.334355828220859,
      "grad_norm": 14.254410002981151,
      "learning_rate": 6.798101911068139e-06,
      "loss": 2.3993,
      "step": 3915
    },
    {
      "epoch": 1.334696659850034,
      "grad_norm": 14.460864690519069,
      "learning_rate": 6.796251361905759e-06,
      "loss": 2.1682,
      "step": 3916
    },
    {
      "epoch": 1.3350374914792091,
      "grad_norm": 16.41066647033498,
      "learning_rate": 6.794400530185889e-06,
      "loss": 2.6489,
      "step": 3917
    },
    {
      "epoch": 1.3353783231083844,
      "grad_norm": 35.791416385958385,
      "learning_rate": 6.7925494161996705e-06,
      "loss": 3.0755,
      "step": 3918
    },
    {
      "epoch": 1.3357191547375598,
      "grad_norm": 22.279546770559367,
      "learning_rate": 6.790698020238294e-06,
      "loss": 2.2229,
      "step": 3919
    },
    {
      "epoch": 1.3360599863667348,
      "grad_norm": 22.73376290380986,
      "learning_rate": 6.788846342592991e-06,
      "loss": 1.4827,
      "step": 3920
    },
    {
      "epoch": 1.33640081799591,
      "grad_norm": 17.332685406293752,
      "learning_rate": 6.786994383555037e-06,
      "loss": 2.261,
      "step": 3921
    },
    {
      "epoch": 1.3367416496250852,
      "grad_norm": 15.355793195837107,
      "learning_rate": 6.785142143415754e-06,
      "loss": 2.3056,
      "step": 3922
    },
    {
      "epoch": 1.3370824812542603,
      "grad_norm": 22.700955769998025,
      "learning_rate": 6.783289622466503e-06,
      "loss": 2.3986,
      "step": 3923
    },
    {
      "epoch": 1.3374233128834356,
      "grad_norm": 17.820343585335078,
      "learning_rate": 6.781436820998697e-06,
      "loss": 2.7847,
      "step": 3924
    },
    {
      "epoch": 1.3377641445126107,
      "grad_norm": 22.028047698722347,
      "learning_rate": 6.779583739303789e-06,
      "loss": 2.0335,
      "step": 3925
    },
    {
      "epoch": 1.338104976141786,
      "grad_norm": 28.16750353305819,
      "learning_rate": 6.7777303776732725e-06,
      "loss": 1.6513,
      "step": 3926
    },
    {
      "epoch": 1.3384458077709611,
      "grad_norm": 29.432767640020277,
      "learning_rate": 6.775876736398692e-06,
      "loss": 2.673,
      "step": 3927
    },
    {
      "epoch": 1.3387866394001362,
      "grad_norm": 15.915401777218342,
      "learning_rate": 6.774022815771631e-06,
      "loss": 2.4791,
      "step": 3928
    },
    {
      "epoch": 1.3391274710293115,
      "grad_norm": 17.05504661342197,
      "learning_rate": 6.77216861608372e-06,
      "loss": 2.0134,
      "step": 3929
    },
    {
      "epoch": 1.3394683026584868,
      "grad_norm": 13.095722629978235,
      "learning_rate": 6.770314137626631e-06,
      "loss": 1.9494,
      "step": 3930
    },
    {
      "epoch": 1.339809134287662,
      "grad_norm": 15.702277118978483,
      "learning_rate": 6.768459380692079e-06,
      "loss": 2.5766,
      "step": 3931
    },
    {
      "epoch": 1.340149965916837,
      "grad_norm": 15.427214706424712,
      "learning_rate": 6.766604345571827e-06,
      "loss": 2.5998,
      "step": 3932
    },
    {
      "epoch": 1.3404907975460123,
      "grad_norm": 21.935174795316502,
      "learning_rate": 6.764749032557683e-06,
      "loss": 2.4382,
      "step": 3933
    },
    {
      "epoch": 1.3408316291751874,
      "grad_norm": 31.962066301241855,
      "learning_rate": 6.762893441941487e-06,
      "loss": 2.9641,
      "step": 3934
    },
    {
      "epoch": 1.3411724608043627,
      "grad_norm": 19.176846574776288,
      "learning_rate": 6.761037574015136e-06,
      "loss": 2.5409,
      "step": 3935
    },
    {
      "epoch": 1.3415132924335378,
      "grad_norm": 13.45491074271513,
      "learning_rate": 6.7591814290705645e-06,
      "loss": 2.0225,
      "step": 3936
    },
    {
      "epoch": 1.341854124062713,
      "grad_norm": 14.1703637418656,
      "learning_rate": 6.757325007399751e-06,
      "loss": 2.0603,
      "step": 3937
    },
    {
      "epoch": 1.3421949556918882,
      "grad_norm": 26.590833448257772,
      "learning_rate": 6.755468309294718e-06,
      "loss": 2.1169,
      "step": 3938
    },
    {
      "epoch": 1.3425357873210633,
      "grad_norm": 22.340231165867152,
      "learning_rate": 6.753611335047533e-06,
      "loss": 2.5642,
      "step": 3939
    },
    {
      "epoch": 1.3428766189502386,
      "grad_norm": 20.21293242497697,
      "learning_rate": 6.7517540849503025e-06,
      "loss": 2.3219,
      "step": 3940
    },
    {
      "epoch": 1.3432174505794139,
      "grad_norm": 16.041895635972786,
      "learning_rate": 6.749896559295183e-06,
      "loss": 2.5965,
      "step": 3941
    },
    {
      "epoch": 1.343558282208589,
      "grad_norm": 17.038333298579545,
      "learning_rate": 6.748038758374368e-06,
      "loss": 2.4708,
      "step": 3942
    },
    {
      "epoch": 1.343899113837764,
      "grad_norm": 19.392688586225063,
      "learning_rate": 6.746180682480098e-06,
      "loss": 2.1638,
      "step": 3943
    },
    {
      "epoch": 1.3442399454669394,
      "grad_norm": 28.262957873147588,
      "learning_rate": 6.744322331904655e-06,
      "loss": 2.4218,
      "step": 3944
    },
    {
      "epoch": 1.3445807770961145,
      "grad_norm": 15.507477541579416,
      "learning_rate": 6.7424637069403655e-06,
      "loss": 1.9211,
      "step": 3945
    },
    {
      "epoch": 1.3449216087252898,
      "grad_norm": 17.261159988348105,
      "learning_rate": 6.7406048078795985e-06,
      "loss": 2.2782,
      "step": 3946
    },
    {
      "epoch": 1.3452624403544649,
      "grad_norm": 16.982298690851398,
      "learning_rate": 6.738745635014767e-06,
      "loss": 2.6624,
      "step": 3947
    },
    {
      "epoch": 1.3456032719836402,
      "grad_norm": 20.186247990272243,
      "learning_rate": 6.736886188638324e-06,
      "loss": 2.3136,
      "step": 3948
    },
    {
      "epoch": 1.3459441036128152,
      "grad_norm": 19.272146544234623,
      "learning_rate": 6.735026469042771e-06,
      "loss": 2.5187,
      "step": 3949
    },
    {
      "epoch": 1.3462849352419903,
      "grad_norm": 14.42585529234335,
      "learning_rate": 6.7331664765206476e-06,
      "loss": 2.1949,
      "step": 3950
    },
    {
      "epoch": 1.3466257668711656,
      "grad_norm": 16.66782917254955,
      "learning_rate": 6.731306211364538e-06,
      "loss": 2.724,
      "step": 3951
    },
    {
      "epoch": 1.346966598500341,
      "grad_norm": 19.790437080407226,
      "learning_rate": 6.729445673867069e-06,
      "loss": 2.2639,
      "step": 3952
    },
    {
      "epoch": 1.347307430129516,
      "grad_norm": 14.128910958377219,
      "learning_rate": 6.727584864320911e-06,
      "loss": 2.3215,
      "step": 3953
    },
    {
      "epoch": 1.3476482617586911,
      "grad_norm": 16.375063237189135,
      "learning_rate": 6.725723783018776e-06,
      "loss": 2.1036,
      "step": 3954
    },
    {
      "epoch": 1.3479890933878664,
      "grad_norm": 18.948694573825097,
      "learning_rate": 6.723862430253421e-06,
      "loss": 2.558,
      "step": 3955
    },
    {
      "epoch": 1.3483299250170415,
      "grad_norm": 21.44175995027011,
      "learning_rate": 6.722000806317645e-06,
      "loss": 1.9551,
      "step": 3956
    },
    {
      "epoch": 1.3486707566462168,
      "grad_norm": 15.110184244331426,
      "learning_rate": 6.720138911504283e-06,
      "loss": 2.3549,
      "step": 3957
    },
    {
      "epoch": 1.349011588275392,
      "grad_norm": 2742.972761314463,
      "learning_rate": 6.718276746106227e-06,
      "loss": 2.0309,
      "step": 3958
    },
    {
      "epoch": 1.3493524199045672,
      "grad_norm": 16.26635429272673,
      "learning_rate": 6.716414310416397e-06,
      "loss": 2.2225,
      "step": 3959
    },
    {
      "epoch": 1.3496932515337423,
      "grad_norm": 14.302994101674273,
      "learning_rate": 6.714551604727763e-06,
      "loss": 2.281,
      "step": 3960
    },
    {
      "epoch": 1.3500340831629174,
      "grad_norm": 25.484301440705355,
      "learning_rate": 6.712688629333337e-06,
      "loss": 2.3777,
      "step": 3961
    },
    {
      "epoch": 1.3503749147920927,
      "grad_norm": 28.818765837355418,
      "learning_rate": 6.710825384526172e-06,
      "loss": 2.0625,
      "step": 3962
    },
    {
      "epoch": 1.350715746421268,
      "grad_norm": 18.786343937076815,
      "learning_rate": 6.7089618705993635e-06,
      "loss": 2.3281,
      "step": 3963
    },
    {
      "epoch": 1.351056578050443,
      "grad_norm": 15.584873240743278,
      "learning_rate": 6.7070980878460505e-06,
      "loss": 3.188,
      "step": 3964
    },
    {
      "epoch": 1.3513974096796182,
      "grad_norm": 20.971025709596386,
      "learning_rate": 6.70523403655941e-06,
      "loss": 1.5928,
      "step": 3965
    },
    {
      "epoch": 1.3517382413087935,
      "grad_norm": 36.43797068226448,
      "learning_rate": 6.7033697170326694e-06,
      "loss": 2.7645,
      "step": 3966
    },
    {
      "epoch": 1.3520790729379686,
      "grad_norm": 16.086409505250273,
      "learning_rate": 6.701505129559093e-06,
      "loss": 2.4911,
      "step": 3967
    },
    {
      "epoch": 1.352419904567144,
      "grad_norm": 19.901051839506227,
      "learning_rate": 6.699640274431985e-06,
      "loss": 2.8267,
      "step": 3968
    },
    {
      "epoch": 1.352760736196319,
      "grad_norm": 15.027273247288564,
      "learning_rate": 6.697775151944694e-06,
      "loss": 2.5942,
      "step": 3969
    },
    {
      "epoch": 1.3531015678254943,
      "grad_norm": 23.15630792601581,
      "learning_rate": 6.695909762390615e-06,
      "loss": 2.1881,
      "step": 3970
    },
    {
      "epoch": 1.3534423994546694,
      "grad_norm": 20.70328240891375,
      "learning_rate": 6.69404410606318e-06,
      "loss": 2.7542,
      "step": 3971
    },
    {
      "epoch": 1.3537832310838445,
      "grad_norm": 23.98601888556263,
      "learning_rate": 6.692178183255862e-06,
      "loss": 2.2346,
      "step": 3972
    },
    {
      "epoch": 1.3541240627130198,
      "grad_norm": 19.851549329193475,
      "learning_rate": 6.690311994262182e-06,
      "loss": 2.2817,
      "step": 3973
    },
    {
      "epoch": 1.354464894342195,
      "grad_norm": 11.376590899206189,
      "learning_rate": 6.688445539375693e-06,
      "loss": 2.0361,
      "step": 3974
    },
    {
      "epoch": 1.3548057259713702,
      "grad_norm": 20.623996979381854,
      "learning_rate": 6.686578818890003e-06,
      "loss": 2.0397,
      "step": 3975
    },
    {
      "epoch": 1.3551465576005453,
      "grad_norm": 19.366699725711303,
      "learning_rate": 6.6847118330987486e-06,
      "loss": 2.5811,
      "step": 3976
    },
    {
      "epoch": 1.3554873892297206,
      "grad_norm": 15.148130041279943,
      "learning_rate": 6.682844582295617e-06,
      "loss": 2.6696,
      "step": 3977
    },
    {
      "epoch": 1.3558282208588956,
      "grad_norm": 17.173217148497073,
      "learning_rate": 6.6809770667743335e-06,
      "loss": 1.4998,
      "step": 3978
    },
    {
      "epoch": 1.356169052488071,
      "grad_norm": 16.33245211118213,
      "learning_rate": 6.679109286828665e-06,
      "loss": 2.5324,
      "step": 3979
    },
    {
      "epoch": 1.356509884117246,
      "grad_norm": 20.59107767209744,
      "learning_rate": 6.6772412427524224e-06,
      "loss": 2.1547,
      "step": 3980
    },
    {
      "epoch": 1.3568507157464214,
      "grad_norm": 19.849801984701337,
      "learning_rate": 6.675372934839457e-06,
      "loss": 2.3262,
      "step": 3981
    },
    {
      "epoch": 1.3571915473755964,
      "grad_norm": 32.871637031805676,
      "learning_rate": 6.673504363383657e-06,
      "loss": 2.7198,
      "step": 3982
    },
    {
      "epoch": 1.3575323790047715,
      "grad_norm": 23.358786323113634,
      "learning_rate": 6.6716355286789595e-06,
      "loss": 2.9496,
      "step": 3983
    },
    {
      "epoch": 1.3578732106339468,
      "grad_norm": 14.108882060454794,
      "learning_rate": 6.669766431019341e-06,
      "loss": 1.4995,
      "step": 3984
    },
    {
      "epoch": 1.3582140422631221,
      "grad_norm": 16.868898199158863,
      "learning_rate": 6.667897070698815e-06,
      "loss": 2.2347,
      "step": 3985
    },
    {
      "epoch": 1.3585548738922972,
      "grad_norm": 16.681616599083593,
      "learning_rate": 6.666027448011438e-06,
      "loss": 2.6437,
      "step": 3986
    },
    {
      "epoch": 1.3588957055214723,
      "grad_norm": 18.816473097242017,
      "learning_rate": 6.664157563251313e-06,
      "loss": 2.8454,
      "step": 3987
    },
    {
      "epoch": 1.3592365371506476,
      "grad_norm": 19.894184538749062,
      "learning_rate": 6.66228741671258e-06,
      "loss": 1.868,
      "step": 3988
    },
    {
      "epoch": 1.3595773687798227,
      "grad_norm": 16.723783289051973,
      "learning_rate": 6.660417008689417e-06,
      "loss": 2.4035,
      "step": 3989
    },
    {
      "epoch": 1.359918200408998,
      "grad_norm": 12.590346941430214,
      "learning_rate": 6.65854633947605e-06,
      "loss": 2.4324,
      "step": 3990
    },
    {
      "epoch": 1.360259032038173,
      "grad_norm": 21.77209743162936,
      "learning_rate": 6.656675409366741e-06,
      "loss": 2.4762,
      "step": 3991
    },
    {
      "epoch": 1.3605998636673484,
      "grad_norm": 13.44267242681174,
      "learning_rate": 6.654804218655796e-06,
      "loss": 2.2917,
      "step": 3992
    },
    {
      "epoch": 1.3609406952965235,
      "grad_norm": 17.701274496013788,
      "learning_rate": 6.65293276763756e-06,
      "loss": 2.6656,
      "step": 3993
    },
    {
      "epoch": 1.3612815269256986,
      "grad_norm": 17.890984882122222,
      "learning_rate": 6.651061056606418e-06,
      "loss": 1.7633,
      "step": 3994
    },
    {
      "epoch": 1.361622358554874,
      "grad_norm": 19.18213471432306,
      "learning_rate": 6.649189085856801e-06,
      "loss": 2.1932,
      "step": 3995
    },
    {
      "epoch": 1.3619631901840492,
      "grad_norm": 14.686212209876405,
      "learning_rate": 6.647316855683175e-06,
      "loss": 1.9913,
      "step": 3996
    },
    {
      "epoch": 1.3623040218132243,
      "grad_norm": 18.68365824248722,
      "learning_rate": 6.64544436638005e-06,
      "loss": 1.912,
      "step": 3997
    },
    {
      "epoch": 1.3626448534423994,
      "grad_norm": 30.75097274977735,
      "learning_rate": 6.643571618241977e-06,
      "loss": 1.9378,
      "step": 3998
    },
    {
      "epoch": 1.3629856850715747,
      "grad_norm": 17.292137457572448,
      "learning_rate": 6.641698611563545e-06,
      "loss": 2.7281,
      "step": 3999
    },
    {
      "epoch": 1.3633265167007498,
      "grad_norm": 17.67170547869491,
      "learning_rate": 6.639825346639386e-06,
      "loss": 2.323,
      "step": 4000
    },
    {
      "epoch": 1.363667348329925,
      "grad_norm": 12.712052027131477,
      "learning_rate": 6.637951823764173e-06,
      "loss": 2.081,
      "step": 4001
    },
    {
      "epoch": 1.3640081799591002,
      "grad_norm": 68.13401727721093,
      "learning_rate": 6.636078043232616e-06,
      "loss": 2.0664,
      "step": 4002
    },
    {
      "epoch": 1.3643490115882755,
      "grad_norm": 20.217286208560058,
      "learning_rate": 6.634204005339473e-06,
      "loss": 1.9044,
      "step": 4003
    },
    {
      "epoch": 1.3646898432174506,
      "grad_norm": 14.782250362795386,
      "learning_rate": 6.632329710379532e-06,
      "loss": 2.5283,
      "step": 4004
    },
    {
      "epoch": 1.3650306748466257,
      "grad_norm": 16.375843687473143,
      "learning_rate": 6.630455158647631e-06,
      "loss": 2.3356,
      "step": 4005
    },
    {
      "epoch": 1.365371506475801,
      "grad_norm": 21.534898197466504,
      "learning_rate": 6.6285803504386415e-06,
      "loss": 1.8482,
      "step": 4006
    },
    {
      "epoch": 1.3657123381049763,
      "grad_norm": 14.761266922225166,
      "learning_rate": 6.626705286047482e-06,
      "loss": 1.8946,
      "step": 4007
    },
    {
      "epoch": 1.3660531697341514,
      "grad_norm": 16.50662413342029,
      "learning_rate": 6.624829965769103e-06,
      "loss": 2.1722,
      "step": 4008
    },
    {
      "epoch": 1.3663940013633264,
      "grad_norm": 19.251580403279423,
      "learning_rate": 6.622954389898505e-06,
      "loss": 2.7596,
      "step": 4009
    },
    {
      "epoch": 1.3667348329925018,
      "grad_norm": 12.056685919100195,
      "learning_rate": 6.621078558730719e-06,
      "loss": 2.3449,
      "step": 4010
    },
    {
      "epoch": 1.3670756646216768,
      "grad_norm": 29.862114602729264,
      "learning_rate": 6.619202472560823e-06,
      "loss": 2.1358,
      "step": 4011
    },
    {
      "epoch": 1.3674164962508522,
      "grad_norm": 20.6706124252245,
      "learning_rate": 6.617326131683932e-06,
      "loss": 1.9836,
      "step": 4012
    },
    {
      "epoch": 1.3677573278800272,
      "grad_norm": 11.97283836544764,
      "learning_rate": 6.615449536395203e-06,
      "loss": 2.0862,
      "step": 4013
    },
    {
      "epoch": 1.3680981595092025,
      "grad_norm": 16.762829392396153,
      "learning_rate": 6.613572686989831e-06,
      "loss": 2.3208,
      "step": 4014
    },
    {
      "epoch": 1.3684389911383776,
      "grad_norm": 15.465443861048868,
      "learning_rate": 6.611695583763053e-06,
      "loss": 1.9743,
      "step": 4015
    },
    {
      "epoch": 1.3687798227675527,
      "grad_norm": 11.35089050475214,
      "learning_rate": 6.6098182270101415e-06,
      "loss": 1.6009,
      "step": 4016
    },
    {
      "epoch": 1.369120654396728,
      "grad_norm": 22.18151261938851,
      "learning_rate": 6.607940617026416e-06,
      "loss": 1.9887,
      "step": 4017
    },
    {
      "epoch": 1.3694614860259033,
      "grad_norm": 18.843116936258617,
      "learning_rate": 6.606062754107232e-06,
      "loss": 2.4372,
      "step": 4018
    },
    {
      "epoch": 1.3698023176550784,
      "grad_norm": 16.628988007952554,
      "learning_rate": 6.604184638547981e-06,
      "loss": 1.9778,
      "step": 4019
    },
    {
      "epoch": 1.3701431492842535,
      "grad_norm": 14.94171855064862,
      "learning_rate": 6.602306270644101e-06,
      "loss": 2.3785,
      "step": 4020
    },
    {
      "epoch": 1.3704839809134288,
      "grad_norm": 13.593436378567938,
      "learning_rate": 6.6004276506910684e-06,
      "loss": 1.897,
      "step": 4021
    },
    {
      "epoch": 1.370824812542604,
      "grad_norm": 18.592813023255882,
      "learning_rate": 6.598548778984393e-06,
      "loss": 2.5464,
      "step": 4022
    },
    {
      "epoch": 1.3711656441717792,
      "grad_norm": 10.853595315923817,
      "learning_rate": 6.596669655819632e-06,
      "loss": 2.2601,
      "step": 4023
    },
    {
      "epoch": 1.3715064758009543,
      "grad_norm": 20.042366304423293,
      "learning_rate": 6.594790281492378e-06,
      "loss": 2.5854,
      "step": 4024
    },
    {
      "epoch": 1.3718473074301296,
      "grad_norm": 16.513404280483122,
      "learning_rate": 6.592910656298264e-06,
      "loss": 1.8443,
      "step": 4025
    },
    {
      "epoch": 1.3721881390593047,
      "grad_norm": 17.564937681026382,
      "learning_rate": 6.591030780532963e-06,
      "loss": 2.2527,
      "step": 4026
    },
    {
      "epoch": 1.3725289706884798,
      "grad_norm": 29.576600324453846,
      "learning_rate": 6.589150654492187e-06,
      "loss": 2.2793,
      "step": 4027
    },
    {
      "epoch": 1.372869802317655,
      "grad_norm": 16.38545823734742,
      "learning_rate": 6.587270278471687e-06,
      "loss": 2.4684,
      "step": 4028
    },
    {
      "epoch": 1.3732106339468302,
      "grad_norm": 15.6548697439115,
      "learning_rate": 6.585389652767253e-06,
      "loss": 2.9369,
      "step": 4029
    },
    {
      "epoch": 1.3735514655760055,
      "grad_norm": 14.032668236161104,
      "learning_rate": 6.5835087776747165e-06,
      "loss": 2.5064,
      "step": 4030
    },
    {
      "epoch": 1.3738922972051806,
      "grad_norm": 17.14349040280224,
      "learning_rate": 6.581627653489945e-06,
      "loss": 1.3395,
      "step": 4031
    },
    {
      "epoch": 1.3742331288343559,
      "grad_norm": 17.82702243603661,
      "learning_rate": 6.5797462805088475e-06,
      "loss": 2.2301,
      "step": 4032
    },
    {
      "epoch": 1.374573960463531,
      "grad_norm": 18.486420089066666,
      "learning_rate": 6.577864659027371e-06,
      "loss": 2.3194,
      "step": 4033
    },
    {
      "epoch": 1.3749147920927063,
      "grad_norm": 20.87687558313877,
      "learning_rate": 6.575982789341503e-06,
      "loss": 2.3011,
      "step": 4034
    },
    {
      "epoch": 1.3752556237218814,
      "grad_norm": 13.728471907029181,
      "learning_rate": 6.57410067174727e-06,
      "loss": 2.0545,
      "step": 4035
    },
    {
      "epoch": 1.3755964553510567,
      "grad_norm": 17.448253526879572,
      "learning_rate": 6.5722183065407345e-06,
      "loss": 2.4056,
      "step": 4036
    },
    {
      "epoch": 1.3759372869802318,
      "grad_norm": 25.150875228570023,
      "learning_rate": 6.5703356940179974e-06,
      "loss": 2.3097,
      "step": 4037
    },
    {
      "epoch": 1.3762781186094069,
      "grad_norm": 15.127248435831987,
      "learning_rate": 6.568452834475209e-06,
      "loss": 2.7579,
      "step": 4038
    },
    {
      "epoch": 1.3766189502385822,
      "grad_norm": 19.238802074282056,
      "learning_rate": 6.566569728208544e-06,
      "loss": 2.7834,
      "step": 4039
    },
    {
      "epoch": 1.3769597818677572,
      "grad_norm": 14.85067762157095,
      "learning_rate": 6.564686375514226e-06,
      "loss": 1.9136,
      "step": 4040
    },
    {
      "epoch": 1.3773006134969326,
      "grad_norm": 17.02647148358015,
      "learning_rate": 6.562802776688511e-06,
      "loss": 2.1712,
      "step": 4041
    },
    {
      "epoch": 1.3776414451261076,
      "grad_norm": 13.08960960764839,
      "learning_rate": 6.560918932027698e-06,
      "loss": 2.2136,
      "step": 4042
    },
    {
      "epoch": 1.377982276755283,
      "grad_norm": 16.24222422853547,
      "learning_rate": 6.5590348418281245e-06,
      "loss": 2.0946,
      "step": 4043
    },
    {
      "epoch": 1.378323108384458,
      "grad_norm": 18.285014953506455,
      "learning_rate": 6.557150506386163e-06,
      "loss": 2.1049,
      "step": 4044
    },
    {
      "epoch": 1.3786639400136331,
      "grad_norm": 13.814695341223445,
      "learning_rate": 6.555265925998229e-06,
      "loss": 1.9096,
      "step": 4045
    },
    {
      "epoch": 1.3790047716428084,
      "grad_norm": 13.02718234434643,
      "learning_rate": 6.553381100960774e-06,
      "loss": 2.119,
      "step": 4046
    },
    {
      "epoch": 1.3793456032719837,
      "grad_norm": 23.033395614848438,
      "learning_rate": 6.5514960315702865e-06,
      "loss": 2.4611,
      "step": 4047
    },
    {
      "epoch": 1.3796864349011588,
      "grad_norm": 21.55558739668274,
      "learning_rate": 6.549610718123298e-06,
      "loss": 1.5679,
      "step": 4048
    },
    {
      "epoch": 1.380027266530334,
      "grad_norm": 11.811051260671974,
      "learning_rate": 6.547725160916375e-06,
      "loss": 2.0442,
      "step": 4049
    },
    {
      "epoch": 1.3803680981595092,
      "grad_norm": 23.316635812133164,
      "learning_rate": 6.545839360246124e-06,
      "loss": 2.1392,
      "step": 4050
    },
    {
      "epoch": 1.3807089297886843,
      "grad_norm": 14.994687417617131,
      "learning_rate": 6.543953316409186e-06,
      "loss": 2.2499,
      "step": 4051
    },
    {
      "epoch": 1.3810497614178596,
      "grad_norm": 19.089471706555113,
      "learning_rate": 6.542067029702248e-06,
      "loss": 2.325,
      "step": 4052
    },
    {
      "epoch": 1.3813905930470347,
      "grad_norm": 15.07025300026965,
      "learning_rate": 6.540180500422025e-06,
      "loss": 2.3912,
      "step": 4053
    },
    {
      "epoch": 1.38173142467621,
      "grad_norm": 18.368759588493422,
      "learning_rate": 6.538293728865278e-06,
      "loss": 2.2036,
      "step": 4054
    },
    {
      "epoch": 1.382072256305385,
      "grad_norm": 416.18742037813195,
      "learning_rate": 6.536406715328807e-06,
      "loss": 2.4325,
      "step": 4055
    },
    {
      "epoch": 1.3824130879345602,
      "grad_norm": 23.636985591705283,
      "learning_rate": 6.534519460109441e-06,
      "loss": 2.5124,
      "step": 4056
    },
    {
      "epoch": 1.3827539195637355,
      "grad_norm": 14.717799705904612,
      "learning_rate": 6.5326319635040565e-06,
      "loss": 2.3464,
      "step": 4057
    },
    {
      "epoch": 1.3830947511929108,
      "grad_norm": 16.18448222390933,
      "learning_rate": 6.530744225809562e-06,
      "loss": 2.2514,
      "step": 4058
    },
    {
      "epoch": 1.383435582822086,
      "grad_norm": 17.816015744328855,
      "learning_rate": 6.528856247322908e-06,
      "loss": 2.0599,
      "step": 4059
    },
    {
      "epoch": 1.383776414451261,
      "grad_norm": 17.90812231147436,
      "learning_rate": 6.526968028341081e-06,
      "loss": 1.8762,
      "step": 4060
    },
    {
      "epoch": 1.3841172460804363,
      "grad_norm": 19.74809377383867,
      "learning_rate": 6.5250795691611034e-06,
      "loss": 2.1113,
      "step": 4061
    },
    {
      "epoch": 1.3844580777096114,
      "grad_norm": 15.059273287793626,
      "learning_rate": 6.5231908700800385e-06,
      "loss": 2.4504,
      "step": 4062
    },
    {
      "epoch": 1.3847989093387867,
      "grad_norm": 17.00077566657081,
      "learning_rate": 6.521301931394987e-06,
      "loss": 2.382,
      "step": 4063
    },
    {
      "epoch": 1.3851397409679618,
      "grad_norm": 14.740718791593597,
      "learning_rate": 6.519412753403085e-06,
      "loss": 1.8498,
      "step": 4064
    },
    {
      "epoch": 1.385480572597137,
      "grad_norm": 20.584209527588705,
      "learning_rate": 6.517523336401508e-06,
      "loss": 1.9443,
      "step": 4065
    },
    {
      "epoch": 1.3858214042263122,
      "grad_norm": 25.597056907611815,
      "learning_rate": 6.51563368068747e-06,
      "loss": 2.4438,
      "step": 4066
    },
    {
      "epoch": 1.3861622358554873,
      "grad_norm": 16.961414728238708,
      "learning_rate": 6.513743786558221e-06,
      "loss": 2.2819,
      "step": 4067
    },
    {
      "epoch": 1.3865030674846626,
      "grad_norm": 17.1046155410895,
      "learning_rate": 6.5118536543110475e-06,
      "loss": 2.5748,
      "step": 4068
    },
    {
      "epoch": 1.3868438991138379,
      "grad_norm": 20.39730259755839,
      "learning_rate": 6.509963284243278e-06,
      "loss": 1.798,
      "step": 4069
    },
    {
      "epoch": 1.387184730743013,
      "grad_norm": 13.157044170022662,
      "learning_rate": 6.508072676652272e-06,
      "loss": 2.152,
      "step": 4070
    },
    {
      "epoch": 1.387525562372188,
      "grad_norm": 9.977540315981646,
      "learning_rate": 6.506181831835429e-06,
      "loss": 1.7747,
      "step": 4071
    },
    {
      "epoch": 1.3878663940013634,
      "grad_norm": 51.838248201196144,
      "learning_rate": 6.50429075009019e-06,
      "loss": 2.3495,
      "step": 4072
    },
    {
      "epoch": 1.3882072256305384,
      "grad_norm": 16.006541506248055,
      "learning_rate": 6.502399431714027e-06,
      "loss": 2.174,
      "step": 4073
    },
    {
      "epoch": 1.3885480572597138,
      "grad_norm": 22.3515317539378,
      "learning_rate": 6.500507877004454e-06,
      "loss": 2.2558,
      "step": 4074
    },
    {
      "epoch": 1.3888888888888888,
      "grad_norm": 16.676019688673847,
      "learning_rate": 6.498616086259017e-06,
      "loss": 2.6179,
      "step": 4075
    },
    {
      "epoch": 1.3892297205180641,
      "grad_norm": 12.160248574830206,
      "learning_rate": 6.496724059775304e-06,
      "loss": 2.2462,
      "step": 4076
    },
    {
      "epoch": 1.3895705521472392,
      "grad_norm": 19.89994411306739,
      "learning_rate": 6.49483179785094e-06,
      "loss": 1.9553,
      "step": 4077
    },
    {
      "epoch": 1.3899113837764143,
      "grad_norm": 22.876641473526966,
      "learning_rate": 6.492939300783582e-06,
      "loss": 2.2315,
      "step": 4078
    },
    {
      "epoch": 1.3902522154055896,
      "grad_norm": 14.991333300072895,
      "learning_rate": 6.491046568870929e-06,
      "loss": 2.5941,
      "step": 4079
    },
    {
      "epoch": 1.390593047034765,
      "grad_norm": 15.914751322769293,
      "learning_rate": 6.489153602410716e-06,
      "loss": 2.7665,
      "step": 4080
    },
    {
      "epoch": 1.39093387866394,
      "grad_norm": 15.681191988725045,
      "learning_rate": 6.487260401700713e-06,
      "loss": 2.4018,
      "step": 4081
    },
    {
      "epoch": 1.3912747102931151,
      "grad_norm": 15.123392258219752,
      "learning_rate": 6.485366967038729e-06,
      "loss": 2.5182,
      "step": 4082
    },
    {
      "epoch": 1.3916155419222904,
      "grad_norm": 19.742107821568283,
      "learning_rate": 6.4834732987226065e-06,
      "loss": 2.185,
      "step": 4083
    },
    {
      "epoch": 1.3919563735514655,
      "grad_norm": 14.659445679105813,
      "learning_rate": 6.4815793970502295e-06,
      "loss": 1.7686,
      "step": 4084
    },
    {
      "epoch": 1.3922972051806408,
      "grad_norm": 21.95513690617308,
      "learning_rate": 6.479685262319516e-06,
      "loss": 2.5647,
      "step": 4085
    },
    {
      "epoch": 1.392638036809816,
      "grad_norm": 23.005480644990797,
      "learning_rate": 6.477790894828422e-06,
      "loss": 2.4519,
      "step": 4086
    },
    {
      "epoch": 1.3929788684389912,
      "grad_norm": 25.810367871797613,
      "learning_rate": 6.475896294874935e-06,
      "loss": 1.7914,
      "step": 4087
    },
    {
      "epoch": 1.3933197000681663,
      "grad_norm": 13.445514289278618,
      "learning_rate": 6.474001462757087e-06,
      "loss": 2.4808,
      "step": 4088
    },
    {
      "epoch": 1.3936605316973414,
      "grad_norm": 12.119944412943243,
      "learning_rate": 6.4721063987729395e-06,
      "loss": 1.9577,
      "step": 4089
    },
    {
      "epoch": 1.3940013633265167,
      "grad_norm": 19.149250303003743,
      "learning_rate": 6.470211103220596e-06,
      "loss": 2.6152,
      "step": 4090
    },
    {
      "epoch": 1.394342194955692,
      "grad_norm": 15.937716949155401,
      "learning_rate": 6.4683155763981946e-06,
      "loss": 2.748,
      "step": 4091
    },
    {
      "epoch": 1.394683026584867,
      "grad_norm": 13.322521154554362,
      "learning_rate": 6.4664198186039075e-06,
      "loss": 2.3051,
      "step": 4092
    },
    {
      "epoch": 1.3950238582140422,
      "grad_norm": 34.771514194464906,
      "learning_rate": 6.4645238301359445e-06,
      "loss": 1.9937,
      "step": 4093
    },
    {
      "epoch": 1.3953646898432175,
      "grad_norm": 14.464032841418115,
      "learning_rate": 6.462627611292556e-06,
      "loss": 2.4611,
      "step": 4094
    },
    {
      "epoch": 1.3957055214723926,
      "grad_norm": 12.120505301708688,
      "learning_rate": 6.4607311623720186e-06,
      "loss": 1.5425,
      "step": 4095
    },
    {
      "epoch": 1.3960463531015679,
      "grad_norm": 15.231063034211667,
      "learning_rate": 6.458834483672655e-06,
      "loss": 2.3392,
      "step": 4096
    },
    {
      "epoch": 1.396387184730743,
      "grad_norm": 29.135546116577828,
      "learning_rate": 6.456937575492821e-06,
      "loss": 2.8971,
      "step": 4097
    },
    {
      "epoch": 1.3967280163599183,
      "grad_norm": 17.201598008768457,
      "learning_rate": 6.455040438130906e-06,
      "loss": 2.7372,
      "step": 4098
    },
    {
      "epoch": 1.3970688479890934,
      "grad_norm": 14.106023149445445,
      "learning_rate": 6.453143071885337e-06,
      "loss": 2.5583,
      "step": 4099
    },
    {
      "epoch": 1.3974096796182685,
      "grad_norm": 13.251687855890035,
      "learning_rate": 6.451245477054579e-06,
      "loss": 2.0171,
      "step": 4100
    },
    {
      "epoch": 1.3977505112474438,
      "grad_norm": 17.61178936693371,
      "learning_rate": 6.449347653937131e-06,
      "loss": 2.3619,
      "step": 4101
    },
    {
      "epoch": 1.398091342876619,
      "grad_norm": 18.694930775981494,
      "learning_rate": 6.447449602831527e-06,
      "loss": 2.5008,
      "step": 4102
    },
    {
      "epoch": 1.3984321745057942,
      "grad_norm": 19.209663012577874,
      "learning_rate": 6.44555132403634e-06,
      "loss": 2.062,
      "step": 4103
    },
    {
      "epoch": 1.3987730061349692,
      "grad_norm": 18.24571456352589,
      "learning_rate": 6.443652817850173e-06,
      "loss": 2.5617,
      "step": 4104
    },
    {
      "epoch": 1.3991138377641446,
      "grad_norm": 14.06117034829641,
      "learning_rate": 6.441754084571672e-06,
      "loss": 2.27,
      "step": 4105
    },
    {
      "epoch": 1.3994546693933196,
      "grad_norm": 18.658003717348084,
      "learning_rate": 6.439855124499514e-06,
      "loss": 2.2428,
      "step": 4106
    },
    {
      "epoch": 1.399795501022495,
      "grad_norm": 16.510613573780113,
      "learning_rate": 6.437955937932413e-06,
      "loss": 1.8777,
      "step": 4107
    },
    {
      "epoch": 1.40013633265167,
      "grad_norm": 15.16941048725686,
      "learning_rate": 6.4360565251691184e-06,
      "loss": 1.909,
      "step": 4108
    },
    {
      "epoch": 1.4004771642808453,
      "grad_norm": 14.247384478164912,
      "learning_rate": 6.434156886508417e-06,
      "loss": 2.4311,
      "step": 4109
    },
    {
      "epoch": 1.4008179959100204,
      "grad_norm": 20.163006926665293,
      "learning_rate": 6.432257022249128e-06,
      "loss": 2.3731,
      "step": 4110
    },
    {
      "epoch": 1.4011588275391955,
      "grad_norm": 12.611694157383754,
      "learning_rate": 6.430356932690109e-06,
      "loss": 2.2791,
      "step": 4111
    },
    {
      "epoch": 1.4014996591683708,
      "grad_norm": 12.608135200737784,
      "learning_rate": 6.428456618130248e-06,
      "loss": 2.0172,
      "step": 4112
    },
    {
      "epoch": 1.4018404907975461,
      "grad_norm": 12.1409340760151,
      "learning_rate": 6.426556078868475e-06,
      "loss": 2.4051,
      "step": 4113
    },
    {
      "epoch": 1.4021813224267212,
      "grad_norm": 17.848624619061592,
      "learning_rate": 6.424655315203754e-06,
      "loss": 2.1296,
      "step": 4114
    },
    {
      "epoch": 1.4025221540558963,
      "grad_norm": 16.416457531910574,
      "learning_rate": 6.422754327435079e-06,
      "loss": 2.4566,
      "step": 4115
    },
    {
      "epoch": 1.4028629856850716,
      "grad_norm": 18.817352908618325,
      "learning_rate": 6.420853115861485e-06,
      "loss": 2.4618,
      "step": 4116
    },
    {
      "epoch": 1.4032038173142467,
      "grad_norm": 23.18760941071446,
      "learning_rate": 6.418951680782039e-06,
      "loss": 2.5911,
      "step": 4117
    },
    {
      "epoch": 1.403544648943422,
      "grad_norm": 15.039375027779045,
      "learning_rate": 6.4170500224958464e-06,
      "loss": 2.3492,
      "step": 4118
    },
    {
      "epoch": 1.403885480572597,
      "grad_norm": 18.429429967019285,
      "learning_rate": 6.415148141302043e-06,
      "loss": 2.0229,
      "step": 4119
    },
    {
      "epoch": 1.4042263122017724,
      "grad_norm": 14.230769749092802,
      "learning_rate": 6.413246037499806e-06,
      "loss": 2.01,
      "step": 4120
    },
    {
      "epoch": 1.4045671438309475,
      "grad_norm": 23.54049759151644,
      "learning_rate": 6.411343711388341e-06,
      "loss": 2.2121,
      "step": 4121
    },
    {
      "epoch": 1.4049079754601226,
      "grad_norm": 15.64969096467556,
      "learning_rate": 6.409441163266889e-06,
      "loss": 2.141,
      "step": 4122
    },
    {
      "epoch": 1.405248807089298,
      "grad_norm": 19.55806493353418,
      "learning_rate": 6.4075383934347345e-06,
      "loss": 2.2184,
      "step": 4123
    },
    {
      "epoch": 1.4055896387184732,
      "grad_norm": 21.745485225788897,
      "learning_rate": 6.405635402191186e-06,
      "loss": 1.9559,
      "step": 4124
    },
    {
      "epoch": 1.4059304703476483,
      "grad_norm": 20.175399725427596,
      "learning_rate": 6.403732189835595e-06,
      "loss": 2.6058,
      "step": 4125
    },
    {
      "epoch": 1.4062713019768234,
      "grad_norm": 17.92500363332867,
      "learning_rate": 6.40182875666734e-06,
      "loss": 2.3318,
      "step": 4126
    },
    {
      "epoch": 1.4066121336059987,
      "grad_norm": 15.682507790664987,
      "learning_rate": 6.399925102985842e-06,
      "loss": 2.3658,
      "step": 4127
    },
    {
      "epoch": 1.4069529652351738,
      "grad_norm": 19.114881402399035,
      "learning_rate": 6.398021229090554e-06,
      "loss": 2.1565,
      "step": 4128
    },
    {
      "epoch": 1.407293796864349,
      "grad_norm": 16.004587300449412,
      "learning_rate": 6.3961171352809605e-06,
      "loss": 2.281,
      "step": 4129
    },
    {
      "epoch": 1.4076346284935242,
      "grad_norm": 12.614857262550869,
      "learning_rate": 6.394212821856583e-06,
      "loss": 1.9879,
      "step": 4130
    },
    {
      "epoch": 1.4079754601226995,
      "grad_norm": 13.41797626183804,
      "learning_rate": 6.39230828911698e-06,
      "loss": 2.3234,
      "step": 4131
    },
    {
      "epoch": 1.4083162917518746,
      "grad_norm": 13.342986940035779,
      "learning_rate": 6.39040353736174e-06,
      "loss": 2.2872,
      "step": 4132
    },
    {
      "epoch": 1.4086571233810496,
      "grad_norm": 14.750294524433418,
      "learning_rate": 6.388498566890488e-06,
      "loss": 2.2263,
      "step": 4133
    },
    {
      "epoch": 1.408997955010225,
      "grad_norm": 18.953844613403813,
      "learning_rate": 6.386593378002885e-06,
      "loss": 1.6709,
      "step": 4134
    },
    {
      "epoch": 1.4093387866394003,
      "grad_norm": 20.982031110468785,
      "learning_rate": 6.384687970998625e-06,
      "loss": 2.2389,
      "step": 4135
    },
    {
      "epoch": 1.4096796182685754,
      "grad_norm": 352.22498613594826,
      "learning_rate": 6.382782346177434e-06,
      "loss": 2.3374,
      "step": 4136
    },
    {
      "epoch": 1.4100204498977504,
      "grad_norm": 22.00672607503255,
      "learning_rate": 6.380876503839076e-06,
      "loss": 2.0881,
      "step": 4137
    },
    {
      "epoch": 1.4103612815269257,
      "grad_norm": 16.25052418380546,
      "learning_rate": 6.378970444283345e-06,
      "loss": 2.1543,
      "step": 4138
    },
    {
      "epoch": 1.4107021131561008,
      "grad_norm": 21.62533553118729,
      "learning_rate": 6.3770641678100765e-06,
      "loss": 2.2178,
      "step": 4139
    },
    {
      "epoch": 1.4110429447852761,
      "grad_norm": 13.759621322740271,
      "learning_rate": 6.375157674719131e-06,
      "loss": 2.5971,
      "step": 4140
    },
    {
      "epoch": 1.4113837764144512,
      "grad_norm": 14.579761079181587,
      "learning_rate": 6.373250965310409e-06,
      "loss": 2.4856,
      "step": 4141
    },
    {
      "epoch": 1.4117246080436265,
      "grad_norm": 14.074023293999941,
      "learning_rate": 6.371344039883845e-06,
      "loss": 1.9761,
      "step": 4142
    },
    {
      "epoch": 1.4120654396728016,
      "grad_norm": 22.726269742860328,
      "learning_rate": 6.369436898739403e-06,
      "loss": 2.0237,
      "step": 4143
    },
    {
      "epoch": 1.4124062713019767,
      "grad_norm": 22.528727466340133,
      "learning_rate": 6.367529542177085e-06,
      "loss": 2.9489,
      "step": 4144
    },
    {
      "epoch": 1.412747102931152,
      "grad_norm": 14.825812036375087,
      "learning_rate": 6.365621970496929e-06,
      "loss": 2.2372,
      "step": 4145
    },
    {
      "epoch": 1.4130879345603273,
      "grad_norm": 15.416948315380447,
      "learning_rate": 6.363714183998998e-06,
      "loss": 2.0254,
      "step": 4146
    },
    {
      "epoch": 1.4134287661895024,
      "grad_norm": 14.698314649298018,
      "learning_rate": 6.361806182983396e-06,
      "loss": 1.8583,
      "step": 4147
    },
    {
      "epoch": 1.4137695978186775,
      "grad_norm": 18.45917086961421,
      "learning_rate": 6.359897967750265e-06,
      "loss": 2.0,
      "step": 4148
    },
    {
      "epoch": 1.4141104294478528,
      "grad_norm": 29.24096146733855,
      "learning_rate": 6.357989538599767e-06,
      "loss": 2.4595,
      "step": 4149
    },
    {
      "epoch": 1.414451261077028,
      "grad_norm": 10.253272857884088,
      "learning_rate": 6.35608089583211e-06,
      "loss": 1.7902,
      "step": 4150
    },
    {
      "epoch": 1.4147920927062032,
      "grad_norm": 22.76363082770309,
      "learning_rate": 6.354172039747529e-06,
      "loss": 2.4769,
      "step": 4151
    },
    {
      "epoch": 1.4151329243353783,
      "grad_norm": 14.634767613237905,
      "learning_rate": 6.352262970646296e-06,
      "loss": 2.0706,
      "step": 4152
    },
    {
      "epoch": 1.4154737559645536,
      "grad_norm": 14.680772971891189,
      "learning_rate": 6.350353688828714e-06,
      "loss": 2.0739,
      "step": 4153
    },
    {
      "epoch": 1.4158145875937287,
      "grad_norm": 14.720155225670668,
      "learning_rate": 6.348444194595123e-06,
      "loss": 2.6389,
      "step": 4154
    },
    {
      "epoch": 1.4161554192229038,
      "grad_norm": 12.181347793025616,
      "learning_rate": 6.34653448824589e-06,
      "loss": 2.2509,
      "step": 4155
    },
    {
      "epoch": 1.416496250852079,
      "grad_norm": 17.994703425404,
      "learning_rate": 6.344624570081424e-06,
      "loss": 2.3743,
      "step": 4156
    },
    {
      "epoch": 1.4168370824812544,
      "grad_norm": 19.367981681632944,
      "learning_rate": 6.342714440402158e-06,
      "loss": 2.241,
      "step": 4157
    },
    {
      "epoch": 1.4171779141104295,
      "grad_norm": 24.063527378896463,
      "learning_rate": 6.340804099508567e-06,
      "loss": 2.4286,
      "step": 4158
    },
    {
      "epoch": 1.4175187457396046,
      "grad_norm": 28.305408740774908,
      "learning_rate": 6.3388935477011535e-06,
      "loss": 2.6028,
      "step": 4159
    },
    {
      "epoch": 1.4178595773687799,
      "grad_norm": 15.262134528043909,
      "learning_rate": 6.336982785280455e-06,
      "loss": 2.2715,
      "step": 4160
    },
    {
      "epoch": 1.418200408997955,
      "grad_norm": 16.111081824053684,
      "learning_rate": 6.335071812547042e-06,
      "loss": 1.7806,
      "step": 4161
    },
    {
      "epoch": 1.4185412406271303,
      "grad_norm": 35.480247532982006,
      "learning_rate": 6.33316062980152e-06,
      "loss": 2.0574,
      "step": 4162
    },
    {
      "epoch": 1.4188820722563054,
      "grad_norm": 19.60758558617339,
      "learning_rate": 6.3312492373445224e-06,
      "loss": 2.4616,
      "step": 4163
    },
    {
      "epoch": 1.4192229038854807,
      "grad_norm": 24.902592795990603,
      "learning_rate": 6.32933763547672e-06,
      "loss": 2.2607,
      "step": 4164
    },
    {
      "epoch": 1.4195637355146558,
      "grad_norm": 17.778563612495017,
      "learning_rate": 6.327425824498816e-06,
      "loss": 2.1846,
      "step": 4165
    },
    {
      "epoch": 1.4199045671438308,
      "grad_norm": 11.81758538025014,
      "learning_rate": 6.325513804711546e-06,
      "loss": 1.9197,
      "step": 4166
    },
    {
      "epoch": 1.4202453987730062,
      "grad_norm": 33.00935052970937,
      "learning_rate": 6.323601576415678e-06,
      "loss": 2.0023,
      "step": 4167
    },
    {
      "epoch": 1.4205862304021815,
      "grad_norm": 14.208665203697588,
      "learning_rate": 6.321689139912013e-06,
      "loss": 2.0178,
      "step": 4168
    },
    {
      "epoch": 1.4209270620313565,
      "grad_norm": 16.816914787292436,
      "learning_rate": 6.319776495501384e-06,
      "loss": 1.8852,
      "step": 4169
    },
    {
      "epoch": 1.4212678936605316,
      "grad_norm": 15.407702196022148,
      "learning_rate": 6.317863643484661e-06,
      "loss": 2.6873,
      "step": 4170
    },
    {
      "epoch": 1.421608725289707,
      "grad_norm": 14.719541047486363,
      "learning_rate": 6.315950584162739e-06,
      "loss": 1.6438,
      "step": 4171
    },
    {
      "epoch": 1.421949556918882,
      "grad_norm": 15.973866953295847,
      "learning_rate": 6.31403731783655e-06,
      "loss": 2.408,
      "step": 4172
    },
    {
      "epoch": 1.4222903885480573,
      "grad_norm": 17.527285050790386,
      "learning_rate": 6.312123844807063e-06,
      "loss": 2.5421,
      "step": 4173
    },
    {
      "epoch": 1.4226312201772324,
      "grad_norm": 17.50606732492775,
      "learning_rate": 6.310210165375271e-06,
      "loss": 2.2525,
      "step": 4174
    },
    {
      "epoch": 1.4229720518064077,
      "grad_norm": 15.138694469002937,
      "learning_rate": 6.308296279842204e-06,
      "loss": 2.6332,
      "step": 4175
    },
    {
      "epoch": 1.4233128834355828,
      "grad_norm": 10.92006462807346,
      "learning_rate": 6.306382188508924e-06,
      "loss": 2.3816,
      "step": 4176
    },
    {
      "epoch": 1.423653715064758,
      "grad_norm": 16.6847621628487,
      "learning_rate": 6.304467891676525e-06,
      "loss": 2.3162,
      "step": 4177
    },
    {
      "epoch": 1.4239945466939332,
      "grad_norm": 18.386561194723605,
      "learning_rate": 6.302553389646135e-06,
      "loss": 2.5523,
      "step": 4178
    },
    {
      "epoch": 1.4243353783231083,
      "grad_norm": 13.787999506991124,
      "learning_rate": 6.300638682718911e-06,
      "loss": 2.6583,
      "step": 4179
    },
    {
      "epoch": 1.4246762099522836,
      "grad_norm": 30.698395412153527,
      "learning_rate": 6.298723771196046e-06,
      "loss": 2.3897,
      "step": 4180
    },
    {
      "epoch": 1.4250170415814587,
      "grad_norm": 16.967694134526244,
      "learning_rate": 6.2968086553787586e-06,
      "loss": 2.3993,
      "step": 4181
    },
    {
      "epoch": 1.425357873210634,
      "grad_norm": 24.62544892765898,
      "learning_rate": 6.2948933355683105e-06,
      "loss": 2.3652,
      "step": 4182
    },
    {
      "epoch": 1.425698704839809,
      "grad_norm": 22.460790262252434,
      "learning_rate": 6.292977812065986e-06,
      "loss": 1.8015,
      "step": 4183
    },
    {
      "epoch": 1.4260395364689844,
      "grad_norm": 10.554610217745585,
      "learning_rate": 6.291062085173103e-06,
      "loss": 2.5285,
      "step": 4184
    },
    {
      "epoch": 1.4263803680981595,
      "grad_norm": 20.87489133084229,
      "learning_rate": 6.289146155191015e-06,
      "loss": 1.9355,
      "step": 4185
    },
    {
      "epoch": 1.4267211997273348,
      "grad_norm": 14.359916100258236,
      "learning_rate": 6.2872300224211045e-06,
      "loss": 2.2342,
      "step": 4186
    },
    {
      "epoch": 1.4270620313565099,
      "grad_norm": 14.163856174852588,
      "learning_rate": 6.285313687164789e-06,
      "loss": 1.656,
      "step": 4187
    },
    {
      "epoch": 1.427402862985685,
      "grad_norm": 20.08092368138605,
      "learning_rate": 6.283397149723514e-06,
      "loss": 2.6091,
      "step": 4188
    },
    {
      "epoch": 1.4277436946148603,
      "grad_norm": 16.85974387576498,
      "learning_rate": 6.281480410398755e-06,
      "loss": 2.333,
      "step": 4189
    },
    {
      "epoch": 1.4280845262440354,
      "grad_norm": 21.511297075794307,
      "learning_rate": 6.279563469492029e-06,
      "loss": 2.5258,
      "step": 4190
    },
    {
      "epoch": 1.4284253578732107,
      "grad_norm": 12.692910719790788,
      "learning_rate": 6.277646327304876e-06,
      "loss": 2.1101,
      "step": 4191
    },
    {
      "epoch": 1.4287661895023858,
      "grad_norm": 24.867981489883274,
      "learning_rate": 6.275728984138869e-06,
      "loss": 1.9843,
      "step": 4192
    },
    {
      "epoch": 1.429107021131561,
      "grad_norm": 25.95256324955016,
      "learning_rate": 6.2738114402956145e-06,
      "loss": 2.5161,
      "step": 4193
    },
    {
      "epoch": 1.4294478527607362,
      "grad_norm": 15.191864365946055,
      "learning_rate": 6.271893696076749e-06,
      "loss": 2.4124,
      "step": 4194
    },
    {
      "epoch": 1.4297886843899112,
      "grad_norm": 14.332128678671753,
      "learning_rate": 6.2699757517839425e-06,
      "loss": 1.7101,
      "step": 4195
    },
    {
      "epoch": 1.4301295160190866,
      "grad_norm": 28.87104246552696,
      "learning_rate": 6.268057607718895e-06,
      "loss": 2.4512,
      "step": 4196
    },
    {
      "epoch": 1.4304703476482619,
      "grad_norm": 18.251559084713737,
      "learning_rate": 6.266139264183336e-06,
      "loss": 2.158,
      "step": 4197
    },
    {
      "epoch": 1.430811179277437,
      "grad_norm": 13.43547482878226,
      "learning_rate": 6.264220721479033e-06,
      "loss": 2.475,
      "step": 4198
    },
    {
      "epoch": 1.431152010906612,
      "grad_norm": 19.84823892724352,
      "learning_rate": 6.262301979907778e-06,
      "loss": 2.691,
      "step": 4199
    },
    {
      "epoch": 1.4314928425357873,
      "grad_norm": 13.209139208067842,
      "learning_rate": 6.260383039771395e-06,
      "loss": 2.122,
      "step": 4200
    },
    {
      "epoch": 1.4318336741649624,
      "grad_norm": 13.331149417705998,
      "learning_rate": 6.258463901371743e-06,
      "loss": 2.3425,
      "step": 4201
    },
    {
      "epoch": 1.4321745057941377,
      "grad_norm": 14.860728690564288,
      "learning_rate": 6.256544565010711e-06,
      "loss": 1.9794,
      "step": 4202
    },
    {
      "epoch": 1.4325153374233128,
      "grad_norm": 17.35000519217022,
      "learning_rate": 6.254625030990216e-06,
      "loss": 1.4361,
      "step": 4203
    },
    {
      "epoch": 1.4328561690524881,
      "grad_norm": 15.692009710966701,
      "learning_rate": 6.252705299612209e-06,
      "loss": 2.5015,
      "step": 4204
    },
    {
      "epoch": 1.4331970006816632,
      "grad_norm": 20.874088560140365,
      "learning_rate": 6.250785371178671e-06,
      "loss": 1.75,
      "step": 4205
    },
    {
      "epoch": 1.4335378323108383,
      "grad_norm": 17.885714389417462,
      "learning_rate": 6.248865245991615e-06,
      "loss": 1.8058,
      "step": 4206
    },
    {
      "epoch": 1.4338786639400136,
      "grad_norm": 16.59138856263439,
      "learning_rate": 6.246944924353087e-06,
      "loss": 2.3612,
      "step": 4207
    },
    {
      "epoch": 1.434219495569189,
      "grad_norm": 26.265938065698933,
      "learning_rate": 6.245024406565157e-06,
      "loss": 3.1253,
      "step": 4208
    },
    {
      "epoch": 1.434560327198364,
      "grad_norm": 30.418837941984698,
      "learning_rate": 6.24310369292993e-06,
      "loss": 2.8319,
      "step": 4209
    },
    {
      "epoch": 1.434901158827539,
      "grad_norm": 15.492970370052923,
      "learning_rate": 6.241182783749545e-06,
      "loss": 2.4145,
      "step": 4210
    },
    {
      "epoch": 1.4352419904567144,
      "grad_norm": 14.654970623711524,
      "learning_rate": 6.239261679326167e-06,
      "loss": 2.386,
      "step": 4211
    },
    {
      "epoch": 1.4355828220858895,
      "grad_norm": 15.581792357055546,
      "learning_rate": 6.237340379961994e-06,
      "loss": 2.4502,
      "step": 4212
    },
    {
      "epoch": 1.4359236537150648,
      "grad_norm": 22.66362302585363,
      "learning_rate": 6.235418885959255e-06,
      "loss": 2.7681,
      "step": 4213
    },
    {
      "epoch": 1.43626448534424,
      "grad_norm": 13.330091029818776,
      "learning_rate": 6.233497197620206e-06,
      "loss": 1.9599,
      "step": 4214
    },
    {
      "epoch": 1.4366053169734152,
      "grad_norm": 15.905976949849602,
      "learning_rate": 6.2315753152471355e-06,
      "loss": 2.5152,
      "step": 4215
    },
    {
      "epoch": 1.4369461486025903,
      "grad_norm": 15.054554940467366,
      "learning_rate": 6.2296532391423694e-06,
      "loss": 1.906,
      "step": 4216
    },
    {
      "epoch": 1.4372869802317654,
      "grad_norm": 22.857127906898214,
      "learning_rate": 6.227730969608251e-06,
      "loss": 2.1621,
      "step": 4217
    },
    {
      "epoch": 1.4376278118609407,
      "grad_norm": 14.427961984621895,
      "learning_rate": 6.225808506947164e-06,
      "loss": 2.1973,
      "step": 4218
    },
    {
      "epoch": 1.437968643490116,
      "grad_norm": 15.03534402302923,
      "learning_rate": 6.22388585146152e-06,
      "loss": 2.1455,
      "step": 4219
    },
    {
      "epoch": 1.438309475119291,
      "grad_norm": 24.40634167255601,
      "learning_rate": 6.221963003453759e-06,
      "loss": 2.9244,
      "step": 4220
    },
    {
      "epoch": 1.4386503067484662,
      "grad_norm": 22.047401670869874,
      "learning_rate": 6.220039963226354e-06,
      "loss": 2.0966,
      "step": 4221
    },
    {
      "epoch": 1.4389911383776415,
      "grad_norm": 21.34540651239847,
      "learning_rate": 6.218116731081806e-06,
      "loss": 1.5897,
      "step": 4222
    },
    {
      "epoch": 1.4393319700068166,
      "grad_norm": 22.059410684659365,
      "learning_rate": 6.216193307322645e-06,
      "loss": 2.2273,
      "step": 4223
    },
    {
      "epoch": 1.4396728016359919,
      "grad_norm": 16.436266967787553,
      "learning_rate": 6.214269692251438e-06,
      "loss": 1.9336,
      "step": 4224
    },
    {
      "epoch": 1.440013633265167,
      "grad_norm": 15.98754581407864,
      "learning_rate": 6.212345886170774e-06,
      "loss": 2.0265,
      "step": 4225
    },
    {
      "epoch": 1.4403544648943423,
      "grad_norm": 23.0872174940283,
      "learning_rate": 6.210421889383276e-06,
      "loss": 2.2666,
      "step": 4226
    },
    {
      "epoch": 1.4406952965235174,
      "grad_norm": 16.52843861759327,
      "learning_rate": 6.208497702191595e-06,
      "loss": 2.8351,
      "step": 4227
    },
    {
      "epoch": 1.4410361281526924,
      "grad_norm": 22.298666109855876,
      "learning_rate": 6.206573324898416e-06,
      "loss": 2.2592,
      "step": 4228
    },
    {
      "epoch": 1.4413769597818678,
      "grad_norm": 14.978301648140484,
      "learning_rate": 6.2046487578064495e-06,
      "loss": 1.9295,
      "step": 4229
    },
    {
      "epoch": 1.441717791411043,
      "grad_norm": 17.90480972558633,
      "learning_rate": 6.202724001218439e-06,
      "loss": 2.1672,
      "step": 4230
    },
    {
      "epoch": 1.4420586230402181,
      "grad_norm": 12.585361350641351,
      "learning_rate": 6.200799055437153e-06,
      "loss": 2.2879,
      "step": 4231
    },
    {
      "epoch": 1.4423994546693932,
      "grad_norm": 20.109374456354946,
      "learning_rate": 6.198873920765397e-06,
      "loss": 2.4619,
      "step": 4232
    },
    {
      "epoch": 1.4427402862985685,
      "grad_norm": 13.080141725853899,
      "learning_rate": 6.196948597506002e-06,
      "loss": 2.164,
      "step": 4233
    },
    {
      "epoch": 1.4430811179277436,
      "grad_norm": 17.755071481631937,
      "learning_rate": 6.195023085961827e-06,
      "loss": 2.4733,
      "step": 4234
    },
    {
      "epoch": 1.443421949556919,
      "grad_norm": 16.867412611086863,
      "learning_rate": 6.1930973864357625e-06,
      "loss": 2.06,
      "step": 4235
    },
    {
      "epoch": 1.443762781186094,
      "grad_norm": 22.103999448089247,
      "learning_rate": 6.19117149923073e-06,
      "loss": 2.2865,
      "step": 4236
    },
    {
      "epoch": 1.4441036128152693,
      "grad_norm": 18.21101645243261,
      "learning_rate": 6.189245424649679e-06,
      "loss": 1.973,
      "step": 4237
    },
    {
      "epoch": 1.4444444444444444,
      "grad_norm": 15.2564358912888,
      "learning_rate": 6.18731916299559e-06,
      "loss": 2.0041,
      "step": 4238
    },
    {
      "epoch": 1.4447852760736195,
      "grad_norm": 17.879490826712384,
      "learning_rate": 6.1853927145714674e-06,
      "loss": 2.718,
      "step": 4239
    },
    {
      "epoch": 1.4451261077027948,
      "grad_norm": 20.174451687470043,
      "learning_rate": 6.1834660796803534e-06,
      "loss": 1.5841,
      "step": 4240
    },
    {
      "epoch": 1.4454669393319701,
      "grad_norm": 12.987564516108856,
      "learning_rate": 6.181539258625315e-06,
      "loss": 2.6677,
      "step": 4241
    },
    {
      "epoch": 1.4458077709611452,
      "grad_norm": 20.09110286861504,
      "learning_rate": 6.179612251709448e-06,
      "loss": 2.0094,
      "step": 4242
    },
    {
      "epoch": 1.4461486025903203,
      "grad_norm": 12.987812387415309,
      "learning_rate": 6.177685059235876e-06,
      "loss": 2.1286,
      "step": 4243
    },
    {
      "epoch": 1.4464894342194956,
      "grad_norm": 14.3545352644821,
      "learning_rate": 6.175757681507758e-06,
      "loss": 2.3371,
      "step": 4244
    },
    {
      "epoch": 1.4468302658486707,
      "grad_norm": 22.288155204113032,
      "learning_rate": 6.173830118828275e-06,
      "loss": 2.8501,
      "step": 4245
    },
    {
      "epoch": 1.447171097477846,
      "grad_norm": 29.64320304280237,
      "learning_rate": 6.171902371500642e-06,
      "loss": 2.8582,
      "step": 4246
    },
    {
      "epoch": 1.447511929107021,
      "grad_norm": 19.564205282650125,
      "learning_rate": 6.169974439828101e-06,
      "loss": 2.3193,
      "step": 4247
    },
    {
      "epoch": 1.4478527607361964,
      "grad_norm": 29.452910982062246,
      "learning_rate": 6.1680463241139225e-06,
      "loss": 2.0912,
      "step": 4248
    },
    {
      "epoch": 1.4481935923653715,
      "grad_norm": 17.1178976348185,
      "learning_rate": 6.166118024661407e-06,
      "loss": 2.25,
      "step": 4249
    },
    {
      "epoch": 1.4485344239945466,
      "grad_norm": 15.664826094723566,
      "learning_rate": 6.164189541773886e-06,
      "loss": 1.5707,
      "step": 4250
    },
    {
      "epoch": 1.4488752556237219,
      "grad_norm": 12.220811830922289,
      "learning_rate": 6.162260875754716e-06,
      "loss": 2.1177,
      "step": 4251
    },
    {
      "epoch": 1.4492160872528972,
      "grad_norm": 23.794005563244202,
      "learning_rate": 6.160332026907282e-06,
      "loss": 1.8313,
      "step": 4252
    },
    {
      "epoch": 1.4495569188820723,
      "grad_norm": 13.6298634651217,
      "learning_rate": 6.158402995535001e-06,
      "loss": 2.3747,
      "step": 4253
    },
    {
      "epoch": 1.4498977505112474,
      "grad_norm": 15.813918315891687,
      "learning_rate": 6.156473781941319e-06,
      "loss": 2.2417,
      "step": 4254
    },
    {
      "epoch": 1.4502385821404227,
      "grad_norm": 13.47904160091913,
      "learning_rate": 6.154544386429707e-06,
      "loss": 2.0697,
      "step": 4255
    },
    {
      "epoch": 1.4505794137695978,
      "grad_norm": 23.011362407020368,
      "learning_rate": 6.1526148093036656e-06,
      "loss": 2.2636,
      "step": 4256
    },
    {
      "epoch": 1.450920245398773,
      "grad_norm": 22.889794017241012,
      "learning_rate": 6.150685050866727e-06,
      "loss": 2.0932,
      "step": 4257
    },
    {
      "epoch": 1.4512610770279482,
      "grad_norm": 30.82330838810119,
      "learning_rate": 6.148755111422451e-06,
      "loss": 2.3957,
      "step": 4258
    },
    {
      "epoch": 1.4516019086571235,
      "grad_norm": 23.935686965115497,
      "learning_rate": 6.146824991274421e-06,
      "loss": 2.7841,
      "step": 4259
    },
    {
      "epoch": 1.4519427402862985,
      "grad_norm": 17.61780993992046,
      "learning_rate": 6.144894690726256e-06,
      "loss": 2.561,
      "step": 4260
    },
    {
      "epoch": 1.4522835719154736,
      "grad_norm": 242.80002039823913,
      "learning_rate": 6.142964210081597e-06,
      "loss": 2.0443,
      "step": 4261
    },
    {
      "epoch": 1.452624403544649,
      "grad_norm": 11.70289154237094,
      "learning_rate": 6.141033549644118e-06,
      "loss": 1.9113,
      "step": 4262
    },
    {
      "epoch": 1.4529652351738243,
      "grad_norm": 21.911882038558463,
      "learning_rate": 6.1391027097175205e-06,
      "loss": 2.6845,
      "step": 4263
    },
    {
      "epoch": 1.4533060668029993,
      "grad_norm": 16.62583260436707,
      "learning_rate": 6.1371716906055336e-06,
      "loss": 2.2245,
      "step": 4264
    },
    {
      "epoch": 1.4536468984321744,
      "grad_norm": 22.64149010667539,
      "learning_rate": 6.13524049261191e-06,
      "loss": 2.8715,
      "step": 4265
    },
    {
      "epoch": 1.4539877300613497,
      "grad_norm": 14.089539580086985,
      "learning_rate": 6.133309116040439e-06,
      "loss": 2.1364,
      "step": 4266
    },
    {
      "epoch": 1.4543285616905248,
      "grad_norm": 14.505852779115182,
      "learning_rate": 6.131377561194933e-06,
      "loss": 1.5217,
      "step": 4267
    },
    {
      "epoch": 1.4546693933197001,
      "grad_norm": 17.185176707948177,
      "learning_rate": 6.129445828379233e-06,
      "loss": 2.7124,
      "step": 4268
    },
    {
      "epoch": 1.4550102249488752,
      "grad_norm": 14.290284085140701,
      "learning_rate": 6.1275139178972075e-06,
      "loss": 2.3308,
      "step": 4269
    },
    {
      "epoch": 1.4553510565780505,
      "grad_norm": 19.096161598925374,
      "learning_rate": 6.125581830052755e-06,
      "loss": 1.9929,
      "step": 4270
    },
    {
      "epoch": 1.4556918882072256,
      "grad_norm": 18.12013683008587,
      "learning_rate": 6.123649565149799e-06,
      "loss": 2.4343,
      "step": 4271
    },
    {
      "epoch": 1.4560327198364007,
      "grad_norm": 13.038962764134952,
      "learning_rate": 6.1217171234922944e-06,
      "loss": 2.2847,
      "step": 4272
    },
    {
      "epoch": 1.456373551465576,
      "grad_norm": 17.360416312255598,
      "learning_rate": 6.1197845053842195e-06,
      "loss": 2.2635,
      "step": 4273
    },
    {
      "epoch": 1.4567143830947513,
      "grad_norm": 25.484879142860926,
      "learning_rate": 6.117851711129585e-06,
      "loss": 2.7283,
      "step": 4274
    },
    {
      "epoch": 1.4570552147239264,
      "grad_norm": 29.376893548359316,
      "learning_rate": 6.115918741032428e-06,
      "loss": 2.437,
      "step": 4275
    },
    {
      "epoch": 1.4573960463531015,
      "grad_norm": 16.16409466806491,
      "learning_rate": 6.113985595396809e-06,
      "loss": 2.2009,
      "step": 4276
    },
    {
      "epoch": 1.4577368779822768,
      "grad_norm": 13.56511507686784,
      "learning_rate": 6.112052274526823e-06,
      "loss": 2.1692,
      "step": 4277
    },
    {
      "epoch": 1.4580777096114519,
      "grad_norm": 13.535814029694661,
      "learning_rate": 6.110118778726586e-06,
      "loss": 2.0236,
      "step": 4278
    },
    {
      "epoch": 1.4584185412406272,
      "grad_norm": 25.073024018092053,
      "learning_rate": 6.108185108300245e-06,
      "loss": 1.8582,
      "step": 4279
    },
    {
      "epoch": 1.4587593728698023,
      "grad_norm": 12.572283803837419,
      "learning_rate": 6.106251263551977e-06,
      "loss": 1.88,
      "step": 4280
    },
    {
      "epoch": 1.4591002044989776,
      "grad_norm": 30.7971577453711,
      "learning_rate": 6.104317244785982e-06,
      "loss": 2.1154,
      "step": 4281
    },
    {
      "epoch": 1.4594410361281527,
      "grad_norm": 13.143771012195034,
      "learning_rate": 6.1023830523064855e-06,
      "loss": 2.2885,
      "step": 4282
    },
    {
      "epoch": 1.4597818677573278,
      "grad_norm": 14.586642640200965,
      "learning_rate": 6.100448686417749e-06,
      "loss": 2.3554,
      "step": 4283
    },
    {
      "epoch": 1.460122699386503,
      "grad_norm": 12.715610464070068,
      "learning_rate": 6.0985141474240525e-06,
      "loss": 2.215,
      "step": 4284
    },
    {
      "epoch": 1.4604635310156784,
      "grad_norm": 20.95063636163122,
      "learning_rate": 6.096579435629708e-06,
      "loss": 2.2391,
      "step": 4285
    },
    {
      "epoch": 1.4608043626448535,
      "grad_norm": 57.65964542070164,
      "learning_rate": 6.0946445513390525e-06,
      "loss": 1.8572,
      "step": 4286
    },
    {
      "epoch": 1.4611451942740286,
      "grad_norm": 14.78284054788824,
      "learning_rate": 6.092709494856453e-06,
      "loss": 2.1532,
      "step": 4287
    },
    {
      "epoch": 1.4614860259032039,
      "grad_norm": 11.99865197803567,
      "learning_rate": 6.0907742664862975e-06,
      "loss": 2.2437,
      "step": 4288
    },
    {
      "epoch": 1.461826857532379,
      "grad_norm": 21.329583554622452,
      "learning_rate": 6.088838866533011e-06,
      "loss": 2.1765,
      "step": 4289
    },
    {
      "epoch": 1.4621676891615543,
      "grad_norm": 11.701458568737111,
      "learning_rate": 6.086903295301033e-06,
      "loss": 1.8296,
      "step": 4290
    },
    {
      "epoch": 1.4625085207907293,
      "grad_norm": 45.92196539672999,
      "learning_rate": 6.084967553094843e-06,
      "loss": 1.7079,
      "step": 4291
    },
    {
      "epoch": 1.4628493524199047,
      "grad_norm": 10.392438969979086,
      "learning_rate": 6.083031640218937e-06,
      "loss": 2.0963,
      "step": 4292
    },
    {
      "epoch": 1.4631901840490797,
      "grad_norm": 17.247349808683282,
      "learning_rate": 6.081095556977842e-06,
      "loss": 2.5887,
      "step": 4293
    },
    {
      "epoch": 1.4635310156782548,
      "grad_norm": 19.810562764139124,
      "learning_rate": 6.079159303676112e-06,
      "loss": 2.334,
      "step": 4294
    },
    {
      "epoch": 1.4638718473074301,
      "grad_norm": 14.540947116198309,
      "learning_rate": 6.077222880618329e-06,
      "loss": 2.2917,
      "step": 4295
    },
    {
      "epoch": 1.4642126789366054,
      "grad_norm": 26.13807109684002,
      "learning_rate": 6.0752862881090965e-06,
      "loss": 2.25,
      "step": 4296
    },
    {
      "epoch": 1.4645535105657805,
      "grad_norm": 15.124923711554166,
      "learning_rate": 6.073349526453052e-06,
      "loss": 2.4132,
      "step": 4297
    },
    {
      "epoch": 1.4648943421949556,
      "grad_norm": 20.550593339355665,
      "learning_rate": 6.071412595954854e-06,
      "loss": 2.3937,
      "step": 4298
    },
    {
      "epoch": 1.465235173824131,
      "grad_norm": 18.914122404323038,
      "learning_rate": 6.069475496919189e-06,
      "loss": 2.3307,
      "step": 4299
    },
    {
      "epoch": 1.465576005453306,
      "grad_norm": 15.970941530060696,
      "learning_rate": 6.067538229650773e-06,
      "loss": 2.0849,
      "step": 4300
    },
    {
      "epoch": 1.4659168370824813,
      "grad_norm": 16.837603364078372,
      "learning_rate": 6.065600794454343e-06,
      "loss": 1.9744,
      "step": 4301
    },
    {
      "epoch": 1.4662576687116564,
      "grad_norm": 19.51989939916051,
      "learning_rate": 6.063663191634666e-06,
      "loss": 2.4512,
      "step": 4302
    },
    {
      "epoch": 1.4665985003408317,
      "grad_norm": 15.077583682172724,
      "learning_rate": 6.061725421496534e-06,
      "loss": 2.5204,
      "step": 4303
    },
    {
      "epoch": 1.4669393319700068,
      "grad_norm": 14.124409501399741,
      "learning_rate": 6.059787484344767e-06,
      "loss": 2.0642,
      "step": 4304
    },
    {
      "epoch": 1.467280163599182,
      "grad_norm": 19.38218886199291,
      "learning_rate": 6.05784938048421e-06,
      "loss": 2.2292,
      "step": 4305
    },
    {
      "epoch": 1.4676209952283572,
      "grad_norm": 16.299590382758964,
      "learning_rate": 6.0559111102197345e-06,
      "loss": 2.1577,
      "step": 4306
    },
    {
      "epoch": 1.4679618268575325,
      "grad_norm": 12.589899971029206,
      "learning_rate": 6.053972673856236e-06,
      "loss": 1.9149,
      "step": 4307
    },
    {
      "epoch": 1.4683026584867076,
      "grad_norm": 14.212808604878427,
      "learning_rate": 6.052034071698644e-06,
      "loss": 2.2989,
      "step": 4308
    },
    {
      "epoch": 1.4686434901158827,
      "grad_norm": 29.39616079600534,
      "learning_rate": 6.050095304051903e-06,
      "loss": 2.0846,
      "step": 4309
    },
    {
      "epoch": 1.468984321745058,
      "grad_norm": 20.69304951942108,
      "learning_rate": 6.04815637122099e-06,
      "loss": 2.5216,
      "step": 4310
    },
    {
      "epoch": 1.469325153374233,
      "grad_norm": 14.192350957687575,
      "learning_rate": 6.046217273510908e-06,
      "loss": 2.4299,
      "step": 4311
    },
    {
      "epoch": 1.4696659850034084,
      "grad_norm": 18.85416948409746,
      "learning_rate": 6.044278011226683e-06,
      "loss": 1.7272,
      "step": 4312
    },
    {
      "epoch": 1.4700068166325835,
      "grad_norm": 23.232350282919203,
      "learning_rate": 6.042338584673371e-06,
      "loss": 2.3484,
      "step": 4313
    },
    {
      "epoch": 1.4703476482617588,
      "grad_norm": 19.635254522626155,
      "learning_rate": 6.040398994156049e-06,
      "loss": 2.1765,
      "step": 4314
    },
    {
      "epoch": 1.4706884798909339,
      "grad_norm": 13.076979793285888,
      "learning_rate": 6.038459239979826e-06,
      "loss": 2.3731,
      "step": 4315
    },
    {
      "epoch": 1.471029311520109,
      "grad_norm": 14.393060088279091,
      "learning_rate": 6.036519322449829e-06,
      "loss": 1.9468,
      "step": 4316
    },
    {
      "epoch": 1.4713701431492843,
      "grad_norm": 12.839167963294821,
      "learning_rate": 6.034579241871217e-06,
      "loss": 2.023,
      "step": 4317
    },
    {
      "epoch": 1.4717109747784596,
      "grad_norm": 17.56449554808446,
      "learning_rate": 6.032638998549172e-06,
      "loss": 2.484,
      "step": 4318
    },
    {
      "epoch": 1.4720518064076347,
      "grad_norm": 16.404348030226192,
      "learning_rate": 6.030698592788901e-06,
      "loss": 1.9406,
      "step": 4319
    },
    {
      "epoch": 1.4723926380368098,
      "grad_norm": 15.499986313430371,
      "learning_rate": 6.028758024895638e-06,
      "loss": 1.8789,
      "step": 4320
    },
    {
      "epoch": 1.472733469665985,
      "grad_norm": 37.41188864338542,
      "learning_rate": 6.026817295174643e-06,
      "loss": 2.3985,
      "step": 4321
    },
    {
      "epoch": 1.4730743012951601,
      "grad_norm": 20.44075541887155,
      "learning_rate": 6.0248764039312e-06,
      "loss": 2.281,
      "step": 4322
    },
    {
      "epoch": 1.4734151329243355,
      "grad_norm": 12.267937133202398,
      "learning_rate": 6.02293535147062e-06,
      "loss": 2.3547,
      "step": 4323
    },
    {
      "epoch": 1.4737559645535105,
      "grad_norm": 16.086559265614447,
      "learning_rate": 6.020994138098235e-06,
      "loss": 2.0084,
      "step": 4324
    },
    {
      "epoch": 1.4740967961826859,
      "grad_norm": 11.087196037558313,
      "learning_rate": 6.019052764119409e-06,
      "loss": 2.0872,
      "step": 4325
    },
    {
      "epoch": 1.474437627811861,
      "grad_norm": 17.846364859924826,
      "learning_rate": 6.017111229839528e-06,
      "loss": 2.3159,
      "step": 4326
    },
    {
      "epoch": 1.474778459441036,
      "grad_norm": 14.490141538655365,
      "learning_rate": 6.015169535563999e-06,
      "loss": 2.4355,
      "step": 4327
    },
    {
      "epoch": 1.4751192910702113,
      "grad_norm": 11.639946560469058,
      "learning_rate": 6.013227681598263e-06,
      "loss": 1.9089,
      "step": 4328
    },
    {
      "epoch": 1.4754601226993864,
      "grad_norm": 14.155393277264535,
      "learning_rate": 6.011285668247777e-06,
      "loss": 1.9716,
      "step": 4329
    },
    {
      "epoch": 1.4758009543285617,
      "grad_norm": 16.367862802599,
      "learning_rate": 6.009343495818031e-06,
      "loss": 2.4176,
      "step": 4330
    },
    {
      "epoch": 1.4761417859577368,
      "grad_norm": 16.09009333025595,
      "learning_rate": 6.007401164614535e-06,
      "loss": 2.3468,
      "step": 4331
    },
    {
      "epoch": 1.4764826175869121,
      "grad_norm": 15.633988974105367,
      "learning_rate": 6.005458674942826e-06,
      "loss": 2.5215,
      "step": 4332
    },
    {
      "epoch": 1.4768234492160872,
      "grad_norm": 16.624538682470327,
      "learning_rate": 6.003516027108465e-06,
      "loss": 2.5859,
      "step": 4333
    },
    {
      "epoch": 1.4771642808452625,
      "grad_norm": 22.964766823710335,
      "learning_rate": 6.001573221417037e-06,
      "loss": 2.1781,
      "step": 4334
    },
    {
      "epoch": 1.4775051124744376,
      "grad_norm": 14.01060032774743,
      "learning_rate": 5.999630258174156e-06,
      "loss": 1.9674,
      "step": 4335
    },
    {
      "epoch": 1.477845944103613,
      "grad_norm": 16.506220272520224,
      "learning_rate": 5.997687137685455e-06,
      "loss": 2.2624,
      "step": 4336
    },
    {
      "epoch": 1.478186775732788,
      "grad_norm": 14.404370658515393,
      "learning_rate": 5.995743860256595e-06,
      "loss": 2.0539,
      "step": 4337
    },
    {
      "epoch": 1.478527607361963,
      "grad_norm": 10.44174277671755,
      "learning_rate": 5.993800426193264e-06,
      "loss": 2.0565,
      "step": 4338
    },
    {
      "epoch": 1.4788684389911384,
      "grad_norm": 22.494032498142996,
      "learning_rate": 5.991856835801167e-06,
      "loss": 2.3975,
      "step": 4339
    },
    {
      "epoch": 1.4792092706203135,
      "grad_norm": 14.635242832998575,
      "learning_rate": 5.989913089386045e-06,
      "loss": 2.2482,
      "step": 4340
    },
    {
      "epoch": 1.4795501022494888,
      "grad_norm": 12.302199216684343,
      "learning_rate": 5.987969187253649e-06,
      "loss": 2.2157,
      "step": 4341
    },
    {
      "epoch": 1.4798909338786639,
      "grad_norm": 16.301296609549595,
      "learning_rate": 5.9860251297097695e-06,
      "loss": 2.3824,
      "step": 4342
    },
    {
      "epoch": 1.4802317655078392,
      "grad_norm": 21.84649136079427,
      "learning_rate": 5.984080917060213e-06,
      "loss": 2.3244,
      "step": 4343
    },
    {
      "epoch": 1.4805725971370143,
      "grad_norm": 26.51710485272529,
      "learning_rate": 5.98213654961081e-06,
      "loss": 2.2449,
      "step": 4344
    },
    {
      "epoch": 1.4809134287661894,
      "grad_norm": 14.56186016740109,
      "learning_rate": 5.980192027667416e-06,
      "loss": 2.6343,
      "step": 4345
    },
    {
      "epoch": 1.4812542603953647,
      "grad_norm": 17.193353823222914,
      "learning_rate": 5.978247351535917e-06,
      "loss": 2.3574,
      "step": 4346
    },
    {
      "epoch": 1.48159509202454,
      "grad_norm": 15.853976602905274,
      "learning_rate": 5.9763025215222144e-06,
      "loss": 2.4601,
      "step": 4347
    },
    {
      "epoch": 1.481935923653715,
      "grad_norm": 19.708376564723228,
      "learning_rate": 5.97435753793224e-06,
      "loss": 2.0711,
      "step": 4348
    },
    {
      "epoch": 1.4822767552828902,
      "grad_norm": 19.26244784265896,
      "learning_rate": 5.972412401071943e-06,
      "loss": 1.8038,
      "step": 4349
    },
    {
      "epoch": 1.4826175869120655,
      "grad_norm": 19.675499374713446,
      "learning_rate": 5.970467111247307e-06,
      "loss": 2.3348,
      "step": 4350
    },
    {
      "epoch": 1.4829584185412406,
      "grad_norm": 19.183839445836153,
      "learning_rate": 5.968521668764331e-06,
      "loss": 3.1425,
      "step": 4351
    },
    {
      "epoch": 1.4832992501704159,
      "grad_norm": 15.853785278498082,
      "learning_rate": 5.96657607392904e-06,
      "loss": 2.3034,
      "step": 4352
    },
    {
      "epoch": 1.483640081799591,
      "grad_norm": 12.174878318202902,
      "learning_rate": 5.964630327047485e-06,
      "loss": 1.9542,
      "step": 4353
    },
    {
      "epoch": 1.4839809134287663,
      "grad_norm": 24.327380907425574,
      "learning_rate": 5.96268442842574e-06,
      "loss": 2.6644,
      "step": 4354
    },
    {
      "epoch": 1.4843217450579413,
      "grad_norm": 17.68424270666283,
      "learning_rate": 5.9607383783699014e-06,
      "loss": 2.3073,
      "step": 4355
    },
    {
      "epoch": 1.4846625766871164,
      "grad_norm": 34.12590406465019,
      "learning_rate": 5.958792177186093e-06,
      "loss": 2.3565,
      "step": 4356
    },
    {
      "epoch": 1.4850034083162917,
      "grad_norm": 15.672661659068368,
      "learning_rate": 5.956845825180457e-06,
      "loss": 1.9451,
      "step": 4357
    },
    {
      "epoch": 1.485344239945467,
      "grad_norm": 19.245374370862002,
      "learning_rate": 5.954899322659163e-06,
      "loss": 1.821,
      "step": 4358
    },
    {
      "epoch": 1.4856850715746421,
      "grad_norm": 15.983421880457763,
      "learning_rate": 5.9529526699284054e-06,
      "loss": 2.3557,
      "step": 4359
    },
    {
      "epoch": 1.4860259032038172,
      "grad_norm": 20.991168062847308,
      "learning_rate": 5.951005867294401e-06,
      "loss": 2.4095,
      "step": 4360
    },
    {
      "epoch": 1.4863667348329925,
      "grad_norm": 20.864067691311067,
      "learning_rate": 5.949058915063386e-06,
      "loss": 2.6486,
      "step": 4361
    },
    {
      "epoch": 1.4867075664621676,
      "grad_norm": 14.669008420087733,
      "learning_rate": 5.947111813541625e-06,
      "loss": 2.3718,
      "step": 4362
    },
    {
      "epoch": 1.487048398091343,
      "grad_norm": 14.614261738980627,
      "learning_rate": 5.945164563035408e-06,
      "loss": 2.4114,
      "step": 4363
    },
    {
      "epoch": 1.487389229720518,
      "grad_norm": 17.210871030497227,
      "learning_rate": 5.943217163851041e-06,
      "loss": 2.5233,
      "step": 4364
    },
    {
      "epoch": 1.4877300613496933,
      "grad_norm": 16.214764175071146,
      "learning_rate": 5.94126961629486e-06,
      "loss": 1.2758,
      "step": 4365
    },
    {
      "epoch": 1.4880708929788684,
      "grad_norm": 16.01496250527842,
      "learning_rate": 5.93932192067322e-06,
      "loss": 1.6968,
      "step": 4366
    },
    {
      "epoch": 1.4884117246080435,
      "grad_norm": 17.22485266381545,
      "learning_rate": 5.937374077292503e-06,
      "loss": 2.4804,
      "step": 4367
    },
    {
      "epoch": 1.4887525562372188,
      "grad_norm": 19.867666478342336,
      "learning_rate": 5.9354260864591155e-06,
      "loss": 1.8169,
      "step": 4368
    },
    {
      "epoch": 1.4890933878663941,
      "grad_norm": 16.638604739152306,
      "learning_rate": 5.933477948479478e-06,
      "loss": 2.3097,
      "step": 4369
    },
    {
      "epoch": 1.4894342194955692,
      "grad_norm": 14.995261878171839,
      "learning_rate": 5.931529663660042e-06,
      "loss": 2.5542,
      "step": 4370
    },
    {
      "epoch": 1.4897750511247443,
      "grad_norm": 12.212560788723833,
      "learning_rate": 5.9295812323072845e-06,
      "loss": 2.1694,
      "step": 4371
    },
    {
      "epoch": 1.4901158827539196,
      "grad_norm": 21.273236827636545,
      "learning_rate": 5.927632654727696e-06,
      "loss": 2.0328,
      "step": 4372
    },
    {
      "epoch": 1.4904567143830947,
      "grad_norm": 29.05755149922163,
      "learning_rate": 5.9256839312278e-06,
      "loss": 1.3758,
      "step": 4373
    },
    {
      "epoch": 1.49079754601227,
      "grad_norm": 15.995402679235733,
      "learning_rate": 5.923735062114137e-06,
      "loss": 2.612,
      "step": 4374
    },
    {
      "epoch": 1.491138377641445,
      "grad_norm": 20.392348800079166,
      "learning_rate": 5.92178604769327e-06,
      "loss": 2.2273,
      "step": 4375
    },
    {
      "epoch": 1.4914792092706204,
      "grad_norm": 12.539207363167089,
      "learning_rate": 5.91983688827179e-06,
      "loss": 2.2712,
      "step": 4376
    },
    {
      "epoch": 1.4918200408997955,
      "grad_norm": 14.859699208246981,
      "learning_rate": 5.917887584156305e-06,
      "loss": 2.3026,
      "step": 4377
    },
    {
      "epoch": 1.4921608725289706,
      "grad_norm": 18.51362982762398,
      "learning_rate": 5.91593813565345e-06,
      "loss": 2.0117,
      "step": 4378
    },
    {
      "epoch": 1.4925017041581459,
      "grad_norm": 21.970984023736694,
      "learning_rate": 5.913988543069878e-06,
      "loss": 2.2254,
      "step": 4379
    },
    {
      "epoch": 1.4928425357873212,
      "grad_norm": 27.294980308504538,
      "learning_rate": 5.9120388067122715e-06,
      "loss": 2.2178,
      "step": 4380
    },
    {
      "epoch": 1.4931833674164963,
      "grad_norm": 18.42680947737345,
      "learning_rate": 5.91008892688733e-06,
      "loss": 2.8541,
      "step": 4381
    },
    {
      "epoch": 1.4935241990456714,
      "grad_norm": 19.22236118660401,
      "learning_rate": 5.908138903901776e-06,
      "loss": 2.2953,
      "step": 4382
    },
    {
      "epoch": 1.4938650306748467,
      "grad_norm": 17.854036861615693,
      "learning_rate": 5.9061887380623585e-06,
      "loss": 2.1056,
      "step": 4383
    },
    {
      "epoch": 1.4942058623040217,
      "grad_norm": 13.975789898285342,
      "learning_rate": 5.904238429675845e-06,
      "loss": 2.375,
      "step": 4384
    },
    {
      "epoch": 1.494546693933197,
      "grad_norm": 18.122727077258578,
      "learning_rate": 5.902287979049027e-06,
      "loss": 2.4284,
      "step": 4385
    },
    {
      "epoch": 1.4948875255623721,
      "grad_norm": 16.10644014774808,
      "learning_rate": 5.9003373864887185e-06,
      "loss": 1.8998,
      "step": 4386
    },
    {
      "epoch": 1.4952283571915475,
      "grad_norm": 15.610424663536032,
      "learning_rate": 5.898386652301754e-06,
      "loss": 2.267,
      "step": 4387
    },
    {
      "epoch": 1.4955691888207225,
      "grad_norm": 16.316840219254782,
      "learning_rate": 5.896435776794994e-06,
      "loss": 2.1361,
      "step": 4388
    },
    {
      "epoch": 1.4959100204498976,
      "grad_norm": 14.158062634043883,
      "learning_rate": 5.894484760275318e-06,
      "loss": 2.387,
      "step": 4389
    },
    {
      "epoch": 1.496250852079073,
      "grad_norm": 20.83430941653588,
      "learning_rate": 5.892533603049628e-06,
      "loss": 2.5654,
      "step": 4390
    },
    {
      "epoch": 1.4965916837082482,
      "grad_norm": 16.18184828947139,
      "learning_rate": 5.890582305424848e-06,
      "loss": 2.7041,
      "step": 4391
    },
    {
      "epoch": 1.4969325153374233,
      "grad_norm": 14.648440214088051,
      "learning_rate": 5.8886308677079265e-06,
      "loss": 1.5609,
      "step": 4392
    },
    {
      "epoch": 1.4972733469665984,
      "grad_norm": 22.247200744273147,
      "learning_rate": 5.886679290205832e-06,
      "loss": 2.7497,
      "step": 4393
    },
    {
      "epoch": 1.4976141785957737,
      "grad_norm": 14.874886673186973,
      "learning_rate": 5.884727573225558e-06,
      "loss": 2.2733,
      "step": 4394
    },
    {
      "epoch": 1.4979550102249488,
      "grad_norm": 18.928202482097987,
      "learning_rate": 5.8827757170741126e-06,
      "loss": 2.7269,
      "step": 4395
    },
    {
      "epoch": 1.4982958418541241,
      "grad_norm": 17.73729242999824,
      "learning_rate": 5.880823722058533e-06,
      "loss": 2.6446,
      "step": 4396
    },
    {
      "epoch": 1.4986366734832992,
      "grad_norm": 21.13625665848129,
      "learning_rate": 5.878871588485876e-06,
      "loss": 1.8042,
      "step": 4397
    },
    {
      "epoch": 1.4989775051124745,
      "grad_norm": 18.73452948178375,
      "learning_rate": 5.876919316663219e-06,
      "loss": 2.5779,
      "step": 4398
    },
    {
      "epoch": 1.4993183367416496,
      "grad_norm": 20.13286226542239,
      "learning_rate": 5.874966906897663e-06,
      "loss": 2.6525,
      "step": 4399
    },
    {
      "epoch": 1.4996591683708247,
      "grad_norm": 24.69860815147935,
      "learning_rate": 5.8730143594963296e-06,
      "loss": 2.2239,
      "step": 4400
    },
    {
      "epoch": 1.5,
      "grad_norm": 25.558244928201383,
      "learning_rate": 5.871061674766362e-06,
      "loss": 2.7978,
      "step": 4401
    },
    {
      "epoch": 1.5003408316291753,
      "grad_norm": 14.886890223038504,
      "learning_rate": 5.8691088530149265e-06,
      "loss": 2.2302,
      "step": 4402
    },
    {
      "epoch": 1.5006816632583504,
      "grad_norm": 19.412116256120086,
      "learning_rate": 5.867155894549208e-06,
      "loss": 1.939,
      "step": 4403
    },
    {
      "epoch": 1.5010224948875255,
      "grad_norm": 30.85734612653365,
      "learning_rate": 5.865202799676414e-06,
      "loss": 2.2362,
      "step": 4404
    },
    {
      "epoch": 1.5013633265167008,
      "grad_norm": 14.391035080093314,
      "learning_rate": 5.863249568703777e-06,
      "loss": 2.5675,
      "step": 4405
    },
    {
      "epoch": 1.501704158145876,
      "grad_norm": 15.091927908563555,
      "learning_rate": 5.861296201938547e-06,
      "loss": 2.529,
      "step": 4406
    },
    {
      "epoch": 1.502044989775051,
      "grad_norm": 12.909622535084232,
      "learning_rate": 5.859342699687993e-06,
      "loss": 2.3284,
      "step": 4407
    },
    {
      "epoch": 1.5023858214042263,
      "grad_norm": 18.254061869263904,
      "learning_rate": 5.857389062259414e-06,
      "loss": 2.459,
      "step": 4408
    },
    {
      "epoch": 1.5027266530334016,
      "grad_norm": 13.521037819599774,
      "learning_rate": 5.855435289960122e-06,
      "loss": 2.5122,
      "step": 4409
    },
    {
      "epoch": 1.5030674846625767,
      "grad_norm": 12.627369573483533,
      "learning_rate": 5.853481383097453e-06,
      "loss": 2.3909,
      "step": 4410
    },
    {
      "epoch": 1.5034083162917518,
      "grad_norm": 24.500293654417543,
      "learning_rate": 5.851527341978767e-06,
      "loss": 2.7882,
      "step": 4411
    },
    {
      "epoch": 1.503749147920927,
      "grad_norm": 14.593590357615406,
      "learning_rate": 5.849573166911437e-06,
      "loss": 2.2719,
      "step": 4412
    },
    {
      "epoch": 1.5040899795501024,
      "grad_norm": 16.741907622236518,
      "learning_rate": 5.847618858202868e-06,
      "loss": 2.2827,
      "step": 4413
    },
    {
      "epoch": 1.5044308111792775,
      "grad_norm": 41.12631586840671,
      "learning_rate": 5.845664416160478e-06,
      "loss": 2.8361,
      "step": 4414
    },
    {
      "epoch": 1.5047716428084525,
      "grad_norm": 10.02700999534083,
      "learning_rate": 5.843709841091708e-06,
      "loss": 2.1223,
      "step": 4415
    },
    {
      "epoch": 1.5051124744376279,
      "grad_norm": 12.124044832579232,
      "learning_rate": 5.8417551333040225e-06,
      "loss": 1.7476,
      "step": 4416
    },
    {
      "epoch": 1.5054533060668032,
      "grad_norm": 29.226746138162905,
      "learning_rate": 5.839800293104904e-06,
      "loss": 2.2365,
      "step": 4417
    },
    {
      "epoch": 1.505794137695978,
      "grad_norm": 18.21900625925857,
      "learning_rate": 5.837845320801855e-06,
      "loss": 1.6272,
      "step": 4418
    },
    {
      "epoch": 1.5061349693251533,
      "grad_norm": 36.45758020967617,
      "learning_rate": 5.835890216702403e-06,
      "loss": 2.6181,
      "step": 4419
    },
    {
      "epoch": 1.5064758009543286,
      "grad_norm": 14.937692253151669,
      "learning_rate": 5.8339349811140935e-06,
      "loss": 2.2365,
      "step": 4420
    },
    {
      "epoch": 1.5068166325835037,
      "grad_norm": 13.856883177491705,
      "learning_rate": 5.8319796143444885e-06,
      "loss": 2.1825,
      "step": 4421
    },
    {
      "epoch": 1.5071574642126788,
      "grad_norm": 18.66235581193284,
      "learning_rate": 5.8300241167011805e-06,
      "loss": 2.4865,
      "step": 4422
    },
    {
      "epoch": 1.5074982958418541,
      "grad_norm": 11.980513085267878,
      "learning_rate": 5.828068488491775e-06,
      "loss": 2.0243,
      "step": 4423
    },
    {
      "epoch": 1.5078391274710294,
      "grad_norm": 20.508395866238647,
      "learning_rate": 5.826112730023898e-06,
      "loss": 2.3111,
      "step": 4424
    },
    {
      "epoch": 1.5081799591002045,
      "grad_norm": 15.153418986583143,
      "learning_rate": 5.824156841605202e-06,
      "loss": 2.0755,
      "step": 4425
    },
    {
      "epoch": 1.5085207907293796,
      "grad_norm": 15.703733243239974,
      "learning_rate": 5.822200823543354e-06,
      "loss": 2.7779,
      "step": 4426
    },
    {
      "epoch": 1.508861622358555,
      "grad_norm": 14.251157260829073,
      "learning_rate": 5.820244676146042e-06,
      "loss": 2.0868,
      "step": 4427
    },
    {
      "epoch": 1.50920245398773,
      "grad_norm": 14.149203967149003,
      "learning_rate": 5.818288399720978e-06,
      "loss": 2.164,
      "step": 4428
    },
    {
      "epoch": 1.509543285616905,
      "grad_norm": 21.21821425587182,
      "learning_rate": 5.816331994575892e-06,
      "loss": 2.3317,
      "step": 4429
    },
    {
      "epoch": 1.5098841172460804,
      "grad_norm": 13.863420653658345,
      "learning_rate": 5.814375461018532e-06,
      "loss": 1.749,
      "step": 4430
    },
    {
      "epoch": 1.5102249488752557,
      "grad_norm": 21.93916276195974,
      "learning_rate": 5.812418799356672e-06,
      "loss": 2.3073,
      "step": 4431
    },
    {
      "epoch": 1.5105657805044308,
      "grad_norm": 20.39953958683771,
      "learning_rate": 5.810462009898099e-06,
      "loss": 2.9469,
      "step": 4432
    },
    {
      "epoch": 1.5109066121336059,
      "grad_norm": 18.24309487640441,
      "learning_rate": 5.808505092950627e-06,
      "loss": 2.4919,
      "step": 4433
    },
    {
      "epoch": 1.5112474437627812,
      "grad_norm": 15.680008778556585,
      "learning_rate": 5.8065480488220845e-06,
      "loss": 1.8797,
      "step": 4434
    },
    {
      "epoch": 1.5115882753919565,
      "grad_norm": 19.04827567637642,
      "learning_rate": 5.804590877820323e-06,
      "loss": 2.4124,
      "step": 4435
    },
    {
      "epoch": 1.5119291070211316,
      "grad_norm": 20.685789229847273,
      "learning_rate": 5.802633580253215e-06,
      "loss": 2.3996,
      "step": 4436
    },
    {
      "epoch": 1.5122699386503067,
      "grad_norm": 14.90527949181147,
      "learning_rate": 5.800676156428649e-06,
      "loss": 2.1288,
      "step": 4437
    },
    {
      "epoch": 1.512610770279482,
      "grad_norm": 20.944916764061936,
      "learning_rate": 5.798718606654535e-06,
      "loss": 2.9461,
      "step": 4438
    },
    {
      "epoch": 1.512951601908657,
      "grad_norm": 18.103054874099506,
      "learning_rate": 5.796760931238807e-06,
      "loss": 2.944,
      "step": 4439
    },
    {
      "epoch": 1.5132924335378322,
      "grad_norm": 23.60172217259511,
      "learning_rate": 5.7948031304894115e-06,
      "loss": 2.4022,
      "step": 4440
    },
    {
      "epoch": 1.5136332651670075,
      "grad_norm": 14.86106548121626,
      "learning_rate": 5.792845204714319e-06,
      "loss": 2.0864,
      "step": 4441
    },
    {
      "epoch": 1.5139740967961828,
      "grad_norm": 12.03073156570767,
      "learning_rate": 5.790887154221521e-06,
      "loss": 2.4186,
      "step": 4442
    },
    {
      "epoch": 1.5143149284253579,
      "grad_norm": 15.847178466827643,
      "learning_rate": 5.788928979319024e-06,
      "loss": 2.1119,
      "step": 4443
    },
    {
      "epoch": 1.514655760054533,
      "grad_norm": 18.95602963578717,
      "learning_rate": 5.7869706803148584e-06,
      "loss": 2.4698,
      "step": 4444
    },
    {
      "epoch": 1.5149965916837083,
      "grad_norm": 16.074294736576917,
      "learning_rate": 5.785012257517074e-06,
      "loss": 2.4274,
      "step": 4445
    },
    {
      "epoch": 1.5153374233128836,
      "grad_norm": 11.793597161533356,
      "learning_rate": 5.783053711233734e-06,
      "loss": 1.5285,
      "step": 4446
    },
    {
      "epoch": 1.5156782549420587,
      "grad_norm": 23.456431194985754,
      "learning_rate": 5.781095041772929e-06,
      "loss": 2.0796,
      "step": 4447
    },
    {
      "epoch": 1.5160190865712337,
      "grad_norm": 20.218628181745157,
      "learning_rate": 5.779136249442765e-06,
      "loss": 2.52,
      "step": 4448
    },
    {
      "epoch": 1.516359918200409,
      "grad_norm": 13.703867807308157,
      "learning_rate": 5.777177334551367e-06,
      "loss": 2.0232,
      "step": 4449
    },
    {
      "epoch": 1.5167007498295841,
      "grad_norm": 14.411004856013218,
      "learning_rate": 5.775218297406879e-06,
      "loss": 2.4637,
      "step": 4450
    },
    {
      "epoch": 1.5170415814587592,
      "grad_norm": 14.601776749000834,
      "learning_rate": 5.773259138317469e-06,
      "loss": 2.1378,
      "step": 4451
    },
    {
      "epoch": 1.5173824130879345,
      "grad_norm": 14.986078956973593,
      "learning_rate": 5.771299857591317e-06,
      "loss": 2.1091,
      "step": 4452
    },
    {
      "epoch": 1.5177232447171098,
      "grad_norm": 31.542911771964874,
      "learning_rate": 5.769340455536629e-06,
      "loss": 2.3165,
      "step": 4453
    },
    {
      "epoch": 1.518064076346285,
      "grad_norm": 20.42977572361904,
      "learning_rate": 5.7673809324616235e-06,
      "loss": 2.4137,
      "step": 4454
    },
    {
      "epoch": 1.51840490797546,
      "grad_norm": 14.559900918365898,
      "learning_rate": 5.765421288674542e-06,
      "loss": 1.7066,
      "step": 4455
    },
    {
      "epoch": 1.5187457396046353,
      "grad_norm": 14.772606808981097,
      "learning_rate": 5.763461524483645e-06,
      "loss": 1.9501,
      "step": 4456
    },
    {
      "epoch": 1.5190865712338106,
      "grad_norm": 23.03491619089363,
      "learning_rate": 5.761501640197212e-06,
      "loss": 2.4569,
      "step": 4457
    },
    {
      "epoch": 1.5194274028629857,
      "grad_norm": 16.848824042349236,
      "learning_rate": 5.759541636123539e-06,
      "loss": 1.6339,
      "step": 4458
    },
    {
      "epoch": 1.5197682344921608,
      "grad_norm": 18.124254777033162,
      "learning_rate": 5.757581512570943e-06,
      "loss": 2.5538,
      "step": 4459
    },
    {
      "epoch": 1.5201090661213361,
      "grad_norm": 17.856634659117717,
      "learning_rate": 5.755621269847758e-06,
      "loss": 2.2541,
      "step": 4460
    },
    {
      "epoch": 1.5204498977505112,
      "grad_norm": 13.849110966944009,
      "learning_rate": 5.753660908262342e-06,
      "loss": 2.3696,
      "step": 4461
    },
    {
      "epoch": 1.5207907293796863,
      "grad_norm": 13.584582524788413,
      "learning_rate": 5.751700428123065e-06,
      "loss": 2.6209,
      "step": 4462
    },
    {
      "epoch": 1.5211315610088616,
      "grad_norm": 19.112757412219054,
      "learning_rate": 5.749739829738316e-06,
      "loss": 2.2952,
      "step": 4463
    },
    {
      "epoch": 1.521472392638037,
      "grad_norm": 18.088438281388893,
      "learning_rate": 5.747779113416508e-06,
      "loss": 2.183,
      "step": 4464
    },
    {
      "epoch": 1.521813224267212,
      "grad_norm": 11.992419791784215,
      "learning_rate": 5.745818279466069e-06,
      "loss": 1.9801,
      "step": 4465
    },
    {
      "epoch": 1.522154055896387,
      "grad_norm": 18.736318127353996,
      "learning_rate": 5.743857328195446e-06,
      "loss": 2.3415,
      "step": 4466
    },
    {
      "epoch": 1.5224948875255624,
      "grad_norm": 21.830839218996527,
      "learning_rate": 5.741896259913103e-06,
      "loss": 2.2487,
      "step": 4467
    },
    {
      "epoch": 1.5228357191547377,
      "grad_norm": 18.333678034586118,
      "learning_rate": 5.739935074927526e-06,
      "loss": 2.6311,
      "step": 4468
    },
    {
      "epoch": 1.5231765507839128,
      "grad_norm": 20.283494017826694,
      "learning_rate": 5.737973773547216e-06,
      "loss": 2.303,
      "step": 4469
    },
    {
      "epoch": 1.5235173824130879,
      "grad_norm": 14.959179322809163,
      "learning_rate": 5.7360123560806944e-06,
      "loss": 2.2666,
      "step": 4470
    },
    {
      "epoch": 1.5238582140422632,
      "grad_norm": 25.75322318743068,
      "learning_rate": 5.734050822836499e-06,
      "loss": 2.2817,
      "step": 4471
    },
    {
      "epoch": 1.5241990456714383,
      "grad_norm": 15.714489051219761,
      "learning_rate": 5.732089174123185e-06,
      "loss": 2.0088,
      "step": 4472
    },
    {
      "epoch": 1.5245398773006134,
      "grad_norm": 20.199216484723046,
      "learning_rate": 5.730127410249333e-06,
      "loss": 2.7705,
      "step": 4473
    },
    {
      "epoch": 1.5248807089297887,
      "grad_norm": 15.680619179307195,
      "learning_rate": 5.728165531523531e-06,
      "loss": 2.5124,
      "step": 4474
    },
    {
      "epoch": 1.525221540558964,
      "grad_norm": 12.090547151637967,
      "learning_rate": 5.726203538254392e-06,
      "loss": 2.209,
      "step": 4475
    },
    {
      "epoch": 1.525562372188139,
      "grad_norm": 30.17350445026114,
      "learning_rate": 5.7242414307505465e-06,
      "loss": 2.1134,
      "step": 4476
    },
    {
      "epoch": 1.5259032038173141,
      "grad_norm": 16.987742033332104,
      "learning_rate": 5.722279209320641e-06,
      "loss": 2.3002,
      "step": 4477
    },
    {
      "epoch": 1.5262440354464895,
      "grad_norm": 31.554420522679724,
      "learning_rate": 5.72031687427334e-06,
      "loss": 2.105,
      "step": 4478
    },
    {
      "epoch": 1.5265848670756648,
      "grad_norm": 19.561010236949638,
      "learning_rate": 5.718354425917328e-06,
      "loss": 2.1272,
      "step": 4479
    },
    {
      "epoch": 1.5269256987048399,
      "grad_norm": 12.530186208826217,
      "learning_rate": 5.716391864561303e-06,
      "loss": 2.3438,
      "step": 4480
    },
    {
      "epoch": 1.527266530334015,
      "grad_norm": 16.035803145145216,
      "learning_rate": 5.714429190513988e-06,
      "loss": 2.6538,
      "step": 4481
    },
    {
      "epoch": 1.5276073619631902,
      "grad_norm": 22.89299402574627,
      "learning_rate": 5.712466404084117e-06,
      "loss": 2.4911,
      "step": 4482
    },
    {
      "epoch": 1.5279481935923653,
      "grad_norm": 16.503772399703138,
      "learning_rate": 5.710503505580445e-06,
      "loss": 2.7619,
      "step": 4483
    },
    {
      "epoch": 1.5282890252215404,
      "grad_norm": 15.326450524968614,
      "learning_rate": 5.708540495311742e-06,
      "loss": 2.0846,
      "step": 4484
    },
    {
      "epoch": 1.5286298568507157,
      "grad_norm": 16.319309335572246,
      "learning_rate": 5.706577373586799e-06,
      "loss": 2.3781,
      "step": 4485
    },
    {
      "epoch": 1.528970688479891,
      "grad_norm": 18.99701068828645,
      "learning_rate": 5.704614140714422e-06,
      "loss": 1.5151,
      "step": 4486
    },
    {
      "epoch": 1.5293115201090661,
      "grad_norm": 18.491769616070897,
      "learning_rate": 5.702650797003437e-06,
      "loss": 2.55,
      "step": 4487
    },
    {
      "epoch": 1.5296523517382412,
      "grad_norm": 18.316054523060785,
      "learning_rate": 5.700687342762684e-06,
      "loss": 2.3734,
      "step": 4488
    },
    {
      "epoch": 1.5299931833674165,
      "grad_norm": 24.661487887213248,
      "learning_rate": 5.698723778301021e-06,
      "loss": 2.1193,
      "step": 4489
    },
    {
      "epoch": 1.5303340149965918,
      "grad_norm": 16.256841104142268,
      "learning_rate": 5.696760103927329e-06,
      "loss": 2.3262,
      "step": 4490
    },
    {
      "epoch": 1.530674846625767,
      "grad_norm": 26.452010638481248,
      "learning_rate": 5.694796319950497e-06,
      "loss": 1.6503,
      "step": 4491
    },
    {
      "epoch": 1.531015678254942,
      "grad_norm": 22.348801509536344,
      "learning_rate": 5.692832426679439e-06,
      "loss": 2.2185,
      "step": 4492
    },
    {
      "epoch": 1.5313565098841173,
      "grad_norm": 19.338725045868447,
      "learning_rate": 5.690868424423082e-06,
      "loss": 2.8113,
      "step": 4493
    },
    {
      "epoch": 1.5316973415132924,
      "grad_norm": 11.846456876975783,
      "learning_rate": 5.688904313490372e-06,
      "loss": 2.2021,
      "step": 4494
    },
    {
      "epoch": 1.5320381731424675,
      "grad_norm": 16.395202551988312,
      "learning_rate": 5.686940094190272e-06,
      "loss": 1.9343,
      "step": 4495
    },
    {
      "epoch": 1.5323790047716428,
      "grad_norm": 15.877035115543848,
      "learning_rate": 5.6849757668317586e-06,
      "loss": 2.2147,
      "step": 4496
    },
    {
      "epoch": 1.532719836400818,
      "grad_norm": 20.97390855976674,
      "learning_rate": 5.68301133172383e-06,
      "loss": 1.6291,
      "step": 4497
    },
    {
      "epoch": 1.5330606680299932,
      "grad_norm": 18.647775445006904,
      "learning_rate": 5.681046789175502e-06,
      "loss": 2.2285,
      "step": 4498
    },
    {
      "epoch": 1.5334014996591683,
      "grad_norm": 16.943944498590685,
      "learning_rate": 5.679082139495803e-06,
      "loss": 2.3699,
      "step": 4499
    },
    {
      "epoch": 1.5337423312883436,
      "grad_norm": 17.522932187070786,
      "learning_rate": 5.677117382993779e-06,
      "loss": 2.6339,
      "step": 4500
    },
    {
      "epoch": 1.534083162917519,
      "grad_norm": 25.636283146035158,
      "learning_rate": 5.6751525199784956e-06,
      "loss": 2.3638,
      "step": 4501
    },
    {
      "epoch": 1.534423994546694,
      "grad_norm": 22.947926318488395,
      "learning_rate": 5.673187550759033e-06,
      "loss": 2.1207,
      "step": 4502
    },
    {
      "epoch": 1.534764826175869,
      "grad_norm": 16.298488879302703,
      "learning_rate": 5.671222475644489e-06,
      "loss": 2.4745,
      "step": 4503
    },
    {
      "epoch": 1.5351056578050444,
      "grad_norm": 16.892881503603896,
      "learning_rate": 5.669257294943978e-06,
      "loss": 2.5625,
      "step": 4504
    },
    {
      "epoch": 1.5354464894342195,
      "grad_norm": 18.810721396712914,
      "learning_rate": 5.6672920089666295e-06,
      "loss": 2.7655,
      "step": 4505
    },
    {
      "epoch": 1.5357873210633946,
      "grad_norm": 16.23708019399656,
      "learning_rate": 5.665326618021591e-06,
      "loss": 2.2891,
      "step": 4506
    },
    {
      "epoch": 1.5361281526925699,
      "grad_norm": 20.71895409802873,
      "learning_rate": 5.6633611224180295e-06,
      "loss": 2.6315,
      "step": 4507
    },
    {
      "epoch": 1.5364689843217452,
      "grad_norm": 11.060093325691811,
      "learning_rate": 5.66139552246512e-06,
      "loss": 2.0289,
      "step": 4508
    },
    {
      "epoch": 1.5368098159509203,
      "grad_norm": 26.31080175708086,
      "learning_rate": 5.659429818472064e-06,
      "loss": 2.2056,
      "step": 4509
    },
    {
      "epoch": 1.5371506475800953,
      "grad_norm": 17.966651799965245,
      "learning_rate": 5.6574640107480725e-06,
      "loss": 3.057,
      "step": 4510
    },
    {
      "epoch": 1.5374914792092707,
      "grad_norm": 10.471145168779648,
      "learning_rate": 5.655498099602374e-06,
      "loss": 2.0043,
      "step": 4511
    },
    {
      "epoch": 1.537832310838446,
      "grad_norm": 13.880479499215832,
      "learning_rate": 5.653532085344215e-06,
      "loss": 2.4699,
      "step": 4512
    },
    {
      "epoch": 1.538173142467621,
      "grad_norm": 20.74585367647219,
      "learning_rate": 5.651565968282859e-06,
      "loss": 2.3074,
      "step": 4513
    },
    {
      "epoch": 1.5385139740967961,
      "grad_norm": 18.36012233788483,
      "learning_rate": 5.649599748727579e-06,
      "loss": 2.1297,
      "step": 4514
    },
    {
      "epoch": 1.5388548057259714,
      "grad_norm": 38.80231104375928,
      "learning_rate": 5.647633426987677e-06,
      "loss": 2.6303,
      "step": 4515
    },
    {
      "epoch": 1.5391956373551465,
      "grad_norm": 13.863756370189114,
      "learning_rate": 5.6456670033724555e-06,
      "loss": 1.8361,
      "step": 4516
    },
    {
      "epoch": 1.5395364689843216,
      "grad_norm": 19.17303323161461,
      "learning_rate": 5.643700478191245e-06,
      "loss": 1.7585,
      "step": 4517
    },
    {
      "epoch": 1.539877300613497,
      "grad_norm": 17.723271151182608,
      "learning_rate": 5.641733851753387e-06,
      "loss": 2.7757,
      "step": 4518
    },
    {
      "epoch": 1.5402181322426722,
      "grad_norm": 17.37096971489388,
      "learning_rate": 5.6397671243682395e-06,
      "loss": 2.5183,
      "step": 4519
    },
    {
      "epoch": 1.5405589638718473,
      "grad_norm": 17.128054717352043,
      "learning_rate": 5.637800296345178e-06,
      "loss": 1.7092,
      "step": 4520
    },
    {
      "epoch": 1.5408997955010224,
      "grad_norm": 12.594532367964533,
      "learning_rate": 5.63583336799359e-06,
      "loss": 1.7114,
      "step": 4521
    },
    {
      "epoch": 1.5412406271301977,
      "grad_norm": 38.01902789877474,
      "learning_rate": 5.63386633962288e-06,
      "loss": 1.62,
      "step": 4522
    },
    {
      "epoch": 1.541581458759373,
      "grad_norm": 15.085500985553569,
      "learning_rate": 5.631899211542473e-06,
      "loss": 2.8388,
      "step": 4523
    },
    {
      "epoch": 1.5419222903885481,
      "grad_norm": 22.339557268709473,
      "learning_rate": 5.629931984061806e-06,
      "loss": 2.2281,
      "step": 4524
    },
    {
      "epoch": 1.5422631220177232,
      "grad_norm": 15.71327914815617,
      "learning_rate": 5.627964657490329e-06,
      "loss": 2.2207,
      "step": 4525
    },
    {
      "epoch": 1.5426039536468985,
      "grad_norm": 15.207207161489809,
      "learning_rate": 5.625997232137511e-06,
      "loss": 2.4799,
      "step": 4526
    },
    {
      "epoch": 1.5429447852760736,
      "grad_norm": 13.905384900383336,
      "learning_rate": 5.624029708312837e-06,
      "loss": 2.3252,
      "step": 4527
    },
    {
      "epoch": 1.5432856169052487,
      "grad_norm": 15.207839827584456,
      "learning_rate": 5.622062086325806e-06,
      "loss": 2.3561,
      "step": 4528
    },
    {
      "epoch": 1.543626448534424,
      "grad_norm": 22.383305603381622,
      "learning_rate": 5.620094366485933e-06,
      "loss": 2.1148,
      "step": 4529
    },
    {
      "epoch": 1.5439672801635993,
      "grad_norm": 18.063635031186525,
      "learning_rate": 5.618126549102747e-06,
      "loss": 2.5071,
      "step": 4530
    },
    {
      "epoch": 1.5443081117927744,
      "grad_norm": 17.36417222482685,
      "learning_rate": 5.616158634485793e-06,
      "loss": 1.9633,
      "step": 4531
    },
    {
      "epoch": 1.5446489434219495,
      "grad_norm": 15.368398933692372,
      "learning_rate": 5.614190622944636e-06,
      "loss": 2.2394,
      "step": 4532
    },
    {
      "epoch": 1.5449897750511248,
      "grad_norm": 20.66995014101257,
      "learning_rate": 5.612222514788847e-06,
      "loss": 2.1755,
      "step": 4533
    },
    {
      "epoch": 1.5453306066803,
      "grad_norm": 13.883159489876338,
      "learning_rate": 5.61025431032802e-06,
      "loss": 1.8324,
      "step": 4534
    },
    {
      "epoch": 1.545671438309475,
      "grad_norm": 17.069285077286047,
      "learning_rate": 5.608286009871761e-06,
      "loss": 2.1934,
      "step": 4535
    },
    {
      "epoch": 1.5460122699386503,
      "grad_norm": 20.161207577776928,
      "learning_rate": 5.6063176137296895e-06,
      "loss": 2.2412,
      "step": 4536
    },
    {
      "epoch": 1.5463531015678256,
      "grad_norm": 14.954737516201321,
      "learning_rate": 5.604349122211446e-06,
      "loss": 1.7689,
      "step": 4537
    },
    {
      "epoch": 1.5466939331970007,
      "grad_norm": 12.859945054608637,
      "learning_rate": 5.602380535626679e-06,
      "loss": 2.607,
      "step": 4538
    },
    {
      "epoch": 1.5470347648261757,
      "grad_norm": 17.59051769094524,
      "learning_rate": 5.6004118542850535e-06,
      "loss": 1.8486,
      "step": 4539
    },
    {
      "epoch": 1.547375596455351,
      "grad_norm": 16.160519540205215,
      "learning_rate": 5.598443078496255e-06,
      "loss": 2.4315,
      "step": 4540
    },
    {
      "epoch": 1.5477164280845264,
      "grad_norm": 14.925796151242583,
      "learning_rate": 5.59647420856998e-06,
      "loss": 2.2024,
      "step": 4541
    },
    {
      "epoch": 1.5480572597137015,
      "grad_norm": 18.899566168539746,
      "learning_rate": 5.594505244815935e-06,
      "loss": 2.3097,
      "step": 4542
    },
    {
      "epoch": 1.5483980913428765,
      "grad_norm": 19.554760601109123,
      "learning_rate": 5.592536187543849e-06,
      "loss": 2.2554,
      "step": 4543
    },
    {
      "epoch": 1.5487389229720518,
      "grad_norm": 13.591426221566362,
      "learning_rate": 5.590567037063462e-06,
      "loss": 2.1538,
      "step": 4544
    },
    {
      "epoch": 1.5490797546012272,
      "grad_norm": 13.829188094516626,
      "learning_rate": 5.588597793684529e-06,
      "loss": 2.1956,
      "step": 4545
    },
    {
      "epoch": 1.549420586230402,
      "grad_norm": 17.095589028937535,
      "learning_rate": 5.586628457716821e-06,
      "loss": 2.5356,
      "step": 4546
    },
    {
      "epoch": 1.5497614178595773,
      "grad_norm": 12.634154219209861,
      "learning_rate": 5.584659029470121e-06,
      "loss": 1.8715,
      "step": 4547
    },
    {
      "epoch": 1.5501022494887526,
      "grad_norm": 14.973753590959701,
      "learning_rate": 5.582689509254227e-06,
      "loss": 2.2156,
      "step": 4548
    },
    {
      "epoch": 1.5504430811179277,
      "grad_norm": 17.98834271483902,
      "learning_rate": 5.580719897378955e-06,
      "loss": 2.2146,
      "step": 4549
    },
    {
      "epoch": 1.5507839127471028,
      "grad_norm": 19.35957377640842,
      "learning_rate": 5.5787501941541314e-06,
      "loss": 2.0851,
      "step": 4550
    },
    {
      "epoch": 1.5511247443762781,
      "grad_norm": 19.96868296802965,
      "learning_rate": 5.5767803998895975e-06,
      "loss": 2.2074,
      "step": 4551
    },
    {
      "epoch": 1.5514655760054534,
      "grad_norm": 20.663593731404152,
      "learning_rate": 5.5748105148952115e-06,
      "loss": 2.3345,
      "step": 4552
    },
    {
      "epoch": 1.5518064076346285,
      "grad_norm": 16.690297506648037,
      "learning_rate": 5.572840539480843e-06,
      "loss": 2.1432,
      "step": 4553
    },
    {
      "epoch": 1.5521472392638036,
      "grad_norm": 14.826384953917485,
      "learning_rate": 5.570870473956377e-06,
      "loss": 2.2964,
      "step": 4554
    },
    {
      "epoch": 1.552488070892979,
      "grad_norm": 14.071220323692726,
      "learning_rate": 5.568900318631714e-06,
      "loss": 2.0338,
      "step": 4555
    },
    {
      "epoch": 1.5528289025221542,
      "grad_norm": 26.697938344253778,
      "learning_rate": 5.566930073816764e-06,
      "loss": 2.4725,
      "step": 4556
    },
    {
      "epoch": 1.553169734151329,
      "grad_norm": 17.798551277004922,
      "learning_rate": 5.5649597398214575e-06,
      "loss": 2.6123,
      "step": 4557
    },
    {
      "epoch": 1.5535105657805044,
      "grad_norm": 12.287513049906446,
      "learning_rate": 5.562989316955737e-06,
      "loss": 2.5775,
      "step": 4558
    },
    {
      "epoch": 1.5538513974096797,
      "grad_norm": 19.007803119440304,
      "learning_rate": 5.561018805529553e-06,
      "loss": 2.0729,
      "step": 4559
    },
    {
      "epoch": 1.5541922290388548,
      "grad_norm": 13.006272864799202,
      "learning_rate": 5.559048205852879e-06,
      "loss": 2.4404,
      "step": 4560
    },
    {
      "epoch": 1.5545330606680299,
      "grad_norm": 14.586298957211246,
      "learning_rate": 5.557077518235698e-06,
      "loss": 1.6723,
      "step": 4561
    },
    {
      "epoch": 1.5548738922972052,
      "grad_norm": 13.36579397886948,
      "learning_rate": 5.5551067429880034e-06,
      "loss": 2.175,
      "step": 4562
    },
    {
      "epoch": 1.5552147239263805,
      "grad_norm": 20.743318942079622,
      "learning_rate": 5.55313588041981e-06,
      "loss": 2.5129,
      "step": 4563
    },
    {
      "epoch": 1.5555555555555556,
      "grad_norm": 20.264671916343385,
      "learning_rate": 5.551164930841138e-06,
      "loss": 2.3455,
      "step": 4564
    },
    {
      "epoch": 1.5558963871847307,
      "grad_norm": 20.13533189770474,
      "learning_rate": 5.549193894562032e-06,
      "loss": 2.3851,
      "step": 4565
    },
    {
      "epoch": 1.556237218813906,
      "grad_norm": 14.090510022935085,
      "learning_rate": 5.547222771892539e-06,
      "loss": 2.1433,
      "step": 4566
    },
    {
      "epoch": 1.556578050443081,
      "grad_norm": 14.82879322602214,
      "learning_rate": 5.545251563142725e-06,
      "loss": 2.8661,
      "step": 4567
    },
    {
      "epoch": 1.5569188820722561,
      "grad_norm": 26.995243757317944,
      "learning_rate": 5.543280268622669e-06,
      "loss": 2.3495,
      "step": 4568
    },
    {
      "epoch": 1.5572597137014315,
      "grad_norm": 19.255254756246035,
      "learning_rate": 5.541308888642465e-06,
      "loss": 2.1623,
      "step": 4569
    },
    {
      "epoch": 1.5576005453306068,
      "grad_norm": 16.961854471561328,
      "learning_rate": 5.539337423512217e-06,
      "loss": 2.4926,
      "step": 4570
    },
    {
      "epoch": 1.5579413769597819,
      "grad_norm": 14.01331064459722,
      "learning_rate": 5.537365873542047e-06,
      "loss": 2.0223,
      "step": 4571
    },
    {
      "epoch": 1.558282208588957,
      "grad_norm": 12.123596527398922,
      "learning_rate": 5.535394239042086e-06,
      "loss": 2.3812,
      "step": 4572
    },
    {
      "epoch": 1.5586230402181322,
      "grad_norm": 14.752387093153903,
      "learning_rate": 5.533422520322478e-06,
      "loss": 2.2265,
      "step": 4573
    },
    {
      "epoch": 1.5589638718473076,
      "grad_norm": 14.940809183515194,
      "learning_rate": 5.5314507176933856e-06,
      "loss": 2.4791,
      "step": 4574
    },
    {
      "epoch": 1.5593047034764826,
      "grad_norm": 19.28870371571758,
      "learning_rate": 5.52947883146498e-06,
      "loss": 2.6672,
      "step": 4575
    },
    {
      "epoch": 1.5596455351056577,
      "grad_norm": 20.435038679828807,
      "learning_rate": 5.527506861947445e-06,
      "loss": 2.2547,
      "step": 4576
    },
    {
      "epoch": 1.559986366734833,
      "grad_norm": 16.507803246680247,
      "learning_rate": 5.525534809450981e-06,
      "loss": 2.3596,
      "step": 4577
    },
    {
      "epoch": 1.5603271983640081,
      "grad_norm": 18.418613911185275,
      "learning_rate": 5.5235626742858e-06,
      "loss": 1.9977,
      "step": 4578
    },
    {
      "epoch": 1.5606680299931832,
      "grad_norm": 15.44972011545337,
      "learning_rate": 5.521590456762126e-06,
      "loss": 2.7591,
      "step": 4579
    },
    {
      "epoch": 1.5610088616223585,
      "grad_norm": 15.814121777675343,
      "learning_rate": 5.519618157190197e-06,
      "loss": 2.2601,
      "step": 4580
    },
    {
      "epoch": 1.5613496932515338,
      "grad_norm": 12.295472619648839,
      "learning_rate": 5.517645775880263e-06,
      "loss": 1.8493,
      "step": 4581
    },
    {
      "epoch": 1.561690524880709,
      "grad_norm": 14.770501002788462,
      "learning_rate": 5.5156733131425875e-06,
      "loss": 1.5913,
      "step": 4582
    },
    {
      "epoch": 1.562031356509884,
      "grad_norm": 14.511594336066459,
      "learning_rate": 5.513700769287449e-06,
      "loss": 2.335,
      "step": 4583
    },
    {
      "epoch": 1.5623721881390593,
      "grad_norm": 21.453927189682354,
      "learning_rate": 5.511728144625133e-06,
      "loss": 1.9614,
      "step": 4584
    },
    {
      "epoch": 1.5627130197682346,
      "grad_norm": 21.877374074891087,
      "learning_rate": 5.509755439465943e-06,
      "loss": 2.1801,
      "step": 4585
    },
    {
      "epoch": 1.5630538513974097,
      "grad_norm": 14.702030830553168,
      "learning_rate": 5.507782654120193e-06,
      "loss": 2.2213,
      "step": 4586
    },
    {
      "epoch": 1.5633946830265848,
      "grad_norm": 14.000137805129189,
      "learning_rate": 5.50580978889821e-06,
      "loss": 2.6444,
      "step": 4587
    },
    {
      "epoch": 1.56373551465576,
      "grad_norm": 22.51301905484996,
      "learning_rate": 5.503836844110334e-06,
      "loss": 2.0408,
      "step": 4588
    },
    {
      "epoch": 1.5640763462849352,
      "grad_norm": 12.647322205959572,
      "learning_rate": 5.501863820066918e-06,
      "loss": 2.0935,
      "step": 4589
    },
    {
      "epoch": 1.5644171779141103,
      "grad_norm": 16.00808512692911,
      "learning_rate": 5.499890717078322e-06,
      "loss": 2.3375,
      "step": 4590
    },
    {
      "epoch": 1.5647580095432856,
      "grad_norm": 16.040408127764202,
      "learning_rate": 5.49791753545493e-06,
      "loss": 2.6628,
      "step": 4591
    },
    {
      "epoch": 1.565098841172461,
      "grad_norm": 18.37728286925699,
      "learning_rate": 5.495944275507125e-06,
      "loss": 2.3967,
      "step": 4592
    },
    {
      "epoch": 1.565439672801636,
      "grad_norm": 16.39960534739319,
      "learning_rate": 5.493970937545313e-06,
      "loss": 1.7704,
      "step": 4593
    },
    {
      "epoch": 1.565780504430811,
      "grad_norm": 26.325334377958335,
      "learning_rate": 5.4919975218799045e-06,
      "loss": 2.2746,
      "step": 4594
    },
    {
      "epoch": 1.5661213360599864,
      "grad_norm": 20.420164168115935,
      "learning_rate": 5.490024028821328e-06,
      "loss": 2.6453,
      "step": 4595
    },
    {
      "epoch": 1.5664621676891617,
      "grad_norm": 14.143236931366648,
      "learning_rate": 5.488050458680021e-06,
      "loss": 1.6972,
      "step": 4596
    },
    {
      "epoch": 1.5668029993183368,
      "grad_norm": 17.89774372202911,
      "learning_rate": 5.486076811766436e-06,
      "loss": 2.3785,
      "step": 4597
    },
    {
      "epoch": 1.5671438309475119,
      "grad_norm": 19.283277481883054,
      "learning_rate": 5.484103088391029e-06,
      "loss": 2.1286,
      "step": 4598
    },
    {
      "epoch": 1.5674846625766872,
      "grad_norm": 25.421014864645883,
      "learning_rate": 5.482129288864283e-06,
      "loss": 1.8287,
      "step": 4599
    },
    {
      "epoch": 1.5678254942058623,
      "grad_norm": 14.646514939845268,
      "learning_rate": 5.480155413496679e-06,
      "loss": 2.3781,
      "step": 4600
    },
    {
      "epoch": 1.5681663258350373,
      "grad_norm": 13.98732859142388,
      "learning_rate": 5.478181462598719e-06,
      "loss": 2.2064,
      "step": 4601
    },
    {
      "epoch": 1.5685071574642127,
      "grad_norm": 32.41951813025049,
      "learning_rate": 5.476207436480909e-06,
      "loss": 2.4675,
      "step": 4602
    },
    {
      "epoch": 1.568847989093388,
      "grad_norm": 21.941174274870537,
      "learning_rate": 5.4742333354537745e-06,
      "loss": 2.6162,
      "step": 4603
    },
    {
      "epoch": 1.569188820722563,
      "grad_norm": 11.793073749845188,
      "learning_rate": 5.472259159827848e-06,
      "loss": 1.8835,
      "step": 4604
    },
    {
      "epoch": 1.5695296523517381,
      "grad_norm": 17.346573750732396,
      "learning_rate": 5.470284909913677e-06,
      "loss": 2.4639,
      "step": 4605
    },
    {
      "epoch": 1.5698704839809134,
      "grad_norm": 17.42582630901005,
      "learning_rate": 5.468310586021817e-06,
      "loss": 3.0802,
      "step": 4606
    },
    {
      "epoch": 1.5702113156100888,
      "grad_norm": 15.401741676969037,
      "learning_rate": 5.466336188462837e-06,
      "loss": 1.8691,
      "step": 4607
    },
    {
      "epoch": 1.5705521472392638,
      "grad_norm": 11.690478941968317,
      "learning_rate": 5.4643617175473185e-06,
      "loss": 2.6479,
      "step": 4608
    },
    {
      "epoch": 1.570892978868439,
      "grad_norm": 14.777477674091458,
      "learning_rate": 5.462387173585853e-06,
      "loss": 2.5138,
      "step": 4609
    },
    {
      "epoch": 1.5712338104976142,
      "grad_norm": 27.373978050325842,
      "learning_rate": 5.4604125568890454e-06,
      "loss": 2.1911,
      "step": 4610
    },
    {
      "epoch": 1.5715746421267893,
      "grad_norm": 16.599439860731973,
      "learning_rate": 5.458437867767509e-06,
      "loss": 1.7195,
      "step": 4611
    },
    {
      "epoch": 1.5719154737559644,
      "grad_norm": 36.449446557991905,
      "learning_rate": 5.456463106531871e-06,
      "loss": 2.358,
      "step": 4612
    },
    {
      "epoch": 1.5722563053851397,
      "grad_norm": 19.085110698107933,
      "learning_rate": 5.454488273492769e-06,
      "loss": 2.5286,
      "step": 4613
    },
    {
      "epoch": 1.572597137014315,
      "grad_norm": 18.224970945778153,
      "learning_rate": 5.452513368960855e-06,
      "loss": 2.4288,
      "step": 4614
    },
    {
      "epoch": 1.5729379686434901,
      "grad_norm": 16.048249133512822,
      "learning_rate": 5.450538393246783e-06,
      "loss": 2.6852,
      "step": 4615
    },
    {
      "epoch": 1.5732788002726652,
      "grad_norm": 14.636668346392778,
      "learning_rate": 5.448563346661228e-06,
      "loss": 1.8733,
      "step": 4616
    },
    {
      "epoch": 1.5736196319018405,
      "grad_norm": 15.540391698782331,
      "learning_rate": 5.446588229514877e-06,
      "loss": 2.1636,
      "step": 4617
    },
    {
      "epoch": 1.5739604635310158,
      "grad_norm": 16.037902516657606,
      "learning_rate": 5.444613042118418e-06,
      "loss": 2.3338,
      "step": 4618
    },
    {
      "epoch": 1.574301295160191,
      "grad_norm": 20.03005865690343,
      "learning_rate": 5.442637784782556e-06,
      "loss": 2.5549,
      "step": 4619
    },
    {
      "epoch": 1.574642126789366,
      "grad_norm": 22.534853996436805,
      "learning_rate": 5.44066245781801e-06,
      "loss": 1.5909,
      "step": 4620
    },
    {
      "epoch": 1.5749829584185413,
      "grad_norm": 14.190853960766699,
      "learning_rate": 5.438687061535503e-06,
      "loss": 2.2622,
      "step": 4621
    },
    {
      "epoch": 1.5753237900477164,
      "grad_norm": 22.014241252830185,
      "learning_rate": 5.436711596245777e-06,
      "loss": 2.1066,
      "step": 4622
    },
    {
      "epoch": 1.5756646216768915,
      "grad_norm": 14.455333023933067,
      "learning_rate": 5.434736062259579e-06,
      "loss": 2.4928,
      "step": 4623
    },
    {
      "epoch": 1.5760054533060668,
      "grad_norm": 17.76242840541652,
      "learning_rate": 5.432760459887666e-06,
      "loss": 2.1982,
      "step": 4624
    },
    {
      "epoch": 1.576346284935242,
      "grad_norm": 33.272882083104676,
      "learning_rate": 5.4307847894408116e-06,
      "loss": 1.9724,
      "step": 4625
    },
    {
      "epoch": 1.5766871165644172,
      "grad_norm": 15.24615022442608,
      "learning_rate": 5.4288090512297945e-06,
      "loss": 2.5744,
      "step": 4626
    },
    {
      "epoch": 1.5770279481935923,
      "grad_norm": 22.122656958167198,
      "learning_rate": 5.426833245565406e-06,
      "loss": 2.3038,
      "step": 4627
    },
    {
      "epoch": 1.5773687798227676,
      "grad_norm": 14.14748460551022,
      "learning_rate": 5.42485737275845e-06,
      "loss": 2.3185,
      "step": 4628
    },
    {
      "epoch": 1.5777096114519429,
      "grad_norm": 14.687321056910124,
      "learning_rate": 5.422881433119737e-06,
      "loss": 2.5036,
      "step": 4629
    },
    {
      "epoch": 1.578050443081118,
      "grad_norm": 31.663759619728303,
      "learning_rate": 5.42090542696009e-06,
      "loss": 2.3224,
      "step": 4630
    },
    {
      "epoch": 1.578391274710293,
      "grad_norm": 21.960932119217066,
      "learning_rate": 5.418929354590347e-06,
      "loss": 2.1939,
      "step": 4631
    },
    {
      "epoch": 1.5787321063394684,
      "grad_norm": 20.72785634735985,
      "learning_rate": 5.416953216321345e-06,
      "loss": 2.6083,
      "step": 4632
    },
    {
      "epoch": 1.5790729379686435,
      "grad_norm": 31.731885029212318,
      "learning_rate": 5.414977012463945e-06,
      "loss": 2.1992,
      "step": 4633
    },
    {
      "epoch": 1.5794137695978185,
      "grad_norm": 12.276380973376998,
      "learning_rate": 5.413000743329009e-06,
      "loss": 1.9109,
      "step": 4634
    },
    {
      "epoch": 1.5797546012269938,
      "grad_norm": 20.470569860085885,
      "learning_rate": 5.411024409227411e-06,
      "loss": 2.52,
      "step": 4635
    },
    {
      "epoch": 1.5800954328561692,
      "grad_norm": 20.962358531567894,
      "learning_rate": 5.409048010470039e-06,
      "loss": 2.204,
      "step": 4636
    },
    {
      "epoch": 1.5804362644853442,
      "grad_norm": 22.81665795611026,
      "learning_rate": 5.407071547367784e-06,
      "loss": 2.3528,
      "step": 4637
    },
    {
      "epoch": 1.5807770961145193,
      "grad_norm": 12.939003824448685,
      "learning_rate": 5.405095020231555e-06,
      "loss": 2.1259,
      "step": 4638
    },
    {
      "epoch": 1.5811179277436946,
      "grad_norm": 29.888831792634384,
      "learning_rate": 5.403118429372267e-06,
      "loss": 2.421,
      "step": 4639
    },
    {
      "epoch": 1.58145875937287,
      "grad_norm": 15.632711955291112,
      "learning_rate": 5.4011417751008455e-06,
      "loss": 2.2353,
      "step": 4640
    },
    {
      "epoch": 1.581799591002045,
      "grad_norm": 17.4122137130586,
      "learning_rate": 5.399165057728226e-06,
      "loss": 2.4716,
      "step": 4641
    },
    {
      "epoch": 1.5821404226312201,
      "grad_norm": 19.174409827525633,
      "learning_rate": 5.397188277565355e-06,
      "loss": 2.6457,
      "step": 4642
    },
    {
      "epoch": 1.5824812542603954,
      "grad_norm": 22.00045746583584,
      "learning_rate": 5.395211434923186e-06,
      "loss": 1.8662,
      "step": 4643
    },
    {
      "epoch": 1.5828220858895705,
      "grad_norm": 18.913033993964564,
      "learning_rate": 5.393234530112684e-06,
      "loss": 2.5742,
      "step": 4644
    },
    {
      "epoch": 1.5831629175187456,
      "grad_norm": 17.012479972847277,
      "learning_rate": 5.391257563444825e-06,
      "loss": 2.387,
      "step": 4645
    },
    {
      "epoch": 1.583503749147921,
      "grad_norm": 33.54248938150719,
      "learning_rate": 5.389280535230594e-06,
      "loss": 2.368,
      "step": 4646
    },
    {
      "epoch": 1.5838445807770962,
      "grad_norm": 16.515161711463605,
      "learning_rate": 5.387303445780986e-06,
      "loss": 2.2156,
      "step": 4647
    },
    {
      "epoch": 1.5841854124062713,
      "grad_norm": 26.625199851611374,
      "learning_rate": 5.385326295407005e-06,
      "loss": 2.7548,
      "step": 4648
    },
    {
      "epoch": 1.5845262440354464,
      "grad_norm": 14.752083247463844,
      "learning_rate": 5.383349084419661e-06,
      "loss": 2.8358,
      "step": 4649
    },
    {
      "epoch": 1.5848670756646217,
      "grad_norm": 23.351771785959752,
      "learning_rate": 5.381371813129982e-06,
      "loss": 1.723,
      "step": 4650
    },
    {
      "epoch": 1.585207907293797,
      "grad_norm": 16.545803286328656,
      "learning_rate": 5.379394481848999e-06,
      "loss": 2.2993,
      "step": 4651
    },
    {
      "epoch": 1.585548738922972,
      "grad_norm": 20.851525567762618,
      "learning_rate": 5.3774170908877535e-06,
      "loss": 2.6296,
      "step": 4652
    },
    {
      "epoch": 1.5858895705521472,
      "grad_norm": 20.6575013287245,
      "learning_rate": 5.375439640557295e-06,
      "loss": 2.0242,
      "step": 4653
    },
    {
      "epoch": 1.5862304021813225,
      "grad_norm": 19.784085820823893,
      "learning_rate": 5.373462131168689e-06,
      "loss": 1.8094,
      "step": 4654
    },
    {
      "epoch": 1.5865712338104976,
      "grad_norm": 34.98160824540084,
      "learning_rate": 5.371484563033002e-06,
      "loss": 3.028,
      "step": 4655
    },
    {
      "epoch": 1.5869120654396727,
      "grad_norm": 13.747895588667932,
      "learning_rate": 5.369506936461316e-06,
      "loss": 2.2318,
      "step": 4656
    },
    {
      "epoch": 1.587252897068848,
      "grad_norm": 28.88492809379754,
      "learning_rate": 5.367529251764716e-06,
      "loss": 2.1059,
      "step": 4657
    },
    {
      "epoch": 1.5875937286980233,
      "grad_norm": 20.939272519636326,
      "learning_rate": 5.365551509254302e-06,
      "loss": 2.8905,
      "step": 4658
    },
    {
      "epoch": 1.5879345603271984,
      "grad_norm": 19.68541811777297,
      "learning_rate": 5.3635737092411825e-06,
      "loss": 2.6345,
      "step": 4659
    },
    {
      "epoch": 1.5882753919563735,
      "grad_norm": 14.97019151365539,
      "learning_rate": 5.361595852036468e-06,
      "loss": 2.1119,
      "step": 4660
    },
    {
      "epoch": 1.5886162235855488,
      "grad_norm": 24.36875551307099,
      "learning_rate": 5.359617937951289e-06,
      "loss": 2.3617,
      "step": 4661
    },
    {
      "epoch": 1.588957055214724,
      "grad_norm": 17.26557903885738,
      "learning_rate": 5.357639967296776e-06,
      "loss": 1.9659,
      "step": 4662
    },
    {
      "epoch": 1.5892978868438992,
      "grad_norm": 26.069391860040454,
      "learning_rate": 5.355661940384072e-06,
      "loss": 1.9836,
      "step": 4663
    },
    {
      "epoch": 1.5896387184730743,
      "grad_norm": 12.805942372836391,
      "learning_rate": 5.353683857524329e-06,
      "loss": 2.2034,
      "step": 4664
    },
    {
      "epoch": 1.5899795501022496,
      "grad_norm": 14.647399733685965,
      "learning_rate": 5.351705719028708e-06,
      "loss": 1.9891,
      "step": 4665
    },
    {
      "epoch": 1.5903203817314246,
      "grad_norm": 14.79863326128338,
      "learning_rate": 5.349727525208377e-06,
      "loss": 2.9953,
      "step": 4666
    },
    {
      "epoch": 1.5906612133605997,
      "grad_norm": 17.321718464026656,
      "learning_rate": 5.347749276374513e-06,
      "loss": 2.4116,
      "step": 4667
    },
    {
      "epoch": 1.591002044989775,
      "grad_norm": 17.237722513848905,
      "learning_rate": 5.345770972838306e-06,
      "loss": 2.251,
      "step": 4668
    },
    {
      "epoch": 1.5913428766189504,
      "grad_norm": 14.746371598225174,
      "learning_rate": 5.343792614910947e-06,
      "loss": 2.0302,
      "step": 4669
    },
    {
      "epoch": 1.5916837082481254,
      "grad_norm": 11.27656665031137,
      "learning_rate": 5.34181420290364e-06,
      "loss": 1.6276,
      "step": 4670
    },
    {
      "epoch": 1.5920245398773005,
      "grad_norm": 15.013949050076866,
      "learning_rate": 5.3398357371276e-06,
      "loss": 2.7505,
      "step": 4671
    },
    {
      "epoch": 1.5923653715064758,
      "grad_norm": 32.66609629689877,
      "learning_rate": 5.337857217894046e-06,
      "loss": 2.338,
      "step": 4672
    },
    {
      "epoch": 1.5927062031356511,
      "grad_norm": 15.020324791586102,
      "learning_rate": 5.335878645514206e-06,
      "loss": 2.6731,
      "step": 4673
    },
    {
      "epoch": 1.5930470347648262,
      "grad_norm": 19.34260333220306,
      "learning_rate": 5.333900020299316e-06,
      "loss": 2.5834,
      "step": 4674
    },
    {
      "epoch": 1.5933878663940013,
      "grad_norm": 12.654032536471469,
      "learning_rate": 5.331921342560627e-06,
      "loss": 2.0597,
      "step": 4675
    },
    {
      "epoch": 1.5937286980231766,
      "grad_norm": 12.134939730321133,
      "learning_rate": 5.329942612609388e-06,
      "loss": 1.7519,
      "step": 4676
    },
    {
      "epoch": 1.5940695296523517,
      "grad_norm": 20.158133418284404,
      "learning_rate": 5.327963830756863e-06,
      "loss": 1.9644,
      "step": 4677
    },
    {
      "epoch": 1.5944103612815268,
      "grad_norm": 20.96096656109158,
      "learning_rate": 5.32598499731432e-06,
      "loss": 2.0334,
      "step": 4678
    },
    {
      "epoch": 1.594751192910702,
      "grad_norm": 18.42856920966636,
      "learning_rate": 5.324006112593041e-06,
      "loss": 2.3241,
      "step": 4679
    },
    {
      "epoch": 1.5950920245398774,
      "grad_norm": 14.498728384850468,
      "learning_rate": 5.3220271769043105e-06,
      "loss": 2.239,
      "step": 4680
    },
    {
      "epoch": 1.5954328561690525,
      "grad_norm": 18.04582973732253,
      "learning_rate": 5.320048190559423e-06,
      "loss": 2.1997,
      "step": 4681
    },
    {
      "epoch": 1.5957736877982276,
      "grad_norm": 12.940081024612457,
      "learning_rate": 5.318069153869682e-06,
      "loss": 2.0214,
      "step": 4682
    },
    {
      "epoch": 1.596114519427403,
      "grad_norm": 18.342847186490165,
      "learning_rate": 5.3160900671463955e-06,
      "loss": 1.6898,
      "step": 4683
    },
    {
      "epoch": 1.5964553510565782,
      "grad_norm": 18.918886888436464,
      "learning_rate": 5.3141109307008835e-06,
      "loss": 2.7664,
      "step": 4684
    },
    {
      "epoch": 1.596796182685753,
      "grad_norm": 19.993758593123385,
      "learning_rate": 5.3121317448444735e-06,
      "loss": 1.767,
      "step": 4685
    },
    {
      "epoch": 1.5971370143149284,
      "grad_norm": 15.957796262172817,
      "learning_rate": 5.310152509888495e-06,
      "loss": 2.2264,
      "step": 4686
    },
    {
      "epoch": 1.5974778459441037,
      "grad_norm": 20.088795351504174,
      "learning_rate": 5.3081732261442916e-06,
      "loss": 2.0241,
      "step": 4687
    },
    {
      "epoch": 1.5978186775732788,
      "grad_norm": 19.615194506479,
      "learning_rate": 5.306193893923216e-06,
      "loss": 3.0074,
      "step": 4688
    },
    {
      "epoch": 1.5981595092024539,
      "grad_norm": 15.600081064851654,
      "learning_rate": 5.3042145135366195e-06,
      "loss": 2.6235,
      "step": 4689
    },
    {
      "epoch": 1.5985003408316292,
      "grad_norm": 17.106532332801283,
      "learning_rate": 5.302235085295871e-06,
      "loss": 1.7402,
      "step": 4690
    },
    {
      "epoch": 1.5988411724608045,
      "grad_norm": 18.17030682858321,
      "learning_rate": 5.300255609512338e-06,
      "loss": 2.705,
      "step": 4691
    },
    {
      "epoch": 1.5991820040899796,
      "grad_norm": 16.894671729406653,
      "learning_rate": 5.298276086497404e-06,
      "loss": 1.7992,
      "step": 4692
    },
    {
      "epoch": 1.5995228357191547,
      "grad_norm": 12.681007445466138,
      "learning_rate": 5.2962965165624545e-06,
      "loss": 1.9953,
      "step": 4693
    },
    {
      "epoch": 1.59986366734833,
      "grad_norm": 10.741763659900377,
      "learning_rate": 5.294316900018883e-06,
      "loss": 2.1954,
      "step": 4694
    },
    {
      "epoch": 1.6002044989775053,
      "grad_norm": 15.027571840301112,
      "learning_rate": 5.292337237178092e-06,
      "loss": 2.4663,
      "step": 4695
    },
    {
      "epoch": 1.6005453306066801,
      "grad_norm": 16.59236091297349,
      "learning_rate": 5.2903575283514885e-06,
      "loss": 2.7809,
      "step": 4696
    },
    {
      "epoch": 1.6008861622358554,
      "grad_norm": 33.35124859797374,
      "learning_rate": 5.28837777385049e-06,
      "loss": 2.1735,
      "step": 4697
    },
    {
      "epoch": 1.6012269938650308,
      "grad_norm": 21.708288789890354,
      "learning_rate": 5.2863979739865216e-06,
      "loss": 2.1864,
      "step": 4698
    },
    {
      "epoch": 1.6015678254942058,
      "grad_norm": 14.857351977769206,
      "learning_rate": 5.284418129071011e-06,
      "loss": 2.9296,
      "step": 4699
    },
    {
      "epoch": 1.601908657123381,
      "grad_norm": 14.074763473061298,
      "learning_rate": 5.282438239415395e-06,
      "loss": 1.9322,
      "step": 4700
    },
    {
      "epoch": 1.6022494887525562,
      "grad_norm": 14.481627065090272,
      "learning_rate": 5.280458305331121e-06,
      "loss": 1.9119,
      "step": 4701
    },
    {
      "epoch": 1.6025903203817315,
      "grad_norm": 33.67325225332307,
      "learning_rate": 5.2784783271296394e-06,
      "loss": 2.4168,
      "step": 4702
    },
    {
      "epoch": 1.6029311520109066,
      "grad_norm": 24.132126788456738,
      "learning_rate": 5.2764983051224084e-06,
      "loss": 2.4514,
      "step": 4703
    },
    {
      "epoch": 1.6032719836400817,
      "grad_norm": 37.02300922469965,
      "learning_rate": 5.274518239620892e-06,
      "loss": 2.2424,
      "step": 4704
    },
    {
      "epoch": 1.603612815269257,
      "grad_norm": 18.993172517000364,
      "learning_rate": 5.272538130936564e-06,
      "loss": 1.7784,
      "step": 4705
    },
    {
      "epoch": 1.6039536468984323,
      "grad_norm": 16.241374015988928,
      "learning_rate": 5.270557979380903e-06,
      "loss": 1.9841,
      "step": 4706
    },
    {
      "epoch": 1.6042944785276072,
      "grad_norm": 20.99607196944566,
      "learning_rate": 5.2685777852653975e-06,
      "loss": 2.4077,
      "step": 4707
    },
    {
      "epoch": 1.6046353101567825,
      "grad_norm": 36.59788171488439,
      "learning_rate": 5.266597548901534e-06,
      "loss": 2.3502,
      "step": 4708
    },
    {
      "epoch": 1.6049761417859578,
      "grad_norm": 24.28572392413443,
      "learning_rate": 5.2646172706008154e-06,
      "loss": 2.2109,
      "step": 4709
    },
    {
      "epoch": 1.605316973415133,
      "grad_norm": 25.415939668100158,
      "learning_rate": 5.262636950674749e-06,
      "loss": 2.0621,
      "step": 4710
    },
    {
      "epoch": 1.605657805044308,
      "grad_norm": 13.56751998705234,
      "learning_rate": 5.260656589434842e-06,
      "loss": 2.2551,
      "step": 4711
    },
    {
      "epoch": 1.6059986366734833,
      "grad_norm": 19.900139700215078,
      "learning_rate": 5.258676187192616e-06,
      "loss": 2.456,
      "step": 4712
    },
    {
      "epoch": 1.6063394683026586,
      "grad_norm": 16.221156369863387,
      "learning_rate": 5.256695744259598e-06,
      "loss": 2.9659,
      "step": 4713
    },
    {
      "epoch": 1.6066802999318337,
      "grad_norm": 17.818954171216124,
      "learning_rate": 5.254715260947316e-06,
      "loss": 2.2736,
      "step": 4714
    },
    {
      "epoch": 1.6070211315610088,
      "grad_norm": 19.47428759931484,
      "learning_rate": 5.252734737567311e-06,
      "loss": 1.7805,
      "step": 4715
    },
    {
      "epoch": 1.607361963190184,
      "grad_norm": 21.472074002066456,
      "learning_rate": 5.250754174431124e-06,
      "loss": 2.4977,
      "step": 4716
    },
    {
      "epoch": 1.6077027948193592,
      "grad_norm": 20.971111915184963,
      "learning_rate": 5.248773571850307e-06,
      "loss": 2.0834,
      "step": 4717
    },
    {
      "epoch": 1.6080436264485343,
      "grad_norm": 15.12183463491661,
      "learning_rate": 5.246792930136419e-06,
      "loss": 2.3588,
      "step": 4718
    },
    {
      "epoch": 1.6083844580777096,
      "grad_norm": 15.667656363243958,
      "learning_rate": 5.24481224960102e-06,
      "loss": 2.1863,
      "step": 4719
    },
    {
      "epoch": 1.6087252897068849,
      "grad_norm": 17.900783769826695,
      "learning_rate": 5.242831530555679e-06,
      "loss": 2.4117,
      "step": 4720
    },
    {
      "epoch": 1.60906612133606,
      "grad_norm": 19.178168114243654,
      "learning_rate": 5.240850773311972e-06,
      "loss": 2.1345,
      "step": 4721
    },
    {
      "epoch": 1.609406952965235,
      "grad_norm": 15.34408154091941,
      "learning_rate": 5.238869978181478e-06,
      "loss": 2.3254,
      "step": 4722
    },
    {
      "epoch": 1.6097477845944104,
      "grad_norm": 18.133759110979035,
      "learning_rate": 5.2368891454757866e-06,
      "loss": 2.9138,
      "step": 4723
    },
    {
      "epoch": 1.6100886162235857,
      "grad_norm": 20.243321198268298,
      "learning_rate": 5.234908275506489e-06,
      "loss": 2.0969,
      "step": 4724
    },
    {
      "epoch": 1.6104294478527608,
      "grad_norm": 15.572955209296648,
      "learning_rate": 5.232927368585186e-06,
      "loss": 2.2005,
      "step": 4725
    },
    {
      "epoch": 1.6107702794819359,
      "grad_norm": 14.8809692043145,
      "learning_rate": 5.23094642502348e-06,
      "loss": 2.4485,
      "step": 4726
    },
    {
      "epoch": 1.6111111111111112,
      "grad_norm": 16.777786829735565,
      "learning_rate": 5.228965445132983e-06,
      "loss": 2.8474,
      "step": 4727
    },
    {
      "epoch": 1.6114519427402862,
      "grad_norm": 33.17230873085475,
      "learning_rate": 5.226984429225309e-06,
      "loss": 2.2256,
      "step": 4728
    },
    {
      "epoch": 1.6117927743694613,
      "grad_norm": 19.142008093520378,
      "learning_rate": 5.225003377612079e-06,
      "loss": 2.038,
      "step": 4729
    },
    {
      "epoch": 1.6121336059986366,
      "grad_norm": 24.293424113706873,
      "learning_rate": 5.223022290604925e-06,
      "loss": 2.5534,
      "step": 4730
    },
    {
      "epoch": 1.612474437627812,
      "grad_norm": 13.196059410576522,
      "learning_rate": 5.2210411685154745e-06,
      "loss": 2.6308,
      "step": 4731
    },
    {
      "epoch": 1.612815269256987,
      "grad_norm": 20.6360949933755,
      "learning_rate": 5.21906001165537e-06,
      "loss": 2.0029,
      "step": 4732
    },
    {
      "epoch": 1.6131561008861621,
      "grad_norm": 19.212019041253082,
      "learning_rate": 5.217078820336253e-06,
      "loss": 2.4892,
      "step": 4733
    },
    {
      "epoch": 1.6134969325153374,
      "grad_norm": 17.321423995659565,
      "learning_rate": 5.215097594869774e-06,
      "loss": 2.7709,
      "step": 4734
    },
    {
      "epoch": 1.6138377641445127,
      "grad_norm": 20.38551578703487,
      "learning_rate": 5.213116335567586e-06,
      "loss": 2.4508,
      "step": 4735
    },
    {
      "epoch": 1.6141785957736878,
      "grad_norm": 19.86845065363426,
      "learning_rate": 5.211135042741351e-06,
      "loss": 2.3064,
      "step": 4736
    },
    {
      "epoch": 1.614519427402863,
      "grad_norm": 13.841332929360192,
      "learning_rate": 5.209153716702733e-06,
      "loss": 2.1819,
      "step": 4737
    },
    {
      "epoch": 1.6148602590320382,
      "grad_norm": 18.24039413893503,
      "learning_rate": 5.207172357763402e-06,
      "loss": 2.4102,
      "step": 4738
    },
    {
      "epoch": 1.6152010906612133,
      "grad_norm": 14.881981072411888,
      "learning_rate": 5.205190966235035e-06,
      "loss": 2.5261,
      "step": 4739
    },
    {
      "epoch": 1.6155419222903884,
      "grad_norm": 20.551582022853243,
      "learning_rate": 5.203209542429312e-06,
      "loss": 1.6278,
      "step": 4740
    },
    {
      "epoch": 1.6158827539195637,
      "grad_norm": 13.474641849243861,
      "learning_rate": 5.20122808665792e-06,
      "loss": 2.7306,
      "step": 4741
    },
    {
      "epoch": 1.616223585548739,
      "grad_norm": 18.43494496861583,
      "learning_rate": 5.199246599232548e-06,
      "loss": 2.3161,
      "step": 4742
    },
    {
      "epoch": 1.616564417177914,
      "grad_norm": 16.36862930952595,
      "learning_rate": 5.197265080464893e-06,
      "loss": 2.2114,
      "step": 4743
    },
    {
      "epoch": 1.6169052488070892,
      "grad_norm": 12.164881095641489,
      "learning_rate": 5.195283530666656e-06,
      "loss": 2.2103,
      "step": 4744
    },
    {
      "epoch": 1.6172460804362645,
      "grad_norm": 19.062949925329484,
      "learning_rate": 5.1933019501495415e-06,
      "loss": 2.2634,
      "step": 4745
    },
    {
      "epoch": 1.6175869120654398,
      "grad_norm": 13.967171043749913,
      "learning_rate": 5.19132033922526e-06,
      "loss": 2.4176,
      "step": 4746
    },
    {
      "epoch": 1.617927743694615,
      "grad_norm": 15.521960410970243,
      "learning_rate": 5.189338698205529e-06,
      "loss": 2.2158,
      "step": 4747
    },
    {
      "epoch": 1.61826857532379,
      "grad_norm": 17.011347431662664,
      "learning_rate": 5.187357027402067e-06,
      "loss": 2.1872,
      "step": 4748
    },
    {
      "epoch": 1.6186094069529653,
      "grad_norm": 20.060964160219566,
      "learning_rate": 5.185375327126599e-06,
      "loss": 2.2628,
      "step": 4749
    },
    {
      "epoch": 1.6189502385821404,
      "grad_norm": 32.72687994805059,
      "learning_rate": 5.183393597690853e-06,
      "loss": 2.9579,
      "step": 4750
    },
    {
      "epoch": 1.6192910702113155,
      "grad_norm": 24.785079292444326,
      "learning_rate": 5.181411839406565e-06,
      "loss": 2.3804,
      "step": 4751
    },
    {
      "epoch": 1.6196319018404908,
      "grad_norm": 17.37546739977131,
      "learning_rate": 5.179430052585471e-06,
      "loss": 2.47,
      "step": 4752
    },
    {
      "epoch": 1.619972733469666,
      "grad_norm": 15.909225882420248,
      "learning_rate": 5.177448237539318e-06,
      "loss": 1.9115,
      "step": 4753
    },
    {
      "epoch": 1.6203135650988412,
      "grad_norm": 17.368994538762525,
      "learning_rate": 5.175466394579849e-06,
      "loss": 1.8512,
      "step": 4754
    },
    {
      "epoch": 1.6206543967280163,
      "grad_norm": 15.512697461623949,
      "learning_rate": 5.173484524018818e-06,
      "loss": 2.7518,
      "step": 4755
    },
    {
      "epoch": 1.6209952283571916,
      "grad_norm": 37.44817097395937,
      "learning_rate": 5.171502626167981e-06,
      "loss": 2.3608,
      "step": 4756
    },
    {
      "epoch": 1.6213360599863669,
      "grad_norm": 19.241435576494794,
      "learning_rate": 5.169520701339098e-06,
      "loss": 1.745,
      "step": 4757
    },
    {
      "epoch": 1.621676891615542,
      "grad_norm": 15.731827138648343,
      "learning_rate": 5.167538749843934e-06,
      "loss": 2.3703,
      "step": 4758
    },
    {
      "epoch": 1.622017723244717,
      "grad_norm": 25.469653034956696,
      "learning_rate": 5.165556771994257e-06,
      "loss": 2.2461,
      "step": 4759
    },
    {
      "epoch": 1.6223585548738924,
      "grad_norm": 13.365726569577935,
      "learning_rate": 5.163574768101841e-06,
      "loss": 2.3313,
      "step": 4760
    },
    {
      "epoch": 1.6226993865030674,
      "grad_norm": 15.686460452573593,
      "learning_rate": 5.161592738478464e-06,
      "loss": 2.2375,
      "step": 4761
    },
    {
      "epoch": 1.6230402181322425,
      "grad_norm": 22.673660663741778,
      "learning_rate": 5.159610683435905e-06,
      "loss": 2.5936,
      "step": 4762
    },
    {
      "epoch": 1.6233810497614178,
      "grad_norm": 13.212635133517308,
      "learning_rate": 5.157628603285949e-06,
      "loss": 2.3545,
      "step": 4763
    },
    {
      "epoch": 1.6237218813905931,
      "grad_norm": 12.90440764387199,
      "learning_rate": 5.155646498340389e-06,
      "loss": 2.1459,
      "step": 4764
    },
    {
      "epoch": 1.6240627130197682,
      "grad_norm": 13.298021234710733,
      "learning_rate": 5.153664368911013e-06,
      "loss": 2.3657,
      "step": 4765
    },
    {
      "epoch": 1.6244035446489433,
      "grad_norm": 16.65360138319781,
      "learning_rate": 5.15168221530962e-06,
      "loss": 2.2958,
      "step": 4766
    },
    {
      "epoch": 1.6247443762781186,
      "grad_norm": 16.485407916263572,
      "learning_rate": 5.149700037848012e-06,
      "loss": 2.3042,
      "step": 4767
    },
    {
      "epoch": 1.625085207907294,
      "grad_norm": 15.845620752654689,
      "learning_rate": 5.147717836837991e-06,
      "loss": 2.0949,
      "step": 4768
    },
    {
      "epoch": 1.625426039536469,
      "grad_norm": 19.804021256504576,
      "learning_rate": 5.145735612591367e-06,
      "loss": 2.1761,
      "step": 4769
    },
    {
      "epoch": 1.6257668711656441,
      "grad_norm": 24.754072417566196,
      "learning_rate": 5.143753365419952e-06,
      "loss": 2.7028,
      "step": 4770
    },
    {
      "epoch": 1.6261077027948194,
      "grad_norm": 19.53357521391298,
      "learning_rate": 5.141771095635558e-06,
      "loss": 2.6186,
      "step": 4771
    },
    {
      "epoch": 1.6264485344239945,
      "grad_norm": 23.82577299945505,
      "learning_rate": 5.139788803550008e-06,
      "loss": 2.4368,
      "step": 4772
    },
    {
      "epoch": 1.6267893660531696,
      "grad_norm": 30.264751988445088,
      "learning_rate": 5.137806489475122e-06,
      "loss": 2.3501,
      "step": 4773
    },
    {
      "epoch": 1.627130197682345,
      "grad_norm": 17.12254932437324,
      "learning_rate": 5.135824153722727e-06,
      "loss": 2.3244,
      "step": 4774
    },
    {
      "epoch": 1.6274710293115202,
      "grad_norm": 15.659519394492966,
      "learning_rate": 5.133841796604651e-06,
      "loss": 2.7478,
      "step": 4775
    },
    {
      "epoch": 1.6278118609406953,
      "grad_norm": 15.007658451170252,
      "learning_rate": 5.131859418432729e-06,
      "loss": 2.4151,
      "step": 4776
    },
    {
      "epoch": 1.6281526925698704,
      "grad_norm": 16.940219180002245,
      "learning_rate": 5.1298770195187945e-06,
      "loss": 2.4073,
      "step": 4777
    },
    {
      "epoch": 1.6284935241990457,
      "grad_norm": 20.768181147983828,
      "learning_rate": 5.12789460017469e-06,
      "loss": 2.0284,
      "step": 4778
    },
    {
      "epoch": 1.628834355828221,
      "grad_norm": 12.282241845207508,
      "learning_rate": 5.1259121607122535e-06,
      "loss": 2.1163,
      "step": 4779
    },
    {
      "epoch": 1.629175187457396,
      "grad_norm": 16.62626946948372,
      "learning_rate": 5.123929701443332e-06,
      "loss": 2.6893,
      "step": 4780
    },
    {
      "epoch": 1.6295160190865712,
      "grad_norm": 14.978821886382468,
      "learning_rate": 5.121947222679777e-06,
      "loss": 2.4665,
      "step": 4781
    },
    {
      "epoch": 1.6298568507157465,
      "grad_norm": 12.037444231150701,
      "learning_rate": 5.119964724733439e-06,
      "loss": 1.9648,
      "step": 4782
    },
    {
      "epoch": 1.6301976823449216,
      "grad_norm": 15.013474527442463,
      "learning_rate": 5.11798220791617e-06,
      "loss": 1.8592,
      "step": 4783
    },
    {
      "epoch": 1.6305385139740967,
      "grad_norm": 13.809771764110652,
      "learning_rate": 5.115999672539832e-06,
      "loss": 2.0466,
      "step": 4784
    },
    {
      "epoch": 1.630879345603272,
      "grad_norm": 17.602000552611656,
      "learning_rate": 5.114017118916282e-06,
      "loss": 2.5019,
      "step": 4785
    },
    {
      "epoch": 1.6312201772324473,
      "grad_norm": 13.219674369637023,
      "learning_rate": 5.112034547357387e-06,
      "loss": 2.0652,
      "step": 4786
    },
    {
      "epoch": 1.6315610088616224,
      "grad_norm": 19.54180231760908,
      "learning_rate": 5.110051958175012e-06,
      "loss": 2.1537,
      "step": 4787
    },
    {
      "epoch": 1.6319018404907975,
      "grad_norm": 21.620645232618184,
      "learning_rate": 5.1080693516810235e-06,
      "loss": 2.5667,
      "step": 4788
    },
    {
      "epoch": 1.6322426721199728,
      "grad_norm": 14.595476524823887,
      "learning_rate": 5.106086728187299e-06,
      "loss": 2.4769,
      "step": 4789
    },
    {
      "epoch": 1.632583503749148,
      "grad_norm": 23.228747608346424,
      "learning_rate": 5.104104088005707e-06,
      "loss": 2.1188,
      "step": 4790
    },
    {
      "epoch": 1.6329243353783232,
      "grad_norm": 23.10978846429566,
      "learning_rate": 5.102121431448128e-06,
      "loss": 2.699,
      "step": 4791
    },
    {
      "epoch": 1.6332651670074982,
      "grad_norm": 15.725475298502737,
      "learning_rate": 5.100138758826443e-06,
      "loss": 1.7101,
      "step": 4792
    },
    {
      "epoch": 1.6336059986366736,
      "grad_norm": 17.439045581009527,
      "learning_rate": 5.09815607045253e-06,
      "loss": 2.1544,
      "step": 4793
    },
    {
      "epoch": 1.6339468302658486,
      "grad_norm": 18.260646434981513,
      "learning_rate": 5.096173366638278e-06,
      "loss": 1.599,
      "step": 4794
    },
    {
      "epoch": 1.6342876618950237,
      "grad_norm": 15.470465157883865,
      "learning_rate": 5.094190647695574e-06,
      "loss": 2.0468,
      "step": 4795
    },
    {
      "epoch": 1.634628493524199,
      "grad_norm": 11.945594353890595,
      "learning_rate": 5.092207913936305e-06,
      "loss": 1.7252,
      "step": 4796
    },
    {
      "epoch": 1.6349693251533743,
      "grad_norm": 16.725441609671837,
      "learning_rate": 5.090225165672363e-06,
      "loss": 2.1198,
      "step": 4797
    },
    {
      "epoch": 1.6353101567825494,
      "grad_norm": 22.729403300419904,
      "learning_rate": 5.088242403215644e-06,
      "loss": 2.2904,
      "step": 4798
    },
    {
      "epoch": 1.6356509884117245,
      "grad_norm": 15.897247407412719,
      "learning_rate": 5.086259626878044e-06,
      "loss": 2.122,
      "step": 4799
    },
    {
      "epoch": 1.6359918200408998,
      "grad_norm": 20.711543053587665,
      "learning_rate": 5.08427683697146e-06,
      "loss": 2.9403,
      "step": 4800
    },
    {
      "epoch": 1.6363326516700751,
      "grad_norm": 22.522529841669037,
      "learning_rate": 5.082294033807795e-06,
      "loss": 2.6746,
      "step": 4801
    },
    {
      "epoch": 1.6366734832992502,
      "grad_norm": 22.101987064961193,
      "learning_rate": 5.080311217698951e-06,
      "loss": 2.4564,
      "step": 4802
    },
    {
      "epoch": 1.6370143149284253,
      "grad_norm": 32.90558934216906,
      "learning_rate": 5.078328388956833e-06,
      "loss": 1.9571,
      "step": 4803
    },
    {
      "epoch": 1.6373551465576006,
      "grad_norm": 19.14793507257964,
      "learning_rate": 5.076345547893348e-06,
      "loss": 2.1909,
      "step": 4804
    },
    {
      "epoch": 1.6376959781867757,
      "grad_norm": 18.352457655197064,
      "learning_rate": 5.074362694820403e-06,
      "loss": 2.0345,
      "step": 4805
    },
    {
      "epoch": 1.6380368098159508,
      "grad_norm": 15.089267865227239,
      "learning_rate": 5.072379830049911e-06,
      "loss": 2.1916,
      "step": 4806
    },
    {
      "epoch": 1.638377641445126,
      "grad_norm": 23.966183511737448,
      "learning_rate": 5.070396953893783e-06,
      "loss": 1.9454,
      "step": 4807
    },
    {
      "epoch": 1.6387184730743014,
      "grad_norm": 28.210483462763236,
      "learning_rate": 5.0684140666639345e-06,
      "loss": 2.1839,
      "step": 4808
    },
    {
      "epoch": 1.6390593047034765,
      "grad_norm": 24.737500107936814,
      "learning_rate": 5.0664311686722815e-06,
      "loss": 2.0775,
      "step": 4809
    },
    {
      "epoch": 1.6394001363326516,
      "grad_norm": 19.993980958664725,
      "learning_rate": 5.06444826023074e-06,
      "loss": 2.6643,
      "step": 4810
    },
    {
      "epoch": 1.639740967961827,
      "grad_norm": 14.833239156404161,
      "learning_rate": 5.062465341651232e-06,
      "loss": 2.2405,
      "step": 4811
    },
    {
      "epoch": 1.6400817995910022,
      "grad_norm": 15.821212108749355,
      "learning_rate": 5.0604824132456765e-06,
      "loss": 2.1622,
      "step": 4812
    },
    {
      "epoch": 1.6404226312201773,
      "grad_norm": 23.560491151451604,
      "learning_rate": 5.0584994753259966e-06,
      "loss": 2.3103,
      "step": 4813
    },
    {
      "epoch": 1.6407634628493524,
      "grad_norm": 16.852690996721776,
      "learning_rate": 5.056516528204114e-06,
      "loss": 2.3291,
      "step": 4814
    },
    {
      "epoch": 1.6411042944785277,
      "grad_norm": 24.007437716883278,
      "learning_rate": 5.0545335721919596e-06,
      "loss": 2.849,
      "step": 4815
    },
    {
      "epoch": 1.6414451261077028,
      "grad_norm": 24.396171592605988,
      "learning_rate": 5.052550607601457e-06,
      "loss": 2.054,
      "step": 4816
    },
    {
      "epoch": 1.6417859577368779,
      "grad_norm": 15.383905501698818,
      "learning_rate": 5.050567634744533e-06,
      "loss": 2.5014,
      "step": 4817
    },
    {
      "epoch": 1.6421267893660532,
      "grad_norm": 16.73178780697449,
      "learning_rate": 5.04858465393312e-06,
      "loss": 2.1528,
      "step": 4818
    },
    {
      "epoch": 1.6424676209952285,
      "grad_norm": 16.02832814173762,
      "learning_rate": 5.046601665479148e-06,
      "loss": 2.1172,
      "step": 4819
    },
    {
      "epoch": 1.6428084526244036,
      "grad_norm": 19.603000541777494,
      "learning_rate": 5.044618669694547e-06,
      "loss": 2.5521,
      "step": 4820
    },
    {
      "epoch": 1.6431492842535786,
      "grad_norm": 13.568436781013848,
      "learning_rate": 5.0426356668912525e-06,
      "loss": 2.435,
      "step": 4821
    },
    {
      "epoch": 1.643490115882754,
      "grad_norm": 14.93610055903479,
      "learning_rate": 5.0406526573811966e-06,
      "loss": 2.1941,
      "step": 4822
    },
    {
      "epoch": 1.6438309475119293,
      "grad_norm": 23.322803826595266,
      "learning_rate": 5.038669641476318e-06,
      "loss": 1.9353,
      "step": 4823
    },
    {
      "epoch": 1.6441717791411041,
      "grad_norm": 16.156614640526254,
      "learning_rate": 5.03668661948855e-06,
      "loss": 2.4958,
      "step": 4824
    },
    {
      "epoch": 1.6445126107702794,
      "grad_norm": 25.82073162175326,
      "learning_rate": 5.03470359172983e-06,
      "loss": 3.171,
      "step": 4825
    },
    {
      "epoch": 1.6448534423994547,
      "grad_norm": 30.310682296511363,
      "learning_rate": 5.032720558512098e-06,
      "loss": 2.5145,
      "step": 4826
    },
    {
      "epoch": 1.6451942740286298,
      "grad_norm": 15.585772932654487,
      "learning_rate": 5.030737520147291e-06,
      "loss": 2.2444,
      "step": 4827
    },
    {
      "epoch": 1.645535105657805,
      "grad_norm": 14.101216544507487,
      "learning_rate": 5.028754476947349e-06,
      "loss": 2.4341,
      "step": 4828
    },
    {
      "epoch": 1.6458759372869802,
      "grad_norm": 18.895827940972733,
      "learning_rate": 5.026771429224217e-06,
      "loss": 2.3236,
      "step": 4829
    },
    {
      "epoch": 1.6462167689161555,
      "grad_norm": 17.835604882777584,
      "learning_rate": 5.024788377289828e-06,
      "loss": 1.96,
      "step": 4830
    },
    {
      "epoch": 1.6465576005453306,
      "grad_norm": 17.673852774192852,
      "learning_rate": 5.02280532145613e-06,
      "loss": 2.7025,
      "step": 4831
    },
    {
      "epoch": 1.6468984321745057,
      "grad_norm": 15.741174190530963,
      "learning_rate": 5.020822262035065e-06,
      "loss": 2.1089,
      "step": 4832
    },
    {
      "epoch": 1.647239263803681,
      "grad_norm": 17.225769688678888,
      "learning_rate": 5.018839199338576e-06,
      "loss": 1.8117,
      "step": 4833
    },
    {
      "epoch": 1.6475800954328563,
      "grad_norm": 24.384903442323818,
      "learning_rate": 5.016856133678605e-06,
      "loss": 2.4974,
      "step": 4834
    },
    {
      "epoch": 1.6479209270620312,
      "grad_norm": 22.484074069174945,
      "learning_rate": 5.0148730653670965e-06,
      "loss": 1.9383,
      "step": 4835
    },
    {
      "epoch": 1.6482617586912065,
      "grad_norm": 31.134001950738526,
      "learning_rate": 5.012889994715996e-06,
      "loss": 2.5408,
      "step": 4836
    },
    {
      "epoch": 1.6486025903203818,
      "grad_norm": 21.10057525019606,
      "learning_rate": 5.01090692203725e-06,
      "loss": 2.4163,
      "step": 4837
    },
    {
      "epoch": 1.648943421949557,
      "grad_norm": 15.171681661219187,
      "learning_rate": 5.008923847642799e-06,
      "loss": 1.9838,
      "step": 4838
    },
    {
      "epoch": 1.649284253578732,
      "grad_norm": 22.634330841862422,
      "learning_rate": 5.006940771844591e-06,
      "loss": 2.2104,
      "step": 4839
    },
    {
      "epoch": 1.6496250852079073,
      "grad_norm": 23.89552132949375,
      "learning_rate": 5.004957694954574e-06,
      "loss": 1.7793,
      "step": 4840
    },
    {
      "epoch": 1.6499659168370826,
      "grad_norm": 15.868331440515032,
      "learning_rate": 5.00297461728469e-06,
      "loss": 2.1969,
      "step": 4841
    },
    {
      "epoch": 1.6503067484662577,
      "grad_norm": 30.840714463703495,
      "learning_rate": 5.000991539146889e-06,
      "loss": 1.8217,
      "step": 4842
    },
    {
      "epoch": 1.6506475800954328,
      "grad_norm": 11.655584201273205,
      "learning_rate": 4.9990084608531135e-06,
      "loss": 1.8678,
      "step": 4843
    },
    {
      "epoch": 1.650988411724608,
      "grad_norm": 17.745952633838357,
      "learning_rate": 4.9970253827153106e-06,
      "loss": 2.3801,
      "step": 4844
    },
    {
      "epoch": 1.6513292433537834,
      "grad_norm": 13.256273736349959,
      "learning_rate": 4.995042305045429e-06,
      "loss": 2.1993,
      "step": 4845
    },
    {
      "epoch": 1.6516700749829583,
      "grad_norm": 17.24608164002074,
      "learning_rate": 4.99305922815541e-06,
      "loss": 2.0378,
      "step": 4846
    },
    {
      "epoch": 1.6520109066121336,
      "grad_norm": 18.912909329330212,
      "learning_rate": 4.9910761523572025e-06,
      "loss": 1.8693,
      "step": 4847
    },
    {
      "epoch": 1.6523517382413089,
      "grad_norm": 13.849034057001004,
      "learning_rate": 4.989093077962753e-06,
      "loss": 1.8478,
      "step": 4848
    },
    {
      "epoch": 1.652692569870484,
      "grad_norm": 17.74369819876824,
      "learning_rate": 4.987110005284004e-06,
      "loss": 2.4935,
      "step": 4849
    },
    {
      "epoch": 1.653033401499659,
      "grad_norm": 15.250830906937846,
      "learning_rate": 4.985126934632905e-06,
      "loss": 2.4647,
      "step": 4850
    },
    {
      "epoch": 1.6533742331288344,
      "grad_norm": 15.632411789692027,
      "learning_rate": 4.983143866321397e-06,
      "loss": 1.8764,
      "step": 4851
    },
    {
      "epoch": 1.6537150647580097,
      "grad_norm": 22.09949710095985,
      "learning_rate": 4.981160800661424e-06,
      "loss": 2.0234,
      "step": 4852
    },
    {
      "epoch": 1.6540558963871848,
      "grad_norm": 15.61977871313864,
      "learning_rate": 4.979177737964935e-06,
      "loss": 2.1148,
      "step": 4853
    },
    {
      "epoch": 1.6543967280163598,
      "grad_norm": 13.089205702701424,
      "learning_rate": 4.977194678543871e-06,
      "loss": 2.0548,
      "step": 4854
    },
    {
      "epoch": 1.6547375596455351,
      "grad_norm": 23.514249481844537,
      "learning_rate": 4.975211622710173e-06,
      "loss": 1.567,
      "step": 4855
    },
    {
      "epoch": 1.6550783912747105,
      "grad_norm": 17.69959200609305,
      "learning_rate": 4.973228570775786e-06,
      "loss": 2.813,
      "step": 4856
    },
    {
      "epoch": 1.6554192229038853,
      "grad_norm": 17.352051255263866,
      "learning_rate": 4.9712455230526516e-06,
      "loss": 2.2817,
      "step": 4857
    },
    {
      "epoch": 1.6557600545330606,
      "grad_norm": 20.436005830405563,
      "learning_rate": 4.96926247985271e-06,
      "loss": 2.1565,
      "step": 4858
    },
    {
      "epoch": 1.656100886162236,
      "grad_norm": 20.400297859309894,
      "learning_rate": 4.967279441487904e-06,
      "loss": 3.1033,
      "step": 4859
    },
    {
      "epoch": 1.656441717791411,
      "grad_norm": 21.393142278212867,
      "learning_rate": 4.965296408270171e-06,
      "loss": 2.2849,
      "step": 4860
    },
    {
      "epoch": 1.6567825494205861,
      "grad_norm": 14.433747193069165,
      "learning_rate": 4.963313380511452e-06,
      "loss": 1.9965,
      "step": 4861
    },
    {
      "epoch": 1.6571233810497614,
      "grad_norm": 26.98246370705561,
      "learning_rate": 4.961330358523684e-06,
      "loss": 2.3008,
      "step": 4862
    },
    {
      "epoch": 1.6574642126789367,
      "grad_norm": 24.332708900590536,
      "learning_rate": 4.959347342618804e-06,
      "loss": 3.3244,
      "step": 4863
    },
    {
      "epoch": 1.6578050443081118,
      "grad_norm": 16.542521763664105,
      "learning_rate": 4.957364333108749e-06,
      "loss": 2.402,
      "step": 4864
    },
    {
      "epoch": 1.658145875937287,
      "grad_norm": 17.46628913074591,
      "learning_rate": 4.9553813303054535e-06,
      "loss": 1.4389,
      "step": 4865
    },
    {
      "epoch": 1.6584867075664622,
      "grad_norm": 18.656604709073857,
      "learning_rate": 4.953398334520855e-06,
      "loss": 2.146,
      "step": 4866
    },
    {
      "epoch": 1.6588275391956373,
      "grad_norm": 11.509538978170022,
      "learning_rate": 4.951415346066883e-06,
      "loss": 2.1461,
      "step": 4867
    },
    {
      "epoch": 1.6591683708248124,
      "grad_norm": 17.168822705537114,
      "learning_rate": 4.9494323652554685e-06,
      "loss": 2.1975,
      "step": 4868
    },
    {
      "epoch": 1.6595092024539877,
      "grad_norm": 23.197580602784573,
      "learning_rate": 4.947449392398545e-06,
      "loss": 1.3506,
      "step": 4869
    },
    {
      "epoch": 1.659850034083163,
      "grad_norm": 15.456505731961155,
      "learning_rate": 4.945466427808043e-06,
      "loss": 2.1847,
      "step": 4870
    },
    {
      "epoch": 1.660190865712338,
      "grad_norm": 21.809131500036912,
      "learning_rate": 4.943483471795887e-06,
      "loss": 1.7611,
      "step": 4871
    },
    {
      "epoch": 1.6605316973415132,
      "grad_norm": 12.523364374652981,
      "learning_rate": 4.941500524674006e-06,
      "loss": 1.8492,
      "step": 4872
    },
    {
      "epoch": 1.6608725289706885,
      "grad_norm": 16.024018531676717,
      "learning_rate": 4.939517586754326e-06,
      "loss": 2.1231,
      "step": 4873
    },
    {
      "epoch": 1.6612133605998638,
      "grad_norm": 16.975222806112605,
      "learning_rate": 4.937534658348769e-06,
      "loss": 2.1385,
      "step": 4874
    },
    {
      "epoch": 1.6615541922290389,
      "grad_norm": 15.541730796333379,
      "learning_rate": 4.935551739769261e-06,
      "loss": 2.2677,
      "step": 4875
    },
    {
      "epoch": 1.661895023858214,
      "grad_norm": 14.048063722393826,
      "learning_rate": 4.93356883132772e-06,
      "loss": 2.2127,
      "step": 4876
    },
    {
      "epoch": 1.6622358554873893,
      "grad_norm": 11.877094847879551,
      "learning_rate": 4.9315859333360654e-06,
      "loss": 1.9457,
      "step": 4877
    },
    {
      "epoch": 1.6625766871165644,
      "grad_norm": 49.192800440008504,
      "learning_rate": 4.9296030461062175e-06,
      "loss": 2.2148,
      "step": 4878
    },
    {
      "epoch": 1.6629175187457395,
      "grad_norm": 17.765849953388358,
      "learning_rate": 4.927620169950091e-06,
      "loss": 1.7352,
      "step": 4879
    },
    {
      "epoch": 1.6632583503749148,
      "grad_norm": 12.929007551086661,
      "learning_rate": 4.925637305179597e-06,
      "loss": 2.176,
      "step": 4880
    },
    {
      "epoch": 1.66359918200409,
      "grad_norm": 15.967093257551834,
      "learning_rate": 4.923654452106653e-06,
      "loss": 2.2764,
      "step": 4881
    },
    {
      "epoch": 1.6639400136332652,
      "grad_norm": 23.99230646759936,
      "learning_rate": 4.9216716110431685e-06,
      "loss": 2.6787,
      "step": 4882
    },
    {
      "epoch": 1.6642808452624402,
      "grad_norm": 17.7134454644471,
      "learning_rate": 4.919688782301049e-06,
      "loss": 2.0142,
      "step": 4883
    },
    {
      "epoch": 1.6646216768916156,
      "grad_norm": 22.315477637627293,
      "learning_rate": 4.917705966192206e-06,
      "loss": 2.2634,
      "step": 4884
    },
    {
      "epoch": 1.6649625085207909,
      "grad_norm": 31.273217038731897,
      "learning_rate": 4.915723163028541e-06,
      "loss": 1.419,
      "step": 4885
    },
    {
      "epoch": 1.665303340149966,
      "grad_norm": 13.477989815617509,
      "learning_rate": 4.9137403731219565e-06,
      "loss": 2.0024,
      "step": 4886
    },
    {
      "epoch": 1.665644171779141,
      "grad_norm": 20.655423807096867,
      "learning_rate": 4.911757596784358e-06,
      "loss": 1.5946,
      "step": 4887
    },
    {
      "epoch": 1.6659850034083163,
      "grad_norm": 14.724358176768634,
      "learning_rate": 4.90977483432764e-06,
      "loss": 1.7564,
      "step": 4888
    },
    {
      "epoch": 1.6663258350374914,
      "grad_norm": 18.843859304057467,
      "learning_rate": 4.907792086063697e-06,
      "loss": 2.1621,
      "step": 4889
    },
    {
      "epoch": 1.6666666666666665,
      "grad_norm": 15.310754677905447,
      "learning_rate": 4.905809352304428e-06,
      "loss": 2.4078,
      "step": 4890
    },
    {
      "epoch": 1.6670074982958418,
      "grad_norm": 15.969084549118236,
      "learning_rate": 4.903826633361723e-06,
      "loss": 2.5587,
      "step": 4891
    },
    {
      "epoch": 1.6673483299250171,
      "grad_norm": 16.978205620895732,
      "learning_rate": 4.901843929547472e-06,
      "loss": 2.4837,
      "step": 4892
    },
    {
      "epoch": 1.6676891615541922,
      "grad_norm": 15.272938264457052,
      "learning_rate": 4.899861241173559e-06,
      "loss": 1.775,
      "step": 4893
    },
    {
      "epoch": 1.6680299931833673,
      "grad_norm": 16.264205845074425,
      "learning_rate": 4.897878568551873e-06,
      "loss": 2.1323,
      "step": 4894
    },
    {
      "epoch": 1.6683708248125426,
      "grad_norm": 19.52549980377032,
      "learning_rate": 4.895895911994296e-06,
      "loss": 2.5831,
      "step": 4895
    },
    {
      "epoch": 1.668711656441718,
      "grad_norm": 14.199883066482922,
      "learning_rate": 4.893913271812704e-06,
      "loss": 2.3024,
      "step": 4896
    },
    {
      "epoch": 1.669052488070893,
      "grad_norm": 20.4515055655972,
      "learning_rate": 4.891930648318978e-06,
      "loss": 2.4019,
      "step": 4897
    },
    {
      "epoch": 1.669393319700068,
      "grad_norm": 14.167507544485932,
      "learning_rate": 4.88994804182499e-06,
      "loss": 2.2696,
      "step": 4898
    },
    {
      "epoch": 1.6697341513292434,
      "grad_norm": 27.947327597860966,
      "learning_rate": 4.8879654526426135e-06,
      "loss": 2.5619,
      "step": 4899
    },
    {
      "epoch": 1.6700749829584185,
      "grad_norm": 15.295747060212172,
      "learning_rate": 4.885982881083719e-06,
      "loss": 2.163,
      "step": 4900
    },
    {
      "epoch": 1.6704158145875936,
      "grad_norm": 35.158550609064946,
      "learning_rate": 4.88400032746017e-06,
      "loss": 2.7459,
      "step": 4901
    },
    {
      "epoch": 1.670756646216769,
      "grad_norm": 29.894273263985298,
      "learning_rate": 4.882017792083831e-06,
      "loss": 2.5929,
      "step": 4902
    },
    {
      "epoch": 1.6710974778459442,
      "grad_norm": 15.93299947517147,
      "learning_rate": 4.880035275266563e-06,
      "loss": 2.3599,
      "step": 4903
    },
    {
      "epoch": 1.6714383094751193,
      "grad_norm": 16.075908713881255,
      "learning_rate": 4.878052777320225e-06,
      "loss": 2.4784,
      "step": 4904
    },
    {
      "epoch": 1.6717791411042944,
      "grad_norm": 13.742616981462417,
      "learning_rate": 4.876070298556669e-06,
      "loss": 1.8688,
      "step": 4905
    },
    {
      "epoch": 1.6721199727334697,
      "grad_norm": 18.26330189049245,
      "learning_rate": 4.874087839287748e-06,
      "loss": 2.3571,
      "step": 4906
    },
    {
      "epoch": 1.672460804362645,
      "grad_norm": 20.842041067755485,
      "learning_rate": 4.8721053998253134e-06,
      "loss": 3.1098,
      "step": 4907
    },
    {
      "epoch": 1.67280163599182,
      "grad_norm": 14.76931700122803,
      "learning_rate": 4.8701229804812055e-06,
      "loss": 1.9781,
      "step": 4908
    },
    {
      "epoch": 1.6731424676209952,
      "grad_norm": 23.298843285178013,
      "learning_rate": 4.868140581567273e-06,
      "loss": 2.2187,
      "step": 4909
    },
    {
      "epoch": 1.6734832992501705,
      "grad_norm": 13.559466353436344,
      "learning_rate": 4.8661582033953494e-06,
      "loss": 1.8774,
      "step": 4910
    },
    {
      "epoch": 1.6738241308793456,
      "grad_norm": 29.85295285903798,
      "learning_rate": 4.8641758462772735e-06,
      "loss": 2.2068,
      "step": 4911
    },
    {
      "epoch": 1.6741649625085206,
      "grad_norm": 15.342913604091422,
      "learning_rate": 4.862193510524879e-06,
      "loss": 2.3586,
      "step": 4912
    },
    {
      "epoch": 1.674505794137696,
      "grad_norm": 16.76933419345027,
      "learning_rate": 4.860211196449994e-06,
      "loss": 2.0511,
      "step": 4913
    },
    {
      "epoch": 1.6748466257668713,
      "grad_norm": 28.02869085349051,
      "learning_rate": 4.858228904364443e-06,
      "loss": 2.4141,
      "step": 4914
    },
    {
      "epoch": 1.6751874573960464,
      "grad_norm": 17.689455764666764,
      "learning_rate": 4.85624663458005e-06,
      "loss": 2.5473,
      "step": 4915
    },
    {
      "epoch": 1.6755282890252214,
      "grad_norm": 26.303027360194918,
      "learning_rate": 4.854264387408634e-06,
      "loss": 2.1478,
      "step": 4916
    },
    {
      "epoch": 1.6758691206543967,
      "grad_norm": 12.569721863256435,
      "learning_rate": 4.8522821631620085e-06,
      "loss": 2.3867,
      "step": 4917
    },
    {
      "epoch": 1.676209952283572,
      "grad_norm": 14.926095315457456,
      "learning_rate": 4.8502999621519895e-06,
      "loss": 2.4966,
      "step": 4918
    },
    {
      "epoch": 1.6765507839127471,
      "grad_norm": 16.883433173491202,
      "learning_rate": 4.848317784690381e-06,
      "loss": 2.8784,
      "step": 4919
    },
    {
      "epoch": 1.6768916155419222,
      "grad_norm": 20.769540457335854,
      "learning_rate": 4.8463356310889874e-06,
      "loss": 2.2929,
      "step": 4920
    },
    {
      "epoch": 1.6772324471710975,
      "grad_norm": 14.65173236100568,
      "learning_rate": 4.844353501659613e-06,
      "loss": 2.1918,
      "step": 4921
    },
    {
      "epoch": 1.6775732788002726,
      "grad_norm": 14.57057199930505,
      "learning_rate": 4.842371396714053e-06,
      "loss": 2.2541,
      "step": 4922
    },
    {
      "epoch": 1.6779141104294477,
      "grad_norm": 13.404638649321324,
      "learning_rate": 4.840389316564096e-06,
      "loss": 2.5274,
      "step": 4923
    },
    {
      "epoch": 1.678254942058623,
      "grad_norm": 20.777193294147885,
      "learning_rate": 4.838407261521537e-06,
      "loss": 2.61,
      "step": 4924
    },
    {
      "epoch": 1.6785957736877983,
      "grad_norm": 15.855991366823298,
      "learning_rate": 4.83642523189816e-06,
      "loss": 2.5258,
      "step": 4925
    },
    {
      "epoch": 1.6789366053169734,
      "grad_norm": 19.642929434680763,
      "learning_rate": 4.834443228005746e-06,
      "loss": 2.6445,
      "step": 4926
    },
    {
      "epoch": 1.6792774369461485,
      "grad_norm": 10.675065410072174,
      "learning_rate": 4.832461250156068e-06,
      "loss": 1.8268,
      "step": 4927
    },
    {
      "epoch": 1.6796182685753238,
      "grad_norm": 15.263062645879616,
      "learning_rate": 4.830479298660904e-06,
      "loss": 2.2715,
      "step": 4928
    },
    {
      "epoch": 1.6799591002044991,
      "grad_norm": 24.402358715476627,
      "learning_rate": 4.828497373832022e-06,
      "loss": 1.9165,
      "step": 4929
    },
    {
      "epoch": 1.6802999318336742,
      "grad_norm": 13.881597047055982,
      "learning_rate": 4.826515475981183e-06,
      "loss": 2.2159,
      "step": 4930
    },
    {
      "epoch": 1.6806407634628493,
      "grad_norm": 22.106003916923964,
      "learning_rate": 4.824533605420152e-06,
      "loss": 2.1401,
      "step": 4931
    },
    {
      "epoch": 1.6809815950920246,
      "grad_norm": 13.382691782157409,
      "learning_rate": 4.822551762460685e-06,
      "loss": 1.8347,
      "step": 4932
    },
    {
      "epoch": 1.6813224267211997,
      "grad_norm": 17.2861720002287,
      "learning_rate": 4.820569947414529e-06,
      "loss": 2.2032,
      "step": 4933
    },
    {
      "epoch": 1.6816632583503748,
      "grad_norm": 17.890966198420244,
      "learning_rate": 4.818588160593437e-06,
      "loss": 2.076,
      "step": 4934
    },
    {
      "epoch": 1.68200408997955,
      "grad_norm": 21.049524315489254,
      "learning_rate": 4.816606402309148e-06,
      "loss": 2.2251,
      "step": 4935
    },
    {
      "epoch": 1.6823449216087254,
      "grad_norm": 15.818198016967274,
      "learning_rate": 4.814624672873401e-06,
      "loss": 1.809,
      "step": 4936
    },
    {
      "epoch": 1.6826857532379005,
      "grad_norm": 13.797913196634136,
      "learning_rate": 4.812642972597934e-06,
      "loss": 2.0723,
      "step": 4937
    },
    {
      "epoch": 1.6830265848670756,
      "grad_norm": 15.487075623181802,
      "learning_rate": 4.8106613017944725e-06,
      "loss": 1.7367,
      "step": 4938
    },
    {
      "epoch": 1.6833674164962509,
      "grad_norm": 14.990970216205517,
      "learning_rate": 4.80867966077474e-06,
      "loss": 2.6722,
      "step": 4939
    },
    {
      "epoch": 1.6837082481254262,
      "grad_norm": 17.345450390318447,
      "learning_rate": 4.806698049850459e-06,
      "loss": 2.1805,
      "step": 4940
    },
    {
      "epoch": 1.6840490797546013,
      "grad_norm": 15.835540681307156,
      "learning_rate": 4.804716469333347e-06,
      "loss": 2.2509,
      "step": 4941
    },
    {
      "epoch": 1.6843899113837764,
      "grad_norm": 18.751722228196975,
      "learning_rate": 4.802734919535108e-06,
      "loss": 2.594,
      "step": 4942
    },
    {
      "epoch": 1.6847307430129517,
      "grad_norm": 26.41975549366376,
      "learning_rate": 4.8007534007674536e-06,
      "loss": 1.7505,
      "step": 4943
    },
    {
      "epoch": 1.6850715746421268,
      "grad_norm": 16.01807754570169,
      "learning_rate": 4.798771913342082e-06,
      "loss": 2.241,
      "step": 4944
    },
    {
      "epoch": 1.6854124062713018,
      "grad_norm": 14.544618701257948,
      "learning_rate": 4.796790457570688e-06,
      "loss": 2.0941,
      "step": 4945
    },
    {
      "epoch": 1.6857532379004772,
      "grad_norm": 20.67356058323263,
      "learning_rate": 4.794809033764966e-06,
      "loss": 2.8077,
      "step": 4946
    },
    {
      "epoch": 1.6860940695296525,
      "grad_norm": 12.96968645630812,
      "learning_rate": 4.7928276422366e-06,
      "loss": 2.3481,
      "step": 4947
    },
    {
      "epoch": 1.6864349011588275,
      "grad_norm": 17.324300850863715,
      "learning_rate": 4.790846283297268e-06,
      "loss": 2.6171,
      "step": 4948
    },
    {
      "epoch": 1.6867757327880026,
      "grad_norm": 12.510058045961754,
      "learning_rate": 4.788864957258651e-06,
      "loss": 2.2917,
      "step": 4949
    },
    {
      "epoch": 1.687116564417178,
      "grad_norm": 12.173541806070626,
      "learning_rate": 4.786883664432415e-06,
      "loss": 1.9895,
      "step": 4950
    },
    {
      "epoch": 1.6874573960463533,
      "grad_norm": 19.065045230317388,
      "learning_rate": 4.784902405130227e-06,
      "loss": 2.2342,
      "step": 4951
    },
    {
      "epoch": 1.6877982276755283,
      "grad_norm": 16.725864813280253,
      "learning_rate": 4.782921179663749e-06,
      "loss": 2.2417,
      "step": 4952
    },
    {
      "epoch": 1.6881390593047034,
      "grad_norm": 14.96871736121806,
      "learning_rate": 4.7809399883446315e-06,
      "loss": 2.708,
      "step": 4953
    },
    {
      "epoch": 1.6884798909338787,
      "grad_norm": 16.13075201561084,
      "learning_rate": 4.778958831484525e-06,
      "loss": 2.1985,
      "step": 4954
    },
    {
      "epoch": 1.6888207225630538,
      "grad_norm": 16.394742029215042,
      "learning_rate": 4.776977709395076e-06,
      "loss": 1.9046,
      "step": 4955
    },
    {
      "epoch": 1.689161554192229,
      "grad_norm": 18.456901939349812,
      "learning_rate": 4.774996622387922e-06,
      "loss": 2.6707,
      "step": 4956
    },
    {
      "epoch": 1.6895023858214042,
      "grad_norm": 13.803279661159113,
      "learning_rate": 4.773015570774694e-06,
      "loss": 2.5971,
      "step": 4957
    },
    {
      "epoch": 1.6898432174505795,
      "grad_norm": 16.893487807995882,
      "learning_rate": 4.771034554867018e-06,
      "loss": 1.7319,
      "step": 4958
    },
    {
      "epoch": 1.6901840490797546,
      "grad_norm": 15.23408464182785,
      "learning_rate": 4.7690535749765216e-06,
      "loss": 2.0159,
      "step": 4959
    },
    {
      "epoch": 1.6905248807089297,
      "grad_norm": 26.036005707837028,
      "learning_rate": 4.767072631414815e-06,
      "loss": 2.3289,
      "step": 4960
    },
    {
      "epoch": 1.690865712338105,
      "grad_norm": 17.77381576271729,
      "learning_rate": 4.7650917244935114e-06,
      "loss": 2.6536,
      "step": 4961
    },
    {
      "epoch": 1.6912065439672803,
      "grad_norm": 15.955281095686487,
      "learning_rate": 4.763110854524214e-06,
      "loss": 2.1367,
      "step": 4962
    },
    {
      "epoch": 1.6915473755964554,
      "grad_norm": 18.95732938108048,
      "learning_rate": 4.761130021818524e-06,
      "loss": 2.116,
      "step": 4963
    },
    {
      "epoch": 1.6918882072256305,
      "grad_norm": 17.18260962579657,
      "learning_rate": 4.75914922668803e-06,
      "loss": 2.9141,
      "step": 4964
    },
    {
      "epoch": 1.6922290388548058,
      "grad_norm": 23.765438512231132,
      "learning_rate": 4.757168469444323e-06,
      "loss": 2.5122,
      "step": 4965
    },
    {
      "epoch": 1.6925698704839809,
      "grad_norm": 27.221711048126924,
      "learning_rate": 4.755187750398983e-06,
      "loss": 1.6651,
      "step": 4966
    },
    {
      "epoch": 1.692910702113156,
      "grad_norm": 17.26529528283423,
      "learning_rate": 4.753207069863582e-06,
      "loss": 2.8065,
      "step": 4967
    },
    {
      "epoch": 1.6932515337423313,
      "grad_norm": 14.766002676509075,
      "learning_rate": 4.751226428149694e-06,
      "loss": 2.0708,
      "step": 4968
    },
    {
      "epoch": 1.6935923653715066,
      "grad_norm": 14.490491003803482,
      "learning_rate": 4.749245825568877e-06,
      "loss": 1.8389,
      "step": 4969
    },
    {
      "epoch": 1.6939331970006817,
      "grad_norm": 16.311430643856674,
      "learning_rate": 4.74726526243269e-06,
      "loss": 2.2167,
      "step": 4970
    },
    {
      "epoch": 1.6942740286298568,
      "grad_norm": 20.51224267725315,
      "learning_rate": 4.7452847390526845e-06,
      "loss": 2.4753,
      "step": 4971
    },
    {
      "epoch": 1.694614860259032,
      "grad_norm": 15.998872104301114,
      "learning_rate": 4.743304255740404e-06,
      "loss": 2.3924,
      "step": 4972
    },
    {
      "epoch": 1.6949556918882074,
      "grad_norm": 21.969704044336932,
      "learning_rate": 4.741323812807384e-06,
      "loss": 2.1038,
      "step": 4973
    },
    {
      "epoch": 1.6952965235173822,
      "grad_norm": 20.857320622433484,
      "learning_rate": 4.739343410565158e-06,
      "loss": 2.6294,
      "step": 4974
    },
    {
      "epoch": 1.6956373551465576,
      "grad_norm": 15.33479390703286,
      "learning_rate": 4.737363049325254e-06,
      "loss": 2.1678,
      "step": 4975
    },
    {
      "epoch": 1.6959781867757329,
      "grad_norm": 15.779572425444512,
      "learning_rate": 4.7353827293991845e-06,
      "loss": 2.3927,
      "step": 4976
    },
    {
      "epoch": 1.696319018404908,
      "grad_norm": 17.975629766021278,
      "learning_rate": 4.7334024510984675e-06,
      "loss": 1.9814,
      "step": 4977
    },
    {
      "epoch": 1.696659850034083,
      "grad_norm": 18.992852492961397,
      "learning_rate": 4.731422214734605e-06,
      "loss": 2.3055,
      "step": 4978
    },
    {
      "epoch": 1.6970006816632583,
      "grad_norm": 17.111598955194136,
      "learning_rate": 4.729442020619096e-06,
      "loss": 1.9773,
      "step": 4979
    },
    {
      "epoch": 1.6973415132924337,
      "grad_norm": 12.678859464766719,
      "learning_rate": 4.7274618690634365e-06,
      "loss": 2.8256,
      "step": 4980
    },
    {
      "epoch": 1.6976823449216087,
      "grad_norm": 15.496882671231207,
      "learning_rate": 4.725481760379111e-06,
      "loss": 2.8654,
      "step": 4981
    },
    {
      "epoch": 1.6980231765507838,
      "grad_norm": 14.312427312094934,
      "learning_rate": 4.723501694877593e-06,
      "loss": 2.6084,
      "step": 4982
    },
    {
      "epoch": 1.6983640081799591,
      "grad_norm": 14.61087637856195,
      "learning_rate": 4.721521672870361e-06,
      "loss": 2.4197,
      "step": 4983
    },
    {
      "epoch": 1.6987048398091344,
      "grad_norm": 31.190890536501406,
      "learning_rate": 4.719541694668881e-06,
      "loss": 2.3004,
      "step": 4984
    },
    {
      "epoch": 1.6990456714383093,
      "grad_norm": 16.89284468185952,
      "learning_rate": 4.717561760584605e-06,
      "loss": 1.7995,
      "step": 4985
    },
    {
      "epoch": 1.6993865030674846,
      "grad_norm": 13.264087949690238,
      "learning_rate": 4.715581870928992e-06,
      "loss": 2.2271,
      "step": 4986
    },
    {
      "epoch": 1.69972733469666,
      "grad_norm": 24.090322489434865,
      "learning_rate": 4.71360202601348e-06,
      "loss": 2.3944,
      "step": 4987
    },
    {
      "epoch": 1.700068166325835,
      "grad_norm": 12.281494142033734,
      "learning_rate": 4.71162222614951e-06,
      "loss": 2.0376,
      "step": 4988
    },
    {
      "epoch": 1.70040899795501,
      "grad_norm": 19.42608334168245,
      "learning_rate": 4.709642471648513e-06,
      "loss": 2.4892,
      "step": 4989
    },
    {
      "epoch": 1.7007498295841854,
      "grad_norm": 13.840633157436393,
      "learning_rate": 4.707662762821909e-06,
      "loss": 2.0881,
      "step": 4990
    },
    {
      "epoch": 1.7010906612133607,
      "grad_norm": 17.785708219836415,
      "learning_rate": 4.7056830999811194e-06,
      "loss": 1.8074,
      "step": 4991
    },
    {
      "epoch": 1.7014314928425358,
      "grad_norm": 12.881354906407255,
      "learning_rate": 4.703703483437546e-06,
      "loss": 2.0515,
      "step": 4992
    },
    {
      "epoch": 1.701772324471711,
      "grad_norm": 18.704130560543963,
      "learning_rate": 4.701723913502597e-06,
      "loss": 1.7811,
      "step": 4993
    },
    {
      "epoch": 1.7021131561008862,
      "grad_norm": 16.111017852117204,
      "learning_rate": 4.699744390487663e-06,
      "loss": 2.4018,
      "step": 4994
    },
    {
      "epoch": 1.7024539877300615,
      "grad_norm": 24.382318597748345,
      "learning_rate": 4.69776491470413e-06,
      "loss": 2.2085,
      "step": 4995
    },
    {
      "epoch": 1.7027948193592364,
      "grad_norm": 19.045664746262155,
      "learning_rate": 4.695785486463381e-06,
      "loss": 1.8228,
      "step": 4996
    },
    {
      "epoch": 1.7031356509884117,
      "grad_norm": 19.9124084507416,
      "learning_rate": 4.693806106076787e-06,
      "loss": 2.9119,
      "step": 4997
    },
    {
      "epoch": 1.703476482617587,
      "grad_norm": 16.318265263110035,
      "learning_rate": 4.691826773855708e-06,
      "loss": 2.4178,
      "step": 4998
    },
    {
      "epoch": 1.703817314246762,
      "grad_norm": 15.000251074367135,
      "learning_rate": 4.6898474901115065e-06,
      "loss": 2.6046,
      "step": 4999
    },
    {
      "epoch": 1.7041581458759372,
      "grad_norm": 15.794854278668598,
      "learning_rate": 4.68786825515553e-06,
      "loss": 1.9026,
      "step": 5000
    },
    {
      "epoch": 1.7044989775051125,
      "grad_norm": 14.811668104385472,
      "learning_rate": 4.685889069299117e-06,
      "loss": 2.115,
      "step": 5001
    },
    {
      "epoch": 1.7048398091342878,
      "grad_norm": 23.465623936253554,
      "learning_rate": 4.683909932853606e-06,
      "loss": 2.4674,
      "step": 5002
    },
    {
      "epoch": 1.7051806407634629,
      "grad_norm": 14.274206156385405,
      "learning_rate": 4.6819308461303206e-06,
      "loss": 2.7018,
      "step": 5003
    },
    {
      "epoch": 1.705521472392638,
      "grad_norm": 14.571891207388248,
      "learning_rate": 4.6799518094405775e-06,
      "loss": 2.3773,
      "step": 5004
    },
    {
      "epoch": 1.7058623040218133,
      "grad_norm": 30.605502195564902,
      "learning_rate": 4.67797282309569e-06,
      "loss": 1.8841,
      "step": 5005
    },
    {
      "epoch": 1.7062031356509886,
      "grad_norm": 18.595630235523746,
      "learning_rate": 4.675993887406961e-06,
      "loss": 2.4681,
      "step": 5006
    },
    {
      "epoch": 1.7065439672801634,
      "grad_norm": 18.71228450597498,
      "learning_rate": 4.6740150026856804e-06,
      "loss": 2.4879,
      "step": 5007
    },
    {
      "epoch": 1.7068847989093388,
      "grad_norm": 17.00728349067932,
      "learning_rate": 4.672036169243139e-06,
      "loss": 1.6327,
      "step": 5008
    },
    {
      "epoch": 1.707225630538514,
      "grad_norm": 16.422401488061585,
      "learning_rate": 4.670057387390614e-06,
      "loss": 1.6663,
      "step": 5009
    },
    {
      "epoch": 1.7075664621676891,
      "grad_norm": 19.40166610647862,
      "learning_rate": 4.668078657439374e-06,
      "loss": 2.9595,
      "step": 5010
    },
    {
      "epoch": 1.7079072937968642,
      "grad_norm": 12.997921374983479,
      "learning_rate": 4.666099979700684e-06,
      "loss": 1.9456,
      "step": 5011
    },
    {
      "epoch": 1.7082481254260395,
      "grad_norm": 21.721912000367276,
      "learning_rate": 4.664121354485796e-06,
      "loss": 2.3154,
      "step": 5012
    },
    {
      "epoch": 1.7085889570552149,
      "grad_norm": 19.584310106391953,
      "learning_rate": 4.662142782105955e-06,
      "loss": 2.4619,
      "step": 5013
    },
    {
      "epoch": 1.70892978868439,
      "grad_norm": 19.742934072927405,
      "learning_rate": 4.660164262872402e-06,
      "loss": 2.3353,
      "step": 5014
    },
    {
      "epoch": 1.709270620313565,
      "grad_norm": 19.88429579683072,
      "learning_rate": 4.658185797096362e-06,
      "loss": 2.8665,
      "step": 5015
    },
    {
      "epoch": 1.7096114519427403,
      "grad_norm": 20.84082344588187,
      "learning_rate": 4.656207385089054e-06,
      "loss": 2.3705,
      "step": 5016
    },
    {
      "epoch": 1.7099522835719154,
      "grad_norm": 16.091225729103108,
      "learning_rate": 4.654229027161696e-06,
      "loss": 2.118,
      "step": 5017
    },
    {
      "epoch": 1.7102931152010905,
      "grad_norm": 12.272798253103634,
      "learning_rate": 4.6522507236254885e-06,
      "loss": 2.1762,
      "step": 5018
    },
    {
      "epoch": 1.7106339468302658,
      "grad_norm": 13.0740553228587,
      "learning_rate": 4.6502724747916235e-06,
      "loss": 2.3089,
      "step": 5019
    },
    {
      "epoch": 1.7109747784594411,
      "grad_norm": 20.516167750461612,
      "learning_rate": 4.648294280971294e-06,
      "loss": 2.0399,
      "step": 5020
    },
    {
      "epoch": 1.7113156100886162,
      "grad_norm": 18.057423938125282,
      "learning_rate": 4.646316142475673e-06,
      "loss": 2.3713,
      "step": 5021
    },
    {
      "epoch": 1.7116564417177913,
      "grad_norm": 11.453226918672748,
      "learning_rate": 4.644338059615931e-06,
      "loss": 2.1004,
      "step": 5022
    },
    {
      "epoch": 1.7119972733469666,
      "grad_norm": 14.106917564161913,
      "learning_rate": 4.642360032703226e-06,
      "loss": 1.9267,
      "step": 5023
    },
    {
      "epoch": 1.712338104976142,
      "grad_norm": 13.490179943198866,
      "learning_rate": 4.640382062048713e-06,
      "loss": 1.7777,
      "step": 5024
    },
    {
      "epoch": 1.712678936605317,
      "grad_norm": 15.513063791719384,
      "learning_rate": 4.638404147963533e-06,
      "loss": 2.8815,
      "step": 5025
    },
    {
      "epoch": 1.713019768234492,
      "grad_norm": 17.657979011829852,
      "learning_rate": 4.636426290758819e-06,
      "loss": 2.1548,
      "step": 5026
    },
    {
      "epoch": 1.7133605998636674,
      "grad_norm": 15.514592007902005,
      "learning_rate": 4.634448490745699e-06,
      "loss": 1.9357,
      "step": 5027
    },
    {
      "epoch": 1.7137014314928425,
      "grad_norm": 37.90727319544818,
      "learning_rate": 4.632470748235286e-06,
      "loss": 2.0881,
      "step": 5028
    },
    {
      "epoch": 1.7140422631220176,
      "grad_norm": 12.488995256178825,
      "learning_rate": 4.630493063538685e-06,
      "loss": 1.5465,
      "step": 5029
    },
    {
      "epoch": 1.7143830947511929,
      "grad_norm": 13.113242810998464,
      "learning_rate": 4.6285154369669986e-06,
      "loss": 2.2363,
      "step": 5030
    },
    {
      "epoch": 1.7147239263803682,
      "grad_norm": 20.892187126514376,
      "learning_rate": 4.626537868831313e-06,
      "loss": 2.6393,
      "step": 5031
    },
    {
      "epoch": 1.7150647580095433,
      "grad_norm": 17.955107157160644,
      "learning_rate": 4.624560359442706e-06,
      "loss": 2.3694,
      "step": 5032
    },
    {
      "epoch": 1.7154055896387184,
      "grad_norm": 31.69603686618882,
      "learning_rate": 4.622582909112248e-06,
      "loss": 1.6855,
      "step": 5033
    },
    {
      "epoch": 1.7157464212678937,
      "grad_norm": 20.92571105598751,
      "learning_rate": 4.620605518151003e-06,
      "loss": 2.4929,
      "step": 5034
    },
    {
      "epoch": 1.716087252897069,
      "grad_norm": 22.37325239597225,
      "learning_rate": 4.6186281868700186e-06,
      "loss": 1.8494,
      "step": 5035
    },
    {
      "epoch": 1.716428084526244,
      "grad_norm": 17.78941433253117,
      "learning_rate": 4.6166509155803405e-06,
      "loss": 2.7589,
      "step": 5036
    },
    {
      "epoch": 1.7167689161554192,
      "grad_norm": 15.806373273595014,
      "learning_rate": 4.614673704592997e-06,
      "loss": 2.3025,
      "step": 5037
    },
    {
      "epoch": 1.7171097477845945,
      "grad_norm": 22.65381248418739,
      "learning_rate": 4.612696554219014e-06,
      "loss": 2.4517,
      "step": 5038
    },
    {
      "epoch": 1.7174505794137696,
      "grad_norm": 18.069697979530222,
      "learning_rate": 4.610719464769407e-06,
      "loss": 2.1864,
      "step": 5039
    },
    {
      "epoch": 1.7177914110429446,
      "grad_norm": 14.413616985708764,
      "learning_rate": 4.6087424365551765e-06,
      "loss": 2.3897,
      "step": 5040
    },
    {
      "epoch": 1.71813224267212,
      "grad_norm": 12.540587597005697,
      "learning_rate": 4.6067654698873175e-06,
      "loss": 2.3287,
      "step": 5041
    },
    {
      "epoch": 1.7184730743012953,
      "grad_norm": 18.424506304556694,
      "learning_rate": 4.604788565076817e-06,
      "loss": 2.4246,
      "step": 5042
    },
    {
      "epoch": 1.7188139059304703,
      "grad_norm": 20.343867475132917,
      "learning_rate": 4.602811722434648e-06,
      "loss": 2.8961,
      "step": 5043
    },
    {
      "epoch": 1.7191547375596454,
      "grad_norm": 19.633169297161047,
      "learning_rate": 4.600834942271775e-06,
      "loss": 1.9568,
      "step": 5044
    },
    {
      "epoch": 1.7194955691888207,
      "grad_norm": 17.669711128594532,
      "learning_rate": 4.598858224899156e-06,
      "loss": 1.9764,
      "step": 5045
    },
    {
      "epoch": 1.719836400817996,
      "grad_norm": 19.882448214897664,
      "learning_rate": 4.596881570627734e-06,
      "loss": 1.6391,
      "step": 5046
    },
    {
      "epoch": 1.7201772324471711,
      "grad_norm": 15.135827143042716,
      "learning_rate": 4.594904979768445e-06,
      "loss": 2.1068,
      "step": 5047
    },
    {
      "epoch": 1.7205180640763462,
      "grad_norm": 30.83329562692499,
      "learning_rate": 4.592928452632217e-06,
      "loss": 2.8664,
      "step": 5048
    },
    {
      "epoch": 1.7208588957055215,
      "grad_norm": 13.926476902004083,
      "learning_rate": 4.590951989529963e-06,
      "loss": 2.0289,
      "step": 5049
    },
    {
      "epoch": 1.7211997273346966,
      "grad_norm": 17.219471378700067,
      "learning_rate": 4.58897559077259e-06,
      "loss": 2.6694,
      "step": 5050
    },
    {
      "epoch": 1.7215405589638717,
      "grad_norm": 18.041916460440913,
      "learning_rate": 4.586999256670992e-06,
      "loss": 2.3146,
      "step": 5051
    },
    {
      "epoch": 1.721881390593047,
      "grad_norm": 13.742144857733097,
      "learning_rate": 4.5850229875360564e-06,
      "loss": 2.5892,
      "step": 5052
    },
    {
      "epoch": 1.7222222222222223,
      "grad_norm": 20.579810856198975,
      "learning_rate": 4.583046783678655e-06,
      "loss": 2.1115,
      "step": 5053
    },
    {
      "epoch": 1.7225630538513974,
      "grad_norm": 19.563694425929132,
      "learning_rate": 4.581070645409654e-06,
      "loss": 2.4788,
      "step": 5054
    },
    {
      "epoch": 1.7229038854805725,
      "grad_norm": 15.276617602723613,
      "learning_rate": 4.579094573039911e-06,
      "loss": 2.7262,
      "step": 5055
    },
    {
      "epoch": 1.7232447171097478,
      "grad_norm": 11.497066579589347,
      "learning_rate": 4.577118566880266e-06,
      "loss": 2.1753,
      "step": 5056
    },
    {
      "epoch": 1.7235855487389231,
      "grad_norm": 18.98928769432887,
      "learning_rate": 4.575142627241552e-06,
      "loss": 1.7823,
      "step": 5057
    },
    {
      "epoch": 1.7239263803680982,
      "grad_norm": 19.16138287217517,
      "learning_rate": 4.573166754434595e-06,
      "loss": 2.3013,
      "step": 5058
    },
    {
      "epoch": 1.7242672119972733,
      "grad_norm": 14.437691159203107,
      "learning_rate": 4.571190948770208e-06,
      "loss": 1.9086,
      "step": 5059
    },
    {
      "epoch": 1.7246080436264486,
      "grad_norm": 15.124520467508404,
      "learning_rate": 4.569215210559189e-06,
      "loss": 2.6679,
      "step": 5060
    },
    {
      "epoch": 1.7249488752556237,
      "grad_norm": 15.383353424039155,
      "learning_rate": 4.567239540112335e-06,
      "loss": 1.9739,
      "step": 5061
    },
    {
      "epoch": 1.7252897068847988,
      "grad_norm": 13.037316172348396,
      "learning_rate": 4.565263937740423e-06,
      "loss": 2.4938,
      "step": 5062
    },
    {
      "epoch": 1.725630538513974,
      "grad_norm": 16.78592772262329,
      "learning_rate": 4.563288403754222e-06,
      "loss": 2.2603,
      "step": 5063
    },
    {
      "epoch": 1.7259713701431494,
      "grad_norm": 15.480306618240919,
      "learning_rate": 4.561312938464498e-06,
      "loss": 3.0272,
      "step": 5064
    },
    {
      "epoch": 1.7263122017723245,
      "grad_norm": 20.7587876265301,
      "learning_rate": 4.559337542181993e-06,
      "loss": 2.4091,
      "step": 5065
    },
    {
      "epoch": 1.7266530334014996,
      "grad_norm": 14.343024602029418,
      "learning_rate": 4.557362215217445e-06,
      "loss": 2.3403,
      "step": 5066
    },
    {
      "epoch": 1.7269938650306749,
      "grad_norm": 20.968824031383352,
      "learning_rate": 4.555386957881584e-06,
      "loss": 2.0663,
      "step": 5067
    },
    {
      "epoch": 1.7273346966598502,
      "grad_norm": 10.862283418542374,
      "learning_rate": 4.553411770485126e-06,
      "loss": 2.0419,
      "step": 5068
    },
    {
      "epoch": 1.7276755282890253,
      "grad_norm": 27.69303715825103,
      "learning_rate": 4.551436653338771e-06,
      "loss": 2.069,
      "step": 5069
    },
    {
      "epoch": 1.7280163599182004,
      "grad_norm": 19.847169454817625,
      "learning_rate": 4.549461606753219e-06,
      "loss": 2.5615,
      "step": 5070
    },
    {
      "epoch": 1.7283571915473757,
      "grad_norm": 19.052175638891185,
      "learning_rate": 4.5474866310391486e-06,
      "loss": 2.3054,
      "step": 5071
    },
    {
      "epoch": 1.7286980231765507,
      "grad_norm": 15.497954106404382,
      "learning_rate": 4.545511726507232e-06,
      "loss": 2.1702,
      "step": 5072
    },
    {
      "epoch": 1.7290388548057258,
      "grad_norm": 13.697898763224496,
      "learning_rate": 4.543536893468131e-06,
      "loss": 2.1937,
      "step": 5073
    },
    {
      "epoch": 1.7293796864349011,
      "grad_norm": 13.6275565220429,
      "learning_rate": 4.541562132232494e-06,
      "loss": 2.1822,
      "step": 5074
    },
    {
      "epoch": 1.7297205180640765,
      "grad_norm": 21.36465248286139,
      "learning_rate": 4.539587443110956e-06,
      "loss": 2.3368,
      "step": 5075
    },
    {
      "epoch": 1.7300613496932515,
      "grad_norm": 21.821921037291453,
      "learning_rate": 4.537612826414148e-06,
      "loss": 2.0527,
      "step": 5076
    },
    {
      "epoch": 1.7304021813224266,
      "grad_norm": 20.809950401087153,
      "learning_rate": 4.535638282452684e-06,
      "loss": 2.0027,
      "step": 5077
    },
    {
      "epoch": 1.730743012951602,
      "grad_norm": 12.477891139395208,
      "learning_rate": 4.533663811537164e-06,
      "loss": 2.2599,
      "step": 5078
    },
    {
      "epoch": 1.7310838445807772,
      "grad_norm": 21.696442768999127,
      "learning_rate": 4.531689413978184e-06,
      "loss": 2.5802,
      "step": 5079
    },
    {
      "epoch": 1.7314246762099523,
      "grad_norm": 14.04987685028848,
      "learning_rate": 4.529715090086325e-06,
      "loss": 2.4352,
      "step": 5080
    },
    {
      "epoch": 1.7317655078391274,
      "grad_norm": 14.948690478854681,
      "learning_rate": 4.5277408401721525e-06,
      "loss": 2.3183,
      "step": 5081
    },
    {
      "epoch": 1.7321063394683027,
      "grad_norm": 14.555405661810765,
      "learning_rate": 4.525766664546227e-06,
      "loss": 1.766,
      "step": 5082
    },
    {
      "epoch": 1.7324471710974778,
      "grad_norm": 14.941222976219215,
      "learning_rate": 4.523792563519092e-06,
      "loss": 2.1094,
      "step": 5083
    },
    {
      "epoch": 1.732788002726653,
      "grad_norm": 22.63886615610604,
      "learning_rate": 4.521818537401282e-06,
      "loss": 2.2758,
      "step": 5084
    },
    {
      "epoch": 1.7331288343558282,
      "grad_norm": 14.699611729937146,
      "learning_rate": 4.5198445865033214e-06,
      "loss": 2.3905,
      "step": 5085
    },
    {
      "epoch": 1.7334696659850035,
      "grad_norm": 11.468955696632042,
      "learning_rate": 4.517870711135719e-06,
      "loss": 1.6515,
      "step": 5086
    },
    {
      "epoch": 1.7338104976141786,
      "grad_norm": 18.347654772365313,
      "learning_rate": 4.5158969116089715e-06,
      "loss": 2.4654,
      "step": 5087
    },
    {
      "epoch": 1.7341513292433537,
      "grad_norm": 18.069941495307585,
      "learning_rate": 4.513923188233566e-06,
      "loss": 2.2095,
      "step": 5088
    },
    {
      "epoch": 1.734492160872529,
      "grad_norm": 15.530963778042679,
      "learning_rate": 4.51194954131998e-06,
      "loss": 2.5167,
      "step": 5089
    },
    {
      "epoch": 1.7348329925017043,
      "grad_norm": 15.063219057545643,
      "learning_rate": 4.509975971178674e-06,
      "loss": 2.3769,
      "step": 5090
    },
    {
      "epoch": 1.7351738241308794,
      "grad_norm": 14.095217808063346,
      "learning_rate": 4.508002478120096e-06,
      "loss": 2.1875,
      "step": 5091
    },
    {
      "epoch": 1.7355146557600545,
      "grad_norm": 14.976659562729596,
      "learning_rate": 4.5060290624546885e-06,
      "loss": 2.1296,
      "step": 5092
    },
    {
      "epoch": 1.7358554873892298,
      "grad_norm": 15.333036972093163,
      "learning_rate": 4.504055724492877e-06,
      "loss": 1.9961,
      "step": 5093
    },
    {
      "epoch": 1.7361963190184049,
      "grad_norm": 18.0779770594952,
      "learning_rate": 4.502082464545072e-06,
      "loss": 1.8732,
      "step": 5094
    },
    {
      "epoch": 1.73653715064758,
      "grad_norm": 13.860268528008062,
      "learning_rate": 4.5001092829216785e-06,
      "loss": 2.2192,
      "step": 5095
    },
    {
      "epoch": 1.7368779822767553,
      "grad_norm": 19.413565426030022,
      "learning_rate": 4.498136179933085e-06,
      "loss": 3.109,
      "step": 5096
    },
    {
      "epoch": 1.7372188139059306,
      "grad_norm": 25.029378533432237,
      "learning_rate": 4.496163155889666e-06,
      "loss": 2.2057,
      "step": 5097
    },
    {
      "epoch": 1.7375596455351057,
      "grad_norm": 16.178090153596322,
      "learning_rate": 4.494190211101791e-06,
      "loss": 2.1624,
      "step": 5098
    },
    {
      "epoch": 1.7379004771642808,
      "grad_norm": 15.956750976835927,
      "learning_rate": 4.492217345879809e-06,
      "loss": 2.1793,
      "step": 5099
    },
    {
      "epoch": 1.738241308793456,
      "grad_norm": 13.002218738924466,
      "learning_rate": 4.490244560534058e-06,
      "loss": 2.2065,
      "step": 5100
    },
    {
      "epoch": 1.7385821404226314,
      "grad_norm": 22.17664732490765,
      "learning_rate": 4.488271855374868e-06,
      "loss": 1.6305,
      "step": 5101
    },
    {
      "epoch": 1.7389229720518065,
      "grad_norm": 15.384219379235127,
      "learning_rate": 4.4862992307125534e-06,
      "loss": 2.2589,
      "step": 5102
    },
    {
      "epoch": 1.7392638036809815,
      "grad_norm": 21.911975365077126,
      "learning_rate": 4.4843266868574125e-06,
      "loss": 2.0298,
      "step": 5103
    },
    {
      "epoch": 1.7396046353101569,
      "grad_norm": 14.482871388150002,
      "learning_rate": 4.482354224119739e-06,
      "loss": 2.3263,
      "step": 5104
    },
    {
      "epoch": 1.739945466939332,
      "grad_norm": 19.365151658231678,
      "learning_rate": 4.4803818428098045e-06,
      "loss": 1.9553,
      "step": 5105
    },
    {
      "epoch": 1.740286298568507,
      "grad_norm": 21.871090986840194,
      "learning_rate": 4.4784095432378746e-06,
      "loss": 2.3565,
      "step": 5106
    },
    {
      "epoch": 1.7406271301976823,
      "grad_norm": 19.783678384377726,
      "learning_rate": 4.476437325714201e-06,
      "loss": 2.4907,
      "step": 5107
    },
    {
      "epoch": 1.7409679618268576,
      "grad_norm": 22.93889834467417,
      "learning_rate": 4.47446519054902e-06,
      "loss": 2.9903,
      "step": 5108
    },
    {
      "epoch": 1.7413087934560327,
      "grad_norm": 18.32732181453398,
      "learning_rate": 4.472493138052557e-06,
      "loss": 2.6042,
      "step": 5109
    },
    {
      "epoch": 1.7416496250852078,
      "grad_norm": 17.59660705119435,
      "learning_rate": 4.470521168535022e-06,
      "loss": 2.5598,
      "step": 5110
    },
    {
      "epoch": 1.7419904567143831,
      "grad_norm": 15.021579305173448,
      "learning_rate": 4.468549282306617e-06,
      "loss": 1.9924,
      "step": 5111
    },
    {
      "epoch": 1.7423312883435584,
      "grad_norm": 22.68269095899723,
      "learning_rate": 4.466577479677523e-06,
      "loss": 1.8932,
      "step": 5112
    },
    {
      "epoch": 1.7426721199727335,
      "grad_norm": 18.030860016040503,
      "learning_rate": 4.464605760957916e-06,
      "loss": 2.2986,
      "step": 5113
    },
    {
      "epoch": 1.7430129516019086,
      "grad_norm": 20.25867945223272,
      "learning_rate": 4.462634126457955e-06,
      "loss": 2.0511,
      "step": 5114
    },
    {
      "epoch": 1.743353783231084,
      "grad_norm": 18.87870835845495,
      "learning_rate": 4.460662576487783e-06,
      "loss": 2.4401,
      "step": 5115
    },
    {
      "epoch": 1.743694614860259,
      "grad_norm": 27.6443661365189,
      "learning_rate": 4.458691111357537e-06,
      "loss": 2.1463,
      "step": 5116
    },
    {
      "epoch": 1.744035446489434,
      "grad_norm": 18.73090214809991,
      "learning_rate": 4.456719731377332e-06,
      "loss": 1.9245,
      "step": 5117
    },
    {
      "epoch": 1.7443762781186094,
      "grad_norm": 13.519953612673026,
      "learning_rate": 4.454748436857276e-06,
      "loss": 2.3429,
      "step": 5118
    },
    {
      "epoch": 1.7447171097477847,
      "grad_norm": 15.365371421567367,
      "learning_rate": 4.452777228107463e-06,
      "loss": 2.056,
      "step": 5119
    },
    {
      "epoch": 1.7450579413769598,
      "grad_norm": 12.586322061690149,
      "learning_rate": 4.4508061054379705e-06,
      "loss": 2.3949,
      "step": 5120
    },
    {
      "epoch": 1.7453987730061349,
      "grad_norm": 19.40577934288575,
      "learning_rate": 4.4488350691588624e-06,
      "loss": 2.0839,
      "step": 5121
    },
    {
      "epoch": 1.7457396046353102,
      "grad_norm": 15.795101016865903,
      "learning_rate": 4.446864119580192e-06,
      "loss": 2.3132,
      "step": 5122
    },
    {
      "epoch": 1.7460804362644855,
      "grad_norm": 17.200528177669543,
      "learning_rate": 4.444893257011998e-06,
      "loss": 2.3768,
      "step": 5123
    },
    {
      "epoch": 1.7464212678936604,
      "grad_norm": 20.947912105664297,
      "learning_rate": 4.442922481764306e-06,
      "loss": 2.5519,
      "step": 5124
    },
    {
      "epoch": 1.7467620995228357,
      "grad_norm": 14.015479239236948,
      "learning_rate": 4.440951794147122e-06,
      "loss": 2.5172,
      "step": 5125
    },
    {
      "epoch": 1.747102931152011,
      "grad_norm": 21.549001875029777,
      "learning_rate": 4.4389811944704476e-06,
      "loss": 2.3767,
      "step": 5126
    },
    {
      "epoch": 1.747443762781186,
      "grad_norm": 10.84065752154643,
      "learning_rate": 4.437010683044266e-06,
      "loss": 1.9243,
      "step": 5127
    },
    {
      "epoch": 1.7477845944103612,
      "grad_norm": 17.68132816923046,
      "learning_rate": 4.4350402601785425e-06,
      "loss": 2.6642,
      "step": 5128
    },
    {
      "epoch": 1.7481254260395365,
      "grad_norm": 25.705284468193096,
      "learning_rate": 4.433069926183238e-06,
      "loss": 1.8235,
      "step": 5129
    },
    {
      "epoch": 1.7484662576687118,
      "grad_norm": 15.731930819516988,
      "learning_rate": 4.431099681368289e-06,
      "loss": 2.0522,
      "step": 5130
    },
    {
      "epoch": 1.7488070892978869,
      "grad_norm": 14.2470458239207,
      "learning_rate": 4.429129526043624e-06,
      "loss": 2.63,
      "step": 5131
    },
    {
      "epoch": 1.749147920927062,
      "grad_norm": 17.176678588314427,
      "learning_rate": 4.427159460519159e-06,
      "loss": 2.0559,
      "step": 5132
    },
    {
      "epoch": 1.7494887525562373,
      "grad_norm": 20.2394025632041,
      "learning_rate": 4.425189485104792e-06,
      "loss": 2.4002,
      "step": 5133
    },
    {
      "epoch": 1.7498295841854126,
      "grad_norm": 19.608357194023256,
      "learning_rate": 4.423219600110404e-06,
      "loss": 2.3485,
      "step": 5134
    },
    {
      "epoch": 1.7501704158145874,
      "grad_norm": 17.665367063467652,
      "learning_rate": 4.42124980584587e-06,
      "loss": 2.0905,
      "step": 5135
    },
    {
      "epoch": 1.7505112474437627,
      "grad_norm": 21.075550580754435,
      "learning_rate": 4.419280102621047e-06,
      "loss": 2.2592,
      "step": 5136
    },
    {
      "epoch": 1.750852079072938,
      "grad_norm": 25.87264319750964,
      "learning_rate": 4.417310490745773e-06,
      "loss": 2.3822,
      "step": 5137
    },
    {
      "epoch": 1.7511929107021131,
      "grad_norm": 18.82225129742288,
      "learning_rate": 4.41534097052988e-06,
      "loss": 1.6717,
      "step": 5138
    },
    {
      "epoch": 1.7515337423312882,
      "grad_norm": 19.77707497225093,
      "learning_rate": 4.413371542283181e-06,
      "loss": 2.0733,
      "step": 5139
    },
    {
      "epoch": 1.7518745739604635,
      "grad_norm": 12.374411045699889,
      "learning_rate": 4.411402206315471e-06,
      "loss": 2.1299,
      "step": 5140
    },
    {
      "epoch": 1.7522154055896388,
      "grad_norm": 15.75324790787278,
      "learning_rate": 4.409432962936539e-06,
      "loss": 2.2574,
      "step": 5141
    },
    {
      "epoch": 1.752556237218814,
      "grad_norm": 16.799892575003046,
      "learning_rate": 4.407463812456152e-06,
      "loss": 2.0906,
      "step": 5142
    },
    {
      "epoch": 1.752897068847989,
      "grad_norm": 14.962223270682488,
      "learning_rate": 4.405494755184065e-06,
      "loss": 1.9705,
      "step": 5143
    },
    {
      "epoch": 1.7532379004771643,
      "grad_norm": 18.741642457184337,
      "learning_rate": 4.403525791430022e-06,
      "loss": 2.668,
      "step": 5144
    },
    {
      "epoch": 1.7535787321063396,
      "grad_norm": 17.28072189599227,
      "learning_rate": 4.401556921503746e-06,
      "loss": 2.0305,
      "step": 5145
    },
    {
      "epoch": 1.7539195637355145,
      "grad_norm": 18.818454366900504,
      "learning_rate": 4.3995881457149465e-06,
      "loss": 2.4462,
      "step": 5146
    },
    {
      "epoch": 1.7542603953646898,
      "grad_norm": 24.646895648681475,
      "learning_rate": 4.3976194643733225e-06,
      "loss": 1.8048,
      "step": 5147
    },
    {
      "epoch": 1.7546012269938651,
      "grad_norm": 14.65979795628185,
      "learning_rate": 4.395650877788557e-06,
      "loss": 2.2892,
      "step": 5148
    },
    {
      "epoch": 1.7549420586230402,
      "grad_norm": 15.956902990292656,
      "learning_rate": 4.3936823862703105e-06,
      "loss": 2.2802,
      "step": 5149
    },
    {
      "epoch": 1.7552828902522153,
      "grad_norm": 15.42358398509083,
      "learning_rate": 4.391713990128242e-06,
      "loss": 2.5677,
      "step": 5150
    },
    {
      "epoch": 1.7556237218813906,
      "grad_norm": 14.04965134738658,
      "learning_rate": 4.3897456896719814e-06,
      "loss": 2.1454,
      "step": 5151
    },
    {
      "epoch": 1.755964553510566,
      "grad_norm": 16.070807558357625,
      "learning_rate": 4.387777485211155e-06,
      "loss": 1.787,
      "step": 5152
    },
    {
      "epoch": 1.756305385139741,
      "grad_norm": 20.54092092025563,
      "learning_rate": 4.3858093770553655e-06,
      "loss": 2.2455,
      "step": 5153
    },
    {
      "epoch": 1.756646216768916,
      "grad_norm": 20.200531060464293,
      "learning_rate": 4.383841365514208e-06,
      "loss": 2.3754,
      "step": 5154
    },
    {
      "epoch": 1.7569870483980914,
      "grad_norm": 20.536712314587447,
      "learning_rate": 4.381873450897255e-06,
      "loss": 1.8847,
      "step": 5155
    },
    {
      "epoch": 1.7573278800272665,
      "grad_norm": 18.732254873095805,
      "learning_rate": 4.3799056335140675e-06,
      "loss": 2.3076,
      "step": 5156
    },
    {
      "epoch": 1.7576687116564416,
      "grad_norm": 34.42809005476786,
      "learning_rate": 4.377937913674195e-06,
      "loss": 2.4725,
      "step": 5157
    },
    {
      "epoch": 1.7580095432856169,
      "grad_norm": 26.94952743440283,
      "learning_rate": 4.375970291687165e-06,
      "loss": 2.5146,
      "step": 5158
    },
    {
      "epoch": 1.7583503749147922,
      "grad_norm": 15.891943882405467,
      "learning_rate": 4.37400276786249e-06,
      "loss": 2.1845,
      "step": 5159
    },
    {
      "epoch": 1.7586912065439673,
      "grad_norm": 20.7867133943064,
      "learning_rate": 4.372035342509673e-06,
      "loss": 2.6423,
      "step": 5160
    },
    {
      "epoch": 1.7590320381731424,
      "grad_norm": 19.774075200502747,
      "learning_rate": 4.370068015938197e-06,
      "loss": 2.3917,
      "step": 5161
    },
    {
      "epoch": 1.7593728698023177,
      "grad_norm": 16.178699468410787,
      "learning_rate": 4.368100788457527e-06,
      "loss": 1.9482,
      "step": 5162
    },
    {
      "epoch": 1.759713701431493,
      "grad_norm": 22.762251903076532,
      "learning_rate": 4.366133660377121e-06,
      "loss": 2.5826,
      "step": 5163
    },
    {
      "epoch": 1.760054533060668,
      "grad_norm": 30.49888896936627,
      "learning_rate": 4.364166632006413e-06,
      "loss": 2.5322,
      "step": 5164
    },
    {
      "epoch": 1.7603953646898431,
      "grad_norm": 12.121110306422834,
      "learning_rate": 4.362199703654823e-06,
      "loss": 1.6104,
      "step": 5165
    },
    {
      "epoch": 1.7607361963190185,
      "grad_norm": 14.22227022275941,
      "learning_rate": 4.360232875631762e-06,
      "loss": 2.139,
      "step": 5166
    },
    {
      "epoch": 1.7610770279481935,
      "grad_norm": 14.654260244447787,
      "learning_rate": 4.358266148246614e-06,
      "loss": 2.1818,
      "step": 5167
    },
    {
      "epoch": 1.7614178595773686,
      "grad_norm": 11.98963325863662,
      "learning_rate": 4.356299521808756e-06,
      "loss": 2.2926,
      "step": 5168
    },
    {
      "epoch": 1.761758691206544,
      "grad_norm": 13.372039496101104,
      "learning_rate": 4.354332996627546e-06,
      "loss": 2.13,
      "step": 5169
    },
    {
      "epoch": 1.7620995228357192,
      "grad_norm": 14.627291320675825,
      "learning_rate": 4.352366573012327e-06,
      "loss": 2.2748,
      "step": 5170
    },
    {
      "epoch": 1.7624403544648943,
      "grad_norm": 17.261031419481316,
      "learning_rate": 4.350400251272421e-06,
      "loss": 2.3547,
      "step": 5171
    },
    {
      "epoch": 1.7627811860940694,
      "grad_norm": 14.50724129027168,
      "learning_rate": 4.348434031717143e-06,
      "loss": 1.9419,
      "step": 5172
    },
    {
      "epoch": 1.7631220177232447,
      "grad_norm": 15.601626137015229,
      "learning_rate": 4.3464679146557866e-06,
      "loss": 2.2423,
      "step": 5173
    },
    {
      "epoch": 1.76346284935242,
      "grad_norm": 19.136693101107426,
      "learning_rate": 4.344501900397626e-06,
      "loss": 2.3318,
      "step": 5174
    },
    {
      "epoch": 1.7638036809815951,
      "grad_norm": 18.15853545810415,
      "learning_rate": 4.342535989251929e-06,
      "loss": 2.3646,
      "step": 5175
    },
    {
      "epoch": 1.7641445126107702,
      "grad_norm": 17.88930999113575,
      "learning_rate": 4.3405701815279376e-06,
      "loss": 2.2258,
      "step": 5176
    },
    {
      "epoch": 1.7644853442399455,
      "grad_norm": 23.941915279109775,
      "learning_rate": 4.338604477534879e-06,
      "loss": 2.9619,
      "step": 5177
    },
    {
      "epoch": 1.7648261758691206,
      "grad_norm": 14.588102806271792,
      "learning_rate": 4.336638877581971e-06,
      "loss": 2.4366,
      "step": 5178
    },
    {
      "epoch": 1.7651670074982957,
      "grad_norm": 19.17988694195289,
      "learning_rate": 4.3346733819784095e-06,
      "loss": 2.0189,
      "step": 5179
    },
    {
      "epoch": 1.765507839127471,
      "grad_norm": 16.020822866945014,
      "learning_rate": 4.332707991033371e-06,
      "loss": 1.908,
      "step": 5180
    },
    {
      "epoch": 1.7658486707566463,
      "grad_norm": 18.170819428848134,
      "learning_rate": 4.330742705056024e-06,
      "loss": 2.4809,
      "step": 5181
    },
    {
      "epoch": 1.7661895023858214,
      "grad_norm": 18.003602036465953,
      "learning_rate": 4.328777524355513e-06,
      "loss": 2.3801,
      "step": 5182
    },
    {
      "epoch": 1.7665303340149965,
      "grad_norm": 17.659237854238395,
      "learning_rate": 4.326812449240968e-06,
      "loss": 2.6564,
      "step": 5183
    },
    {
      "epoch": 1.7668711656441718,
      "grad_norm": 18.822568627253037,
      "learning_rate": 4.324847480021506e-06,
      "loss": 2.3391,
      "step": 5184
    },
    {
      "epoch": 1.767211997273347,
      "grad_norm": 23.90287207941454,
      "learning_rate": 4.322882617006223e-06,
      "loss": 2.8046,
      "step": 5185
    },
    {
      "epoch": 1.7675528289025222,
      "grad_norm": 28.061175878522562,
      "learning_rate": 4.3209178605042e-06,
      "loss": 2.169,
      "step": 5186
    },
    {
      "epoch": 1.7678936605316973,
      "grad_norm": 17.6463445950485,
      "learning_rate": 4.318953210824499e-06,
      "loss": 1.682,
      "step": 5187
    },
    {
      "epoch": 1.7682344921608726,
      "grad_norm": 12.805419312607222,
      "learning_rate": 4.316988668276171e-06,
      "loss": 2.1139,
      "step": 5188
    },
    {
      "epoch": 1.7685753237900477,
      "grad_norm": 17.065697198312883,
      "learning_rate": 4.315024233168243e-06,
      "loss": 2.0432,
      "step": 5189
    },
    {
      "epoch": 1.7689161554192228,
      "grad_norm": 21.02337818801944,
      "learning_rate": 4.31305990580973e-06,
      "loss": 2.7886,
      "step": 5190
    },
    {
      "epoch": 1.769256987048398,
      "grad_norm": 16.347835776484548,
      "learning_rate": 4.31109568650963e-06,
      "loss": 2.1585,
      "step": 5191
    },
    {
      "epoch": 1.7695978186775734,
      "grad_norm": 13.483451657497273,
      "learning_rate": 4.30913157557692e-06,
      "loss": 1.878,
      "step": 5192
    },
    {
      "epoch": 1.7699386503067485,
      "grad_norm": 13.950003862031465,
      "learning_rate": 4.307167573320562e-06,
      "loss": 1.7488,
      "step": 5193
    },
    {
      "epoch": 1.7702794819359235,
      "grad_norm": 18.089747462320023,
      "learning_rate": 4.305203680049504e-06,
      "loss": 2.7858,
      "step": 5194
    },
    {
      "epoch": 1.7706203135650989,
      "grad_norm": 22.057050352083678,
      "learning_rate": 4.303239896072673e-06,
      "loss": 2.3185,
      "step": 5195
    },
    {
      "epoch": 1.7709611451942742,
      "grad_norm": 16.60242507065884,
      "learning_rate": 4.301276221698979e-06,
      "loss": 2.5922,
      "step": 5196
    },
    {
      "epoch": 1.7713019768234493,
      "grad_norm": 13.878289940339824,
      "learning_rate": 4.299312657237318e-06,
      "loss": 1.9047,
      "step": 5197
    },
    {
      "epoch": 1.7716428084526243,
      "grad_norm": 128.67943336847222,
      "learning_rate": 4.2973492029965655e-06,
      "loss": 2.7212,
      "step": 5198
    },
    {
      "epoch": 1.7719836400817996,
      "grad_norm": 17.19074704925077,
      "learning_rate": 4.295385859285579e-06,
      "loss": 2.259,
      "step": 5199
    },
    {
      "epoch": 1.7723244717109747,
      "grad_norm": 16.509461245188916,
      "learning_rate": 4.293422626413203e-06,
      "loss": 1.6283,
      "step": 5200
    },
    {
      "epoch": 1.7726653033401498,
      "grad_norm": 12.574671908019113,
      "learning_rate": 4.29145950468826e-06,
      "loss": 2.2783,
      "step": 5201
    },
    {
      "epoch": 1.7730061349693251,
      "grad_norm": 18.039759327999626,
      "learning_rate": 4.289496494419556e-06,
      "loss": 2.091,
      "step": 5202
    },
    {
      "epoch": 1.7733469665985004,
      "grad_norm": 18.363845282649415,
      "learning_rate": 4.287533595915884e-06,
      "loss": 2.1459,
      "step": 5203
    },
    {
      "epoch": 1.7736877982276755,
      "grad_norm": 11.463006065984715,
      "learning_rate": 4.285570809486014e-06,
      "loss": 2.1378,
      "step": 5204
    },
    {
      "epoch": 1.7740286298568506,
      "grad_norm": 12.864385778617457,
      "learning_rate": 4.283608135438697e-06,
      "loss": 1.9613,
      "step": 5205
    },
    {
      "epoch": 1.774369461486026,
      "grad_norm": 14.474677316114839,
      "learning_rate": 4.281645574082674e-06,
      "loss": 2.5954,
      "step": 5206
    },
    {
      "epoch": 1.7747102931152012,
      "grad_norm": 17.219788599854596,
      "learning_rate": 4.279683125726662e-06,
      "loss": 2.2496,
      "step": 5207
    },
    {
      "epoch": 1.7750511247443763,
      "grad_norm": 15.171308871848867,
      "learning_rate": 4.27772079067936e-06,
      "loss": 2.422,
      "step": 5208
    },
    {
      "epoch": 1.7753919563735514,
      "grad_norm": 16.23403309549262,
      "learning_rate": 4.275758569249455e-06,
      "loss": 2.1934,
      "step": 5209
    },
    {
      "epoch": 1.7757327880027267,
      "grad_norm": 16.101807896476554,
      "learning_rate": 4.273796461745609e-06,
      "loss": 2.3911,
      "step": 5210
    },
    {
      "epoch": 1.7760736196319018,
      "grad_norm": 12.643945839126618,
      "learning_rate": 4.271834468476469e-06,
      "loss": 2.2212,
      "step": 5211
    },
    {
      "epoch": 1.7764144512610769,
      "grad_norm": 31.193360884767273,
      "learning_rate": 4.2698725897506685e-06,
      "loss": 2.3907,
      "step": 5212
    },
    {
      "epoch": 1.7767552828902522,
      "grad_norm": 19.402399563064424,
      "learning_rate": 4.267910825876816e-06,
      "loss": 2.265,
      "step": 5213
    },
    {
      "epoch": 1.7770961145194275,
      "grad_norm": 20.210168352226738,
      "learning_rate": 4.265949177163503e-06,
      "loss": 2.1921,
      "step": 5214
    },
    {
      "epoch": 1.7774369461486026,
      "grad_norm": 19.008321367727664,
      "learning_rate": 4.263987643919306e-06,
      "loss": 2.3328,
      "step": 5215
    },
    {
      "epoch": 1.7777777777777777,
      "grad_norm": 16.39675747925913,
      "learning_rate": 4.262026226452785e-06,
      "loss": 3.1314,
      "step": 5216
    },
    {
      "epoch": 1.778118609406953,
      "grad_norm": 13.99975841091573,
      "learning_rate": 4.260064925072476e-06,
      "loss": 2.5995,
      "step": 5217
    },
    {
      "epoch": 1.7784594410361283,
      "grad_norm": 19.816965145059672,
      "learning_rate": 4.258103740086898e-06,
      "loss": 1.8668,
      "step": 5218
    },
    {
      "epoch": 1.7788002726653034,
      "grad_norm": 14.315495182584632,
      "learning_rate": 4.256142671804556e-06,
      "loss": 1.9889,
      "step": 5219
    },
    {
      "epoch": 1.7791411042944785,
      "grad_norm": 15.669368037806887,
      "learning_rate": 4.254181720533933e-06,
      "loss": 2.2456,
      "step": 5220
    },
    {
      "epoch": 1.7794819359236538,
      "grad_norm": 18.290257947410613,
      "learning_rate": 4.252220886583493e-06,
      "loss": 2.5872,
      "step": 5221
    },
    {
      "epoch": 1.7798227675528289,
      "grad_norm": 23.1361964322411,
      "learning_rate": 4.2502601702616865e-06,
      "loss": 2.6008,
      "step": 5222
    },
    {
      "epoch": 1.780163599182004,
      "grad_norm": 25.585345837680407,
      "learning_rate": 4.248299571876938e-06,
      "loss": 1.9261,
      "step": 5223
    },
    {
      "epoch": 1.7805044308111793,
      "grad_norm": 22.67873830928119,
      "learning_rate": 4.246339091737658e-06,
      "loss": 2.6562,
      "step": 5224
    },
    {
      "epoch": 1.7808452624403546,
      "grad_norm": 16.163437960381685,
      "learning_rate": 4.244378730152242e-06,
      "loss": 1.683,
      "step": 5225
    },
    {
      "epoch": 1.7811860940695297,
      "grad_norm": 19.42186530338165,
      "learning_rate": 4.242418487429059e-06,
      "loss": 2.3917,
      "step": 5226
    },
    {
      "epoch": 1.7815269256987047,
      "grad_norm": 21.18168637620133,
      "learning_rate": 4.240458363876462e-06,
      "loss": 2.1382,
      "step": 5227
    },
    {
      "epoch": 1.78186775732788,
      "grad_norm": 15.746780198875369,
      "learning_rate": 4.23849835980279e-06,
      "loss": 2.3668,
      "step": 5228
    },
    {
      "epoch": 1.7822085889570554,
      "grad_norm": 21.914052364045357,
      "learning_rate": 4.236538475516357e-06,
      "loss": 2.6084,
      "step": 5229
    },
    {
      "epoch": 1.7825494205862304,
      "grad_norm": 13.966215805091714,
      "learning_rate": 4.2345787113254585e-06,
      "loss": 2.2779,
      "step": 5230
    },
    {
      "epoch": 1.7828902522154055,
      "grad_norm": 16.99400624637014,
      "learning_rate": 4.232619067538378e-06,
      "loss": 1.3921,
      "step": 5231
    },
    {
      "epoch": 1.7832310838445808,
      "grad_norm": 15.714590273838313,
      "learning_rate": 4.230659544463373e-06,
      "loss": 2.3525,
      "step": 5232
    },
    {
      "epoch": 1.783571915473756,
      "grad_norm": 20.999984552234825,
      "learning_rate": 4.228700142408683e-06,
      "loss": 2.0405,
      "step": 5233
    },
    {
      "epoch": 1.783912747102931,
      "grad_norm": 12.45463133752388,
      "learning_rate": 4.226740861682532e-06,
      "loss": 2.2719,
      "step": 5234
    },
    {
      "epoch": 1.7842535787321063,
      "grad_norm": 10.66365954237188,
      "learning_rate": 4.224781702593122e-06,
      "loss": 2.0158,
      "step": 5235
    },
    {
      "epoch": 1.7845944103612816,
      "grad_norm": 15.753772372338394,
      "learning_rate": 4.2228226654486335e-06,
      "loss": 2.2535,
      "step": 5236
    },
    {
      "epoch": 1.7849352419904567,
      "grad_norm": 12.915126000968876,
      "learning_rate": 4.220863750557237e-06,
      "loss": 2.123,
      "step": 5237
    },
    {
      "epoch": 1.7852760736196318,
      "grad_norm": 18.918818576709995,
      "learning_rate": 4.218904958227073e-06,
      "loss": 2.3984,
      "step": 5238
    },
    {
      "epoch": 1.7856169052488071,
      "grad_norm": 14.686613998264935,
      "learning_rate": 4.216946288766267e-06,
      "loss": 2.3304,
      "step": 5239
    },
    {
      "epoch": 1.7859577368779824,
      "grad_norm": 15.110052167738678,
      "learning_rate": 4.2149877424829275e-06,
      "loss": 2.2871,
      "step": 5240
    },
    {
      "epoch": 1.7862985685071575,
      "grad_norm": 18.221431098019252,
      "learning_rate": 4.213029319685142e-06,
      "loss": 2.3302,
      "step": 5241
    },
    {
      "epoch": 1.7866394001363326,
      "grad_norm": 16.80376866017313,
      "learning_rate": 4.211071020680976e-06,
      "loss": 2.5375,
      "step": 5242
    },
    {
      "epoch": 1.786980231765508,
      "grad_norm": 14.35877419240093,
      "learning_rate": 4.209112845778481e-06,
      "loss": 1.6696,
      "step": 5243
    },
    {
      "epoch": 1.787321063394683,
      "grad_norm": 16.096402117727695,
      "learning_rate": 4.207154795285682e-06,
      "loss": 2.5536,
      "step": 5244
    },
    {
      "epoch": 1.787661895023858,
      "grad_norm": 12.868784718432256,
      "learning_rate": 4.205196869510589e-06,
      "loss": 1.8969,
      "step": 5245
    },
    {
      "epoch": 1.7880027266530334,
      "grad_norm": 21.893972626025967,
      "learning_rate": 4.203239068761194e-06,
      "loss": 2.5371,
      "step": 5246
    },
    {
      "epoch": 1.7883435582822087,
      "grad_norm": 13.732050162989005,
      "learning_rate": 4.201281393345466e-06,
      "loss": 1.6791,
      "step": 5247
    },
    {
      "epoch": 1.7886843899113838,
      "grad_norm": 19.679665231345332,
      "learning_rate": 4.1993238435713525e-06,
      "loss": 2.1028,
      "step": 5248
    },
    {
      "epoch": 1.7890252215405589,
      "grad_norm": 15.032523555979186,
      "learning_rate": 4.197366419746786e-06,
      "loss": 2.2215,
      "step": 5249
    },
    {
      "epoch": 1.7893660531697342,
      "grad_norm": 14.712556490839296,
      "learning_rate": 4.195409122179678e-06,
      "loss": 2.1248,
      "step": 5250
    },
    {
      "epoch": 1.7897068847989095,
      "grad_norm": 15.255715711466312,
      "learning_rate": 4.193451951177917e-06,
      "loss": 2.2773,
      "step": 5251
    },
    {
      "epoch": 1.7900477164280846,
      "grad_norm": 17.34243385537868,
      "learning_rate": 4.191494907049375e-06,
      "loss": 1.967,
      "step": 5252
    },
    {
      "epoch": 1.7903885480572597,
      "grad_norm": 16.371860721254848,
      "learning_rate": 4.1895379901019025e-06,
      "loss": 2.0244,
      "step": 5253
    },
    {
      "epoch": 1.790729379686435,
      "grad_norm": 16.281854762652998,
      "learning_rate": 4.187581200643331e-06,
      "loss": 2.3322,
      "step": 5254
    },
    {
      "epoch": 1.79107021131561,
      "grad_norm": 15.798591614550407,
      "learning_rate": 4.185624538981469e-06,
      "loss": 2.4108,
      "step": 5255
    },
    {
      "epoch": 1.7914110429447851,
      "grad_norm": 16.861541017788415,
      "learning_rate": 4.1836680054241096e-06,
      "loss": 2.6917,
      "step": 5256
    },
    {
      "epoch": 1.7917518745739605,
      "grad_norm": 11.896934024503803,
      "learning_rate": 4.1817116002790235e-06,
      "loss": 1.8779,
      "step": 5257
    },
    {
      "epoch": 1.7920927062031358,
      "grad_norm": 21.159392504399207,
      "learning_rate": 4.179755323853959e-06,
      "loss": 2.9045,
      "step": 5258
    },
    {
      "epoch": 1.7924335378323109,
      "grad_norm": 14.35937225970893,
      "learning_rate": 4.177799176456649e-06,
      "loss": 2.3445,
      "step": 5259
    },
    {
      "epoch": 1.792774369461486,
      "grad_norm": 19.861287432225694,
      "learning_rate": 4.1758431583947995e-06,
      "loss": 2.5559,
      "step": 5260
    },
    {
      "epoch": 1.7931152010906612,
      "grad_norm": 17.56233568179384,
      "learning_rate": 4.1738872699761016e-06,
      "loss": 2.2898,
      "step": 5261
    },
    {
      "epoch": 1.7934560327198366,
      "grad_norm": 12.724758303865363,
      "learning_rate": 4.171931511508227e-06,
      "loss": 1.4939,
      "step": 5262
    },
    {
      "epoch": 1.7937968643490114,
      "grad_norm": 20.873635901506322,
      "learning_rate": 4.169975883298822e-06,
      "loss": 2.1963,
      "step": 5263
    },
    {
      "epoch": 1.7941376959781867,
      "grad_norm": 13.385316658502127,
      "learning_rate": 4.1680203856555115e-06,
      "loss": 1.8913,
      "step": 5264
    },
    {
      "epoch": 1.794478527607362,
      "grad_norm": 17.791482461741296,
      "learning_rate": 4.166065018885909e-06,
      "loss": 2.0954,
      "step": 5265
    },
    {
      "epoch": 1.7948193592365371,
      "grad_norm": 20.332298349375968,
      "learning_rate": 4.164109783297598e-06,
      "loss": 2.1327,
      "step": 5266
    },
    {
      "epoch": 1.7951601908657122,
      "grad_norm": 17.150149802514125,
      "learning_rate": 4.162154679198145e-06,
      "loss": 2.0578,
      "step": 5267
    },
    {
      "epoch": 1.7955010224948875,
      "grad_norm": 20.453349315939715,
      "learning_rate": 4.160199706895098e-06,
      "loss": 2.3716,
      "step": 5268
    },
    {
      "epoch": 1.7958418541240628,
      "grad_norm": 20.471220755625584,
      "learning_rate": 4.158244866695978e-06,
      "loss": 2.6919,
      "step": 5269
    },
    {
      "epoch": 1.796182685753238,
      "grad_norm": 12.811644309882633,
      "learning_rate": 4.156290158908292e-06,
      "loss": 1.5699,
      "step": 5270
    },
    {
      "epoch": 1.796523517382413,
      "grad_norm": 32.259828767005,
      "learning_rate": 4.154335583839524e-06,
      "loss": 2.866,
      "step": 5271
    },
    {
      "epoch": 1.7968643490115883,
      "grad_norm": 16.808700322592195,
      "learning_rate": 4.152381141797135e-06,
      "loss": 2.412,
      "step": 5272
    },
    {
      "epoch": 1.7972051806407636,
      "grad_norm": 15.82385120190494,
      "learning_rate": 4.1504268330885636e-06,
      "loss": 2.6066,
      "step": 5273
    },
    {
      "epoch": 1.7975460122699385,
      "grad_norm": 19.488692851969926,
      "learning_rate": 4.148472658021235e-06,
      "loss": 1.7622,
      "step": 5274
    },
    {
      "epoch": 1.7978868438991138,
      "grad_norm": 11.113578617486324,
      "learning_rate": 4.146518616902548e-06,
      "loss": 1.9224,
      "step": 5275
    },
    {
      "epoch": 1.798227675528289,
      "grad_norm": 18.41089840986063,
      "learning_rate": 4.1445647100398785e-06,
      "loss": 2.8304,
      "step": 5276
    },
    {
      "epoch": 1.7985685071574642,
      "grad_norm": 13.703152454797904,
      "learning_rate": 4.142610937740587e-06,
      "loss": 2.6544,
      "step": 5277
    },
    {
      "epoch": 1.7989093387866393,
      "grad_norm": 26.15031131142314,
      "learning_rate": 4.140657300312008e-06,
      "loss": 2.7237,
      "step": 5278
    },
    {
      "epoch": 1.7992501704158146,
      "grad_norm": 12.447822787596733,
      "learning_rate": 4.138703798061454e-06,
      "loss": 2.0417,
      "step": 5279
    },
    {
      "epoch": 1.79959100204499,
      "grad_norm": 15.319689822618809,
      "learning_rate": 4.136750431296225e-06,
      "loss": 1.8598,
      "step": 5280
    },
    {
      "epoch": 1.799931833674165,
      "grad_norm": 10.990436589457122,
      "learning_rate": 4.1347972003235876e-06,
      "loss": 2.2373,
      "step": 5281
    },
    {
      "epoch": 1.80027266530334,
      "grad_norm": 17.667099407811865,
      "learning_rate": 4.132844105450795e-06,
      "loss": 2.2733,
      "step": 5282
    },
    {
      "epoch": 1.8006134969325154,
      "grad_norm": 13.289553343281133,
      "learning_rate": 4.130891146985075e-06,
      "loss": 2.3071,
      "step": 5283
    },
    {
      "epoch": 1.8009543285616907,
      "grad_norm": 18.135636208020195,
      "learning_rate": 4.128938325233639e-06,
      "loss": 2.42,
      "step": 5284
    },
    {
      "epoch": 1.8012951601908656,
      "grad_norm": 15.821410389087161,
      "learning_rate": 4.126985640503672e-06,
      "loss": 2.0587,
      "step": 5285
    },
    {
      "epoch": 1.8016359918200409,
      "grad_norm": 15.786537814962793,
      "learning_rate": 4.125033093102338e-06,
      "loss": 2.5444,
      "step": 5286
    },
    {
      "epoch": 1.8019768234492162,
      "grad_norm": 18.55302424270161,
      "learning_rate": 4.123080683336782e-06,
      "loss": 2.6826,
      "step": 5287
    },
    {
      "epoch": 1.8023176550783913,
      "grad_norm": 15.181797071750413,
      "learning_rate": 4.121128411514126e-06,
      "loss": 2.1908,
      "step": 5288
    },
    {
      "epoch": 1.8026584867075663,
      "grad_norm": 16.44077128252316,
      "learning_rate": 4.1191762779414675e-06,
      "loss": 2.0888,
      "step": 5289
    },
    {
      "epoch": 1.8029993183367417,
      "grad_norm": 19.120082420509107,
      "learning_rate": 4.117224282925888e-06,
      "loss": 1.6726,
      "step": 5290
    },
    {
      "epoch": 1.803340149965917,
      "grad_norm": 20.112126824679056,
      "learning_rate": 4.115272426774445e-06,
      "loss": 2.3512,
      "step": 5291
    },
    {
      "epoch": 1.803680981595092,
      "grad_norm": 13.068415366263507,
      "learning_rate": 4.113320709794168e-06,
      "loss": 2.2284,
      "step": 5292
    },
    {
      "epoch": 1.8040218132242671,
      "grad_norm": 20.330160940644305,
      "learning_rate": 4.111369132292074e-06,
      "loss": 2.2363,
      "step": 5293
    },
    {
      "epoch": 1.8043626448534424,
      "grad_norm": 20.107232059657008,
      "learning_rate": 4.109417694575154e-06,
      "loss": 2.5534,
      "step": 5294
    },
    {
      "epoch": 1.8047034764826178,
      "grad_norm": 17.293733111459435,
      "learning_rate": 4.107466396950373e-06,
      "loss": 2.1371,
      "step": 5295
    },
    {
      "epoch": 1.8050443081117926,
      "grad_norm": 21.623254958576943,
      "learning_rate": 4.105515239724685e-06,
      "loss": 1.9963,
      "step": 5296
    },
    {
      "epoch": 1.805385139740968,
      "grad_norm": 14.790078173493932,
      "learning_rate": 4.1035642232050086e-06,
      "loss": 2.5188,
      "step": 5297
    },
    {
      "epoch": 1.8057259713701432,
      "grad_norm": 15.860063494542196,
      "learning_rate": 4.101613347698247e-06,
      "loss": 2.681,
      "step": 5298
    },
    {
      "epoch": 1.8060668029993183,
      "grad_norm": 15.588952323625755,
      "learning_rate": 4.099662613511282e-06,
      "loss": 1.6576,
      "step": 5299
    },
    {
      "epoch": 1.8064076346284934,
      "grad_norm": 21.686229632690182,
      "learning_rate": 4.097712020950975e-06,
      "loss": 2.1635,
      "step": 5300
    },
    {
      "epoch": 1.8067484662576687,
      "grad_norm": 13.495327285478396,
      "learning_rate": 4.095761570324156e-06,
      "loss": 2.0185,
      "step": 5301
    },
    {
      "epoch": 1.807089297886844,
      "grad_norm": 22.86670268439134,
      "learning_rate": 4.093811261937643e-06,
      "loss": 2.6338,
      "step": 5302
    },
    {
      "epoch": 1.8074301295160191,
      "grad_norm": 23.482154991723245,
      "learning_rate": 4.091861096098225e-06,
      "loss": 1.6411,
      "step": 5303
    },
    {
      "epoch": 1.8077709611451942,
      "grad_norm": 18.43740163414355,
      "learning_rate": 4.089911073112671e-06,
      "loss": 2.4895,
      "step": 5304
    },
    {
      "epoch": 1.8081117927743695,
      "grad_norm": 19.45332632423275,
      "learning_rate": 4.087961193287731e-06,
      "loss": 2.6011,
      "step": 5305
    },
    {
      "epoch": 1.8084526244035446,
      "grad_norm": 14.951846733006658,
      "learning_rate": 4.086011456930125e-06,
      "loss": 2.9102,
      "step": 5306
    },
    {
      "epoch": 1.8087934560327197,
      "grad_norm": 18.768173228768614,
      "learning_rate": 4.084061864346553e-06,
      "loss": 2.3317,
      "step": 5307
    },
    {
      "epoch": 1.809134287661895,
      "grad_norm": 25.17300467209956,
      "learning_rate": 4.082112415843697e-06,
      "loss": 2.1809,
      "step": 5308
    },
    {
      "epoch": 1.8094751192910703,
      "grad_norm": 23.842191094081535,
      "learning_rate": 4.080163111728213e-06,
      "loss": 2.1674,
      "step": 5309
    },
    {
      "epoch": 1.8098159509202454,
      "grad_norm": 100.1505503005152,
      "learning_rate": 4.07821395230673e-06,
      "loss": 2.4359,
      "step": 5310
    },
    {
      "epoch": 1.8101567825494205,
      "grad_norm": 13.211667676951642,
      "learning_rate": 4.076264937885865e-06,
      "loss": 2.087,
      "step": 5311
    },
    {
      "epoch": 1.8104976141785958,
      "grad_norm": 21.65860853053618,
      "learning_rate": 4.074316068772201e-06,
      "loss": 2.0583,
      "step": 5312
    },
    {
      "epoch": 1.810838445807771,
      "grad_norm": 16.819686289794454,
      "learning_rate": 4.072367345272303e-06,
      "loss": 1.9439,
      "step": 5313
    },
    {
      "epoch": 1.8111792774369462,
      "grad_norm": 17.128048975832634,
      "learning_rate": 4.070418767692718e-06,
      "loss": 2.4549,
      "step": 5314
    },
    {
      "epoch": 1.8115201090661213,
      "grad_norm": 22.152637723928414,
      "learning_rate": 4.0684703363399584e-06,
      "loss": 1.7793,
      "step": 5315
    },
    {
      "epoch": 1.8118609406952966,
      "grad_norm": 18.510399068108867,
      "learning_rate": 4.066522051520525e-06,
      "loss": 2.3308,
      "step": 5316
    },
    {
      "epoch": 1.8122017723244717,
      "grad_norm": 15.957078602891515,
      "learning_rate": 4.064573913540887e-06,
      "loss": 2.5012,
      "step": 5317
    },
    {
      "epoch": 1.8125426039536467,
      "grad_norm": 18.010431511342592,
      "learning_rate": 4.0626259227074975e-06,
      "loss": 2.3271,
      "step": 5318
    },
    {
      "epoch": 1.812883435582822,
      "grad_norm": 18.390821511479828,
      "learning_rate": 4.060678079326781e-06,
      "loss": 1.9537,
      "step": 5319
    },
    {
      "epoch": 1.8132242672119974,
      "grad_norm": 18.167934908606124,
      "learning_rate": 4.058730383705141e-06,
      "loss": 2.2359,
      "step": 5320
    },
    {
      "epoch": 1.8135650988411725,
      "grad_norm": 11.040832987567578,
      "learning_rate": 4.05678283614896e-06,
      "loss": 2.3592,
      "step": 5321
    },
    {
      "epoch": 1.8139059304703475,
      "grad_norm": 21.515480901682555,
      "learning_rate": 4.054835436964595e-06,
      "loss": 2.8566,
      "step": 5322
    },
    {
      "epoch": 1.8142467620995228,
      "grad_norm": 27.822219848635935,
      "learning_rate": 4.052888186458375e-06,
      "loss": 2.5675,
      "step": 5323
    },
    {
      "epoch": 1.8145875937286982,
      "grad_norm": 18.86849546759588,
      "learning_rate": 4.050941084936615e-06,
      "loss": 2.0854,
      "step": 5324
    },
    {
      "epoch": 1.8149284253578732,
      "grad_norm": 36.0872047116249,
      "learning_rate": 4.048994132705601e-06,
      "loss": 2.5225,
      "step": 5325
    },
    {
      "epoch": 1.8152692569870483,
      "grad_norm": 15.116708485695336,
      "learning_rate": 4.047047330071594e-06,
      "loss": 1.5788,
      "step": 5326
    },
    {
      "epoch": 1.8156100886162236,
      "grad_norm": 14.13487345607294,
      "learning_rate": 4.045100677340838e-06,
      "loss": 2.3415,
      "step": 5327
    },
    {
      "epoch": 1.8159509202453987,
      "grad_norm": 23.224817518452195,
      "learning_rate": 4.043154174819545e-06,
      "loss": 2.286,
      "step": 5328
    },
    {
      "epoch": 1.8162917518745738,
      "grad_norm": 15.198661860652349,
      "learning_rate": 4.041207822813908e-06,
      "loss": 1.9213,
      "step": 5329
    },
    {
      "epoch": 1.8166325835037491,
      "grad_norm": 13.078996283781139,
      "learning_rate": 4.039261621630099e-06,
      "loss": 2.2641,
      "step": 5330
    },
    {
      "epoch": 1.8169734151329244,
      "grad_norm": 15.324275388222635,
      "learning_rate": 4.037315571574263e-06,
      "loss": 2.3079,
      "step": 5331
    },
    {
      "epoch": 1.8173142467620995,
      "grad_norm": 13.856035833205768,
      "learning_rate": 4.035369672952516e-06,
      "loss": 2.1398,
      "step": 5332
    },
    {
      "epoch": 1.8176550783912746,
      "grad_norm": 14.632982459192034,
      "learning_rate": 4.033423926070962e-06,
      "loss": 2.0987,
      "step": 5333
    },
    {
      "epoch": 1.81799591002045,
      "grad_norm": 25.098778173110805,
      "learning_rate": 4.031478331235672e-06,
      "loss": 2.5418,
      "step": 5334
    },
    {
      "epoch": 1.8183367416496252,
      "grad_norm": 14.583556901799609,
      "learning_rate": 4.029532888752694e-06,
      "loss": 1.6359,
      "step": 5335
    },
    {
      "epoch": 1.8186775732788003,
      "grad_norm": 13.732222614199445,
      "learning_rate": 4.027587598928058e-06,
      "loss": 2.3258,
      "step": 5336
    },
    {
      "epoch": 1.8190184049079754,
      "grad_norm": 17.62441294535624,
      "learning_rate": 4.025642462067763e-06,
      "loss": 2.3704,
      "step": 5337
    },
    {
      "epoch": 1.8193592365371507,
      "grad_norm": 11.800373942694927,
      "learning_rate": 4.023697478477786e-06,
      "loss": 2.1146,
      "step": 5338
    },
    {
      "epoch": 1.8197000681663258,
      "grad_norm": 20.54721163408385,
      "learning_rate": 4.021752648464085e-06,
      "loss": 2.2792,
      "step": 5339
    },
    {
      "epoch": 1.8200408997955009,
      "grad_norm": 18.003816649829496,
      "learning_rate": 4.019807972332585e-06,
      "loss": 2.0871,
      "step": 5340
    },
    {
      "epoch": 1.8203817314246762,
      "grad_norm": 18.262664236806977,
      "learning_rate": 4.017863450389192e-06,
      "loss": 2.8312,
      "step": 5341
    },
    {
      "epoch": 1.8207225630538515,
      "grad_norm": 13.726829527009878,
      "learning_rate": 4.0159190829397885e-06,
      "loss": 2.1684,
      "step": 5342
    },
    {
      "epoch": 1.8210633946830266,
      "grad_norm": 16.467410620860203,
      "learning_rate": 4.013974870290231e-06,
      "loss": 1.9633,
      "step": 5343
    },
    {
      "epoch": 1.8214042263122017,
      "grad_norm": 13.599438671388898,
      "learning_rate": 4.012030812746351e-06,
      "loss": 2.3141,
      "step": 5344
    },
    {
      "epoch": 1.821745057941377,
      "grad_norm": 35.59953632392444,
      "learning_rate": 4.010086910613957e-06,
      "loss": 2.5423,
      "step": 5345
    },
    {
      "epoch": 1.8220858895705523,
      "grad_norm": 14.04560300190091,
      "learning_rate": 4.0081431641988335e-06,
      "loss": 2.5366,
      "step": 5346
    },
    {
      "epoch": 1.8224267211997274,
      "grad_norm": 18.490754730151576,
      "learning_rate": 4.0061995738067395e-06,
      "loss": 2.9418,
      "step": 5347
    },
    {
      "epoch": 1.8227675528289025,
      "grad_norm": 25.713681527336583,
      "learning_rate": 4.004256139743406e-06,
      "loss": 2.4698,
      "step": 5348
    },
    {
      "epoch": 1.8231083844580778,
      "grad_norm": 15.015974191721984,
      "learning_rate": 4.002312862314547e-06,
      "loss": 2.3324,
      "step": 5349
    },
    {
      "epoch": 1.8234492160872529,
      "grad_norm": 20.8739746475346,
      "learning_rate": 4.000369741825847e-06,
      "loss": 2.5385,
      "step": 5350
    },
    {
      "epoch": 1.823790047716428,
      "grad_norm": 12.5170399684396,
      "learning_rate": 3.998426778582964e-06,
      "loss": 2.0601,
      "step": 5351
    },
    {
      "epoch": 1.8241308793456033,
      "grad_norm": 13.909624554899443,
      "learning_rate": 3.996483972891537e-06,
      "loss": 2.4,
      "step": 5352
    },
    {
      "epoch": 1.8244717109747786,
      "grad_norm": 16.10633641251959,
      "learning_rate": 3.994541325057176e-06,
      "loss": 2.3286,
      "step": 5353
    },
    {
      "epoch": 1.8248125426039536,
      "grad_norm": 17.672322496093564,
      "learning_rate": 3.992598835385465e-06,
      "loss": 2.7052,
      "step": 5354
    },
    {
      "epoch": 1.8251533742331287,
      "grad_norm": 20.859456802996498,
      "learning_rate": 3.99065650418197e-06,
      "loss": 1.9969,
      "step": 5355
    },
    {
      "epoch": 1.825494205862304,
      "grad_norm": 19.639259208324454,
      "learning_rate": 3.9887143317522254e-06,
      "loss": 2.933,
      "step": 5356
    },
    {
      "epoch": 1.8258350374914794,
      "grad_norm": 14.48038345611974,
      "learning_rate": 3.9867723184017395e-06,
      "loss": 1.5502,
      "step": 5357
    },
    {
      "epoch": 1.8261758691206544,
      "grad_norm": 21.771392410132385,
      "learning_rate": 3.984830464436002e-06,
      "loss": 2.111,
      "step": 5358
    },
    {
      "epoch": 1.8265167007498295,
      "grad_norm": 17.84277477046626,
      "learning_rate": 3.982888770160475e-06,
      "loss": 2.8489,
      "step": 5359
    },
    {
      "epoch": 1.8268575323790048,
      "grad_norm": 19.44882956433336,
      "learning_rate": 3.980947235880591e-06,
      "loss": 2.2826,
      "step": 5360
    },
    {
      "epoch": 1.82719836400818,
      "grad_norm": 18.79453260160639,
      "learning_rate": 3.979005861901766e-06,
      "loss": 2.5535,
      "step": 5361
    },
    {
      "epoch": 1.827539195637355,
      "grad_norm": 22.18749627913491,
      "learning_rate": 3.977064648529382e-06,
      "loss": 2.3494,
      "step": 5362
    },
    {
      "epoch": 1.8278800272665303,
      "grad_norm": 18.467923935594293,
      "learning_rate": 3.9751235960688e-06,
      "loss": 2.4694,
      "step": 5363
    },
    {
      "epoch": 1.8282208588957056,
      "grad_norm": 15.567054424990191,
      "learning_rate": 3.973182704825358e-06,
      "loss": 2.4521,
      "step": 5364
    },
    {
      "epoch": 1.8285616905248807,
      "grad_norm": 15.986853982274303,
      "learning_rate": 3.971241975104364e-06,
      "loss": 1.9591,
      "step": 5365
    },
    {
      "epoch": 1.8289025221540558,
      "grad_norm": 11.969094264959878,
      "learning_rate": 3.969301407211101e-06,
      "loss": 1.9309,
      "step": 5366
    },
    {
      "epoch": 1.829243353783231,
      "grad_norm": 20.78736194190897,
      "learning_rate": 3.9673610014508305e-06,
      "loss": 1.9612,
      "step": 5367
    },
    {
      "epoch": 1.8295841854124064,
      "grad_norm": 15.219682106844013,
      "learning_rate": 3.965420758128785e-06,
      "loss": 2.3161,
      "step": 5368
    },
    {
      "epoch": 1.8299250170415815,
      "grad_norm": 24.200964629801554,
      "learning_rate": 3.963480677550172e-06,
      "loss": 2.4834,
      "step": 5369
    },
    {
      "epoch": 1.8302658486707566,
      "grad_norm": 36.8067726690729,
      "learning_rate": 3.961540760020176e-06,
      "loss": 2.1758,
      "step": 5370
    },
    {
      "epoch": 1.830606680299932,
      "grad_norm": 17.944613344852602,
      "learning_rate": 3.959601005843952e-06,
      "loss": 2.3653,
      "step": 5371
    },
    {
      "epoch": 1.830947511929107,
      "grad_norm": 20.634633030620336,
      "learning_rate": 3.957661415326629e-06,
      "loss": 2.4459,
      "step": 5372
    },
    {
      "epoch": 1.831288343558282,
      "grad_norm": 15.482432723208218,
      "learning_rate": 3.955721988773318e-06,
      "loss": 2.3225,
      "step": 5373
    },
    {
      "epoch": 1.8316291751874574,
      "grad_norm": 17.755063846197466,
      "learning_rate": 3.953782726489094e-06,
      "loss": 2.7196,
      "step": 5374
    },
    {
      "epoch": 1.8319700068166327,
      "grad_norm": 18.851793265046783,
      "learning_rate": 3.951843628779011e-06,
      "loss": 1.738,
      "step": 5375
    },
    {
      "epoch": 1.8323108384458078,
      "grad_norm": 33.082844764639724,
      "learning_rate": 3.949904695948098e-06,
      "loss": 1.7723,
      "step": 5376
    },
    {
      "epoch": 1.8326516700749829,
      "grad_norm": 12.650857740735146,
      "learning_rate": 3.9479659283013584e-06,
      "loss": 1.7583,
      "step": 5377
    },
    {
      "epoch": 1.8329925017041582,
      "grad_norm": 19.21770055819301,
      "learning_rate": 3.946027326143763e-06,
      "loss": 2.7085,
      "step": 5378
    },
    {
      "epoch": 1.8333333333333335,
      "grad_norm": 28.158030330742882,
      "learning_rate": 3.944088889780266e-06,
      "loss": 2.2155,
      "step": 5379
    },
    {
      "epoch": 1.8336741649625086,
      "grad_norm": 18.20117206044969,
      "learning_rate": 3.942150619515792e-06,
      "loss": 2.2372,
      "step": 5380
    },
    {
      "epoch": 1.8340149965916837,
      "grad_norm": 22.04513642868806,
      "learning_rate": 3.9402125156552365e-06,
      "loss": 1.7789,
      "step": 5381
    },
    {
      "epoch": 1.834355828220859,
      "grad_norm": 12.762118438532136,
      "learning_rate": 3.938274578503468e-06,
      "loss": 1.9846,
      "step": 5382
    },
    {
      "epoch": 1.834696659850034,
      "grad_norm": 16.561370649275347,
      "learning_rate": 3.936336808365337e-06,
      "loss": 2.1372,
      "step": 5383
    },
    {
      "epoch": 1.8350374914792091,
      "grad_norm": 15.631890059305396,
      "learning_rate": 3.93439920554566e-06,
      "loss": 2.3202,
      "step": 5384
    },
    {
      "epoch": 1.8353783231083844,
      "grad_norm": 18.554455407732313,
      "learning_rate": 3.9324617703492275e-06,
      "loss": 2.021,
      "step": 5385
    },
    {
      "epoch": 1.8357191547375598,
      "grad_norm": 17.09152752168175,
      "learning_rate": 3.9305245030808115e-06,
      "loss": 1.9107,
      "step": 5386
    },
    {
      "epoch": 1.8360599863667348,
      "grad_norm": 15.734942148775465,
      "learning_rate": 3.928587404045147e-06,
      "loss": 2.0328,
      "step": 5387
    },
    {
      "epoch": 1.83640081799591,
      "grad_norm": 17.087738906950293,
      "learning_rate": 3.926650473546948e-06,
      "loss": 2.1572,
      "step": 5388
    },
    {
      "epoch": 1.8367416496250852,
      "grad_norm": 33.23585351812588,
      "learning_rate": 3.924713711890904e-06,
      "loss": 1.883,
      "step": 5389
    },
    {
      "epoch": 1.8370824812542605,
      "grad_norm": 22.60373942993587,
      "learning_rate": 3.922777119381674e-06,
      "loss": 2.0568,
      "step": 5390
    },
    {
      "epoch": 1.8374233128834356,
      "grad_norm": 16.022359033614272,
      "learning_rate": 3.920840696323889e-06,
      "loss": 2.4661,
      "step": 5391
    },
    {
      "epoch": 1.8377641445126107,
      "grad_norm": 12.327123361942673,
      "learning_rate": 3.918904443022159e-06,
      "loss": 2.6154,
      "step": 5392
    },
    {
      "epoch": 1.838104976141786,
      "grad_norm": 20.143150152931838,
      "learning_rate": 3.916968359781066e-06,
      "loss": 2.0089,
      "step": 5393
    },
    {
      "epoch": 1.8384458077709611,
      "grad_norm": 49.41621929185133,
      "learning_rate": 3.915032446905158e-06,
      "loss": 3.0612,
      "step": 5394
    },
    {
      "epoch": 1.8387866394001362,
      "grad_norm": 20.091200356454596,
      "learning_rate": 3.913096704698968e-06,
      "loss": 2.1078,
      "step": 5395
    },
    {
      "epoch": 1.8391274710293115,
      "grad_norm": 18.080879377306367,
      "learning_rate": 3.9111611334669916e-06,
      "loss": 2.247,
      "step": 5396
    },
    {
      "epoch": 1.8394683026584868,
      "grad_norm": 13.655445105830692,
      "learning_rate": 3.909225733513702e-06,
      "loss": 2.1211,
      "step": 5397
    },
    {
      "epoch": 1.839809134287662,
      "grad_norm": 17.77306055611424,
      "learning_rate": 3.90729050514355e-06,
      "loss": 1.5655,
      "step": 5398
    },
    {
      "epoch": 1.840149965916837,
      "grad_norm": 16.773601542190285,
      "learning_rate": 3.905355448660949e-06,
      "loss": 2.4743,
      "step": 5399
    },
    {
      "epoch": 1.8404907975460123,
      "grad_norm": 17.64010912929376,
      "learning_rate": 3.903420564370293e-06,
      "loss": 2.4434,
      "step": 5400
    },
    {
      "epoch": 1.8408316291751876,
      "grad_norm": 17.26033404530369,
      "learning_rate": 3.901485852575948e-06,
      "loss": 2.3027,
      "step": 5401
    },
    {
      "epoch": 1.8411724608043627,
      "grad_norm": 19.486746221207007,
      "learning_rate": 3.899551313582254e-06,
      "loss": 1.5841,
      "step": 5402
    },
    {
      "epoch": 1.8415132924335378,
      "grad_norm": 22.68635468738929,
      "learning_rate": 3.897616947693515e-06,
      "loss": 1.9942,
      "step": 5403
    },
    {
      "epoch": 1.841854124062713,
      "grad_norm": 15.109666593361581,
      "learning_rate": 3.89568275521402e-06,
      "loss": 1.9629,
      "step": 5404
    },
    {
      "epoch": 1.8421949556918882,
      "grad_norm": 22.93100245109458,
      "learning_rate": 3.893748736448025e-06,
      "loss": 2.186,
      "step": 5405
    },
    {
      "epoch": 1.8425357873210633,
      "grad_norm": 30.58765026918194,
      "learning_rate": 3.891814891699755e-06,
      "loss": 2.2841,
      "step": 5406
    },
    {
      "epoch": 1.8428766189502386,
      "grad_norm": 18.320927284719897,
      "learning_rate": 3.889881221273416e-06,
      "loss": 2.8499,
      "step": 5407
    },
    {
      "epoch": 1.8432174505794139,
      "grad_norm": 16.99630439286904,
      "learning_rate": 3.887947725473179e-06,
      "loss": 2.2095,
      "step": 5408
    },
    {
      "epoch": 1.843558282208589,
      "grad_norm": 19.80053212080568,
      "learning_rate": 3.886014404603191e-06,
      "loss": 2.8838,
      "step": 5409
    },
    {
      "epoch": 1.843899113837764,
      "grad_norm": 18.031219010643383,
      "learning_rate": 3.8840812589675736e-06,
      "loss": 1.9273,
      "step": 5410
    },
    {
      "epoch": 1.8442399454669394,
      "grad_norm": 22.53288744734064,
      "learning_rate": 3.882148288870416e-06,
      "loss": 1.8267,
      "step": 5411
    },
    {
      "epoch": 1.8445807770961147,
      "grad_norm": 19.38998088783492,
      "learning_rate": 3.880215494615781e-06,
      "loss": 2.3386,
      "step": 5412
    },
    {
      "epoch": 1.8449216087252895,
      "grad_norm": 20.29531142559162,
      "learning_rate": 3.878282876507706e-06,
      "loss": 2.3972,
      "step": 5413
    },
    {
      "epoch": 1.8452624403544649,
      "grad_norm": 17.189241274637315,
      "learning_rate": 3.876350434850202e-06,
      "loss": 2.3633,
      "step": 5414
    },
    {
      "epoch": 1.8456032719836402,
      "grad_norm": 16.894366859659815,
      "learning_rate": 3.874418169947249e-06,
      "loss": 2.1585,
      "step": 5415
    },
    {
      "epoch": 1.8459441036128152,
      "grad_norm": 20.30972150329246,
      "learning_rate": 3.872486082102794e-06,
      "loss": 2.1053,
      "step": 5416
    },
    {
      "epoch": 1.8462849352419903,
      "grad_norm": 21.02858678578101,
      "learning_rate": 3.870554171620768e-06,
      "loss": 2.7315,
      "step": 5417
    },
    {
      "epoch": 1.8466257668711656,
      "grad_norm": 18.925724526703746,
      "learning_rate": 3.868622438805069e-06,
      "loss": 2.4989,
      "step": 5418
    },
    {
      "epoch": 1.846966598500341,
      "grad_norm": 20.563277971813807,
      "learning_rate": 3.866690883959562e-06,
      "loss": 1.9685,
      "step": 5419
    },
    {
      "epoch": 1.847307430129516,
      "grad_norm": 12.325027604122488,
      "learning_rate": 3.8647595073880914e-06,
      "loss": 2.384,
      "step": 5420
    },
    {
      "epoch": 1.8476482617586911,
      "grad_norm": 10.949704752712739,
      "learning_rate": 3.862828309394469e-06,
      "loss": 1.9477,
      "step": 5421
    },
    {
      "epoch": 1.8479890933878664,
      "grad_norm": 13.40415145401447,
      "learning_rate": 3.8608972902824794e-06,
      "loss": 1.7261,
      "step": 5422
    },
    {
      "epoch": 1.8483299250170417,
      "grad_norm": 12.208993916809291,
      "learning_rate": 3.8589664503558825e-06,
      "loss": 1.8894,
      "step": 5423
    },
    {
      "epoch": 1.8486707566462166,
      "grad_norm": 15.662102525429889,
      "learning_rate": 3.857035789918405e-06,
      "loss": 1.9553,
      "step": 5424
    },
    {
      "epoch": 1.849011588275392,
      "grad_norm": 10.840932719326052,
      "learning_rate": 3.855105309273746e-06,
      "loss": 1.9964,
      "step": 5425
    },
    {
      "epoch": 1.8493524199045672,
      "grad_norm": 17.245730742598422,
      "learning_rate": 3.85317500872558e-06,
      "loss": 2.9205,
      "step": 5426
    },
    {
      "epoch": 1.8496932515337423,
      "grad_norm": 16.829081013746404,
      "learning_rate": 3.851244888577552e-06,
      "loss": 1.4885,
      "step": 5427
    },
    {
      "epoch": 1.8500340831629174,
      "grad_norm": 20.834238005745817,
      "learning_rate": 3.8493149491332735e-06,
      "loss": 2.1753,
      "step": 5428
    },
    {
      "epoch": 1.8503749147920927,
      "grad_norm": 19.974372293410116,
      "learning_rate": 3.847385190696336e-06,
      "loss": 1.7784,
      "step": 5429
    },
    {
      "epoch": 1.850715746421268,
      "grad_norm": 11.251463239238076,
      "learning_rate": 3.845455613570296e-06,
      "loss": 1.925,
      "step": 5430
    },
    {
      "epoch": 1.851056578050443,
      "grad_norm": 59.14605982847282,
      "learning_rate": 3.843526218058682e-06,
      "loss": 2.4949,
      "step": 5431
    },
    {
      "epoch": 1.8513974096796182,
      "grad_norm": 20.797647069241894,
      "learning_rate": 3.841597004464999e-06,
      "loss": 2.0352,
      "step": 5432
    },
    {
      "epoch": 1.8517382413087935,
      "grad_norm": 17.855821203255736,
      "learning_rate": 3.839667973092719e-06,
      "loss": 1.989,
      "step": 5433
    },
    {
      "epoch": 1.8520790729379688,
      "grad_norm": 31.38899931925812,
      "learning_rate": 3.837739124245285e-06,
      "loss": 2.2728,
      "step": 5434
    },
    {
      "epoch": 1.8524199045671437,
      "grad_norm": 13.854555881633399,
      "learning_rate": 3.835810458226115e-06,
      "loss": 2.2048,
      "step": 5435
    },
    {
      "epoch": 1.852760736196319,
      "grad_norm": 21.11692354823943,
      "learning_rate": 3.833881975338594e-06,
      "loss": 2.5982,
      "step": 5436
    },
    {
      "epoch": 1.8531015678254943,
      "grad_norm": 15.97709005493656,
      "learning_rate": 3.831953675886078e-06,
      "loss": 2.4357,
      "step": 5437
    },
    {
      "epoch": 1.8534423994546694,
      "grad_norm": 16.937752780264564,
      "learning_rate": 3.8300255601718996e-06,
      "loss": 2.543,
      "step": 5438
    },
    {
      "epoch": 1.8537832310838445,
      "grad_norm": 19.46908120498656,
      "learning_rate": 3.82809762849936e-06,
      "loss": 2.0911,
      "step": 5439
    },
    {
      "epoch": 1.8541240627130198,
      "grad_norm": 17.583844201177495,
      "learning_rate": 3.826169881171726e-06,
      "loss": 1.7681,
      "step": 5440
    },
    {
      "epoch": 1.854464894342195,
      "grad_norm": 15.657335486424628,
      "learning_rate": 3.824242318492244e-06,
      "loss": 1.6985,
      "step": 5441
    },
    {
      "epoch": 1.8548057259713702,
      "grad_norm": 23.170581142449848,
      "learning_rate": 3.822314940764125e-06,
      "loss": 2.4023,
      "step": 5442
    },
    {
      "epoch": 1.8551465576005453,
      "grad_norm": 10.410521340717503,
      "learning_rate": 3.8203877482905525e-06,
      "loss": 1.7729,
      "step": 5443
    },
    {
      "epoch": 1.8554873892297206,
      "grad_norm": 13.083168093111494,
      "learning_rate": 3.818460741374686e-06,
      "loss": 2.3358,
      "step": 5444
    },
    {
      "epoch": 1.8558282208588959,
      "grad_norm": 19.761140240777944,
      "learning_rate": 3.816533920319647e-06,
      "loss": 2.1771,
      "step": 5445
    },
    {
      "epoch": 1.8561690524880707,
      "grad_norm": 19.84814108260719,
      "learning_rate": 3.8146072854285334e-06,
      "loss": 2.6458,
      "step": 5446
    },
    {
      "epoch": 1.856509884117246,
      "grad_norm": 18.438439881758473,
      "learning_rate": 3.812680837004411e-06,
      "loss": 1.9224,
      "step": 5447
    },
    {
      "epoch": 1.8568507157464214,
      "grad_norm": 12.66382205624291,
      "learning_rate": 3.8107545753503215e-06,
      "loss": 1.8907,
      "step": 5448
    },
    {
      "epoch": 1.8571915473755964,
      "grad_norm": 16.3787238926769,
      "learning_rate": 3.8088285007692723e-06,
      "loss": 2.8603,
      "step": 5449
    },
    {
      "epoch": 1.8575323790047715,
      "grad_norm": 17.82613604414563,
      "learning_rate": 3.8069026135642383e-06,
      "loss": 1.9485,
      "step": 5450
    },
    {
      "epoch": 1.8578732106339468,
      "grad_norm": 23.572689804199577,
      "learning_rate": 3.804976914038175e-06,
      "loss": 2.4331,
      "step": 5451
    },
    {
      "epoch": 1.8582140422631221,
      "grad_norm": 26.40614802022223,
      "learning_rate": 3.8030514024940004e-06,
      "loss": 2.4199,
      "step": 5452
    },
    {
      "epoch": 1.8585548738922972,
      "grad_norm": 20.86655132858453,
      "learning_rate": 3.801126079234603e-06,
      "loss": 2.5514,
      "step": 5453
    },
    {
      "epoch": 1.8588957055214723,
      "grad_norm": 16.493472163791374,
      "learning_rate": 3.799200944562848e-06,
      "loss": 2.6229,
      "step": 5454
    },
    {
      "epoch": 1.8592365371506476,
      "grad_norm": 13.22973088880136,
      "learning_rate": 3.7972759987815632e-06,
      "loss": 2.0509,
      "step": 5455
    },
    {
      "epoch": 1.8595773687798227,
      "grad_norm": 16.665693889711463,
      "learning_rate": 3.7953512421935505e-06,
      "loss": 2.5486,
      "step": 5456
    },
    {
      "epoch": 1.8599182004089978,
      "grad_norm": 11.325189655185142,
      "learning_rate": 3.7934266751015854e-06,
      "loss": 1.7887,
      "step": 5457
    },
    {
      "epoch": 1.860259032038173,
      "grad_norm": 18.23407797556481,
      "learning_rate": 3.791502297808406e-06,
      "loss": 2.629,
      "step": 5458
    },
    {
      "epoch": 1.8605998636673484,
      "grad_norm": 26.211388895607524,
      "learning_rate": 3.789578110616726e-06,
      "loss": 2.2025,
      "step": 5459
    },
    {
      "epoch": 1.8609406952965235,
      "grad_norm": 17.63498517150422,
      "learning_rate": 3.787654113829228e-06,
      "loss": 2.2585,
      "step": 5460
    },
    {
      "epoch": 1.8612815269256986,
      "grad_norm": 24.783256926563674,
      "learning_rate": 3.785730307748564e-06,
      "loss": 2.1737,
      "step": 5461
    },
    {
      "epoch": 1.861622358554874,
      "grad_norm": 13.151557703951811,
      "learning_rate": 3.7838066926773555e-06,
      "loss": 1.8326,
      "step": 5462
    },
    {
      "epoch": 1.8619631901840492,
      "grad_norm": 26.560757097872276,
      "learning_rate": 3.7818832689181956e-06,
      "loss": 2.0837,
      "step": 5463
    },
    {
      "epoch": 1.8623040218132243,
      "grad_norm": 18.6980271318111,
      "learning_rate": 3.7799600367736482e-06,
      "loss": 2.2287,
      "step": 5464
    },
    {
      "epoch": 1.8626448534423994,
      "grad_norm": 15.021311651804993,
      "learning_rate": 3.7780369965462414e-06,
      "loss": 2.0362,
      "step": 5465
    },
    {
      "epoch": 1.8629856850715747,
      "grad_norm": 25.322533654129792,
      "learning_rate": 3.7761141485384815e-06,
      "loss": 2.3419,
      "step": 5466
    },
    {
      "epoch": 1.8633265167007498,
      "grad_norm": 17.04701285225489,
      "learning_rate": 3.7741914930528372e-06,
      "loss": 2.3216,
      "step": 5467
    },
    {
      "epoch": 1.8636673483299249,
      "grad_norm": 20.08467943308246,
      "learning_rate": 3.7722690303917493e-06,
      "loss": 1.8963,
      "step": 5468
    },
    {
      "epoch": 1.8640081799591002,
      "grad_norm": 44.17629037827492,
      "learning_rate": 3.7703467608576326e-06,
      "loss": 2.7933,
      "step": 5469
    },
    {
      "epoch": 1.8643490115882755,
      "grad_norm": 26.887967101671023,
      "learning_rate": 3.768424684752865e-06,
      "loss": 1.7601,
      "step": 5470
    },
    {
      "epoch": 1.8646898432174506,
      "grad_norm": 18.564498926027866,
      "learning_rate": 3.7665028023797957e-06,
      "loss": 2.2128,
      "step": 5471
    },
    {
      "epoch": 1.8650306748466257,
      "grad_norm": 18.644052809913433,
      "learning_rate": 3.7645811140407463e-06,
      "loss": 2.5056,
      "step": 5472
    },
    {
      "epoch": 1.865371506475801,
      "grad_norm": 16.734606504052095,
      "learning_rate": 3.7626596200380073e-06,
      "loss": 2.3172,
      "step": 5473
    },
    {
      "epoch": 1.8657123381049763,
      "grad_norm": 15.531930133042572,
      "learning_rate": 3.7607383206738324e-06,
      "loss": 1.8331,
      "step": 5474
    },
    {
      "epoch": 1.8660531697341514,
      "grad_norm": 21.37672349743786,
      "learning_rate": 3.758817216250456e-06,
      "loss": 2.3419,
      "step": 5475
    },
    {
      "epoch": 1.8663940013633264,
      "grad_norm": 18.300953462532178,
      "learning_rate": 3.756896307070071e-06,
      "loss": 2.2766,
      "step": 5476
    },
    {
      "epoch": 1.8667348329925018,
      "grad_norm": 24.15597856189639,
      "learning_rate": 3.7549755934348463e-06,
      "loss": 1.9065,
      "step": 5477
    },
    {
      "epoch": 1.8670756646216768,
      "grad_norm": 13.484540627907016,
      "learning_rate": 3.7530550756469145e-06,
      "loss": 1.9062,
      "step": 5478
    },
    {
      "epoch": 1.867416496250852,
      "grad_norm": 18.123725643050438,
      "learning_rate": 3.751134754008386e-06,
      "loss": 2.4946,
      "step": 5479
    },
    {
      "epoch": 1.8677573278800272,
      "grad_norm": 15.410509573041058,
      "learning_rate": 3.7492146288213303e-06,
      "loss": 2.3069,
      "step": 5480
    },
    {
      "epoch": 1.8680981595092025,
      "grad_norm": 13.893625253870484,
      "learning_rate": 3.7472947003877924e-06,
      "loss": 1.9052,
      "step": 5481
    },
    {
      "epoch": 1.8684389911383776,
      "grad_norm": 17.871233625845978,
      "learning_rate": 3.745374969009786e-06,
      "loss": 1.8914,
      "step": 5482
    },
    {
      "epoch": 1.8687798227675527,
      "grad_norm": 17.622099451708582,
      "learning_rate": 3.7434554349892923e-06,
      "loss": 2.4595,
      "step": 5483
    },
    {
      "epoch": 1.869120654396728,
      "grad_norm": 15.064124254444001,
      "learning_rate": 3.741536098628258e-06,
      "loss": 2.4738,
      "step": 5484
    },
    {
      "epoch": 1.8694614860259033,
      "grad_norm": 13.780076661073135,
      "learning_rate": 3.739616960228606e-06,
      "loss": 2.1951,
      "step": 5485
    },
    {
      "epoch": 1.8698023176550784,
      "grad_norm": 20.928380804317737,
      "learning_rate": 3.737698020092225e-06,
      "loss": 2.6504,
      "step": 5486
    },
    {
      "epoch": 1.8701431492842535,
      "grad_norm": 16.76941858634153,
      "learning_rate": 3.735779278520968e-06,
      "loss": 2.1112,
      "step": 5487
    },
    {
      "epoch": 1.8704839809134288,
      "grad_norm": 32.62577829670677,
      "learning_rate": 3.733860735816665e-06,
      "loss": 2.0106,
      "step": 5488
    },
    {
      "epoch": 1.870824812542604,
      "grad_norm": 18.93530705333411,
      "learning_rate": 3.7319423922811076e-06,
      "loss": 2.3522,
      "step": 5489
    },
    {
      "epoch": 1.871165644171779,
      "grad_norm": 21.442813411104343,
      "learning_rate": 3.7300242482160588e-06,
      "loss": 1.7603,
      "step": 5490
    },
    {
      "epoch": 1.8715064758009543,
      "grad_norm": 20.179905948582636,
      "learning_rate": 3.728106303923253e-06,
      "loss": 3.0635,
      "step": 5491
    },
    {
      "epoch": 1.8718473074301296,
      "grad_norm": 17.81000144401281,
      "learning_rate": 3.726188559704388e-06,
      "loss": 1.8042,
      "step": 5492
    },
    {
      "epoch": 1.8721881390593047,
      "grad_norm": 16.617553961970074,
      "learning_rate": 3.7242710158611327e-06,
      "loss": 2.2456,
      "step": 5493
    },
    {
      "epoch": 1.8725289706884798,
      "grad_norm": 22.77100741407869,
      "learning_rate": 3.722353672695126e-06,
      "loss": 2.3352,
      "step": 5494
    },
    {
      "epoch": 1.872869802317655,
      "grad_norm": 19.270056466659234,
      "learning_rate": 3.7204365305079726e-06,
      "loss": 1.7213,
      "step": 5495
    },
    {
      "epoch": 1.8732106339468304,
      "grad_norm": 24.127906696032294,
      "learning_rate": 3.718519589601245e-06,
      "loss": 1.9903,
      "step": 5496
    },
    {
      "epoch": 1.8735514655760055,
      "grad_norm": 23.55448152214235,
      "learning_rate": 3.7166028502764886e-06,
      "loss": 2.5481,
      "step": 5497
    },
    {
      "epoch": 1.8738922972051806,
      "grad_norm": 18.057559134261453,
      "learning_rate": 3.7146863128352135e-06,
      "loss": 2.5345,
      "step": 5498
    },
    {
      "epoch": 1.8742331288343559,
      "grad_norm": 17.134444421177733,
      "learning_rate": 3.7127699775788954e-06,
      "loss": 2.4588,
      "step": 5499
    },
    {
      "epoch": 1.874573960463531,
      "grad_norm": 20.60055181645457,
      "learning_rate": 3.710853844808987e-06,
      "loss": 2.4044,
      "step": 5500
    },
    {
      "epoch": 1.874914792092706,
      "grad_norm": 15.130182563402412,
      "learning_rate": 3.708937914826899e-06,
      "loss": 1.9754,
      "step": 5501
    },
    {
      "epoch": 1.8752556237218814,
      "grad_norm": 19.587476774282692,
      "learning_rate": 3.707022187934015e-06,
      "loss": 2.0481,
      "step": 5502
    },
    {
      "epoch": 1.8755964553510567,
      "grad_norm": 15.89547911229631,
      "learning_rate": 3.7051066644316907e-06,
      "loss": 2.2115,
      "step": 5503
    },
    {
      "epoch": 1.8759372869802318,
      "grad_norm": 12.172798697848298,
      "learning_rate": 3.7031913446212427e-06,
      "loss": 1.9933,
      "step": 5504
    },
    {
      "epoch": 1.8762781186094069,
      "grad_norm": 17.40168362380706,
      "learning_rate": 3.701276228803956e-06,
      "loss": 1.9696,
      "step": 5505
    },
    {
      "epoch": 1.8766189502385822,
      "grad_norm": 14.111624279443772,
      "learning_rate": 3.6993613172810894e-06,
      "loss": 2.2558,
      "step": 5506
    },
    {
      "epoch": 1.8769597818677575,
      "grad_norm": 32.4253237521203,
      "learning_rate": 3.6974466103538665e-06,
      "loss": 2.2649,
      "step": 5507
    },
    {
      "epoch": 1.8773006134969326,
      "grad_norm": 13.69628430636695,
      "learning_rate": 3.695532108323477e-06,
      "loss": 1.728,
      "step": 5508
    },
    {
      "epoch": 1.8776414451261076,
      "grad_norm": 14.724410674287563,
      "learning_rate": 3.693617811491077e-06,
      "loss": 2.1824,
      "step": 5509
    },
    {
      "epoch": 1.877982276755283,
      "grad_norm": 14.19634359268822,
      "learning_rate": 3.6917037201577977e-06,
      "loss": 2.2161,
      "step": 5510
    },
    {
      "epoch": 1.878323108384458,
      "grad_norm": 13.850615253818324,
      "learning_rate": 3.6897898346247317e-06,
      "loss": 2.3473,
      "step": 5511
    },
    {
      "epoch": 1.8786639400136331,
      "grad_norm": 12.795678142611386,
      "learning_rate": 3.6878761551929383e-06,
      "loss": 1.7457,
      "step": 5512
    },
    {
      "epoch": 1.8790047716428084,
      "grad_norm": 30.238673527129038,
      "learning_rate": 3.685962682163451e-06,
      "loss": 2.3129,
      "step": 5513
    },
    {
      "epoch": 1.8793456032719837,
      "grad_norm": 10.346640886699666,
      "learning_rate": 3.6840494158372637e-06,
      "loss": 1.8949,
      "step": 5514
    },
    {
      "epoch": 1.8796864349011588,
      "grad_norm": 17.081333530839462,
      "learning_rate": 3.68213635651534e-06,
      "loss": 1.8719,
      "step": 5515
    },
    {
      "epoch": 1.880027266530334,
      "grad_norm": 24.15274003870877,
      "learning_rate": 3.6802235044986166e-06,
      "loss": 2.2626,
      "step": 5516
    },
    {
      "epoch": 1.8803680981595092,
      "grad_norm": 14.125268132931238,
      "learning_rate": 3.6783108600879885e-06,
      "loss": 2.3088,
      "step": 5517
    },
    {
      "epoch": 1.8807089297886845,
      "grad_norm": 21.377144442066097,
      "learning_rate": 3.676398423584323e-06,
      "loss": 2.2857,
      "step": 5518
    },
    {
      "epoch": 1.8810497614178596,
      "grad_norm": 14.241330198250836,
      "learning_rate": 3.6744861952884547e-06,
      "loss": 1.901,
      "step": 5519
    },
    {
      "epoch": 1.8813905930470347,
      "grad_norm": 16.17560137049466,
      "learning_rate": 3.672574175501186e-06,
      "loss": 2.4718,
      "step": 5520
    },
    {
      "epoch": 1.88173142467621,
      "grad_norm": 14.491959881134107,
      "learning_rate": 3.670662364523281e-06,
      "loss": 1.6325,
      "step": 5521
    },
    {
      "epoch": 1.882072256305385,
      "grad_norm": 12.618588002554587,
      "learning_rate": 3.6687507626554784e-06,
      "loss": 2.5788,
      "step": 5522
    },
    {
      "epoch": 1.8824130879345602,
      "grad_norm": 18.380739449373838,
      "learning_rate": 3.6668393701984828e-06,
      "loss": 1.769,
      "step": 5523
    },
    {
      "epoch": 1.8827539195637355,
      "grad_norm": 16.82379569706096,
      "learning_rate": 3.6649281874529578e-06,
      "loss": 1.9779,
      "step": 5524
    },
    {
      "epoch": 1.8830947511929108,
      "grad_norm": 16.67131086997878,
      "learning_rate": 3.6630172147195463e-06,
      "loss": 2.3426,
      "step": 5525
    },
    {
      "epoch": 1.883435582822086,
      "grad_norm": 18.313662072453663,
      "learning_rate": 3.6611064522988477e-06,
      "loss": 2.0197,
      "step": 5526
    },
    {
      "epoch": 1.883776414451261,
      "grad_norm": 15.345352349840233,
      "learning_rate": 3.6591959004914333e-06,
      "loss": 2.6541,
      "step": 5527
    },
    {
      "epoch": 1.8841172460804363,
      "grad_norm": 16.818335475135736,
      "learning_rate": 3.657285559597843e-06,
      "loss": 2.4203,
      "step": 5528
    },
    {
      "epoch": 1.8844580777096116,
      "grad_norm": 26.008528569933954,
      "learning_rate": 3.6553754299185796e-06,
      "loss": 2.5216,
      "step": 5529
    },
    {
      "epoch": 1.8847989093387867,
      "grad_norm": 18.415761570424372,
      "learning_rate": 3.6534655117541107e-06,
      "loss": 2.2407,
      "step": 5530
    },
    {
      "epoch": 1.8851397409679618,
      "grad_norm": 23.355329174713432,
      "learning_rate": 3.6515558054048788e-06,
      "loss": 2.0337,
      "step": 5531
    },
    {
      "epoch": 1.885480572597137,
      "grad_norm": 17.644044481262387,
      "learning_rate": 3.6496463111712876e-06,
      "loss": 2.2844,
      "step": 5532
    },
    {
      "epoch": 1.8858214042263122,
      "grad_norm": 19.748746874800066,
      "learning_rate": 3.647737029353705e-06,
      "loss": 2.1656,
      "step": 5533
    },
    {
      "epoch": 1.8861622358554873,
      "grad_norm": 18.93327421693914,
      "learning_rate": 3.6458279602524727e-06,
      "loss": 1.8169,
      "step": 5534
    },
    {
      "epoch": 1.8865030674846626,
      "grad_norm": 20.997847924488294,
      "learning_rate": 3.643919104167892e-06,
      "loss": 2.3983,
      "step": 5535
    },
    {
      "epoch": 1.8868438991138379,
      "grad_norm": 20.719673945368257,
      "learning_rate": 3.6420104614002326e-06,
      "loss": 1.994,
      "step": 5536
    },
    {
      "epoch": 1.887184730743013,
      "grad_norm": 15.474281346169347,
      "learning_rate": 3.640102032249737e-06,
      "loss": 2.1711,
      "step": 5537
    },
    {
      "epoch": 1.887525562372188,
      "grad_norm": 17.659204670184543,
      "learning_rate": 3.638193817016604e-06,
      "loss": 2.977,
      "step": 5538
    },
    {
      "epoch": 1.8878663940013634,
      "grad_norm": 19.08265176381865,
      "learning_rate": 3.636285816001003e-06,
      "loss": 2.5344,
      "step": 5539
    },
    {
      "epoch": 1.8882072256305387,
      "grad_norm": 14.856231534612647,
      "learning_rate": 3.634378029503073e-06,
      "loss": 2.2926,
      "step": 5540
    },
    {
      "epoch": 1.8885480572597138,
      "grad_norm": 15.222727051196117,
      "learning_rate": 3.632470457822916e-06,
      "loss": 2.3596,
      "step": 5541
    },
    {
      "epoch": 1.8888888888888888,
      "grad_norm": 19.889453278519966,
      "learning_rate": 3.6305631012605997e-06,
      "loss": 2.485,
      "step": 5542
    },
    {
      "epoch": 1.8892297205180641,
      "grad_norm": 16.80307559385271,
      "learning_rate": 3.628655960116157e-06,
      "loss": 2.8447,
      "step": 5543
    },
    {
      "epoch": 1.8895705521472392,
      "grad_norm": 19.5781477951071,
      "learning_rate": 3.6267490346895917e-06,
      "loss": 2.6112,
      "step": 5544
    },
    {
      "epoch": 1.8899113837764143,
      "grad_norm": 21.382007540252403,
      "learning_rate": 3.624842325280871e-06,
      "loss": 2.7581,
      "step": 5545
    },
    {
      "epoch": 1.8902522154055896,
      "grad_norm": 20.119814009338796,
      "learning_rate": 3.6229358321899243e-06,
      "loss": 2.2778,
      "step": 5546
    },
    {
      "epoch": 1.890593047034765,
      "grad_norm": 20.176608153244405,
      "learning_rate": 3.621029555716656e-06,
      "loss": 2.4865,
      "step": 5547
    },
    {
      "epoch": 1.89093387866394,
      "grad_norm": 17.949145233942073,
      "learning_rate": 3.619123496160926e-06,
      "loss": 2.4204,
      "step": 5548
    },
    {
      "epoch": 1.8912747102931151,
      "grad_norm": 19.030555249930458,
      "learning_rate": 3.6172176538225667e-06,
      "loss": 2.2029,
      "step": 5549
    },
    {
      "epoch": 1.8916155419222904,
      "grad_norm": 12.279642804269457,
      "learning_rate": 3.6153120290013767e-06,
      "loss": 2.2854,
      "step": 5550
    },
    {
      "epoch": 1.8919563735514657,
      "grad_norm": 21.713804143441553,
      "learning_rate": 3.6134066219971155e-06,
      "loss": 1.8023,
      "step": 5551
    },
    {
      "epoch": 1.8922972051806408,
      "grad_norm": 15.983897393054429,
      "learning_rate": 3.6115014331095113e-06,
      "loss": 2.6156,
      "step": 5552
    },
    {
      "epoch": 1.892638036809816,
      "grad_norm": 20.778649932518086,
      "learning_rate": 3.6095964626382607e-06,
      "loss": 2.2898,
      "step": 5553
    },
    {
      "epoch": 1.8929788684389912,
      "grad_norm": 21.594828585515383,
      "learning_rate": 3.607691710883022e-06,
      "loss": 2.0296,
      "step": 5554
    },
    {
      "epoch": 1.8933197000681663,
      "grad_norm": 20.228634313644406,
      "learning_rate": 3.6057871781434172e-06,
      "loss": 2.4109,
      "step": 5555
    },
    {
      "epoch": 1.8936605316973414,
      "grad_norm": 20.211987189810245,
      "learning_rate": 3.6038828647190404e-06,
      "loss": 2.3998,
      "step": 5556
    },
    {
      "epoch": 1.8940013633265167,
      "grad_norm": 15.105849218354074,
      "learning_rate": 3.6019787709094485e-06,
      "loss": 2.5004,
      "step": 5557
    },
    {
      "epoch": 1.894342194955692,
      "grad_norm": 26.850784959219894,
      "learning_rate": 3.600074897014158e-06,
      "loss": 2.301,
      "step": 5558
    },
    {
      "epoch": 1.894683026584867,
      "grad_norm": 16.479261936083503,
      "learning_rate": 3.5981712433326617e-06,
      "loss": 1.5512,
      "step": 5559
    },
    {
      "epoch": 1.8950238582140422,
      "grad_norm": 18.11122479618633,
      "learning_rate": 3.596267810164408e-06,
      "loss": 1.7308,
      "step": 5560
    },
    {
      "epoch": 1.8953646898432175,
      "grad_norm": 17.06471092515031,
      "learning_rate": 3.5943645978088145e-06,
      "loss": 2.3743,
      "step": 5561
    },
    {
      "epoch": 1.8957055214723928,
      "grad_norm": 20.77541465772056,
      "learning_rate": 3.592461606565267e-06,
      "loss": 2.5879,
      "step": 5562
    },
    {
      "epoch": 1.8960463531015677,
      "grad_norm": 18.053744649087598,
      "learning_rate": 3.5905588367331123e-06,
      "loss": 2.0665,
      "step": 5563
    },
    {
      "epoch": 1.896387184730743,
      "grad_norm": 22.91768731667087,
      "learning_rate": 3.588656288611661e-06,
      "loss": 2.7063,
      "step": 5564
    },
    {
      "epoch": 1.8967280163599183,
      "grad_norm": 14.446698628296824,
      "learning_rate": 3.5867539625001947e-06,
      "loss": 1.9811,
      "step": 5565
    },
    {
      "epoch": 1.8970688479890934,
      "grad_norm": 19.599639879044183,
      "learning_rate": 3.5848518586979575e-06,
      "loss": 2.431,
      "step": 5566
    },
    {
      "epoch": 1.8974096796182685,
      "grad_norm": 16.584948141070257,
      "learning_rate": 3.5829499775041535e-06,
      "loss": 2.153,
      "step": 5567
    },
    {
      "epoch": 1.8977505112474438,
      "grad_norm": 16.63980227457862,
      "learning_rate": 3.5810483192179616e-06,
      "loss": 2.5495,
      "step": 5568
    },
    {
      "epoch": 1.898091342876619,
      "grad_norm": 16.379059629990945,
      "learning_rate": 3.579146884138517e-06,
      "loss": 2.2716,
      "step": 5569
    },
    {
      "epoch": 1.8984321745057942,
      "grad_norm": 16.738709589305877,
      "learning_rate": 3.5772456725649213e-06,
      "loss": 2.0358,
      "step": 5570
    },
    {
      "epoch": 1.8987730061349692,
      "grad_norm": 17.536505857444812,
      "learning_rate": 3.575344684796248e-06,
      "loss": 2.8603,
      "step": 5571
    },
    {
      "epoch": 1.8991138377641446,
      "grad_norm": 15.617976006516926,
      "learning_rate": 3.5734439211315264e-06,
      "loss": 1.8207,
      "step": 5572
    },
    {
      "epoch": 1.8994546693933199,
      "grad_norm": 17.811819914602616,
      "learning_rate": 3.5715433818697543e-06,
      "loss": 2.0093,
      "step": 5573
    },
    {
      "epoch": 1.8997955010224947,
      "grad_norm": 21.128101082923994,
      "learning_rate": 3.569643067309893e-06,
      "loss": 2.2194,
      "step": 5574
    },
    {
      "epoch": 1.90013633265167,
      "grad_norm": 20.555240301422977,
      "learning_rate": 3.5677429777508736e-06,
      "loss": 2.7883,
      "step": 5575
    },
    {
      "epoch": 1.9004771642808453,
      "grad_norm": 15.053221908211968,
      "learning_rate": 3.5658431134915843e-06,
      "loss": 1.6976,
      "step": 5576
    },
    {
      "epoch": 1.9008179959100204,
      "grad_norm": 15.855010866354467,
      "learning_rate": 3.563943474830882e-06,
      "loss": 2.4734,
      "step": 5577
    },
    {
      "epoch": 1.9011588275391955,
      "grad_norm": 12.872288761351575,
      "learning_rate": 3.5620440620675876e-06,
      "loss": 1.9302,
      "step": 5578
    },
    {
      "epoch": 1.9014996591683708,
      "grad_norm": 20.448171569369762,
      "learning_rate": 3.5601448755004888e-06,
      "loss": 2.4691,
      "step": 5579
    },
    {
      "epoch": 1.9018404907975461,
      "grad_norm": 16.73904471512243,
      "learning_rate": 3.5582459154283288e-06,
      "loss": 2.0307,
      "step": 5580
    },
    {
      "epoch": 1.9021813224267212,
      "grad_norm": 16.67133662199536,
      "learning_rate": 3.5563471821498274e-06,
      "loss": 2.299,
      "step": 5581
    },
    {
      "epoch": 1.9025221540558963,
      "grad_norm": 19.915230317972554,
      "learning_rate": 3.5544486759636625e-06,
      "loss": 2.3992,
      "step": 5582
    },
    {
      "epoch": 1.9028629856850716,
      "grad_norm": 20.797949783204267,
      "learning_rate": 3.5525503971684728e-06,
      "loss": 2.6365,
      "step": 5583
    },
    {
      "epoch": 1.903203817314247,
      "grad_norm": 20.91887378369376,
      "learning_rate": 3.55065234606287e-06,
      "loss": 1.9854,
      "step": 5584
    },
    {
      "epoch": 1.9035446489434218,
      "grad_norm": 20.268762795144838,
      "learning_rate": 3.548754522945421e-06,
      "loss": 2.4453,
      "step": 5585
    },
    {
      "epoch": 1.903885480572597,
      "grad_norm": 17.6624068616757,
      "learning_rate": 3.5468569281146623e-06,
      "loss": 2.7625,
      "step": 5586
    },
    {
      "epoch": 1.9042263122017724,
      "grad_norm": 15.114643944708346,
      "learning_rate": 3.5449595618690957e-06,
      "loss": 1.7075,
      "step": 5587
    },
    {
      "epoch": 1.9045671438309475,
      "grad_norm": 27.829708912139864,
      "learning_rate": 3.5430624245071814e-06,
      "loss": 2.5964,
      "step": 5588
    },
    {
      "epoch": 1.9049079754601226,
      "grad_norm": 25.7360027561855,
      "learning_rate": 3.541165516327346e-06,
      "loss": 2.3428,
      "step": 5589
    },
    {
      "epoch": 1.905248807089298,
      "grad_norm": 26.908220138903317,
      "learning_rate": 3.5392688376279827e-06,
      "loss": 2.603,
      "step": 5590
    },
    {
      "epoch": 1.9055896387184732,
      "grad_norm": 16.98849905257533,
      "learning_rate": 3.5373723887074474e-06,
      "loss": 1.9958,
      "step": 5591
    },
    {
      "epoch": 1.9059304703476483,
      "grad_norm": 20.544282178375937,
      "learning_rate": 3.5354761698640555e-06,
      "loss": 2.5946,
      "step": 5592
    },
    {
      "epoch": 1.9062713019768234,
      "grad_norm": 18.079028959040244,
      "learning_rate": 3.5335801813960946e-06,
      "loss": 2.2374,
      "step": 5593
    },
    {
      "epoch": 1.9066121336059987,
      "grad_norm": 22.84246611095971,
      "learning_rate": 3.5316844236018067e-06,
      "loss": 1.9726,
      "step": 5594
    },
    {
      "epoch": 1.9069529652351738,
      "grad_norm": 14.839983385623013,
      "learning_rate": 3.5297888967794037e-06,
      "loss": 2.7813,
      "step": 5595
    },
    {
      "epoch": 1.9072937968643489,
      "grad_norm": 16.65762275551025,
      "learning_rate": 3.5278936012270613e-06,
      "loss": 2.6856,
      "step": 5596
    },
    {
      "epoch": 1.9076346284935242,
      "grad_norm": 20.86005673205567,
      "learning_rate": 3.5259985372429162e-06,
      "loss": 1.8478,
      "step": 5597
    },
    {
      "epoch": 1.9079754601226995,
      "grad_norm": 21.863554416652864,
      "learning_rate": 3.5241037051250666e-06,
      "loss": 1.9022,
      "step": 5598
    },
    {
      "epoch": 1.9083162917518746,
      "grad_norm": 16.78819145312999,
      "learning_rate": 3.5222091051715803e-06,
      "loss": 2.437,
      "step": 5599
    },
    {
      "epoch": 1.9086571233810496,
      "grad_norm": 18.73597889820278,
      "learning_rate": 3.520314737680485e-06,
      "loss": 2.1715,
      "step": 5600
    },
    {
      "epoch": 1.908997955010225,
      "grad_norm": 18.424780847321266,
      "learning_rate": 3.5184206029497705e-06,
      "loss": 2.3903,
      "step": 5601
    },
    {
      "epoch": 1.9093387866394003,
      "grad_norm": 13.974155858847043,
      "learning_rate": 3.516526701277395e-06,
      "loss": 1.7176,
      "step": 5602
    },
    {
      "epoch": 1.9096796182685754,
      "grad_norm": 14.48317957698722,
      "learning_rate": 3.5146330329612733e-06,
      "loss": 2.1846,
      "step": 5603
    },
    {
      "epoch": 1.9100204498977504,
      "grad_norm": 22.94630302878094,
      "learning_rate": 3.5127395982992874e-06,
      "loss": 2.1717,
      "step": 5604
    },
    {
      "epoch": 1.9103612815269257,
      "grad_norm": 16.164038544585825,
      "learning_rate": 3.5108463975892855e-06,
      "loss": 2.0537,
      "step": 5605
    },
    {
      "epoch": 1.9107021131561008,
      "grad_norm": 13.963995110630957,
      "learning_rate": 3.508953431129073e-06,
      "loss": 1.9119,
      "step": 5606
    },
    {
      "epoch": 1.911042944785276,
      "grad_norm": 18.71787540881993,
      "learning_rate": 3.5070606992164204e-06,
      "loss": 2.1594,
      "step": 5607
    },
    {
      "epoch": 1.9113837764144512,
      "grad_norm": 17.517107753734784,
      "learning_rate": 3.5051682021490617e-06,
      "loss": 1.9308,
      "step": 5608
    },
    {
      "epoch": 1.9117246080436265,
      "grad_norm": 12.663197458557725,
      "learning_rate": 3.5032759402246975e-06,
      "loss": 1.8802,
      "step": 5609
    },
    {
      "epoch": 1.9120654396728016,
      "grad_norm": 14.206608682014092,
      "learning_rate": 3.5013839137409857e-06,
      "loss": 2.1909,
      "step": 5610
    },
    {
      "epoch": 1.9124062713019767,
      "grad_norm": 18.48727210669149,
      "learning_rate": 3.4994921229955476e-06,
      "loss": 2.0807,
      "step": 5611
    },
    {
      "epoch": 1.912747102931152,
      "grad_norm": 20.409113496644782,
      "learning_rate": 3.497600568285975e-06,
      "loss": 2.0953,
      "step": 5612
    },
    {
      "epoch": 1.9130879345603273,
      "grad_norm": 23.568047119844575,
      "learning_rate": 3.4957092499098123e-06,
      "loss": 3.1201,
      "step": 5613
    },
    {
      "epoch": 1.9134287661895024,
      "grad_norm": 12.678549786480689,
      "learning_rate": 3.4938181681645723e-06,
      "loss": 2.0361,
      "step": 5614
    },
    {
      "epoch": 1.9137695978186775,
      "grad_norm": 17.456280837494003,
      "learning_rate": 3.4919273233477304e-06,
      "loss": 2.1769,
      "step": 5615
    },
    {
      "epoch": 1.9141104294478528,
      "grad_norm": 17.747463086592244,
      "learning_rate": 3.490036715756725e-06,
      "loss": 1.7644,
      "step": 5616
    },
    {
      "epoch": 1.914451261077028,
      "grad_norm": 18.88619553059399,
      "learning_rate": 3.4881463456889524e-06,
      "loss": 2.1973,
      "step": 5617
    },
    {
      "epoch": 1.914792092706203,
      "grad_norm": 13.806710656722867,
      "learning_rate": 3.486256213441781e-06,
      "loss": 2.3237,
      "step": 5618
    },
    {
      "epoch": 1.9151329243353783,
      "grad_norm": 14.966256240260128,
      "learning_rate": 3.484366319312531e-06,
      "loss": 2.5289,
      "step": 5619
    },
    {
      "epoch": 1.9154737559645536,
      "grad_norm": 15.722141765536367,
      "learning_rate": 3.4824766635984917e-06,
      "loss": 2.1837,
      "step": 5620
    },
    {
      "epoch": 1.9158145875937287,
      "grad_norm": 12.450166107828991,
      "learning_rate": 3.480587246596917e-06,
      "loss": 1.8585,
      "step": 5621
    },
    {
      "epoch": 1.9161554192229038,
      "grad_norm": 17.22469050739553,
      "learning_rate": 3.4786980686050166e-06,
      "loss": 2.5615,
      "step": 5622
    },
    {
      "epoch": 1.916496250852079,
      "grad_norm": 17.15954296891523,
      "learning_rate": 3.4768091299199636e-06,
      "loss": 2.5441,
      "step": 5623
    },
    {
      "epoch": 1.9168370824812544,
      "grad_norm": 20.68934956734518,
      "learning_rate": 3.4749204308388982e-06,
      "loss": 2.1294,
      "step": 5624
    },
    {
      "epoch": 1.9171779141104295,
      "grad_norm": 21.755049329096945,
      "learning_rate": 3.4730319716589222e-06,
      "loss": 1.9404,
      "step": 5625
    },
    {
      "epoch": 1.9175187457396046,
      "grad_norm": 18.68085217194838,
      "learning_rate": 3.471143752677093e-06,
      "loss": 2.7943,
      "step": 5626
    },
    {
      "epoch": 1.9178595773687799,
      "grad_norm": 23.79516780964141,
      "learning_rate": 3.46925577419044e-06,
      "loss": 1.9808,
      "step": 5627
    },
    {
      "epoch": 1.918200408997955,
      "grad_norm": 17.84499092158308,
      "learning_rate": 3.4673680364959455e-06,
      "loss": 2.1692,
      "step": 5628
    },
    {
      "epoch": 1.91854124062713,
      "grad_norm": 22.552971826758533,
      "learning_rate": 3.4654805398905594e-06,
      "loss": 2.4174,
      "step": 5629
    },
    {
      "epoch": 1.9188820722563054,
      "grad_norm": 18.470612978652458,
      "learning_rate": 3.4635932846711952e-06,
      "loss": 2.3553,
      "step": 5630
    },
    {
      "epoch": 1.9192229038854807,
      "grad_norm": 14.860869063311448,
      "learning_rate": 3.4617062711347225e-06,
      "loss": 1.8596,
      "step": 5631
    },
    {
      "epoch": 1.9195637355146558,
      "grad_norm": 18.243754193451856,
      "learning_rate": 3.4598194995779755e-06,
      "loss": 2.7957,
      "step": 5632
    },
    {
      "epoch": 1.9199045671438308,
      "grad_norm": 16.989146862003317,
      "learning_rate": 3.4579329702977536e-06,
      "loss": 2.7325,
      "step": 5633
    },
    {
      "epoch": 1.9202453987730062,
      "grad_norm": 22.62361816317594,
      "learning_rate": 3.4560466835908144e-06,
      "loss": 2.6324,
      "step": 5634
    },
    {
      "epoch": 1.9205862304021815,
      "grad_norm": 26.422617044343443,
      "learning_rate": 3.4541606397538766e-06,
      "loss": 2.3887,
      "step": 5635
    },
    {
      "epoch": 1.9209270620313565,
      "grad_norm": 19.34323904201938,
      "learning_rate": 3.4522748390836257e-06,
      "loss": 2.6277,
      "step": 5636
    },
    {
      "epoch": 1.9212678936605316,
      "grad_norm": 13.792169274596347,
      "learning_rate": 3.450389281876703e-06,
      "loss": 1.7431,
      "step": 5637
    },
    {
      "epoch": 1.921608725289707,
      "grad_norm": 39.34466237148315,
      "learning_rate": 3.4485039684297156e-06,
      "loss": 1.843,
      "step": 5638
    },
    {
      "epoch": 1.921949556918882,
      "grad_norm": 18.347517342433257,
      "learning_rate": 3.446618899039228e-06,
      "loss": 2.0572,
      "step": 5639
    },
    {
      "epoch": 1.9222903885480571,
      "grad_norm": 15.081028577185485,
      "learning_rate": 3.444734074001772e-06,
      "loss": 2.2265,
      "step": 5640
    },
    {
      "epoch": 1.9226312201772324,
      "grad_norm": 11.378503127785025,
      "learning_rate": 3.442849493613839e-06,
      "loss": 1.978,
      "step": 5641
    },
    {
      "epoch": 1.9229720518064077,
      "grad_norm": 12.591626126915639,
      "learning_rate": 3.440965158171877e-06,
      "loss": 1.7706,
      "step": 5642
    },
    {
      "epoch": 1.9233128834355828,
      "grad_norm": 24.208533703360317,
      "learning_rate": 3.4390810679723043e-06,
      "loss": 2.4827,
      "step": 5643
    },
    {
      "epoch": 1.923653715064758,
      "grad_norm": 18.11515509963514,
      "learning_rate": 3.4371972233114914e-06,
      "loss": 2.7296,
      "step": 5644
    },
    {
      "epoch": 1.9239945466939332,
      "grad_norm": 19.735214072388686,
      "learning_rate": 3.4353136244857753e-06,
      "loss": 1.7945,
      "step": 5645
    },
    {
      "epoch": 1.9243353783231085,
      "grad_norm": 15.911458076153854,
      "learning_rate": 3.433430271791457e-06,
      "loss": 1.9612,
      "step": 5646
    },
    {
      "epoch": 1.9246762099522836,
      "grad_norm": 14.144371171976402,
      "learning_rate": 3.4315471655247932e-06,
      "loss": 2.1085,
      "step": 5647
    },
    {
      "epoch": 1.9250170415814587,
      "grad_norm": 16.37502457838329,
      "learning_rate": 3.429664305982002e-06,
      "loss": 2.0092,
      "step": 5648
    },
    {
      "epoch": 1.925357873210634,
      "grad_norm": 16.478007238207738,
      "learning_rate": 3.4277816934592676e-06,
      "loss": 2.0521,
      "step": 5649
    },
    {
      "epoch": 1.925698704839809,
      "grad_norm": 18.450154291478864,
      "learning_rate": 3.4258993282527332e-06,
      "loss": 2.3042,
      "step": 5650
    },
    {
      "epoch": 1.9260395364689842,
      "grad_norm": 16.753240710163613,
      "learning_rate": 3.4240172106584975e-06,
      "loss": 2.6172,
      "step": 5651
    },
    {
      "epoch": 1.9263803680981595,
      "grad_norm": 16.10825816059553,
      "learning_rate": 3.4221353409726303e-06,
      "loss": 2.3384,
      "step": 5652
    },
    {
      "epoch": 1.9267211997273348,
      "grad_norm": 16.758206491052754,
      "learning_rate": 3.4202537194911546e-06,
      "loss": 2.437,
      "step": 5653
    },
    {
      "epoch": 1.9270620313565099,
      "grad_norm": 14.793720060978977,
      "learning_rate": 3.418372346510056e-06,
      "loss": 2.1786,
      "step": 5654
    },
    {
      "epoch": 1.927402862985685,
      "grad_norm": 18.496545379992753,
      "learning_rate": 3.4164912223252856e-06,
      "loss": 2.3196,
      "step": 5655
    },
    {
      "epoch": 1.9277436946148603,
      "grad_norm": 15.178406616159545,
      "learning_rate": 3.4146103472327497e-06,
      "loss": 2.1414,
      "step": 5656
    },
    {
      "epoch": 1.9280845262440356,
      "grad_norm": 16.326226236074,
      "learning_rate": 3.4127297215283144e-06,
      "loss": 2.186,
      "step": 5657
    },
    {
      "epoch": 1.9284253578732107,
      "grad_norm": 41.26069322594157,
      "learning_rate": 3.410849345507814e-06,
      "loss": 2.7666,
      "step": 5658
    },
    {
      "epoch": 1.9287661895023858,
      "grad_norm": 14.90282409509364,
      "learning_rate": 3.408969219467039e-06,
      "loss": 2.293,
      "step": 5659
    },
    {
      "epoch": 1.929107021131561,
      "grad_norm": 20.012160804395556,
      "learning_rate": 3.4070893437017367e-06,
      "loss": 1.6989,
      "step": 5660
    },
    {
      "epoch": 1.9294478527607362,
      "grad_norm": 21.18850908428512,
      "learning_rate": 3.405209718507624e-06,
      "loss": 1.8225,
      "step": 5661
    },
    {
      "epoch": 1.9297886843899112,
      "grad_norm": 23.10664377633577,
      "learning_rate": 3.4033303441803702e-06,
      "loss": 2.3421,
      "step": 5662
    },
    {
      "epoch": 1.9301295160190866,
      "grad_norm": 20.947637106742405,
      "learning_rate": 3.4014512210156082e-06,
      "loss": 2.4207,
      "step": 5663
    },
    {
      "epoch": 1.9304703476482619,
      "grad_norm": 18.99150168444652,
      "learning_rate": 3.399572349308934e-06,
      "loss": 2.1143,
      "step": 5664
    },
    {
      "epoch": 1.930811179277437,
      "grad_norm": 23.28858107985881,
      "learning_rate": 3.397693729355901e-06,
      "loss": 1.8742,
      "step": 5665
    },
    {
      "epoch": 1.931152010906612,
      "grad_norm": 17.94964050659806,
      "learning_rate": 3.3958153614520194e-06,
      "loss": 1.9081,
      "step": 5666
    },
    {
      "epoch": 1.9314928425357873,
      "grad_norm": 20.637905578879224,
      "learning_rate": 3.39393724589277e-06,
      "loss": 1.6868,
      "step": 5667
    },
    {
      "epoch": 1.9318336741649627,
      "grad_norm": 16.446261266672416,
      "learning_rate": 3.3920593829735855e-06,
      "loss": 2.0212,
      "step": 5668
    },
    {
      "epoch": 1.9321745057941377,
      "grad_norm": 17.54566547904794,
      "learning_rate": 3.390181772989859e-06,
      "loss": 2.4183,
      "step": 5669
    },
    {
      "epoch": 1.9325153374233128,
      "grad_norm": 14.974031377006112,
      "learning_rate": 3.3883044162369483e-06,
      "loss": 2.0128,
      "step": 5670
    },
    {
      "epoch": 1.9328561690524881,
      "grad_norm": 14.271743492252666,
      "learning_rate": 3.3864273130101705e-06,
      "loss": 1.8295,
      "step": 5671
    },
    {
      "epoch": 1.9331970006816632,
      "grad_norm": 14.244702205717381,
      "learning_rate": 3.3845504636047995e-06,
      "loss": 2.2481,
      "step": 5672
    },
    {
      "epoch": 1.9335378323108383,
      "grad_norm": 16.18297526822987,
      "learning_rate": 3.3826738683160685e-06,
      "loss": 2.2002,
      "step": 5673
    },
    {
      "epoch": 1.9338786639400136,
      "grad_norm": 19.583610417031558,
      "learning_rate": 3.380797527439178e-06,
      "loss": 2.3968,
      "step": 5674
    },
    {
      "epoch": 1.934219495569189,
      "grad_norm": 12.590362421429102,
      "learning_rate": 3.3789214412692833e-06,
      "loss": 1.7645,
      "step": 5675
    },
    {
      "epoch": 1.934560327198364,
      "grad_norm": 14.891583330713507,
      "learning_rate": 3.3770456101014967e-06,
      "loss": 2.1964,
      "step": 5676
    },
    {
      "epoch": 1.934901158827539,
      "grad_norm": 19.95567816572072,
      "learning_rate": 3.375170034230898e-06,
      "loss": 1.7691,
      "step": 5677
    },
    {
      "epoch": 1.9352419904567144,
      "grad_norm": 14.435772291560511,
      "learning_rate": 3.373294713952521e-06,
      "loss": 2.1553,
      "step": 5678
    },
    {
      "epoch": 1.9355828220858897,
      "grad_norm": 19.678758409587854,
      "learning_rate": 3.3714196495613584e-06,
      "loss": 2.2501,
      "step": 5679
    },
    {
      "epoch": 1.9359236537150648,
      "grad_norm": 25.616739424100103,
      "learning_rate": 3.3695448413523712e-06,
      "loss": 2.2277,
      "step": 5680
    },
    {
      "epoch": 1.93626448534424,
      "grad_norm": 11.715362211935195,
      "learning_rate": 3.3676702896204703e-06,
      "loss": 1.7563,
      "step": 5681
    },
    {
      "epoch": 1.9366053169734152,
      "grad_norm": 23.1252944079966,
      "learning_rate": 3.365795994660529e-06,
      "loss": 2.1098,
      "step": 5682
    },
    {
      "epoch": 1.9369461486025903,
      "grad_norm": 12.998331763938001,
      "learning_rate": 3.363921956767384e-06,
      "loss": 2.0097,
      "step": 5683
    },
    {
      "epoch": 1.9372869802317654,
      "grad_norm": 14.165779186255829,
      "learning_rate": 3.3620481762358292e-06,
      "loss": 2.1563,
      "step": 5684
    },
    {
      "epoch": 1.9376278118609407,
      "grad_norm": 16.72424741903025,
      "learning_rate": 3.360174653360615e-06,
      "loss": 2.0879,
      "step": 5685
    },
    {
      "epoch": 1.937968643490116,
      "grad_norm": 27.203504795620454,
      "learning_rate": 3.3583013884364567e-06,
      "loss": 2.1778,
      "step": 5686
    },
    {
      "epoch": 1.938309475119291,
      "grad_norm": 16.156666480343542,
      "learning_rate": 3.3564283817580247e-06,
      "loss": 1.9652,
      "step": 5687
    },
    {
      "epoch": 1.9386503067484662,
      "grad_norm": 13.606261908620045,
      "learning_rate": 3.35455563361995e-06,
      "loss": 1.7086,
      "step": 5688
    },
    {
      "epoch": 1.9389911383776415,
      "grad_norm": 20.116843998742954,
      "learning_rate": 3.3526831443168263e-06,
      "loss": 1.769,
      "step": 5689
    },
    {
      "epoch": 1.9393319700068168,
      "grad_norm": 19.301714940117613,
      "learning_rate": 3.350810914143201e-06,
      "loss": 2.6504,
      "step": 5690
    },
    {
      "epoch": 1.9396728016359919,
      "grad_norm": 19.43228678922453,
      "learning_rate": 3.3489389433935825e-06,
      "loss": 2.6698,
      "step": 5691
    },
    {
      "epoch": 1.940013633265167,
      "grad_norm": 18.42627404252224,
      "learning_rate": 3.347067232362442e-06,
      "loss": 2.5733,
      "step": 5692
    },
    {
      "epoch": 1.9403544648943423,
      "grad_norm": 12.137564369958367,
      "learning_rate": 3.345195781344206e-06,
      "loss": 1.9686,
      "step": 5693
    },
    {
      "epoch": 1.9406952965235174,
      "grad_norm": 19.09535130945397,
      "learning_rate": 3.3433245906332595e-06,
      "loss": 2.193,
      "step": 5694
    },
    {
      "epoch": 1.9410361281526924,
      "grad_norm": 14.628529685601755,
      "learning_rate": 3.341453660523952e-06,
      "loss": 1.9515,
      "step": 5695
    },
    {
      "epoch": 1.9413769597818678,
      "grad_norm": 27.419302762279617,
      "learning_rate": 3.339582991310584e-06,
      "loss": 1.9686,
      "step": 5696
    },
    {
      "epoch": 1.941717791411043,
      "grad_norm": 14.120642663211887,
      "learning_rate": 3.3377125832874214e-06,
      "loss": 1.8493,
      "step": 5697
    },
    {
      "epoch": 1.9420586230402181,
      "grad_norm": 18.197955075244902,
      "learning_rate": 3.3358424367486874e-06,
      "loss": 1.8572,
      "step": 5698
    },
    {
      "epoch": 1.9423994546693932,
      "grad_norm": 18.080344193269028,
      "learning_rate": 3.3339725519885622e-06,
      "loss": 2.2167,
      "step": 5699
    },
    {
      "epoch": 1.9427402862985685,
      "grad_norm": 15.898188572884655,
      "learning_rate": 3.3321029293011873e-06,
      "loss": 2.4496,
      "step": 5700
    },
    {
      "epoch": 1.9430811179277439,
      "grad_norm": 15.180573342837723,
      "learning_rate": 3.330233568980661e-06,
      "loss": 2.3685,
      "step": 5701
    },
    {
      "epoch": 1.943421949556919,
      "grad_norm": 11.024771534147874,
      "learning_rate": 3.3283644713210417e-06,
      "loss": 2.012,
      "step": 5702
    },
    {
      "epoch": 1.943762781186094,
      "grad_norm": 15.43241077671246,
      "learning_rate": 3.326495636616345e-06,
      "loss": 2.1581,
      "step": 5703
    },
    {
      "epoch": 1.9441036128152693,
      "grad_norm": 26.853148855737334,
      "learning_rate": 3.324627065160545e-06,
      "loss": 2.0485,
      "step": 5704
    },
    {
      "epoch": 1.9444444444444444,
      "grad_norm": 16.047938184512756,
      "learning_rate": 3.322758757247579e-06,
      "loss": 2.4596,
      "step": 5705
    },
    {
      "epoch": 1.9447852760736195,
      "grad_norm": 17.334869400196393,
      "learning_rate": 3.3208907131713373e-06,
      "loss": 2.0285,
      "step": 5706
    },
    {
      "epoch": 1.9451261077027948,
      "grad_norm": 17.47691095332133,
      "learning_rate": 3.3190229332256686e-06,
      "loss": 2.3595,
      "step": 5707
    },
    {
      "epoch": 1.9454669393319701,
      "grad_norm": 22.285971636643016,
      "learning_rate": 3.3171554177043847e-06,
      "loss": 1.8754,
      "step": 5708
    },
    {
      "epoch": 1.9458077709611452,
      "grad_norm": 17.042827274353698,
      "learning_rate": 3.315288166901254e-06,
      "loss": 2.7091,
      "step": 5709
    },
    {
      "epoch": 1.9461486025903203,
      "grad_norm": 18.585611143197024,
      "learning_rate": 3.313421181109999e-06,
      "loss": 2.578,
      "step": 5710
    },
    {
      "epoch": 1.9464894342194956,
      "grad_norm": 16.328094302866184,
      "learning_rate": 3.3115544606243083e-06,
      "loss": 2.116,
      "step": 5711
    },
    {
      "epoch": 1.946830265848671,
      "grad_norm": 21.189043588042637,
      "learning_rate": 3.309688005737821e-06,
      "loss": 1.8614,
      "step": 5712
    },
    {
      "epoch": 1.9471710974778458,
      "grad_norm": 18.16752628850933,
      "learning_rate": 3.3078218167441378e-06,
      "loss": 2.8667,
      "step": 5713
    },
    {
      "epoch": 1.947511929107021,
      "grad_norm": 23.788809618703798,
      "learning_rate": 3.305955893936822e-06,
      "loss": 2.4673,
      "step": 5714
    },
    {
      "epoch": 1.9478527607361964,
      "grad_norm": 21.50942480740741,
      "learning_rate": 3.304090237609387e-06,
      "loss": 1.3951,
      "step": 5715
    },
    {
      "epoch": 1.9481935923653715,
      "grad_norm": 15.126369984680121,
      "learning_rate": 3.3022248480553064e-06,
      "loss": 2.2645,
      "step": 5716
    },
    {
      "epoch": 1.9485344239945466,
      "grad_norm": 18.339461790788306,
      "learning_rate": 3.3003597255680174e-06,
      "loss": 2.2545,
      "step": 5717
    },
    {
      "epoch": 1.9488752556237219,
      "grad_norm": 17.034327207006395,
      "learning_rate": 3.2984948704409104e-06,
      "loss": 1.7873,
      "step": 5718
    },
    {
      "epoch": 1.9492160872528972,
      "grad_norm": 16.35926492535185,
      "learning_rate": 3.296630282967331e-06,
      "loss": 2.2263,
      "step": 5719
    },
    {
      "epoch": 1.9495569188820723,
      "grad_norm": 18.43875159462274,
      "learning_rate": 3.294765963440591e-06,
      "loss": 2.4229,
      "step": 5720
    },
    {
      "epoch": 1.9498977505112474,
      "grad_norm": 14.765969311322797,
      "learning_rate": 3.2929019121539516e-06,
      "loss": 2.184,
      "step": 5721
    },
    {
      "epoch": 1.9502385821404227,
      "grad_norm": 28.806074364653373,
      "learning_rate": 3.2910381294006365e-06,
      "loss": 2.3935,
      "step": 5722
    },
    {
      "epoch": 1.950579413769598,
      "grad_norm": 19.47064146815464,
      "learning_rate": 3.289174615473829e-06,
      "loss": 1.8824,
      "step": 5723
    },
    {
      "epoch": 1.9509202453987728,
      "grad_norm": 23.278854436042312,
      "learning_rate": 3.287311370666664e-06,
      "loss": 2.1099,
      "step": 5724
    },
    {
      "epoch": 1.9512610770279482,
      "grad_norm": 17.732278287311072,
      "learning_rate": 3.285448395272237e-06,
      "loss": 2.501,
      "step": 5725
    },
    {
      "epoch": 1.9516019086571235,
      "grad_norm": 17.833993211164263,
      "learning_rate": 3.283585689583604e-06,
      "loss": 2.5225,
      "step": 5726
    },
    {
      "epoch": 1.9519427402862985,
      "grad_norm": 12.54273241926783,
      "learning_rate": 3.281723253893775e-06,
      "loss": 1.7607,
      "step": 5727
    },
    {
      "epoch": 1.9522835719154736,
      "grad_norm": 18.947616712144374,
      "learning_rate": 3.279861088495716e-06,
      "loss": 2.0575,
      "step": 5728
    },
    {
      "epoch": 1.952624403544649,
      "grad_norm": 10.546948084053541,
      "learning_rate": 3.2779991936823568e-06,
      "loss": 2.0775,
      "step": 5729
    },
    {
      "epoch": 1.9529652351738243,
      "grad_norm": 11.413092740152146,
      "learning_rate": 3.2761375697465804e-06,
      "loss": 1.8894,
      "step": 5730
    },
    {
      "epoch": 1.9533060668029993,
      "grad_norm": 14.696943822839431,
      "learning_rate": 3.2742762169812236e-06,
      "loss": 1.884,
      "step": 5731
    },
    {
      "epoch": 1.9536468984321744,
      "grad_norm": 13.626171079282505,
      "learning_rate": 3.2724151356790906e-06,
      "loss": 2.2697,
      "step": 5732
    },
    {
      "epoch": 1.9539877300613497,
      "grad_norm": 19.9341111742677,
      "learning_rate": 3.2705543261329327e-06,
      "loss": 2.528,
      "step": 5733
    },
    {
      "epoch": 1.954328561690525,
      "grad_norm": 23.15069940763604,
      "learning_rate": 3.2686937886354624e-06,
      "loss": 2.9117,
      "step": 5734
    },
    {
      "epoch": 1.9546693933197,
      "grad_norm": 29.047363316481498,
      "learning_rate": 3.2668335234793537e-06,
      "loss": 2.3789,
      "step": 5735
    },
    {
      "epoch": 1.9550102249488752,
      "grad_norm": 20.478667771172468,
      "learning_rate": 3.2649735309572306e-06,
      "loss": 2.8673,
      "step": 5736
    },
    {
      "epoch": 1.9553510565780505,
      "grad_norm": 17.28018225536224,
      "learning_rate": 3.2631138113616767e-06,
      "loss": 2.1826,
      "step": 5737
    },
    {
      "epoch": 1.9556918882072256,
      "grad_norm": 19.452564874206686,
      "learning_rate": 3.2612543649852336e-06,
      "loss": 2.8529,
      "step": 5738
    },
    {
      "epoch": 1.9560327198364007,
      "grad_norm": 17.231369190378487,
      "learning_rate": 3.2593951921204024e-06,
      "loss": 2.1078,
      "step": 5739
    },
    {
      "epoch": 1.956373551465576,
      "grad_norm": 22.18192376291246,
      "learning_rate": 3.2575362930596366e-06,
      "loss": 1.5022,
      "step": 5740
    },
    {
      "epoch": 1.9567143830947513,
      "grad_norm": 14.863294710865148,
      "learning_rate": 3.255677668095346e-06,
      "loss": 2.4737,
      "step": 5741
    },
    {
      "epoch": 1.9570552147239264,
      "grad_norm": 23.72298850822961,
      "learning_rate": 3.2538193175199035e-06,
      "loss": 2.5175,
      "step": 5742
    },
    {
      "epoch": 1.9573960463531015,
      "grad_norm": 15.549167727169428,
      "learning_rate": 3.251961241625634e-06,
      "loss": 1.9486,
      "step": 5743
    },
    {
      "epoch": 1.9577368779822768,
      "grad_norm": 13.255547340052246,
      "learning_rate": 3.250103440704818e-06,
      "loss": 2.1866,
      "step": 5744
    },
    {
      "epoch": 1.9580777096114519,
      "grad_norm": 15.910939636000897,
      "learning_rate": 3.2482459150496987e-06,
      "loss": 1.9629,
      "step": 5745
    },
    {
      "epoch": 1.958418541240627,
      "grad_norm": 23.591396123758418,
      "learning_rate": 3.246388664952469e-06,
      "loss": 2.8744,
      "step": 5746
    },
    {
      "epoch": 1.9587593728698023,
      "grad_norm": 20.57820224651733,
      "learning_rate": 3.244531690705283e-06,
      "loss": 2.0024,
      "step": 5747
    },
    {
      "epoch": 1.9591002044989776,
      "grad_norm": 17.356657514093797,
      "learning_rate": 3.242674992600251e-06,
      "loss": 3.0643,
      "step": 5748
    },
    {
      "epoch": 1.9594410361281527,
      "grad_norm": 20.278768178567862,
      "learning_rate": 3.240818570929438e-06,
      "loss": 2.1787,
      "step": 5749
    },
    {
      "epoch": 1.9597818677573278,
      "grad_norm": 19.478223208399715,
      "learning_rate": 3.238962425984865e-06,
      "loss": 2.1826,
      "step": 5750
    },
    {
      "epoch": 1.960122699386503,
      "grad_norm": 18.930110855891815,
      "learning_rate": 3.237106558058514e-06,
      "loss": 2.8057,
      "step": 5751
    },
    {
      "epoch": 1.9604635310156784,
      "grad_norm": 14.910385830596676,
      "learning_rate": 3.2352509674423205e-06,
      "loss": 1.9556,
      "step": 5752
    },
    {
      "epoch": 1.9608043626448535,
      "grad_norm": 19.650333671978228,
      "learning_rate": 3.233395654428172e-06,
      "loss": 2.0414,
      "step": 5753
    },
    {
      "epoch": 1.9611451942740286,
      "grad_norm": 38.37434696548279,
      "learning_rate": 3.2315406193079223e-06,
      "loss": 2.2915,
      "step": 5754
    },
    {
      "epoch": 1.9614860259032039,
      "grad_norm": 20.107692601381476,
      "learning_rate": 3.229685862373371e-06,
      "loss": 1.7174,
      "step": 5755
    },
    {
      "epoch": 1.961826857532379,
      "grad_norm": 17.20811445061792,
      "learning_rate": 3.2278313839162807e-06,
      "loss": 1.6098,
      "step": 5756
    },
    {
      "epoch": 1.962167689161554,
      "grad_norm": 26.3328355223654,
      "learning_rate": 3.2259771842283698e-06,
      "loss": 2.3995,
      "step": 5757
    },
    {
      "epoch": 1.9625085207907293,
      "grad_norm": 23.20900071197618,
      "learning_rate": 3.2241232636013087e-06,
      "loss": 2.2084,
      "step": 5758
    },
    {
      "epoch": 1.9628493524199047,
      "grad_norm": 47.758013088167665,
      "learning_rate": 3.222269622326727e-06,
      "loss": 1.8517,
      "step": 5759
    },
    {
      "epoch": 1.9631901840490797,
      "grad_norm": 16.891889705662788,
      "learning_rate": 3.2204162606962126e-06,
      "loss": 2.0093,
      "step": 5760
    },
    {
      "epoch": 1.9635310156782548,
      "grad_norm": 19.50382205363358,
      "learning_rate": 3.218563179001304e-06,
      "loss": 1.7428,
      "step": 5761
    },
    {
      "epoch": 1.9638718473074301,
      "grad_norm": 15.217503932690668,
      "learning_rate": 3.2167103775334972e-06,
      "loss": 2.1608,
      "step": 5762
    },
    {
      "epoch": 1.9642126789366054,
      "grad_norm": 13.727200541448699,
      "learning_rate": 3.2148578565842477e-06,
      "loss": 1.7859,
      "step": 5763
    },
    {
      "epoch": 1.9645535105657805,
      "grad_norm": 19.495935906105302,
      "learning_rate": 3.2130056164449642e-06,
      "loss": 2.155,
      "step": 5764
    },
    {
      "epoch": 1.9648943421949556,
      "grad_norm": 25.76264450147051,
      "learning_rate": 3.211153657407009e-06,
      "loss": 2.7626,
      "step": 5765
    },
    {
      "epoch": 1.965235173824131,
      "grad_norm": 15.932377752946246,
      "learning_rate": 3.2093019797617075e-06,
      "loss": 2.0986,
      "step": 5766
    },
    {
      "epoch": 1.965576005453306,
      "grad_norm": 45.208065379023786,
      "learning_rate": 3.207450583800331e-06,
      "loss": 2.2657,
      "step": 5767
    },
    {
      "epoch": 1.965916837082481,
      "grad_norm": 12.646697955164278,
      "learning_rate": 3.2055994698141146e-06,
      "loss": 2.0711,
      "step": 5768
    },
    {
      "epoch": 1.9662576687116564,
      "grad_norm": 14.213550719421916,
      "learning_rate": 3.203748638094242e-06,
      "loss": 2.4753,
      "step": 5769
    },
    {
      "epoch": 1.9665985003408317,
      "grad_norm": 22.064323664477183,
      "learning_rate": 3.2018980889318617e-06,
      "loss": 2.6925,
      "step": 5770
    },
    {
      "epoch": 1.9669393319700068,
      "grad_norm": 17.002245809385066,
      "learning_rate": 3.2000478226180693e-06,
      "loss": 1.7021,
      "step": 5771
    },
    {
      "epoch": 1.967280163599182,
      "grad_norm": 19.365027048915234,
      "learning_rate": 3.1981978394439186e-06,
      "loss": 2.1416,
      "step": 5772
    },
    {
      "epoch": 1.9676209952283572,
      "grad_norm": 16.78142851571679,
      "learning_rate": 3.1963481397004213e-06,
      "loss": 1.9589,
      "step": 5773
    },
    {
      "epoch": 1.9679618268575325,
      "grad_norm": 24.083165567436765,
      "learning_rate": 3.194498723678542e-06,
      "loss": 2.2409,
      "step": 5774
    },
    {
      "epoch": 1.9683026584867076,
      "grad_norm": 19.090730983097068,
      "learning_rate": 3.192649591669199e-06,
      "loss": 2.2722,
      "step": 5775
    },
    {
      "epoch": 1.9686434901158827,
      "grad_norm": 16.225279861519432,
      "learning_rate": 3.190800743963271e-06,
      "loss": 2.072,
      "step": 5776
    },
    {
      "epoch": 1.968984321745058,
      "grad_norm": 25.66007861051097,
      "learning_rate": 3.1889521808515888e-06,
      "loss": 2.0987,
      "step": 5777
    },
    {
      "epoch": 1.969325153374233,
      "grad_norm": 17.48914165331983,
      "learning_rate": 3.1871039026249357e-06,
      "loss": 2.6635,
      "step": 5778
    },
    {
      "epoch": 1.9696659850034082,
      "grad_norm": 15.066081004326822,
      "learning_rate": 3.185255909574058e-06,
      "loss": 2.5514,
      "step": 5779
    },
    {
      "epoch": 1.9700068166325835,
      "grad_norm": 21.175162333850228,
      "learning_rate": 3.183408201989648e-06,
      "loss": 1.9878,
      "step": 5780
    },
    {
      "epoch": 1.9703476482617588,
      "grad_norm": 14.054304146524911,
      "learning_rate": 3.1815607801623586e-06,
      "loss": 1.9365,
      "step": 5781
    },
    {
      "epoch": 1.9706884798909339,
      "grad_norm": 13.956719905404505,
      "learning_rate": 3.1797136443827994e-06,
      "loss": 2.2665,
      "step": 5782
    },
    {
      "epoch": 1.971029311520109,
      "grad_norm": 16.681645835655956,
      "learning_rate": 3.177866794941529e-06,
      "loss": 2.6672,
      "step": 5783
    },
    {
      "epoch": 1.9713701431492843,
      "grad_norm": 24.69453025659232,
      "learning_rate": 3.1760202321290657e-06,
      "loss": 1.1089,
      "step": 5784
    },
    {
      "epoch": 1.9717109747784596,
      "grad_norm": 19.061346066897997,
      "learning_rate": 3.1741739562358796e-06,
      "loss": 1.6685,
      "step": 5785
    },
    {
      "epoch": 1.9720518064076347,
      "grad_norm": 35.328165613819905,
      "learning_rate": 3.1723279675524006e-06,
      "loss": 2.4507,
      "step": 5786
    },
    {
      "epoch": 1.9723926380368098,
      "grad_norm": 24.193151889118255,
      "learning_rate": 3.170482266369005e-06,
      "loss": 1.7514,
      "step": 5787
    },
    {
      "epoch": 1.972733469665985,
      "grad_norm": 21.51140874434846,
      "learning_rate": 3.168636852976033e-06,
      "loss": 2.5879,
      "step": 5788
    },
    {
      "epoch": 1.9730743012951601,
      "grad_norm": 14.423748220129152,
      "learning_rate": 3.1667917276637753e-06,
      "loss": 2.3045,
      "step": 5789
    },
    {
      "epoch": 1.9734151329243352,
      "grad_norm": 21.45228877039174,
      "learning_rate": 3.1649468907224747e-06,
      "loss": 2.2022,
      "step": 5790
    },
    {
      "epoch": 1.9737559645535105,
      "grad_norm": 18.23047001484623,
      "learning_rate": 3.1631023424423347e-06,
      "loss": 2.8967,
      "step": 5791
    },
    {
      "epoch": 1.9740967961826859,
      "grad_norm": 26.14538127019853,
      "learning_rate": 3.1612580831135077e-06,
      "loss": 2.4539,
      "step": 5792
    },
    {
      "epoch": 1.974437627811861,
      "grad_norm": 19.091662792793464,
      "learning_rate": 3.159414113026103e-06,
      "loss": 2.2058,
      "step": 5793
    },
    {
      "epoch": 1.974778459441036,
      "grad_norm": 21.78570735284556,
      "learning_rate": 3.157570432470187e-06,
      "loss": 2.0414,
      "step": 5794
    },
    {
      "epoch": 1.9751192910702113,
      "grad_norm": 14.323647195712317,
      "learning_rate": 3.1557270417357767e-06,
      "loss": 2.1919,
      "step": 5795
    },
    {
      "epoch": 1.9754601226993866,
      "grad_norm": 16.821743304090177,
      "learning_rate": 3.153883941112842e-06,
      "loss": 2.4257,
      "step": 5796
    },
    {
      "epoch": 1.9758009543285617,
      "grad_norm": 18.597380136317422,
      "learning_rate": 3.1520411308913137e-06,
      "loss": 2.3544,
      "step": 5797
    },
    {
      "epoch": 1.9761417859577368,
      "grad_norm": 15.956738655964285,
      "learning_rate": 3.1501986113610716e-06,
      "loss": 2.0469,
      "step": 5798
    },
    {
      "epoch": 1.9764826175869121,
      "grad_norm": 10.51696138390365,
      "learning_rate": 3.1483563828119507e-06,
      "loss": 1.7813,
      "step": 5799
    },
    {
      "epoch": 1.9768234492160872,
      "grad_norm": 17.064759964947072,
      "learning_rate": 3.146514445533744e-06,
      "loss": 1.6749,
      "step": 5800
    },
    {
      "epoch": 1.9771642808452623,
      "grad_norm": 15.866098263017482,
      "learning_rate": 3.1446727998161915e-06,
      "loss": 2.5238,
      "step": 5801
    },
    {
      "epoch": 1.9775051124744376,
      "grad_norm": 16.21090472417081,
      "learning_rate": 3.142831445948994e-06,
      "loss": 2.7831,
      "step": 5802
    },
    {
      "epoch": 1.977845944103613,
      "grad_norm": 24.384660582288838,
      "learning_rate": 3.1409903842218014e-06,
      "loss": 2.4727,
      "step": 5803
    },
    {
      "epoch": 1.978186775732788,
      "grad_norm": 13.76026811050494,
      "learning_rate": 3.139149614924224e-06,
      "loss": 1.9295,
      "step": 5804
    },
    {
      "epoch": 1.978527607361963,
      "grad_norm": 18.965873107068063,
      "learning_rate": 3.137309138345819e-06,
      "loss": 1.9463,
      "step": 5805
    },
    {
      "epoch": 1.9788684389911384,
      "grad_norm": 16.251860260822273,
      "learning_rate": 3.1354689547760996e-06,
      "loss": 2.3171,
      "step": 5806
    },
    {
      "epoch": 1.9792092706203137,
      "grad_norm": 15.799053116682103,
      "learning_rate": 3.1336290645045385e-06,
      "loss": 1.7163,
      "step": 5807
    },
    {
      "epoch": 1.9795501022494888,
      "grad_norm": 13.37514858381636,
      "learning_rate": 3.131789467820555e-06,
      "loss": 2.1945,
      "step": 5808
    },
    {
      "epoch": 1.9798909338786639,
      "grad_norm": 17.999948210177433,
      "learning_rate": 3.1299501650135234e-06,
      "loss": 2.0215,
      "step": 5809
    },
    {
      "epoch": 1.9802317655078392,
      "grad_norm": 15.009488941506776,
      "learning_rate": 3.128111156372775e-06,
      "loss": 2.5378,
      "step": 5810
    },
    {
      "epoch": 1.9805725971370143,
      "grad_norm": 17.680059808737486,
      "learning_rate": 3.1262724421875956e-06,
      "loss": 2.2799,
      "step": 5811
    },
    {
      "epoch": 1.9809134287661894,
      "grad_norm": 16.22457885747555,
      "learning_rate": 3.124434022747217e-06,
      "loss": 2.2319,
      "step": 5812
    },
    {
      "epoch": 1.9812542603953647,
      "grad_norm": 15.83436911111804,
      "learning_rate": 3.1225958983408356e-06,
      "loss": 2.2219,
      "step": 5813
    },
    {
      "epoch": 1.98159509202454,
      "grad_norm": 15.018952785090322,
      "learning_rate": 3.1207580692575913e-06,
      "loss": 1.6091,
      "step": 5814
    },
    {
      "epoch": 1.981935923653715,
      "grad_norm": 15.15154657744068,
      "learning_rate": 3.1189205357865827e-06,
      "loss": 2.4045,
      "step": 5815
    },
    {
      "epoch": 1.9822767552828902,
      "grad_norm": 19.66513835846893,
      "learning_rate": 3.117083298216864e-06,
      "loss": 2.3999,
      "step": 5816
    },
    {
      "epoch": 1.9826175869120655,
      "grad_norm": 16.399741730416256,
      "learning_rate": 3.1152463568374368e-06,
      "loss": 2.5954,
      "step": 5817
    },
    {
      "epoch": 1.9829584185412408,
      "grad_norm": 16.899856732932477,
      "learning_rate": 3.1134097119372596e-06,
      "loss": 1.7977,
      "step": 5818
    },
    {
      "epoch": 1.9832992501704159,
      "grad_norm": 17.279680797781236,
      "learning_rate": 3.111573363805247e-06,
      "loss": 1.8444,
      "step": 5819
    },
    {
      "epoch": 1.983640081799591,
      "grad_norm": 24.82500988333673,
      "learning_rate": 3.109737312730261e-06,
      "loss": 1.8459,
      "step": 5820
    },
    {
      "epoch": 1.9839809134287663,
      "grad_norm": 21.313812559187685,
      "learning_rate": 3.107901559001119e-06,
      "loss": 2.409,
      "step": 5821
    },
    {
      "epoch": 1.9843217450579413,
      "grad_norm": 14.844109593475453,
      "learning_rate": 3.106066102906595e-06,
      "loss": 2.2521,
      "step": 5822
    },
    {
      "epoch": 1.9846625766871164,
      "grad_norm": 14.410854297147807,
      "learning_rate": 3.1042309447354125e-06,
      "loss": 2.412,
      "step": 5823
    },
    {
      "epoch": 1.9850034083162917,
      "grad_norm": 26.98131284906122,
      "learning_rate": 3.1023960847762476e-06,
      "loss": 2.7353,
      "step": 5824
    },
    {
      "epoch": 1.985344239945467,
      "grad_norm": 14.708311345567425,
      "learning_rate": 3.1005615233177342e-06,
      "loss": 2.472,
      "step": 5825
    },
    {
      "epoch": 1.9856850715746421,
      "grad_norm": 14.235317357297905,
      "learning_rate": 3.0987272606484536e-06,
      "loss": 2.2242,
      "step": 5826
    },
    {
      "epoch": 1.9860259032038172,
      "grad_norm": 15.618919417159342,
      "learning_rate": 3.096893297056942e-06,
      "loss": 2.1691,
      "step": 5827
    },
    {
      "epoch": 1.9863667348329925,
      "grad_norm": 24.31916766371571,
      "learning_rate": 3.0950596328316935e-06,
      "loss": 1.9603,
      "step": 5828
    },
    {
      "epoch": 1.9867075664621678,
      "grad_norm": 14.709506909539577,
      "learning_rate": 3.0932262682611473e-06,
      "loss": 2.0587,
      "step": 5829
    },
    {
      "epoch": 1.987048398091343,
      "grad_norm": 17.242443955886017,
      "learning_rate": 3.091393203633698e-06,
      "loss": 1.6641,
      "step": 5830
    },
    {
      "epoch": 1.987389229720518,
      "grad_norm": 13.971075838564936,
      "learning_rate": 3.0895604392376954e-06,
      "loss": 2.3534,
      "step": 5831
    },
    {
      "epoch": 1.9877300613496933,
      "grad_norm": 12.181546523214486,
      "learning_rate": 3.087727975361443e-06,
      "loss": 1.967,
      "step": 5832
    },
    {
      "epoch": 1.9880708929788684,
      "grad_norm": 17.44368513762168,
      "learning_rate": 3.0858958122931916e-06,
      "loss": 2.3094,
      "step": 5833
    },
    {
      "epoch": 1.9884117246080435,
      "grad_norm": 18.62775781995759,
      "learning_rate": 3.0840639503211467e-06,
      "loss": 1.6119,
      "step": 5834
    },
    {
      "epoch": 1.9887525562372188,
      "grad_norm": 18.804637308630152,
      "learning_rate": 3.0822323897334704e-06,
      "loss": 2.0878,
      "step": 5835
    },
    {
      "epoch": 1.9890933878663941,
      "grad_norm": 19.6721170105978,
      "learning_rate": 3.0804011308182747e-06,
      "loss": 2.7169,
      "step": 5836
    },
    {
      "epoch": 1.9894342194955692,
      "grad_norm": 15.30578367471209,
      "learning_rate": 3.0785701738636197e-06,
      "loss": 2.2699,
      "step": 5837
    },
    {
      "epoch": 1.9897750511247443,
      "grad_norm": 18.396020442070885,
      "learning_rate": 3.076739519157528e-06,
      "loss": 1.6899,
      "step": 5838
    },
    {
      "epoch": 1.9901158827539196,
      "grad_norm": 16.63777053662709,
      "learning_rate": 3.0749091669879644e-06,
      "loss": 2.1342,
      "step": 5839
    },
    {
      "epoch": 1.990456714383095,
      "grad_norm": 17.145648256250215,
      "learning_rate": 3.0730791176428516e-06,
      "loss": 2.3421,
      "step": 5840
    },
    {
      "epoch": 1.99079754601227,
      "grad_norm": 21.136408993985373,
      "learning_rate": 3.071249371410066e-06,
      "loss": 2.2683,
      "step": 5841
    },
    {
      "epoch": 1.991138377641445,
      "grad_norm": 17.099539638031175,
      "learning_rate": 3.0694199285774313e-06,
      "loss": 2.2619,
      "step": 5842
    },
    {
      "epoch": 1.9914792092706204,
      "grad_norm": 24.386718217616597,
      "learning_rate": 3.067590789432727e-06,
      "loss": 1.9913,
      "step": 5843
    },
    {
      "epoch": 1.9918200408997955,
      "grad_norm": 15.223298445531094,
      "learning_rate": 3.065761954263684e-06,
      "loss": 2.2276,
      "step": 5844
    },
    {
      "epoch": 1.9921608725289706,
      "grad_norm": 17.170749936090306,
      "learning_rate": 3.0639334233579864e-06,
      "loss": 2.7601,
      "step": 5845
    },
    {
      "epoch": 1.9925017041581459,
      "grad_norm": 14.674428019745864,
      "learning_rate": 3.062105197003266e-06,
      "loss": 1.9504,
      "step": 5846
    },
    {
      "epoch": 1.9928425357873212,
      "grad_norm": 299.5254054958975,
      "learning_rate": 3.0602772754871134e-06,
      "loss": 1.7415,
      "step": 5847
    },
    {
      "epoch": 1.9931833674164963,
      "grad_norm": 16.213922038071285,
      "learning_rate": 3.0584496590970688e-06,
      "loss": 2.1123,
      "step": 5848
    },
    {
      "epoch": 1.9935241990456714,
      "grad_norm": 19.651975195401057,
      "learning_rate": 3.056622348120619e-06,
      "loss": 2.012,
      "step": 5849
    },
    {
      "epoch": 1.9938650306748467,
      "grad_norm": 20.208402353447077,
      "learning_rate": 3.054795342845212e-06,
      "loss": 1.7364,
      "step": 5850
    },
    {
      "epoch": 1.994205862304022,
      "grad_norm": 15.57343562293611,
      "learning_rate": 3.05296864355824e-06,
      "loss": 1.8256,
      "step": 5851
    },
    {
      "epoch": 1.9945466939331968,
      "grad_norm": 11.566271405367008,
      "learning_rate": 3.05114225054705e-06,
      "loss": 1.6349,
      "step": 5852
    },
    {
      "epoch": 1.9948875255623721,
      "grad_norm": 10.82615789459265,
      "learning_rate": 3.0493161640989444e-06,
      "loss": 2.0746,
      "step": 5853
    },
    {
      "epoch": 1.9952283571915475,
      "grad_norm": 13.99105413937312,
      "learning_rate": 3.047490384501171e-06,
      "loss": 2.3712,
      "step": 5854
    },
    {
      "epoch": 1.9955691888207225,
      "grad_norm": 12.074639327956755,
      "learning_rate": 3.045664912040931e-06,
      "loss": 2.102,
      "step": 5855
    },
    {
      "epoch": 1.9959100204498976,
      "grad_norm": 13.502177407742225,
      "learning_rate": 3.0438397470053806e-06,
      "loss": 1.9314,
      "step": 5856
    },
    {
      "epoch": 1.996250852079073,
      "grad_norm": 14.913298836632139,
      "learning_rate": 3.042014889681626e-06,
      "loss": 1.8047,
      "step": 5857
    },
    {
      "epoch": 1.9965916837082482,
      "grad_norm": 23.494732440905896,
      "learning_rate": 3.0401903403567222e-06,
      "loss": 2.8226,
      "step": 5858
    },
    {
      "epoch": 1.9969325153374233,
      "grad_norm": 12.015646814666866,
      "learning_rate": 3.038366099317681e-06,
      "loss": 1.396,
      "step": 5859
    },
    {
      "epoch": 1.9972733469665984,
      "grad_norm": 11.914490713108812,
      "learning_rate": 3.03654216685146e-06,
      "loss": 1.9757,
      "step": 5860
    },
    {
      "epoch": 1.9976141785957737,
      "grad_norm": 17.56173106992764,
      "learning_rate": 3.034718543244971e-06,
      "loss": 2.7571,
      "step": 5861
    },
    {
      "epoch": 1.997955010224949,
      "grad_norm": 13.703676360891958,
      "learning_rate": 3.0328952287850812e-06,
      "loss": 2.3831,
      "step": 5862
    },
    {
      "epoch": 1.998295841854124,
      "grad_norm": 17.36940272449991,
      "learning_rate": 3.031072223758602e-06,
      "loss": 2.4012,
      "step": 5863
    },
    {
      "epoch": 1.9986366734832992,
      "grad_norm": 15.15020348264035,
      "learning_rate": 3.029249528452298e-06,
      "loss": 1.6634,
      "step": 5864
    },
    {
      "epoch": 1.9989775051124745,
      "grad_norm": 21.78173159763287,
      "learning_rate": 3.0274271431528894e-06,
      "loss": 2.6104,
      "step": 5865
    },
    {
      "epoch": 1.9993183367416496,
      "grad_norm": 13.361806938458646,
      "learning_rate": 3.0256050681470446e-06,
      "loss": 1.9403,
      "step": 5866
    },
    {
      "epoch": 1.9996591683708247,
      "grad_norm": 19.34090170163813,
      "learning_rate": 3.0237833037213822e-06,
      "loss": 2.1582,
      "step": 5867
    },
    {
      "epoch": 2.0,
      "grad_norm": 16.4305595079954,
      "learning_rate": 3.02196185016247e-06,
      "loss": 1.3243,
      "step": 5868
    },
    {
      "epoch": 2.0003408316291753,
      "grad_norm": 17.49196182027437,
      "learning_rate": 3.020140707756836e-06,
      "loss": 1.5072,
      "step": 5869
    },
    {
      "epoch": 2.00068166325835,
      "grad_norm": 17.594830753949804,
      "learning_rate": 3.0183198767909493e-06,
      "loss": 1.4202,
      "step": 5870
    },
    {
      "epoch": 2.0010224948875255,
      "grad_norm": 7.653830446786605,
      "learning_rate": 3.0164993575512325e-06,
      "loss": 1.041,
      "step": 5871
    },
    {
      "epoch": 2.001363326516701,
      "grad_norm": 14.633944500922489,
      "learning_rate": 3.0146791503240656e-06,
      "loss": 1.5661,
      "step": 5872
    },
    {
      "epoch": 2.001704158145876,
      "grad_norm": 12.17944149309846,
      "learning_rate": 3.0128592553957703e-06,
      "loss": 1.8535,
      "step": 5873
    },
    {
      "epoch": 2.002044989775051,
      "grad_norm": 9.731941893972284,
      "learning_rate": 3.0110396730526226e-06,
      "loss": 1.7515,
      "step": 5874
    },
    {
      "epoch": 2.0023858214042263,
      "grad_norm": 10.412821352963755,
      "learning_rate": 3.009220403580855e-06,
      "loss": 1.4629,
      "step": 5875
    },
    {
      "epoch": 2.0027266530334016,
      "grad_norm": 22.6584449548545,
      "learning_rate": 3.0074014472666414e-06,
      "loss": 1.6104,
      "step": 5876
    },
    {
      "epoch": 2.003067484662577,
      "grad_norm": 18.016223455123445,
      "learning_rate": 3.005582804396111e-06,
      "loss": 1.4938,
      "step": 5877
    },
    {
      "epoch": 2.0034083162917518,
      "grad_norm": 12.732893865210198,
      "learning_rate": 3.0037644752553465e-06,
      "loss": 1.4687,
      "step": 5878
    },
    {
      "epoch": 2.003749147920927,
      "grad_norm": 10.442586534086297,
      "learning_rate": 3.001946460130377e-06,
      "loss": 1.5199,
      "step": 5879
    },
    {
      "epoch": 2.0040899795501024,
      "grad_norm": 29.834247509779253,
      "learning_rate": 3.0001287593071817e-06,
      "loss": 1.9859,
      "step": 5880
    },
    {
      "epoch": 2.0044308111792772,
      "grad_norm": 12.60843466704873,
      "learning_rate": 2.9983113730716935e-06,
      "loss": 1.6611,
      "step": 5881
    },
    {
      "epoch": 2.0047716428084525,
      "grad_norm": 14.701526469610043,
      "learning_rate": 2.996494301709796e-06,
      "loss": 1.2581,
      "step": 5882
    },
    {
      "epoch": 2.005112474437628,
      "grad_norm": 15.432775629222135,
      "learning_rate": 2.9946775455073186e-06,
      "loss": 1.8277,
      "step": 5883
    },
    {
      "epoch": 2.005453306066803,
      "grad_norm": 13.35674912828684,
      "learning_rate": 2.992861104750048e-06,
      "loss": 1.4428,
      "step": 5884
    },
    {
      "epoch": 2.005794137695978,
      "grad_norm": 12.668237259288661,
      "learning_rate": 2.991044979723714e-06,
      "loss": 1.4662,
      "step": 5885
    },
    {
      "epoch": 2.0061349693251533,
      "grad_norm": 19.262451123107567,
      "learning_rate": 2.9892291707140008e-06,
      "loss": 1.1924,
      "step": 5886
    },
    {
      "epoch": 2.0064758009543286,
      "grad_norm": 10.048104753670048,
      "learning_rate": 2.9874136780065465e-06,
      "loss": 1.3086,
      "step": 5887
    },
    {
      "epoch": 2.006816632583504,
      "grad_norm": 12.686965039912572,
      "learning_rate": 2.9855985018869314e-06,
      "loss": 1.3283,
      "step": 5888
    },
    {
      "epoch": 2.007157464212679,
      "grad_norm": 19.83647701817355,
      "learning_rate": 2.9837836426406892e-06,
      "loss": 1.2719,
      "step": 5889
    },
    {
      "epoch": 2.007498295841854,
      "grad_norm": 14.359877304684153,
      "learning_rate": 2.981969100553307e-06,
      "loss": 1.8391,
      "step": 5890
    },
    {
      "epoch": 2.0078391274710294,
      "grad_norm": 6.806612880084929,
      "learning_rate": 2.9801548759102197e-06,
      "loss": 0.8837,
      "step": 5891
    },
    {
      "epoch": 2.0081799591002043,
      "grad_norm": 17.884175152993528,
      "learning_rate": 2.978340968996809e-06,
      "loss": 1.8828,
      "step": 5892
    },
    {
      "epoch": 2.0085207907293796,
      "grad_norm": 13.949738768760142,
      "learning_rate": 2.9765273800984134e-06,
      "loss": 1.8214,
      "step": 5893
    },
    {
      "epoch": 2.008861622358555,
      "grad_norm": 17.23643291011136,
      "learning_rate": 2.974714109500315e-06,
      "loss": 1.8594,
      "step": 5894
    },
    {
      "epoch": 2.0092024539877302,
      "grad_norm": 15.877157448978377,
      "learning_rate": 2.9729011574877485e-06,
      "loss": 1.3273,
      "step": 5895
    },
    {
      "epoch": 2.009543285616905,
      "grad_norm": 12.48661549927127,
      "learning_rate": 2.971088524345901e-06,
      "loss": 1.2264,
      "step": 5896
    },
    {
      "epoch": 2.0098841172460804,
      "grad_norm": 14.897193380190433,
      "learning_rate": 2.969276210359906e-06,
      "loss": 1.893,
      "step": 5897
    },
    {
      "epoch": 2.0102249488752557,
      "grad_norm": 16.094292887869155,
      "learning_rate": 2.9674642158148447e-06,
      "loss": 2.1792,
      "step": 5898
    },
    {
      "epoch": 2.010565780504431,
      "grad_norm": 14.468080684023356,
      "learning_rate": 2.9656525409957538e-06,
      "loss": 1.5277,
      "step": 5899
    },
    {
      "epoch": 2.010906612133606,
      "grad_norm": 11.832464069433618,
      "learning_rate": 2.963841186187617e-06,
      "loss": 1.4943,
      "step": 5900
    },
    {
      "epoch": 2.011247443762781,
      "grad_norm": 14.518967868635778,
      "learning_rate": 2.962030151675367e-06,
      "loss": 2.0398,
      "step": 5901
    },
    {
      "epoch": 2.0115882753919565,
      "grad_norm": 11.955859998794185,
      "learning_rate": 2.9602194377438865e-06,
      "loss": 1.3329,
      "step": 5902
    },
    {
      "epoch": 2.0119291070211314,
      "grad_norm": 13.844855042666769,
      "learning_rate": 2.958409044678009e-06,
      "loss": 1.689,
      "step": 5903
    },
    {
      "epoch": 2.0122699386503067,
      "grad_norm": 16.836697715489677,
      "learning_rate": 2.956598972762516e-06,
      "loss": 1.1664,
      "step": 5904
    },
    {
      "epoch": 2.012610770279482,
      "grad_norm": 16.239951306986672,
      "learning_rate": 2.954789222282137e-06,
      "loss": 1.7287,
      "step": 5905
    },
    {
      "epoch": 2.0129516019086573,
      "grad_norm": 11.317480696703344,
      "learning_rate": 2.9529797935215565e-06,
      "loss": 1.4855,
      "step": 5906
    },
    {
      "epoch": 2.013292433537832,
      "grad_norm": 10.697092465237922,
      "learning_rate": 2.951170686765404e-06,
      "loss": 1.1902,
      "step": 5907
    },
    {
      "epoch": 2.0136332651670075,
      "grad_norm": 12.341907337210694,
      "learning_rate": 2.9493619022982557e-06,
      "loss": 1.529,
      "step": 5908
    },
    {
      "epoch": 2.0139740967961828,
      "grad_norm": 16.991995512288433,
      "learning_rate": 2.947553440404646e-06,
      "loss": 1.509,
      "step": 5909
    },
    {
      "epoch": 2.014314928425358,
      "grad_norm": 14.126546520587022,
      "learning_rate": 2.9457453013690496e-06,
      "loss": 0.7812,
      "step": 5910
    },
    {
      "epoch": 2.014655760054533,
      "grad_norm": 12.494546011310195,
      "learning_rate": 2.943937485475893e-06,
      "loss": 1.8191,
      "step": 5911
    },
    {
      "epoch": 2.0149965916837083,
      "grad_norm": 13.087342121501397,
      "learning_rate": 2.942129993009557e-06,
      "loss": 1.2004,
      "step": 5912
    },
    {
      "epoch": 2.0153374233128836,
      "grad_norm": 16.939804567243574,
      "learning_rate": 2.940322824254365e-06,
      "loss": 1.6693,
      "step": 5913
    },
    {
      "epoch": 2.0156782549420584,
      "grad_norm": 14.071113490960164,
      "learning_rate": 2.93851597949459e-06,
      "loss": 1.736,
      "step": 5914
    },
    {
      "epoch": 2.0160190865712337,
      "grad_norm": 15.233768683507082,
      "learning_rate": 2.9367094590144583e-06,
      "loss": 1.5352,
      "step": 5915
    },
    {
      "epoch": 2.016359918200409,
      "grad_norm": 18.837344593481212,
      "learning_rate": 2.9349032630981437e-06,
      "loss": 1.2867,
      "step": 5916
    },
    {
      "epoch": 2.0167007498295844,
      "grad_norm": 10.933406332923548,
      "learning_rate": 2.9330973920297645e-06,
      "loss": 1.4677,
      "step": 5917
    },
    {
      "epoch": 2.0170415814587592,
      "grad_norm": 12.966559877823043,
      "learning_rate": 2.931291846093395e-06,
      "loss": 1.313,
      "step": 5918
    },
    {
      "epoch": 2.0173824130879345,
      "grad_norm": 16.611414141344532,
      "learning_rate": 2.9294866255730527e-06,
      "loss": 1.2212,
      "step": 5919
    },
    {
      "epoch": 2.01772324471711,
      "grad_norm": 14.427731206730764,
      "learning_rate": 2.927681730752705e-06,
      "loss": 1.7153,
      "step": 5920
    },
    {
      "epoch": 2.018064076346285,
      "grad_norm": 11.780427128599952,
      "learning_rate": 2.925877161916273e-06,
      "loss": 1.5987,
      "step": 5921
    },
    {
      "epoch": 2.01840490797546,
      "grad_norm": 10.725819779057913,
      "learning_rate": 2.924072919347619e-06,
      "loss": 1.1893,
      "step": 5922
    },
    {
      "epoch": 2.0187457396046353,
      "grad_norm": 17.579116712848737,
      "learning_rate": 2.9222690033305574e-06,
      "loss": 1.1106,
      "step": 5923
    },
    {
      "epoch": 2.0190865712338106,
      "grad_norm": 16.679341409894658,
      "learning_rate": 2.9204654141488533e-06,
      "loss": 1.8806,
      "step": 5924
    },
    {
      "epoch": 2.0194274028629855,
      "grad_norm": 15.64063676280925,
      "learning_rate": 2.918662152086219e-06,
      "loss": 1.5929,
      "step": 5925
    },
    {
      "epoch": 2.019768234492161,
      "grad_norm": 13.98751452273023,
      "learning_rate": 2.91685921742631e-06,
      "loss": 1.9522,
      "step": 5926
    },
    {
      "epoch": 2.020109066121336,
      "grad_norm": 16.194339241862274,
      "learning_rate": 2.9150566104527427e-06,
      "loss": 1.3364,
      "step": 5927
    },
    {
      "epoch": 2.0204498977505114,
      "grad_norm": 13.775130828961522,
      "learning_rate": 2.9132543314490687e-06,
      "loss": 1.3988,
      "step": 5928
    },
    {
      "epoch": 2.0207907293796863,
      "grad_norm": 17.388030460466943,
      "learning_rate": 2.911452380698795e-06,
      "loss": 1.5626,
      "step": 5929
    },
    {
      "epoch": 2.0211315610088616,
      "grad_norm": 9.410181940610926,
      "learning_rate": 2.909650758485375e-06,
      "loss": 1.2159,
      "step": 5930
    },
    {
      "epoch": 2.021472392638037,
      "grad_norm": 12.46500719942039,
      "learning_rate": 2.9078494650922128e-06,
      "loss": 1.1145,
      "step": 5931
    },
    {
      "epoch": 2.021813224267212,
      "grad_norm": 18.106416333582533,
      "learning_rate": 2.9060485008026602e-06,
      "loss": 1.639,
      "step": 5932
    },
    {
      "epoch": 2.022154055896387,
      "grad_norm": 14.883923441402564,
      "learning_rate": 2.9042478659000093e-06,
      "loss": 2.0705,
      "step": 5933
    },
    {
      "epoch": 2.0224948875255624,
      "grad_norm": 13.446974329878971,
      "learning_rate": 2.9024475606675146e-06,
      "loss": 1.5106,
      "step": 5934
    },
    {
      "epoch": 2.0228357191547377,
      "grad_norm": 12.356651135702604,
      "learning_rate": 2.9006475853883696e-06,
      "loss": 1.491,
      "step": 5935
    },
    {
      "epoch": 2.0231765507839126,
      "grad_norm": 19.501011149278824,
      "learning_rate": 2.898847940345714e-06,
      "loss": 0.8964,
      "step": 5936
    },
    {
      "epoch": 2.023517382413088,
      "grad_norm": 16.124298541506512,
      "learning_rate": 2.897048625822644e-06,
      "loss": 1.7542,
      "step": 5937
    },
    {
      "epoch": 2.023858214042263,
      "grad_norm": 14.581607476617053,
      "learning_rate": 2.8952496421021946e-06,
      "loss": 1.2581,
      "step": 5938
    },
    {
      "epoch": 2.0241990456714385,
      "grad_norm": 14.913194602814318,
      "learning_rate": 2.893450989467354e-06,
      "loss": 1.7744,
      "step": 5939
    },
    {
      "epoch": 2.0245398773006134,
      "grad_norm": 16.75667877400706,
      "learning_rate": 2.8916526682010615e-06,
      "loss": 1.8273,
      "step": 5940
    },
    {
      "epoch": 2.0248807089297887,
      "grad_norm": 17.19260112750406,
      "learning_rate": 2.8898546785861957e-06,
      "loss": 1.1158,
      "step": 5941
    },
    {
      "epoch": 2.025221540558964,
      "grad_norm": 10.341508391087194,
      "learning_rate": 2.888057020905589e-06,
      "loss": 1.1803,
      "step": 5942
    },
    {
      "epoch": 2.025562372188139,
      "grad_norm": 26.34735470889903,
      "learning_rate": 2.8862596954420197e-06,
      "loss": 1.4077,
      "step": 5943
    },
    {
      "epoch": 2.025903203817314,
      "grad_norm": 40.38309636472682,
      "learning_rate": 2.884462702478215e-06,
      "loss": 1.4507,
      "step": 5944
    },
    {
      "epoch": 2.0262440354464895,
      "grad_norm": 17.200781458070555,
      "learning_rate": 2.882666042296849e-06,
      "loss": 1.4572,
      "step": 5945
    },
    {
      "epoch": 2.0265848670756648,
      "grad_norm": 13.659233618427017,
      "learning_rate": 2.880869715180542e-06,
      "loss": 1.6364,
      "step": 5946
    },
    {
      "epoch": 2.0269256987048396,
      "grad_norm": 12.586535167293086,
      "learning_rate": 2.879073721411865e-06,
      "loss": 1.7081,
      "step": 5947
    },
    {
      "epoch": 2.027266530334015,
      "grad_norm": 17.169462281367654,
      "learning_rate": 2.8772780612733353e-06,
      "loss": 1.8755,
      "step": 5948
    },
    {
      "epoch": 2.0276073619631902,
      "grad_norm": 16.130843722745627,
      "learning_rate": 2.8754827350474156e-06,
      "loss": 1.4468,
      "step": 5949
    },
    {
      "epoch": 2.0279481935923656,
      "grad_norm": 18.72158190175183,
      "learning_rate": 2.87368774301652e-06,
      "loss": 1.2191,
      "step": 5950
    },
    {
      "epoch": 2.0282890252215404,
      "grad_norm": 15.240119058873647,
      "learning_rate": 2.8718930854630058e-06,
      "loss": 1.361,
      "step": 5951
    },
    {
      "epoch": 2.0286298568507157,
      "grad_norm": 12.207534379252968,
      "learning_rate": 2.8700987626691812e-06,
      "loss": 1.4238,
      "step": 5952
    },
    {
      "epoch": 2.028970688479891,
      "grad_norm": 14.301160495922447,
      "learning_rate": 2.8683047749173023e-06,
      "loss": 1.4914,
      "step": 5953
    },
    {
      "epoch": 2.029311520109066,
      "grad_norm": 13.687410880607867,
      "learning_rate": 2.866511122489563e-06,
      "loss": 1.691,
      "step": 5954
    },
    {
      "epoch": 2.029652351738241,
      "grad_norm": 13.130999925203884,
      "learning_rate": 2.8647178056681197e-06,
      "loss": 1.8387,
      "step": 5955
    },
    {
      "epoch": 2.0299931833674165,
      "grad_norm": 21.07102286131651,
      "learning_rate": 2.862924824735068e-06,
      "loss": 2.1856,
      "step": 5956
    },
    {
      "epoch": 2.030334014996592,
      "grad_norm": 16.881189544885952,
      "learning_rate": 2.8611321799724433e-06,
      "loss": 1.571,
      "step": 5957
    },
    {
      "epoch": 2.0306748466257667,
      "grad_norm": 19.58728391415735,
      "learning_rate": 2.8593398716622466e-06,
      "loss": 1.8569,
      "step": 5958
    },
    {
      "epoch": 2.031015678254942,
      "grad_norm": 13.61058801036104,
      "learning_rate": 2.857547900086407e-06,
      "loss": 1.5459,
      "step": 5959
    },
    {
      "epoch": 2.0313565098841173,
      "grad_norm": 16.52674447606881,
      "learning_rate": 2.855756265526809e-06,
      "loss": 1.427,
      "step": 5960
    },
    {
      "epoch": 2.0316973415132926,
      "grad_norm": 11.579943498307614,
      "learning_rate": 2.8539649682652904e-06,
      "loss": 1.142,
      "step": 5961
    },
    {
      "epoch": 2.0320381731424675,
      "grad_norm": 10.561175014372564,
      "learning_rate": 2.8521740085836224e-06,
      "loss": 1.3507,
      "step": 5962
    },
    {
      "epoch": 2.032379004771643,
      "grad_norm": 14.39611247094284,
      "learning_rate": 2.8503833867635345e-06,
      "loss": 1.5365,
      "step": 5963
    },
    {
      "epoch": 2.032719836400818,
      "grad_norm": 9.703233820855386,
      "learning_rate": 2.8485931030866953e-06,
      "loss": 0.7759,
      "step": 5964
    },
    {
      "epoch": 2.033060668029993,
      "grad_norm": 15.641864156901267,
      "learning_rate": 2.8468031578347267e-06,
      "loss": 1.6484,
      "step": 5965
    },
    {
      "epoch": 2.0334014996591683,
      "grad_norm": 13.806622697665441,
      "learning_rate": 2.8450135512891936e-06,
      "loss": 1.2991,
      "step": 5966
    },
    {
      "epoch": 2.0337423312883436,
      "grad_norm": 15.159099623616427,
      "learning_rate": 2.8432242837316038e-06,
      "loss": 1.5164,
      "step": 5967
    },
    {
      "epoch": 2.034083162917519,
      "grad_norm": 15.119439544232968,
      "learning_rate": 2.8414353554434214e-06,
      "loss": 1.3432,
      "step": 5968
    },
    {
      "epoch": 2.0344239945466938,
      "grad_norm": 16.740425728177552,
      "learning_rate": 2.8396467667060535e-06,
      "loss": 1.2969,
      "step": 5969
    },
    {
      "epoch": 2.034764826175869,
      "grad_norm": 12.39401512970434,
      "learning_rate": 2.8378585178008444e-06,
      "loss": 1.2345,
      "step": 5970
    },
    {
      "epoch": 2.0351056578050444,
      "grad_norm": 16.3111573980388,
      "learning_rate": 2.836070609009101e-06,
      "loss": 1.3837,
      "step": 5971
    },
    {
      "epoch": 2.0354464894342197,
      "grad_norm": 18.229864512095045,
      "learning_rate": 2.8342830406120635e-06,
      "loss": 0.8862,
      "step": 5972
    },
    {
      "epoch": 2.0357873210633946,
      "grad_norm": 10.036677475721365,
      "learning_rate": 2.832495812890923e-06,
      "loss": 1.2354,
      "step": 5973
    },
    {
      "epoch": 2.03612815269257,
      "grad_norm": 13.136869592550674,
      "learning_rate": 2.8307089261268227e-06,
      "loss": 1.3118,
      "step": 5974
    },
    {
      "epoch": 2.036468984321745,
      "grad_norm": 18.686387038903497,
      "learning_rate": 2.8289223806008424e-06,
      "loss": 2.0248,
      "step": 5975
    },
    {
      "epoch": 2.03680981595092,
      "grad_norm": 13.215247090346043,
      "learning_rate": 2.827136176594014e-06,
      "loss": 1.2512,
      "step": 5976
    },
    {
      "epoch": 2.0371506475800953,
      "grad_norm": 14.079692825992746,
      "learning_rate": 2.825350314387315e-06,
      "loss": 1.0198,
      "step": 5977
    },
    {
      "epoch": 2.0374914792092707,
      "grad_norm": 14.87375738348336,
      "learning_rate": 2.823564794261668e-06,
      "loss": 1.0766,
      "step": 5978
    },
    {
      "epoch": 2.037832310838446,
      "grad_norm": 18.551690033997392,
      "learning_rate": 2.8217796164979427e-06,
      "loss": 1.2613,
      "step": 5979
    },
    {
      "epoch": 2.038173142467621,
      "grad_norm": 12.258252728223658,
      "learning_rate": 2.8199947813769548e-06,
      "loss": 1.1262,
      "step": 5980
    },
    {
      "epoch": 2.038513974096796,
      "grad_norm": 16.314821409772044,
      "learning_rate": 2.818210289179466e-06,
      "loss": 2.0165,
      "step": 5981
    },
    {
      "epoch": 2.0388548057259714,
      "grad_norm": 16.700988755700127,
      "learning_rate": 2.8164261401861836e-06,
      "loss": 1.4338,
      "step": 5982
    },
    {
      "epoch": 2.0391956373551468,
      "grad_norm": 15.256187063023779,
      "learning_rate": 2.8146423346777616e-06,
      "loss": 1.2842,
      "step": 5983
    },
    {
      "epoch": 2.0395364689843216,
      "grad_norm": 36.378421910513715,
      "learning_rate": 2.8128588729348005e-06,
      "loss": 1.5668,
      "step": 5984
    },
    {
      "epoch": 2.039877300613497,
      "grad_norm": 15.710338976545621,
      "learning_rate": 2.811075755237844e-06,
      "loss": 1.3214,
      "step": 5985
    },
    {
      "epoch": 2.0402181322426722,
      "grad_norm": 16.6442838158532,
      "learning_rate": 2.8092929818673865e-06,
      "loss": 2.0606,
      "step": 5986
    },
    {
      "epoch": 2.040558963871847,
      "grad_norm": 12.517315078577818,
      "learning_rate": 2.8075105531038643e-06,
      "loss": 1.3318,
      "step": 5987
    },
    {
      "epoch": 2.0408997955010224,
      "grad_norm": 23.570743962372884,
      "learning_rate": 2.8057284692276565e-06,
      "loss": 1.8938,
      "step": 5988
    },
    {
      "epoch": 2.0412406271301977,
      "grad_norm": 17.860622034729275,
      "learning_rate": 2.8039467305190972e-06,
      "loss": 1.0906,
      "step": 5989
    },
    {
      "epoch": 2.041581458759373,
      "grad_norm": 17.103647678434807,
      "learning_rate": 2.802165337258461e-06,
      "loss": 1.3997,
      "step": 5990
    },
    {
      "epoch": 2.041922290388548,
      "grad_norm": 19.44809455620341,
      "learning_rate": 2.8003842897259627e-06,
      "loss": 1.2619,
      "step": 5991
    },
    {
      "epoch": 2.042263122017723,
      "grad_norm": 10.890480835384933,
      "learning_rate": 2.798603588201775e-06,
      "loss": 1.5004,
      "step": 5992
    },
    {
      "epoch": 2.0426039536468985,
      "grad_norm": 14.487906431303864,
      "learning_rate": 2.796823232966005e-06,
      "loss": 1.1797,
      "step": 5993
    },
    {
      "epoch": 2.042944785276074,
      "grad_norm": 12.506358146725352,
      "learning_rate": 2.795043224298708e-06,
      "loss": 1.1044,
      "step": 5994
    },
    {
      "epoch": 2.0432856169052487,
      "grad_norm": 13.451975735746386,
      "learning_rate": 2.7932635624798943e-06,
      "loss": 1.6392,
      "step": 5995
    },
    {
      "epoch": 2.043626448534424,
      "grad_norm": 14.159872105660597,
      "learning_rate": 2.791484247789504e-06,
      "loss": 1.1043,
      "step": 5996
    },
    {
      "epoch": 2.0439672801635993,
      "grad_norm": 16.094392626913468,
      "learning_rate": 2.789705280507433e-06,
      "loss": 1.369,
      "step": 5997
    },
    {
      "epoch": 2.044308111792774,
      "grad_norm": 30.247163995405867,
      "learning_rate": 2.78792666091352e-06,
      "loss": 1.6155,
      "step": 5998
    },
    {
      "epoch": 2.0446489434219495,
      "grad_norm": 14.0390566112355,
      "learning_rate": 2.7861483892875495e-06,
      "loss": 1.6663,
      "step": 5999
    },
    {
      "epoch": 2.044989775051125,
      "grad_norm": 24.97327276724006,
      "learning_rate": 2.7843704659092525e-06,
      "loss": 1.6387,
      "step": 6000
    },
    {
      "epoch": 2.0453306066803,
      "grad_norm": 24.416186008371508,
      "learning_rate": 2.7825928910582966e-06,
      "loss": 1.7455,
      "step": 6001
    },
    {
      "epoch": 2.045671438309475,
      "grad_norm": 10.577153705337054,
      "learning_rate": 2.780815665014308e-06,
      "loss": 1.2173,
      "step": 6002
    },
    {
      "epoch": 2.0460122699386503,
      "grad_norm": 18.654629260731802,
      "learning_rate": 2.7790387880568517e-06,
      "loss": 1.5048,
      "step": 6003
    },
    {
      "epoch": 2.0463531015678256,
      "grad_norm": 15.156581903066163,
      "learning_rate": 2.77726226046543e-06,
      "loss": 1.7542,
      "step": 6004
    },
    {
      "epoch": 2.046693933197001,
      "grad_norm": 18.443219268211227,
      "learning_rate": 2.775486082519508e-06,
      "loss": 1.1971,
      "step": 6005
    },
    {
      "epoch": 2.0470347648261757,
      "grad_norm": 18.0147478288214,
      "learning_rate": 2.7737102544984784e-06,
      "loss": 1.3608,
      "step": 6006
    },
    {
      "epoch": 2.047375596455351,
      "grad_norm": 13.457844550865897,
      "learning_rate": 2.771934776681686e-06,
      "loss": 1.5057,
      "step": 6007
    },
    {
      "epoch": 2.0477164280845264,
      "grad_norm": 12.400843169819435,
      "learning_rate": 2.7701596493484264e-06,
      "loss": 1.4659,
      "step": 6008
    },
    {
      "epoch": 2.0480572597137012,
      "grad_norm": 19.25518660683634,
      "learning_rate": 2.7683848727779284e-06,
      "loss": 1.3134,
      "step": 6009
    },
    {
      "epoch": 2.0483980913428765,
      "grad_norm": 10.82375029106123,
      "learning_rate": 2.7666104472493714e-06,
      "loss": 0.9353,
      "step": 6010
    },
    {
      "epoch": 2.048738922972052,
      "grad_norm": 11.178447776794355,
      "learning_rate": 2.764836373041886e-06,
      "loss": 0.7616,
      "step": 6011
    },
    {
      "epoch": 2.049079754601227,
      "grad_norm": 15.693882628087273,
      "learning_rate": 2.7630626504345348e-06,
      "loss": 1.5932,
      "step": 6012
    },
    {
      "epoch": 2.049420586230402,
      "grad_norm": 12.485021824154034,
      "learning_rate": 2.7612892797063335e-06,
      "loss": 1.5816,
      "step": 6013
    },
    {
      "epoch": 2.0497614178595773,
      "grad_norm": 13.546241994834437,
      "learning_rate": 2.75951626113624e-06,
      "loss": 1.0455,
      "step": 6014
    },
    {
      "epoch": 2.0501022494887526,
      "grad_norm": 17.688580632893196,
      "learning_rate": 2.757743595003158e-06,
      "loss": 1.4917,
      "step": 6015
    },
    {
      "epoch": 2.050443081117928,
      "grad_norm": 13.344554766177442,
      "learning_rate": 2.755971281585934e-06,
      "loss": 1.2476,
      "step": 6016
    },
    {
      "epoch": 2.050783912747103,
      "grad_norm": 14.113661888596019,
      "learning_rate": 2.75419932116336e-06,
      "loss": 1.8507,
      "step": 6017
    },
    {
      "epoch": 2.051124744376278,
      "grad_norm": 17.021027433294623,
      "learning_rate": 2.752427714014173e-06,
      "loss": 0.9821,
      "step": 6018
    },
    {
      "epoch": 2.0514655760054534,
      "grad_norm": 12.953856401677186,
      "learning_rate": 2.750656460417054e-06,
      "loss": 1.4327,
      "step": 6019
    },
    {
      "epoch": 2.0518064076346283,
      "grad_norm": 13.560919437554038,
      "learning_rate": 2.7488855606506282e-06,
      "loss": 1.455,
      "step": 6020
    },
    {
      "epoch": 2.0521472392638036,
      "grad_norm": 14.799371585388036,
      "learning_rate": 2.7471150149934663e-06,
      "loss": 1.8687,
      "step": 6021
    },
    {
      "epoch": 2.052488070892979,
      "grad_norm": 11.344167565835725,
      "learning_rate": 2.7453448237240767e-06,
      "loss": 1.2249,
      "step": 6022
    },
    {
      "epoch": 2.052828902522154,
      "grad_norm": 25.00752750432449,
      "learning_rate": 2.743574987120924e-06,
      "loss": 0.8313,
      "step": 6023
    },
    {
      "epoch": 2.053169734151329,
      "grad_norm": 22.336977276250384,
      "learning_rate": 2.7418055054624094e-06,
      "loss": 1.6226,
      "step": 6024
    },
    {
      "epoch": 2.0535105657805044,
      "grad_norm": 12.99398817522244,
      "learning_rate": 2.7400363790268746e-06,
      "loss": 1.0414,
      "step": 6025
    },
    {
      "epoch": 2.0538513974096797,
      "grad_norm": 12.414071117350106,
      "learning_rate": 2.738267608092618e-06,
      "loss": 1.2922,
      "step": 6026
    },
    {
      "epoch": 2.054192229038855,
      "grad_norm": 19.809823975390476,
      "learning_rate": 2.7364991929378683e-06,
      "loss": 1.426,
      "step": 6027
    },
    {
      "epoch": 2.05453306066803,
      "grad_norm": 16.900231812489267,
      "learning_rate": 2.7347311338408056e-06,
      "loss": 1.3484,
      "step": 6028
    },
    {
      "epoch": 2.054873892297205,
      "grad_norm": 16.406799376966656,
      "learning_rate": 2.732963431079553e-06,
      "loss": 1.7921,
      "step": 6029
    },
    {
      "epoch": 2.0552147239263805,
      "grad_norm": 16.859931584052653,
      "learning_rate": 2.731196084932177e-06,
      "loss": 0.9953,
      "step": 6030
    },
    {
      "epoch": 2.0555555555555554,
      "grad_norm": 25.677850779193093,
      "learning_rate": 2.729429095676689e-06,
      "loss": 1.7094,
      "step": 6031
    },
    {
      "epoch": 2.0558963871847307,
      "grad_norm": 16.148017266392028,
      "learning_rate": 2.727662463591042e-06,
      "loss": 1.5772,
      "step": 6032
    },
    {
      "epoch": 2.056237218813906,
      "grad_norm": 14.751012811057285,
      "learning_rate": 2.725896188953136e-06,
      "loss": 1.7026,
      "step": 6033
    },
    {
      "epoch": 2.0565780504430813,
      "grad_norm": 28.78919166366769,
      "learning_rate": 2.7241302720408118e-06,
      "loss": 1.3724,
      "step": 6034
    },
    {
      "epoch": 2.056918882072256,
      "grad_norm": 15.802547797600617,
      "learning_rate": 2.722364713131856e-06,
      "loss": 1.9759,
      "step": 6035
    },
    {
      "epoch": 2.0572597137014315,
      "grad_norm": 11.466984607373803,
      "learning_rate": 2.720599512503997e-06,
      "loss": 1.3579,
      "step": 6036
    },
    {
      "epoch": 2.0576005453306068,
      "grad_norm": 14.21223671151292,
      "learning_rate": 2.7188346704349107e-06,
      "loss": 1.5336,
      "step": 6037
    },
    {
      "epoch": 2.057941376959782,
      "grad_norm": 13.292855534573198,
      "learning_rate": 2.7170701872022077e-06,
      "loss": 1.2358,
      "step": 6038
    },
    {
      "epoch": 2.058282208588957,
      "grad_norm": 20.041841894388508,
      "learning_rate": 2.7153060630834536e-06,
      "loss": 1.7245,
      "step": 6039
    },
    {
      "epoch": 2.0586230402181322,
      "grad_norm": 22.40448740227055,
      "learning_rate": 2.7135422983561526e-06,
      "loss": 1.5224,
      "step": 6040
    },
    {
      "epoch": 2.0589638718473076,
      "grad_norm": 12.039356046674603,
      "learning_rate": 2.711778893297746e-06,
      "loss": 1.2668,
      "step": 6041
    },
    {
      "epoch": 2.0593047034764824,
      "grad_norm": 11.956785772094253,
      "learning_rate": 2.710015848185633e-06,
      "loss": 1.6172,
      "step": 6042
    },
    {
      "epoch": 2.0596455351056577,
      "grad_norm": 18.209279678093335,
      "learning_rate": 2.7082531632971404e-06,
      "loss": 2.1429,
      "step": 6043
    },
    {
      "epoch": 2.059986366734833,
      "grad_norm": 15.899561881449745,
      "learning_rate": 2.706490838909547e-06,
      "loss": 1.6744,
      "step": 6044
    },
    {
      "epoch": 2.0603271983640083,
      "grad_norm": 17.72625694971961,
      "learning_rate": 2.704728875300077e-06,
      "loss": 1.3596,
      "step": 6045
    },
    {
      "epoch": 2.060668029993183,
      "grad_norm": 15.840347766720532,
      "learning_rate": 2.702967272745891e-06,
      "loss": 1.336,
      "step": 6046
    },
    {
      "epoch": 2.0610088616223585,
      "grad_norm": 17.273400544838758,
      "learning_rate": 2.701206031524096e-06,
      "loss": 0.9953,
      "step": 6047
    },
    {
      "epoch": 2.061349693251534,
      "grad_norm": 28.844992945399945,
      "learning_rate": 2.699445151911745e-06,
      "loss": 1.5225,
      "step": 6048
    },
    {
      "epoch": 2.061690524880709,
      "grad_norm": 24.2814212972102,
      "learning_rate": 2.6976846341858276e-06,
      "loss": 2.1037,
      "step": 6049
    },
    {
      "epoch": 2.062031356509884,
      "grad_norm": 18.645355438389213,
      "learning_rate": 2.695924478623283e-06,
      "loss": 1.6175,
      "step": 6050
    },
    {
      "epoch": 2.0623721881390593,
      "grad_norm": 11.683049665550653,
      "learning_rate": 2.6941646855009904e-06,
      "loss": 1.2154,
      "step": 6051
    },
    {
      "epoch": 2.0627130197682346,
      "grad_norm": 18.548292024839743,
      "learning_rate": 2.6924052550957714e-06,
      "loss": 1.6116,
      "step": 6052
    },
    {
      "epoch": 2.0630538513974095,
      "grad_norm": 16.74877859613324,
      "learning_rate": 2.6906461876843912e-06,
      "loss": 1.3394,
      "step": 6053
    },
    {
      "epoch": 2.063394683026585,
      "grad_norm": 14.535949513823127,
      "learning_rate": 2.6888874835435595e-06,
      "loss": 1.9623,
      "step": 6054
    },
    {
      "epoch": 2.06373551465576,
      "grad_norm": 12.37969535636808,
      "learning_rate": 2.6871291429499285e-06,
      "loss": 1.26,
      "step": 6055
    },
    {
      "epoch": 2.0640763462849354,
      "grad_norm": 15.788307552707177,
      "learning_rate": 2.6853711661800864e-06,
      "loss": 0.6226,
      "step": 6056
    },
    {
      "epoch": 2.0644171779141103,
      "grad_norm": 13.585270057528318,
      "learning_rate": 2.6836135535105757e-06,
      "loss": 1.4585,
      "step": 6057
    },
    {
      "epoch": 2.0647580095432856,
      "grad_norm": 22.546061525282823,
      "learning_rate": 2.681856305217876e-06,
      "loss": 1.2109,
      "step": 6058
    },
    {
      "epoch": 2.065098841172461,
      "grad_norm": 14.502815395974558,
      "learning_rate": 2.680099421578403e-06,
      "loss": 1.8548,
      "step": 6059
    },
    {
      "epoch": 2.065439672801636,
      "grad_norm": 20.275568642074553,
      "learning_rate": 2.6783429028685305e-06,
      "loss": 1.3786,
      "step": 6060
    },
    {
      "epoch": 2.065780504430811,
      "grad_norm": 56.41157065903689,
      "learning_rate": 2.676586749364559e-06,
      "loss": 1.2619,
      "step": 6061
    },
    {
      "epoch": 2.0661213360599864,
      "grad_norm": 14.766453962083215,
      "learning_rate": 2.6748309613427415e-06,
      "loss": 1.2012,
      "step": 6062
    },
    {
      "epoch": 2.0664621676891617,
      "grad_norm": 15.320608830606233,
      "learning_rate": 2.6730755390792693e-06,
      "loss": 1.8334,
      "step": 6063
    },
    {
      "epoch": 2.0668029993183366,
      "grad_norm": 19.485473023571114,
      "learning_rate": 2.671320482850278e-06,
      "loss": 1.2792,
      "step": 6064
    },
    {
      "epoch": 2.067143830947512,
      "grad_norm": 15.671138089881833,
      "learning_rate": 2.669565792931844e-06,
      "loss": 1.6214,
      "step": 6065
    },
    {
      "epoch": 2.067484662576687,
      "grad_norm": 10.77059076243292,
      "learning_rate": 2.6678114695999887e-06,
      "loss": 1.4354,
      "step": 6066
    },
    {
      "epoch": 2.0678254942058625,
      "grad_norm": 14.518618073019598,
      "learning_rate": 2.666057513130672e-06,
      "loss": 1.5965,
      "step": 6067
    },
    {
      "epoch": 2.0681663258350373,
      "grad_norm": 20.666993107282813,
      "learning_rate": 2.6643039237998003e-06,
      "loss": 2.0369,
      "step": 6068
    },
    {
      "epoch": 2.0685071574642127,
      "grad_norm": 12.800932947795168,
      "learning_rate": 2.6625507018832194e-06,
      "loss": 1.2945,
      "step": 6069
    },
    {
      "epoch": 2.068847989093388,
      "grad_norm": 17.017312271731605,
      "learning_rate": 2.6607978476567177e-06,
      "loss": 1.4761,
      "step": 6070
    },
    {
      "epoch": 2.0691888207225633,
      "grad_norm": 23.388578967087565,
      "learning_rate": 2.6590453613960277e-06,
      "loss": 1.9822,
      "step": 6071
    },
    {
      "epoch": 2.069529652351738,
      "grad_norm": 17.589177001248505,
      "learning_rate": 2.657293243376817e-06,
      "loss": 1.6955,
      "step": 6072
    },
    {
      "epoch": 2.0698704839809134,
      "grad_norm": 15.749406977705906,
      "learning_rate": 2.6555414938747084e-06,
      "loss": 1.5027,
      "step": 6073
    },
    {
      "epoch": 2.0702113156100888,
      "grad_norm": 17.47134661130064,
      "learning_rate": 2.6537901131652565e-06,
      "loss": 0.6588,
      "step": 6074
    },
    {
      "epoch": 2.0705521472392636,
      "grad_norm": 24.25734571791591,
      "learning_rate": 2.6520391015239556e-06,
      "loss": 1.6862,
      "step": 6075
    },
    {
      "epoch": 2.070892978868439,
      "grad_norm": 14.572952244637522,
      "learning_rate": 2.6502884592262544e-06,
      "loss": 1.6116,
      "step": 6076
    },
    {
      "epoch": 2.0712338104976142,
      "grad_norm": 14.083167645037026,
      "learning_rate": 2.6485381865475314e-06,
      "loss": 1.547,
      "step": 6077
    },
    {
      "epoch": 2.0715746421267895,
      "grad_norm": 19.117832908992483,
      "learning_rate": 2.646788283763109e-06,
      "loss": 1.4642,
      "step": 6078
    },
    {
      "epoch": 2.0719154737559644,
      "grad_norm": 19.033240760460487,
      "learning_rate": 2.645038751148262e-06,
      "loss": 1.2163,
      "step": 6079
    },
    {
      "epoch": 2.0722563053851397,
      "grad_norm": 13.465682340100447,
      "learning_rate": 2.643289588978192e-06,
      "loss": 0.9421,
      "step": 6080
    },
    {
      "epoch": 2.072597137014315,
      "grad_norm": 13.651312340434018,
      "learning_rate": 2.6415407975280506e-06,
      "loss": 1.3675,
      "step": 6081
    },
    {
      "epoch": 2.0729379686434903,
      "grad_norm": 10.917648731626931,
      "learning_rate": 2.6397923770729307e-06,
      "loss": 1.1248,
      "step": 6082
    },
    {
      "epoch": 2.073278800272665,
      "grad_norm": 32.789668369243,
      "learning_rate": 2.638044327887865e-06,
      "loss": 1.2036,
      "step": 6083
    },
    {
      "epoch": 2.0736196319018405,
      "grad_norm": 13.592693386183711,
      "learning_rate": 2.6362966502478287e-06,
      "loss": 1.3309,
      "step": 6084
    },
    {
      "epoch": 2.073960463531016,
      "grad_norm": 14.679608985673678,
      "learning_rate": 2.634549344427739e-06,
      "loss": 1.1382,
      "step": 6085
    },
    {
      "epoch": 2.0743012951601907,
      "grad_norm": 13.817074769864439,
      "learning_rate": 2.6328024107024542e-06,
      "loss": 1.6737,
      "step": 6086
    },
    {
      "epoch": 2.074642126789366,
      "grad_norm": 10.14321985410071,
      "learning_rate": 2.631055849346773e-06,
      "loss": 1.0175,
      "step": 6087
    },
    {
      "epoch": 2.0749829584185413,
      "grad_norm": 13.2557421548651,
      "learning_rate": 2.6293096606354363e-06,
      "loss": 1.3459,
      "step": 6088
    },
    {
      "epoch": 2.0753237900477166,
      "grad_norm": 14.582240981110093,
      "learning_rate": 2.62756384484313e-06,
      "loss": 1.4841,
      "step": 6089
    },
    {
      "epoch": 2.0756646216768915,
      "grad_norm": 15.478681318798388,
      "learning_rate": 2.625818402244471e-06,
      "loss": 1.2363,
      "step": 6090
    },
    {
      "epoch": 2.076005453306067,
      "grad_norm": 15.431197323057281,
      "learning_rate": 2.62407333311403e-06,
      "loss": 1.8195,
      "step": 6091
    },
    {
      "epoch": 2.076346284935242,
      "grad_norm": 14.152281696680708,
      "learning_rate": 2.622328637726314e-06,
      "loss": 1.466,
      "step": 6092
    },
    {
      "epoch": 2.0766871165644174,
      "grad_norm": 12.80147956868569,
      "learning_rate": 2.6205843163557663e-06,
      "loss": 1.1673,
      "step": 6093
    },
    {
      "epoch": 2.0770279481935923,
      "grad_norm": 10.96965224821824,
      "learning_rate": 2.6188403692767774e-06,
      "loss": 1.3429,
      "step": 6094
    },
    {
      "epoch": 2.0773687798227676,
      "grad_norm": 19.851841825610535,
      "learning_rate": 2.6170967967636774e-06,
      "loss": 1.2368,
      "step": 6095
    },
    {
      "epoch": 2.077709611451943,
      "grad_norm": 13.563789026178334,
      "learning_rate": 2.615353599090737e-06,
      "loss": 1.2335,
      "step": 6096
    },
    {
      "epoch": 2.0780504430811177,
      "grad_norm": 15.114462971027761,
      "learning_rate": 2.6136107765321683e-06,
      "loss": 0.9032,
      "step": 6097
    },
    {
      "epoch": 2.078391274710293,
      "grad_norm": 17.601266954428198,
      "learning_rate": 2.6118683293621237e-06,
      "loss": 1.4512,
      "step": 6098
    },
    {
      "epoch": 2.0787321063394684,
      "grad_norm": 14.304083719288617,
      "learning_rate": 2.610126257854698e-06,
      "loss": 1.7248,
      "step": 6099
    },
    {
      "epoch": 2.0790729379686437,
      "grad_norm": 16.826779020232355,
      "learning_rate": 2.608384562283925e-06,
      "loss": 1.3466,
      "step": 6100
    },
    {
      "epoch": 2.0794137695978185,
      "grad_norm": 15.05610091767622,
      "learning_rate": 2.606643242923781e-06,
      "loss": 1.5338,
      "step": 6101
    },
    {
      "epoch": 2.079754601226994,
      "grad_norm": 16.033104959977727,
      "learning_rate": 2.6049023000481823e-06,
      "loss": 1.6509,
      "step": 6102
    },
    {
      "epoch": 2.080095432856169,
      "grad_norm": 15.394587954610563,
      "learning_rate": 2.603161733930986e-06,
      "loss": 1.3887,
      "step": 6103
    },
    {
      "epoch": 2.0804362644853445,
      "grad_norm": 15.938537277164354,
      "learning_rate": 2.6014215448459907e-06,
      "loss": 1.7401,
      "step": 6104
    },
    {
      "epoch": 2.0807770961145193,
      "grad_norm": 14.552092212955108,
      "learning_rate": 2.5996817330669368e-06,
      "loss": 1.2871,
      "step": 6105
    },
    {
      "epoch": 2.0811179277436946,
      "grad_norm": 16.65300552767382,
      "learning_rate": 2.5979422988674965e-06,
      "loss": 1.4677,
      "step": 6106
    },
    {
      "epoch": 2.08145875937287,
      "grad_norm": 15.397524332764828,
      "learning_rate": 2.596203242521297e-06,
      "loss": 1.7931,
      "step": 6107
    },
    {
      "epoch": 2.081799591002045,
      "grad_norm": 18.038544111745388,
      "learning_rate": 2.5944645643018983e-06,
      "loss": 1.3578,
      "step": 6108
    },
    {
      "epoch": 2.08214042263122,
      "grad_norm": 27.227305110304606,
      "learning_rate": 2.5927262644827954e-06,
      "loss": 1.3397,
      "step": 6109
    },
    {
      "epoch": 2.0824812542603954,
      "grad_norm": 12.421361058109603,
      "learning_rate": 2.5909883433374385e-06,
      "loss": 1.2993,
      "step": 6110
    },
    {
      "epoch": 2.0828220858895707,
      "grad_norm": 13.945254123826569,
      "learning_rate": 2.5892508011392027e-06,
      "loss": 0.8619,
      "step": 6111
    },
    {
      "epoch": 2.0831629175187456,
      "grad_norm": 13.792160383919931,
      "learning_rate": 2.5875136381614097e-06,
      "loss": 1.6408,
      "step": 6112
    },
    {
      "epoch": 2.083503749147921,
      "grad_norm": 16.425379022122037,
      "learning_rate": 2.58577685467733e-06,
      "loss": 1.7851,
      "step": 6113
    },
    {
      "epoch": 2.0838445807770962,
      "grad_norm": 18.70950358103788,
      "learning_rate": 2.584040450960159e-06,
      "loss": 1.6459,
      "step": 6114
    },
    {
      "epoch": 2.084185412406271,
      "grad_norm": 25.937678828904225,
      "learning_rate": 2.5823044272830427e-06,
      "loss": 1.2769,
      "step": 6115
    },
    {
      "epoch": 2.0845262440354464,
      "grad_norm": 12.226877423438392,
      "learning_rate": 2.580568783919064e-06,
      "loss": 1.0606,
      "step": 6116
    },
    {
      "epoch": 2.0848670756646217,
      "grad_norm": 16.80550334720805,
      "learning_rate": 2.5788335211412475e-06,
      "loss": 1.3376,
      "step": 6117
    },
    {
      "epoch": 2.085207907293797,
      "grad_norm": 18.628170207955108,
      "learning_rate": 2.577098639222555e-06,
      "loss": 1.5323,
      "step": 6118
    },
    {
      "epoch": 2.085548738922972,
      "grad_norm": 15.828134393008575,
      "learning_rate": 2.575364138435893e-06,
      "loss": 1.6429,
      "step": 6119
    },
    {
      "epoch": 2.085889570552147,
      "grad_norm": 13.190746591572411,
      "learning_rate": 2.573630019054103e-06,
      "loss": 1.4296,
      "step": 6120
    },
    {
      "epoch": 2.0862304021813225,
      "grad_norm": 102.84232613873372,
      "learning_rate": 2.5718962813499697e-06,
      "loss": 1.6036,
      "step": 6121
    },
    {
      "epoch": 2.086571233810498,
      "grad_norm": 13.819700736769416,
      "learning_rate": 2.570162925596218e-06,
      "loss": 1.3789,
      "step": 6122
    },
    {
      "epoch": 2.0869120654396727,
      "grad_norm": 11.736756410628727,
      "learning_rate": 2.56842995206551e-06,
      "loss": 1.1257,
      "step": 6123
    },
    {
      "epoch": 2.087252897068848,
      "grad_norm": 19.180986913230555,
      "learning_rate": 2.5666973610304504e-06,
      "loss": 0.8372,
      "step": 6124
    },
    {
      "epoch": 2.0875937286980233,
      "grad_norm": 23.21173431691631,
      "learning_rate": 2.564965152763582e-06,
      "loss": 1.5141,
      "step": 6125
    },
    {
      "epoch": 2.087934560327198,
      "grad_norm": 14.984080398995618,
      "learning_rate": 2.56323332753739e-06,
      "loss": 1.5522,
      "step": 6126
    },
    {
      "epoch": 2.0882753919563735,
      "grad_norm": 11.564028323400402,
      "learning_rate": 2.5615018856242945e-06,
      "loss": 1.0661,
      "step": 6127
    },
    {
      "epoch": 2.0886162235855488,
      "grad_norm": 12.061514091321545,
      "learning_rate": 2.559770827296657e-06,
      "loss": 1.2072,
      "step": 6128
    },
    {
      "epoch": 2.088957055214724,
      "grad_norm": 15.393214033923492,
      "learning_rate": 2.5580401528267866e-06,
      "loss": 1.3872,
      "step": 6129
    },
    {
      "epoch": 2.089297886843899,
      "grad_norm": 21.11444477445145,
      "learning_rate": 2.5563098624869194e-06,
      "loss": 1.2574,
      "step": 6130
    },
    {
      "epoch": 2.0896387184730743,
      "grad_norm": 14.52942521148876,
      "learning_rate": 2.554579956549238e-06,
      "loss": 0.8967,
      "step": 6131
    },
    {
      "epoch": 2.0899795501022496,
      "grad_norm": 17.2646699837619,
      "learning_rate": 2.552850435285864e-06,
      "loss": 1.5475,
      "step": 6132
    },
    {
      "epoch": 2.090320381731425,
      "grad_norm": 16.95524370321703,
      "learning_rate": 2.5511212989688587e-06,
      "loss": 1.9298,
      "step": 6133
    },
    {
      "epoch": 2.0906612133605997,
      "grad_norm": 16.692066294098144,
      "learning_rate": 2.5493925478702216e-06,
      "loss": 1.3164,
      "step": 6134
    },
    {
      "epoch": 2.091002044989775,
      "grad_norm": 18.18426543837245,
      "learning_rate": 2.5476641822618912e-06,
      "loss": 1.6635,
      "step": 6135
    },
    {
      "epoch": 2.0913428766189504,
      "grad_norm": 11.194701611156594,
      "learning_rate": 2.5459362024157474e-06,
      "loss": 1.0975,
      "step": 6136
    },
    {
      "epoch": 2.091683708248125,
      "grad_norm": 15.250106149854359,
      "learning_rate": 2.544208608603609e-06,
      "loss": 1.5212,
      "step": 6137
    },
    {
      "epoch": 2.0920245398773005,
      "grad_norm": 12.93481805034485,
      "learning_rate": 2.542481401097231e-06,
      "loss": 1.4306,
      "step": 6138
    },
    {
      "epoch": 2.092365371506476,
      "grad_norm": 18.168586614664438,
      "learning_rate": 2.540754580168314e-06,
      "loss": 1.7038,
      "step": 6139
    },
    {
      "epoch": 2.092706203135651,
      "grad_norm": 33.07945385121587,
      "learning_rate": 2.5390281460884874e-06,
      "loss": 1.1514,
      "step": 6140
    },
    {
      "epoch": 2.093047034764826,
      "grad_norm": 14.269471215201243,
      "learning_rate": 2.5373020991293317e-06,
      "loss": 1.3484,
      "step": 6141
    },
    {
      "epoch": 2.0933878663940013,
      "grad_norm": 16.395249593298892,
      "learning_rate": 2.535576439562362e-06,
      "loss": 1.9696,
      "step": 6142
    },
    {
      "epoch": 2.0937286980231766,
      "grad_norm": 11.696485683479517,
      "learning_rate": 2.533851167659024e-06,
      "loss": 1.577,
      "step": 6143
    },
    {
      "epoch": 2.094069529652352,
      "grad_norm": 16.25595397449943,
      "learning_rate": 2.5321262836907177e-06,
      "loss": 1.5297,
      "step": 6144
    },
    {
      "epoch": 2.094410361281527,
      "grad_norm": 15.14408564607669,
      "learning_rate": 2.53040178792877e-06,
      "loss": 1.5605,
      "step": 6145
    },
    {
      "epoch": 2.094751192910702,
      "grad_norm": 21.54468077867544,
      "learning_rate": 2.52867768064445e-06,
      "loss": 1.4567,
      "step": 6146
    },
    {
      "epoch": 2.0950920245398774,
      "grad_norm": 13.162594712674546,
      "learning_rate": 2.5269539621089722e-06,
      "loss": 1.8168,
      "step": 6147
    },
    {
      "epoch": 2.0954328561690523,
      "grad_norm": 14.926812246013798,
      "learning_rate": 2.525230632593479e-06,
      "loss": 1.3898,
      "step": 6148
    },
    {
      "epoch": 2.0957736877982276,
      "grad_norm": 17.962923445232153,
      "learning_rate": 2.5235076923690586e-06,
      "loss": 2.0,
      "step": 6149
    },
    {
      "epoch": 2.096114519427403,
      "grad_norm": 12.51456168018112,
      "learning_rate": 2.5217851417067367e-06,
      "loss": 1.7192,
      "step": 6150
    },
    {
      "epoch": 2.096455351056578,
      "grad_norm": 15.538872697580905,
      "learning_rate": 2.520062980877477e-06,
      "loss": 1.5963,
      "step": 6151
    },
    {
      "epoch": 2.096796182685753,
      "grad_norm": 14.662384162596243,
      "learning_rate": 2.5183412101521833e-06,
      "loss": 1.4479,
      "step": 6152
    },
    {
      "epoch": 2.0971370143149284,
      "grad_norm": 20.562380753543863,
      "learning_rate": 2.516619829801695e-06,
      "loss": 1.4838,
      "step": 6153
    },
    {
      "epoch": 2.0974778459441037,
      "grad_norm": 13.266682426061363,
      "learning_rate": 2.5148988400967946e-06,
      "loss": 0.8552,
      "step": 6154
    },
    {
      "epoch": 2.097818677573279,
      "grad_norm": 22.366027180919325,
      "learning_rate": 2.5131782413081977e-06,
      "loss": 1.9986,
      "step": 6155
    },
    {
      "epoch": 2.098159509202454,
      "grad_norm": 12.396119150253158,
      "learning_rate": 2.511458033706564e-06,
      "loss": 1.5425,
      "step": 6156
    },
    {
      "epoch": 2.098500340831629,
      "grad_norm": 16.81921093321046,
      "learning_rate": 2.5097382175624874e-06,
      "loss": 1.7593,
      "step": 6157
    },
    {
      "epoch": 2.0988411724608045,
      "grad_norm": 16.619739109704298,
      "learning_rate": 2.5080187931465044e-06,
      "loss": 1.5801,
      "step": 6158
    },
    {
      "epoch": 2.0991820040899793,
      "grad_norm": 18.402601181166105,
      "learning_rate": 2.506299760729081e-06,
      "loss": 1.8661,
      "step": 6159
    },
    {
      "epoch": 2.0995228357191547,
      "grad_norm": 16.024685125915806,
      "learning_rate": 2.504581120580637e-06,
      "loss": 1.7912,
      "step": 6160
    },
    {
      "epoch": 2.09986366734833,
      "grad_norm": 14.484908768766035,
      "learning_rate": 2.502862872971513e-06,
      "loss": 1.8996,
      "step": 6161
    },
    {
      "epoch": 2.1002044989775053,
      "grad_norm": 17.181903053383806,
      "learning_rate": 2.5011450181719986e-06,
      "loss": 1.4511,
      "step": 6162
    },
    {
      "epoch": 2.10054533060668,
      "grad_norm": 23.966972993226904,
      "learning_rate": 2.4994275564523247e-06,
      "loss": 1.5346,
      "step": 6163
    },
    {
      "epoch": 2.1008861622358554,
      "grad_norm": 16.37616271068201,
      "learning_rate": 2.4977104880826473e-06,
      "loss": 1.1504,
      "step": 6164
    },
    {
      "epoch": 2.1012269938650308,
      "grad_norm": 15.562064829282981,
      "learning_rate": 2.495993813333072e-06,
      "loss": 1.4629,
      "step": 6165
    },
    {
      "epoch": 2.101567825494206,
      "grad_norm": 14.720441695349686,
      "learning_rate": 2.494277532473637e-06,
      "loss": 0.7938,
      "step": 6166
    },
    {
      "epoch": 2.101908657123381,
      "grad_norm": 16.000533015867244,
      "learning_rate": 2.4925616457743213e-06,
      "loss": 1.6223,
      "step": 6167
    },
    {
      "epoch": 2.1022494887525562,
      "grad_norm": 11.686294472381654,
      "learning_rate": 2.49084615350504e-06,
      "loss": 1.5167,
      "step": 6168
    },
    {
      "epoch": 2.1025903203817315,
      "grad_norm": 14.724419173929975,
      "learning_rate": 2.4891310559356475e-06,
      "loss": 1.7129,
      "step": 6169
    },
    {
      "epoch": 2.1029311520109064,
      "grad_norm": 18.170503164137894,
      "learning_rate": 2.4874163533359346e-06,
      "loss": 2.2063,
      "step": 6170
    },
    {
      "epoch": 2.1032719836400817,
      "grad_norm": 13.07394594438088,
      "learning_rate": 2.4857020459756315e-06,
      "loss": 1.59,
      "step": 6171
    },
    {
      "epoch": 2.103612815269257,
      "grad_norm": 15.257974337679112,
      "learning_rate": 2.4839881341244052e-06,
      "loss": 1.4643,
      "step": 6172
    },
    {
      "epoch": 2.1039536468984323,
      "grad_norm": 19.077513556372793,
      "learning_rate": 2.482274618051864e-06,
      "loss": 1.8137,
      "step": 6173
    },
    {
      "epoch": 2.104294478527607,
      "grad_norm": 12.788193917193889,
      "learning_rate": 2.480561498027543e-06,
      "loss": 1.4631,
      "step": 6174
    },
    {
      "epoch": 2.1046353101567825,
      "grad_norm": 10.515969771645617,
      "learning_rate": 2.4788487743209298e-06,
      "loss": 1.2403,
      "step": 6175
    },
    {
      "epoch": 2.104976141785958,
      "grad_norm": 17.287429260340407,
      "learning_rate": 2.4771364472014424e-06,
      "loss": 1.3456,
      "step": 6176
    },
    {
      "epoch": 2.105316973415133,
      "grad_norm": 10.390262782763132,
      "learning_rate": 2.4754245169384312e-06,
      "loss": 1.1472,
      "step": 6177
    },
    {
      "epoch": 2.105657805044308,
      "grad_norm": 10.512995494317291,
      "learning_rate": 2.473712983801197e-06,
      "loss": 1.0583,
      "step": 6178
    },
    {
      "epoch": 2.1059986366734833,
      "grad_norm": 22.069748460228656,
      "learning_rate": 2.4720018480589647e-06,
      "loss": 1.4847,
      "step": 6179
    },
    {
      "epoch": 2.1063394683026586,
      "grad_norm": 16.557476666612466,
      "learning_rate": 2.4702911099809037e-06,
      "loss": 1.6198,
      "step": 6180
    },
    {
      "epoch": 2.1066802999318335,
      "grad_norm": 12.089123404391028,
      "learning_rate": 2.468580769836125e-06,
      "loss": 1.5066,
      "step": 6181
    },
    {
      "epoch": 2.107021131561009,
      "grad_norm": 17.398349067031955,
      "learning_rate": 2.466870827893666e-06,
      "loss": 1.3319,
      "step": 6182
    },
    {
      "epoch": 2.107361963190184,
      "grad_norm": 18.86077421489594,
      "learning_rate": 2.4651612844225096e-06,
      "loss": 1.3355,
      "step": 6183
    },
    {
      "epoch": 2.1077027948193594,
      "grad_norm": 16.583216478293142,
      "learning_rate": 2.4634521396915744e-06,
      "loss": 1.6671,
      "step": 6184
    },
    {
      "epoch": 2.1080436264485343,
      "grad_norm": 16.07739769556006,
      "learning_rate": 2.4617433939697145e-06,
      "loss": 1.622,
      "step": 6185
    },
    {
      "epoch": 2.1083844580777096,
      "grad_norm": 18.322705651788223,
      "learning_rate": 2.4600350475257228e-06,
      "loss": 1.454,
      "step": 6186
    },
    {
      "epoch": 2.108725289706885,
      "grad_norm": 18.998029407520956,
      "learning_rate": 2.4583271006283293e-06,
      "loss": 1.8137,
      "step": 6187
    },
    {
      "epoch": 2.10906612133606,
      "grad_norm": 19.264387821616577,
      "learning_rate": 2.456619553546201e-06,
      "loss": 1.3773,
      "step": 6188
    },
    {
      "epoch": 2.109406952965235,
      "grad_norm": 12.891160070422545,
      "learning_rate": 2.4549124065479413e-06,
      "loss": 1.4889,
      "step": 6189
    },
    {
      "epoch": 2.1097477845944104,
      "grad_norm": 15.677101250559057,
      "learning_rate": 2.4532056599020913e-06,
      "loss": 1.0845,
      "step": 6190
    },
    {
      "epoch": 2.1100886162235857,
      "grad_norm": 16.928932569219533,
      "learning_rate": 2.45149931387713e-06,
      "loss": 1.3542,
      "step": 6191
    },
    {
      "epoch": 2.1104294478527605,
      "grad_norm": 20.07106231808636,
      "learning_rate": 2.4497933687414726e-06,
      "loss": 1.0626,
      "step": 6192
    },
    {
      "epoch": 2.110770279481936,
      "grad_norm": 10.159122480182205,
      "learning_rate": 2.4480878247634666e-06,
      "loss": 1.1576,
      "step": 6193
    },
    {
      "epoch": 2.111111111111111,
      "grad_norm": 17.750598568470892,
      "learning_rate": 2.446382682211408e-06,
      "loss": 1.3528,
      "step": 6194
    },
    {
      "epoch": 2.1114519427402865,
      "grad_norm": 12.254837350061026,
      "learning_rate": 2.4446779413535172e-06,
      "loss": 1.109,
      "step": 6195
    },
    {
      "epoch": 2.1117927743694613,
      "grad_norm": 15.344193710365971,
      "learning_rate": 2.442973602457956e-06,
      "loss": 1.6399,
      "step": 6196
    },
    {
      "epoch": 2.1121336059986366,
      "grad_norm": 15.731890695320631,
      "learning_rate": 2.4412696657928302e-06,
      "loss": 1.4775,
      "step": 6197
    },
    {
      "epoch": 2.112474437627812,
      "grad_norm": 13.690758351425206,
      "learning_rate": 2.43956613162617e-06,
      "loss": 0.8961,
      "step": 6198
    },
    {
      "epoch": 2.1128152692569873,
      "grad_norm": 43.80055795937283,
      "learning_rate": 2.4378630002259495e-06,
      "loss": 1.3721,
      "step": 6199
    },
    {
      "epoch": 2.113156100886162,
      "grad_norm": 27.556066548751897,
      "learning_rate": 2.4361602718600786e-06,
      "loss": 1.5982,
      "step": 6200
    },
    {
      "epoch": 2.1134969325153374,
      "grad_norm": 14.588278483913742,
      "learning_rate": 2.4344579467964023e-06,
      "loss": 1.3398,
      "step": 6201
    },
    {
      "epoch": 2.1138377641445127,
      "grad_norm": 13.978777830892287,
      "learning_rate": 2.4327560253027044e-06,
      "loss": 1.5702,
      "step": 6202
    },
    {
      "epoch": 2.1141785957736876,
      "grad_norm": 16.26471225776593,
      "learning_rate": 2.4310545076467035e-06,
      "loss": 1.4277,
      "step": 6203
    },
    {
      "epoch": 2.114519427402863,
      "grad_norm": 13.284290223879385,
      "learning_rate": 2.4293533940960545e-06,
      "loss": 1.1142,
      "step": 6204
    },
    {
      "epoch": 2.1148602590320382,
      "grad_norm": 12.88971980470685,
      "learning_rate": 2.42765268491835e-06,
      "loss": 1.9101,
      "step": 6205
    },
    {
      "epoch": 2.1152010906612135,
      "grad_norm": 16.949406234675738,
      "learning_rate": 2.425952380381118e-06,
      "loss": 1.4044,
      "step": 6206
    },
    {
      "epoch": 2.1155419222903884,
      "grad_norm": 17.68506171012958,
      "learning_rate": 2.4242524807518254e-06,
      "loss": 1.9936,
      "step": 6207
    },
    {
      "epoch": 2.1158827539195637,
      "grad_norm": 13.088714802518906,
      "learning_rate": 2.422552986297867e-06,
      "loss": 0.9028,
      "step": 6208
    },
    {
      "epoch": 2.116223585548739,
      "grad_norm": 14.857121157685297,
      "learning_rate": 2.420853897286586e-06,
      "loss": 1.0284,
      "step": 6209
    },
    {
      "epoch": 2.116564417177914,
      "grad_norm": 13.993194576467637,
      "learning_rate": 2.4191552139852553e-06,
      "loss": 1.6916,
      "step": 6210
    },
    {
      "epoch": 2.116905248807089,
      "grad_norm": 15.1254759990384,
      "learning_rate": 2.41745693666108e-06,
      "loss": 1.2018,
      "step": 6211
    },
    {
      "epoch": 2.1172460804362645,
      "grad_norm": 19.305121783747566,
      "learning_rate": 2.415759065581213e-06,
      "loss": 1.2828,
      "step": 6212
    },
    {
      "epoch": 2.11758691206544,
      "grad_norm": 16.72236711560976,
      "learning_rate": 2.41406160101273e-06,
      "loss": 2.0018,
      "step": 6213
    },
    {
      "epoch": 2.1179277436946147,
      "grad_norm": 12.793291147294214,
      "learning_rate": 2.4123645432226496e-06,
      "loss": 1.2676,
      "step": 6214
    },
    {
      "epoch": 2.11826857532379,
      "grad_norm": 11.126144617602678,
      "learning_rate": 2.4106678924779308e-06,
      "loss": 1.1408,
      "step": 6215
    },
    {
      "epoch": 2.1186094069529653,
      "grad_norm": 12.635092231867176,
      "learning_rate": 2.4089716490454583e-06,
      "loss": 1.3615,
      "step": 6216
    },
    {
      "epoch": 2.1189502385821406,
      "grad_norm": 13.257218623140941,
      "learning_rate": 2.4072758131920576e-06,
      "loss": 1.4355,
      "step": 6217
    },
    {
      "epoch": 2.1192910702113155,
      "grad_norm": 29.672470108662683,
      "learning_rate": 2.4055803851844967e-06,
      "loss": 1.3088,
      "step": 6218
    },
    {
      "epoch": 2.1196319018404908,
      "grad_norm": 12.286279397293868,
      "learning_rate": 2.4038853652894673e-06,
      "loss": 1.5109,
      "step": 6219
    },
    {
      "epoch": 2.119972733469666,
      "grad_norm": 11.652715837303564,
      "learning_rate": 2.402190753773604e-06,
      "loss": 1.271,
      "step": 6220
    },
    {
      "epoch": 2.120313565098841,
      "grad_norm": 16.916275278310973,
      "learning_rate": 2.400496550903476e-06,
      "loss": 1.3403,
      "step": 6221
    },
    {
      "epoch": 2.1206543967280163,
      "grad_norm": 13.10917799421779,
      "learning_rate": 2.3988027569455895e-06,
      "loss": 1.4327,
      "step": 6222
    },
    {
      "epoch": 2.1209952283571916,
      "grad_norm": 10.65546535078804,
      "learning_rate": 2.3971093721663852e-06,
      "loss": 1.127,
      "step": 6223
    },
    {
      "epoch": 2.121336059986367,
      "grad_norm": 12.51169225601301,
      "learning_rate": 2.3954163968322346e-06,
      "loss": 1.4258,
      "step": 6224
    },
    {
      "epoch": 2.1216768916155417,
      "grad_norm": 29.685097632061186,
      "learning_rate": 2.3937238312094554e-06,
      "loss": 2.1042,
      "step": 6225
    },
    {
      "epoch": 2.122017723244717,
      "grad_norm": 17.567812470614864,
      "learning_rate": 2.392031675564294e-06,
      "loss": 1.6751,
      "step": 6226
    },
    {
      "epoch": 2.1223585548738924,
      "grad_norm": 10.974633101430381,
      "learning_rate": 2.390339930162928e-06,
      "loss": 1.1781,
      "step": 6227
    },
    {
      "epoch": 2.1226993865030677,
      "grad_norm": 19.98576680277969,
      "learning_rate": 2.3886485952714835e-06,
      "loss": 1.8475,
      "step": 6228
    },
    {
      "epoch": 2.1230402181322425,
      "grad_norm": 16.97256159937134,
      "learning_rate": 2.386957671156008e-06,
      "loss": 1.7491,
      "step": 6229
    },
    {
      "epoch": 2.123381049761418,
      "grad_norm": 12.771383893763836,
      "learning_rate": 2.3852671580824908e-06,
      "loss": 1.2656,
      "step": 6230
    },
    {
      "epoch": 2.123721881390593,
      "grad_norm": 13.855489950224936,
      "learning_rate": 2.3835770563168625e-06,
      "loss": 1.5997,
      "step": 6231
    },
    {
      "epoch": 2.124062713019768,
      "grad_norm": 16.841554799089216,
      "learning_rate": 2.3818873661249765e-06,
      "loss": 1.7535,
      "step": 6232
    },
    {
      "epoch": 2.1244035446489433,
      "grad_norm": 20.213317921092244,
      "learning_rate": 2.380198087772629e-06,
      "loss": 1.3475,
      "step": 6233
    },
    {
      "epoch": 2.1247443762781186,
      "grad_norm": 12.420917076683667,
      "learning_rate": 2.3785092215255522e-06,
      "loss": 1.3549,
      "step": 6234
    },
    {
      "epoch": 2.125085207907294,
      "grad_norm": 18.469946030124525,
      "learning_rate": 2.3768207676494103e-06,
      "loss": 0.9973,
      "step": 6235
    },
    {
      "epoch": 2.125426039536469,
      "grad_norm": 13.730821541766405,
      "learning_rate": 2.375132726409804e-06,
      "loss": 1.2372,
      "step": 6236
    },
    {
      "epoch": 2.125766871165644,
      "grad_norm": 15.79464174477173,
      "learning_rate": 2.3734450980722683e-06,
      "loss": 0.9658,
      "step": 6237
    },
    {
      "epoch": 2.1261077027948194,
      "grad_norm": 12.068761785746005,
      "learning_rate": 2.3717578829022745e-06,
      "loss": 1.564,
      "step": 6238
    },
    {
      "epoch": 2.1264485344239947,
      "grad_norm": 23.589222344719598,
      "learning_rate": 2.370071081165228e-06,
      "loss": 1.5356,
      "step": 6239
    },
    {
      "epoch": 2.1267893660531696,
      "grad_norm": 13.054733303692117,
      "learning_rate": 2.3683846931264704e-06,
      "loss": 1.5985,
      "step": 6240
    },
    {
      "epoch": 2.127130197682345,
      "grad_norm": 14.190838531019866,
      "learning_rate": 2.366698719051276e-06,
      "loss": 1.2889,
      "step": 6241
    },
    {
      "epoch": 2.12747102931152,
      "grad_norm": 12.482070505129741,
      "learning_rate": 2.365013159204855e-06,
      "loss": 1.4937,
      "step": 6242
    },
    {
      "epoch": 2.127811860940695,
      "grad_norm": 12.995386860002094,
      "learning_rate": 2.3633280138523546e-06,
      "loss": 1.5395,
      "step": 6243
    },
    {
      "epoch": 2.1281526925698704,
      "grad_norm": 19.66703662457556,
      "learning_rate": 2.361643283258855e-06,
      "loss": 1.8207,
      "step": 6244
    },
    {
      "epoch": 2.1284935241990457,
      "grad_norm": 14.462138820477406,
      "learning_rate": 2.359958967689365e-06,
      "loss": 1.3038,
      "step": 6245
    },
    {
      "epoch": 2.128834355828221,
      "grad_norm": 11.673307518889073,
      "learning_rate": 2.358275067408842e-06,
      "loss": 1.2909,
      "step": 6246
    },
    {
      "epoch": 2.129175187457396,
      "grad_norm": 15.599454766297297,
      "learning_rate": 2.356591582682168e-06,
      "loss": 1.696,
      "step": 6247
    },
    {
      "epoch": 2.129516019086571,
      "grad_norm": 13.802286377834122,
      "learning_rate": 2.354908513774157e-06,
      "loss": 1.4262,
      "step": 6248
    },
    {
      "epoch": 2.1298568507157465,
      "grad_norm": 15.06588519217504,
      "learning_rate": 2.3532258609495705e-06,
      "loss": 1.3545,
      "step": 6249
    },
    {
      "epoch": 2.130197682344922,
      "grad_norm": 13.90662270364312,
      "learning_rate": 2.35154362447309e-06,
      "loss": 1.7569,
      "step": 6250
    },
    {
      "epoch": 2.1305385139740967,
      "grad_norm": 15.898575723083995,
      "learning_rate": 2.3498618046093386e-06,
      "loss": 1.4392,
      "step": 6251
    },
    {
      "epoch": 2.130879345603272,
      "grad_norm": 9.928039767891752,
      "learning_rate": 2.3481804016228786e-06,
      "loss": 1.583,
      "step": 6252
    },
    {
      "epoch": 2.1312201772324473,
      "grad_norm": 14.562711113301557,
      "learning_rate": 2.3464994157781963e-06,
      "loss": 1.6195,
      "step": 6253
    },
    {
      "epoch": 2.131561008861622,
      "grad_norm": 14.767750533405469,
      "learning_rate": 2.344818847339718e-06,
      "loss": 1.6509,
      "step": 6254
    },
    {
      "epoch": 2.1319018404907975,
      "grad_norm": 13.61315302375698,
      "learning_rate": 2.3431386965718055e-06,
      "loss": 0.9522,
      "step": 6255
    },
    {
      "epoch": 2.1322426721199728,
      "grad_norm": 10.272598688221903,
      "learning_rate": 2.341458963738752e-06,
      "loss": 1.2024,
      "step": 6256
    },
    {
      "epoch": 2.132583503749148,
      "grad_norm": 15.202405018658306,
      "learning_rate": 2.3397796491047888e-06,
      "loss": 1.7147,
      "step": 6257
    },
    {
      "epoch": 2.132924335378323,
      "grad_norm": 15.968633579294407,
      "learning_rate": 2.3381007529340722e-06,
      "loss": 1.7109,
      "step": 6258
    },
    {
      "epoch": 2.1332651670074982,
      "grad_norm": 10.025612632097607,
      "learning_rate": 2.3364222754907055e-06,
      "loss": 1.06,
      "step": 6259
    },
    {
      "epoch": 2.1336059986366736,
      "grad_norm": 13.981421572049484,
      "learning_rate": 2.33474421703872e-06,
      "loss": 0.8805,
      "step": 6260
    },
    {
      "epoch": 2.133946830265849,
      "grad_norm": 18.235821659677384,
      "learning_rate": 2.333066577842075e-06,
      "loss": 1.5388,
      "step": 6261
    },
    {
      "epoch": 2.1342876618950237,
      "grad_norm": 16.995087273691716,
      "learning_rate": 2.331389358164678e-06,
      "loss": 1.3812,
      "step": 6262
    },
    {
      "epoch": 2.134628493524199,
      "grad_norm": 9.942863732571093,
      "learning_rate": 2.3297125582703555e-06,
      "loss": 1.2258,
      "step": 6263
    },
    {
      "epoch": 2.1349693251533743,
      "grad_norm": 9.809511492783678,
      "learning_rate": 2.3280361784228763e-06,
      "loss": 1.3287,
      "step": 6264
    },
    {
      "epoch": 2.135310156782549,
      "grad_norm": 26.895011767179426,
      "learning_rate": 2.326360218885946e-06,
      "loss": 0.8062,
      "step": 6265
    },
    {
      "epoch": 2.1356509884117245,
      "grad_norm": 14.662533446715448,
      "learning_rate": 2.3246846799231947e-06,
      "loss": 1.2633,
      "step": 6266
    },
    {
      "epoch": 2.1359918200409,
      "grad_norm": 12.815674344514166,
      "learning_rate": 2.323009561798193e-06,
      "loss": 1.1422,
      "step": 6267
    },
    {
      "epoch": 2.136332651670075,
      "grad_norm": 9.504201943448528,
      "learning_rate": 2.321334864774444e-06,
      "loss": 1.182,
      "step": 6268
    },
    {
      "epoch": 2.13667348329925,
      "grad_norm": 16.282251812498473,
      "learning_rate": 2.3196605891153833e-06,
      "loss": 1.812,
      "step": 6269
    },
    {
      "epoch": 2.1370143149284253,
      "grad_norm": 12.887684177123699,
      "learning_rate": 2.317986735084382e-06,
      "loss": 1.1858,
      "step": 6270
    },
    {
      "epoch": 2.1373551465576006,
      "grad_norm": 18.444917160783973,
      "learning_rate": 2.3163133029447436e-06,
      "loss": 1.6392,
      "step": 6271
    },
    {
      "epoch": 2.137695978186776,
      "grad_norm": 11.305627500062021,
      "learning_rate": 2.314640292959706e-06,
      "loss": 1.3338,
      "step": 6272
    },
    {
      "epoch": 2.138036809815951,
      "grad_norm": 19.619635121942686,
      "learning_rate": 2.3129677053924403e-06,
      "loss": 1.861,
      "step": 6273
    },
    {
      "epoch": 2.138377641445126,
      "grad_norm": 19.612489523792817,
      "learning_rate": 2.3112955405060506e-06,
      "loss": 0.7991,
      "step": 6274
    },
    {
      "epoch": 2.1387184730743014,
      "grad_norm": 15.318109156177933,
      "learning_rate": 2.3096237985635755e-06,
      "loss": 1.2076,
      "step": 6275
    },
    {
      "epoch": 2.1390593047034763,
      "grad_norm": 16.77206580719733,
      "learning_rate": 2.3079524798279873e-06,
      "loss": 1.4533,
      "step": 6276
    },
    {
      "epoch": 2.1394001363326516,
      "grad_norm": 13.052388427426955,
      "learning_rate": 2.30628158456219e-06,
      "loss": 1.1748,
      "step": 6277
    },
    {
      "epoch": 2.139740967961827,
      "grad_norm": 17.27878922728916,
      "learning_rate": 2.3046111130290243e-06,
      "loss": 1.7726,
      "step": 6278
    },
    {
      "epoch": 2.140081799591002,
      "grad_norm": 20.219241755780768,
      "learning_rate": 2.3029410654912565e-06,
      "loss": 2.0337,
      "step": 6279
    },
    {
      "epoch": 2.140422631220177,
      "grad_norm": 18.240753696612053,
      "learning_rate": 2.301271442211597e-06,
      "loss": 0.8721,
      "step": 6280
    },
    {
      "epoch": 2.1407634628493524,
      "grad_norm": 13.795361592816544,
      "learning_rate": 2.2996022434526855e-06,
      "loss": 1.3986,
      "step": 6281
    },
    {
      "epoch": 2.1411042944785277,
      "grad_norm": 16.087069566736456,
      "learning_rate": 2.297933469477086e-06,
      "loss": 1.5859,
      "step": 6282
    },
    {
      "epoch": 2.141445126107703,
      "grad_norm": 13.92242835108406,
      "learning_rate": 2.2962651205473125e-06,
      "loss": 1.1575,
      "step": 6283
    },
    {
      "epoch": 2.141785957736878,
      "grad_norm": 18.097211086291246,
      "learning_rate": 2.2945971969257964e-06,
      "loss": 1.452,
      "step": 6284
    },
    {
      "epoch": 2.142126789366053,
      "grad_norm": 16.579712323263976,
      "learning_rate": 2.292929698874909e-06,
      "loss": 1.6616,
      "step": 6285
    },
    {
      "epoch": 2.1424676209952285,
      "grad_norm": 18.07252041260663,
      "learning_rate": 2.2912626266569597e-06,
      "loss": 1.8844,
      "step": 6286
    },
    {
      "epoch": 2.1428084526244033,
      "grad_norm": 16.699224813486076,
      "learning_rate": 2.2895959805341804e-06,
      "loss": 1.4822,
      "step": 6287
    },
    {
      "epoch": 2.1431492842535786,
      "grad_norm": 15.914625477684664,
      "learning_rate": 2.2879297607687426e-06,
      "loss": 0.7948,
      "step": 6288
    },
    {
      "epoch": 2.143490115882754,
      "grad_norm": 23.769411993802635,
      "learning_rate": 2.28626396762275e-06,
      "loss": 1.0411,
      "step": 6289
    },
    {
      "epoch": 2.1438309475119293,
      "grad_norm": 17.081205802169393,
      "learning_rate": 2.2845986013582378e-06,
      "loss": 1.4327,
      "step": 6290
    },
    {
      "epoch": 2.144171779141104,
      "grad_norm": 14.407140109767882,
      "learning_rate": 2.2829336622371765e-06,
      "loss": 1.5771,
      "step": 6291
    },
    {
      "epoch": 2.1445126107702794,
      "grad_norm": 13.585470377239261,
      "learning_rate": 2.2812691505214633e-06,
      "loss": 1.6108,
      "step": 6292
    },
    {
      "epoch": 2.1448534423994547,
      "grad_norm": 18.459699905032366,
      "learning_rate": 2.2796050664729367e-06,
      "loss": 1.6478,
      "step": 6293
    },
    {
      "epoch": 2.14519427402863,
      "grad_norm": 16.075895915678075,
      "learning_rate": 2.2779414103533643e-06,
      "loss": 1.7872,
      "step": 6294
    },
    {
      "epoch": 2.145535105657805,
      "grad_norm": 22.32423674791049,
      "learning_rate": 2.27627818242444e-06,
      "loss": 1.0986,
      "step": 6295
    },
    {
      "epoch": 2.1458759372869802,
      "grad_norm": 16.330445635158526,
      "learning_rate": 2.274615382947804e-06,
      "loss": 1.503,
      "step": 6296
    },
    {
      "epoch": 2.1462167689161555,
      "grad_norm": 19.648189219501635,
      "learning_rate": 2.2729530121850156e-06,
      "loss": 1.214,
      "step": 6297
    },
    {
      "epoch": 2.1465576005453304,
      "grad_norm": 12.93311872242047,
      "learning_rate": 2.271291070397573e-06,
      "loss": 1.1899,
      "step": 6298
    },
    {
      "epoch": 2.1468984321745057,
      "grad_norm": 18.985718527153487,
      "learning_rate": 2.2696295578469095e-06,
      "loss": 1.0277,
      "step": 6299
    },
    {
      "epoch": 2.147239263803681,
      "grad_norm": 18.714783555084452,
      "learning_rate": 2.2679684747943847e-06,
      "loss": 1.6059,
      "step": 6300
    },
    {
      "epoch": 2.1475800954328563,
      "grad_norm": 18.255945653161383,
      "learning_rate": 2.2663078215012945e-06,
      "loss": 1.4174,
      "step": 6301
    },
    {
      "epoch": 2.147920927062031,
      "grad_norm": 12.614656157512782,
      "learning_rate": 2.2646475982288657e-06,
      "loss": 1.0649,
      "step": 6302
    },
    {
      "epoch": 2.1482617586912065,
      "grad_norm": 17.920239449294478,
      "learning_rate": 2.262987805238259e-06,
      "loss": 1.1351,
      "step": 6303
    },
    {
      "epoch": 2.148602590320382,
      "grad_norm": 14.603666728953467,
      "learning_rate": 2.261328442790566e-06,
      "loss": 1.2174,
      "step": 6304
    },
    {
      "epoch": 2.148943421949557,
      "grad_norm": 14.555673222096987,
      "learning_rate": 2.2596695111468105e-06,
      "loss": 1.0307,
      "step": 6305
    },
    {
      "epoch": 2.149284253578732,
      "grad_norm": 16.56069429707057,
      "learning_rate": 2.2580110105679493e-06,
      "loss": 1.822,
      "step": 6306
    },
    {
      "epoch": 2.1496250852079073,
      "grad_norm": 18.193067309226375,
      "learning_rate": 2.2563529413148718e-06,
      "loss": 2.019,
      "step": 6307
    },
    {
      "epoch": 2.1499659168370826,
      "grad_norm": 29.56503869616025,
      "learning_rate": 2.254695303648398e-06,
      "loss": 1.694,
      "step": 6308
    },
    {
      "epoch": 2.1503067484662575,
      "grad_norm": 14.371603787806224,
      "learning_rate": 2.253038097829282e-06,
      "loss": 1.1391,
      "step": 6309
    },
    {
      "epoch": 2.1506475800954328,
      "grad_norm": 15.74568257949557,
      "learning_rate": 2.2513813241182083e-06,
      "loss": 1.1577,
      "step": 6310
    },
    {
      "epoch": 2.150988411724608,
      "grad_norm": 13.805279977428047,
      "learning_rate": 2.2497249827757933e-06,
      "loss": 1.0711,
      "step": 6311
    },
    {
      "epoch": 2.1513292433537834,
      "grad_norm": 15.430223100886565,
      "learning_rate": 2.248069074062589e-06,
      "loss": 1.798,
      "step": 6312
    },
    {
      "epoch": 2.1516700749829583,
      "grad_norm": 16.055859525457173,
      "learning_rate": 2.2464135982390706e-06,
      "loss": 1.1075,
      "step": 6313
    },
    {
      "epoch": 2.1520109066121336,
      "grad_norm": 18.115216636453244,
      "learning_rate": 2.244758555565656e-06,
      "loss": 1.4734,
      "step": 6314
    },
    {
      "epoch": 2.152351738241309,
      "grad_norm": 12.401228940552873,
      "learning_rate": 2.2431039463026905e-06,
      "loss": 1.4334,
      "step": 6315
    },
    {
      "epoch": 2.152692569870484,
      "grad_norm": 18.352552767460196,
      "learning_rate": 2.2414497707104452e-06,
      "loss": 1.536,
      "step": 6316
    },
    {
      "epoch": 2.153033401499659,
      "grad_norm": 17.58302222505736,
      "learning_rate": 2.239796029049136e-06,
      "loss": 1.8262,
      "step": 6317
    },
    {
      "epoch": 2.1533742331288344,
      "grad_norm": 14.190271732944463,
      "learning_rate": 2.238142721578897e-06,
      "loss": 1.0323,
      "step": 6318
    },
    {
      "epoch": 2.1537150647580097,
      "grad_norm": 12.862328883682183,
      "learning_rate": 2.236489848559801e-06,
      "loss": 1.4266,
      "step": 6319
    },
    {
      "epoch": 2.1540558963871845,
      "grad_norm": 12.906559852559852,
      "learning_rate": 2.2348374102518557e-06,
      "loss": 1.6675,
      "step": 6320
    },
    {
      "epoch": 2.15439672801636,
      "grad_norm": 10.214590012011454,
      "learning_rate": 2.2331854069149915e-06,
      "loss": 1.0753,
      "step": 6321
    },
    {
      "epoch": 2.154737559645535,
      "grad_norm": 27.434029352500055,
      "learning_rate": 2.231533838809077e-06,
      "loss": 1.6348,
      "step": 6322
    },
    {
      "epoch": 2.1550783912747105,
      "grad_norm": 13.111384881797928,
      "learning_rate": 2.2298827061939106e-06,
      "loss": 1.1765,
      "step": 6323
    },
    {
      "epoch": 2.1554192229038853,
      "grad_norm": 19.889375046734504,
      "learning_rate": 2.228232009329221e-06,
      "loss": 1.7678,
      "step": 6324
    },
    {
      "epoch": 2.1557600545330606,
      "grad_norm": 11.892246934634473,
      "learning_rate": 2.226581748474673e-06,
      "loss": 1.4445,
      "step": 6325
    },
    {
      "epoch": 2.156100886162236,
      "grad_norm": 15.438553570371866,
      "learning_rate": 2.2249319238898517e-06,
      "loss": 1.6629,
      "step": 6326
    },
    {
      "epoch": 2.1564417177914113,
      "grad_norm": 15.707225574518112,
      "learning_rate": 2.2232825358342875e-06,
      "loss": 1.8786,
      "step": 6327
    },
    {
      "epoch": 2.156782549420586,
      "grad_norm": 16.563688564829164,
      "learning_rate": 2.2216335845674357e-06,
      "loss": 1.4408,
      "step": 6328
    },
    {
      "epoch": 2.1571233810497614,
      "grad_norm": 15.144408240444719,
      "learning_rate": 2.219985070348678e-06,
      "loss": 1.7102,
      "step": 6329
    },
    {
      "epoch": 2.1574642126789367,
      "grad_norm": 14.794781650821832,
      "learning_rate": 2.2183369934373365e-06,
      "loss": 1.5902,
      "step": 6330
    },
    {
      "epoch": 2.1578050443081116,
      "grad_norm": 27.486051435949136,
      "learning_rate": 2.2166893540926616e-06,
      "loss": 1.386,
      "step": 6331
    },
    {
      "epoch": 2.158145875937287,
      "grad_norm": 11.646321489375513,
      "learning_rate": 2.215042152573827e-06,
      "loss": 1.4574,
      "step": 6332
    },
    {
      "epoch": 2.158486707566462,
      "grad_norm": 20.523745214290635,
      "learning_rate": 2.213395389139952e-06,
      "loss": 1.0599,
      "step": 6333
    },
    {
      "epoch": 2.1588275391956375,
      "grad_norm": 12.162412670277105,
      "learning_rate": 2.2117490640500732e-06,
      "loss": 1.1046,
      "step": 6334
    },
    {
      "epoch": 2.1591683708248124,
      "grad_norm": 10.15939306205793,
      "learning_rate": 2.2101031775631636e-06,
      "loss": 1.0846,
      "step": 6335
    },
    {
      "epoch": 2.1595092024539877,
      "grad_norm": 16.770820483153624,
      "learning_rate": 2.2084577299381343e-06,
      "loss": 1.6056,
      "step": 6336
    },
    {
      "epoch": 2.159850034083163,
      "grad_norm": 19.05263620079242,
      "learning_rate": 2.2068127214338148e-06,
      "loss": 1.9478,
      "step": 6337
    },
    {
      "epoch": 2.1601908657123383,
      "grad_norm": 10.879126531796443,
      "learning_rate": 2.2051681523089736e-06,
      "loss": 1.1732,
      "step": 6338
    },
    {
      "epoch": 2.160531697341513,
      "grad_norm": 15.928616613597537,
      "learning_rate": 2.203524022822307e-06,
      "loss": 1.7771,
      "step": 6339
    },
    {
      "epoch": 2.1608725289706885,
      "grad_norm": 12.442965519365064,
      "learning_rate": 2.2018803332324434e-06,
      "loss": 1.249,
      "step": 6340
    },
    {
      "epoch": 2.161213360599864,
      "grad_norm": 9.440013962507871,
      "learning_rate": 2.2002370837979432e-06,
      "loss": 0.8019,
      "step": 6341
    },
    {
      "epoch": 2.1615541922290387,
      "grad_norm": 13.91365538960959,
      "learning_rate": 2.1985942747772935e-06,
      "loss": 1.2259,
      "step": 6342
    },
    {
      "epoch": 2.161895023858214,
      "grad_norm": 15.122917761091601,
      "learning_rate": 2.1969519064289173e-06,
      "loss": 1.3791,
      "step": 6343
    },
    {
      "epoch": 2.1622358554873893,
      "grad_norm": 18.443374420930702,
      "learning_rate": 2.195309979011164e-06,
      "loss": 0.8839,
      "step": 6344
    },
    {
      "epoch": 2.1625766871165646,
      "grad_norm": 14.890768047774994,
      "learning_rate": 2.193668492782316e-06,
      "loss": 1.4203,
      "step": 6345
    },
    {
      "epoch": 2.1629175187457395,
      "grad_norm": 12.387266316386958,
      "learning_rate": 2.192027448000587e-06,
      "loss": 1.0671,
      "step": 6346
    },
    {
      "epoch": 2.1632583503749148,
      "grad_norm": 16.951911721876236,
      "learning_rate": 2.190386844924115e-06,
      "loss": 2.0304,
      "step": 6347
    },
    {
      "epoch": 2.16359918200409,
      "grad_norm": 12.644114860147614,
      "learning_rate": 2.1887466838109785e-06,
      "loss": 1.415,
      "step": 6348
    },
    {
      "epoch": 2.1639400136332654,
      "grad_norm": 11.46502464980983,
      "learning_rate": 2.187106964919181e-06,
      "loss": 1.5237,
      "step": 6349
    },
    {
      "epoch": 2.1642808452624402,
      "grad_norm": 18.669099735779355,
      "learning_rate": 2.185467688506652e-06,
      "loss": 0.6629,
      "step": 6350
    },
    {
      "epoch": 2.1646216768916156,
      "grad_norm": 13.895882754611515,
      "learning_rate": 2.1838288548312637e-06,
      "loss": 1.4994,
      "step": 6351
    },
    {
      "epoch": 2.164962508520791,
      "grad_norm": 17.44497346450304,
      "learning_rate": 2.182190464150805e-06,
      "loss": 1.7141,
      "step": 6352
    },
    {
      "epoch": 2.1653033401499657,
      "grad_norm": 35.473048109376705,
      "learning_rate": 2.180552516723003e-06,
      "loss": 2.3086,
      "step": 6353
    },
    {
      "epoch": 2.165644171779141,
      "grad_norm": 12.247473240347823,
      "learning_rate": 2.178915012805514e-06,
      "loss": 1.1235,
      "step": 6354
    },
    {
      "epoch": 2.1659850034083163,
      "grad_norm": 14.248074463799268,
      "learning_rate": 2.1772779526559236e-06,
      "loss": 1.8417,
      "step": 6355
    },
    {
      "epoch": 2.1663258350374917,
      "grad_norm": 14.923766009925679,
      "learning_rate": 2.1756413365317474e-06,
      "loss": 1.1367,
      "step": 6356
    },
    {
      "epoch": 2.1666666666666665,
      "grad_norm": 13.147271408064467,
      "learning_rate": 2.1740051646904326e-06,
      "loss": 1.3994,
      "step": 6357
    },
    {
      "epoch": 2.167007498295842,
      "grad_norm": 17.034188150863276,
      "learning_rate": 2.172369437389355e-06,
      "loss": 1.0711,
      "step": 6358
    },
    {
      "epoch": 2.167348329925017,
      "grad_norm": 40.54572385022194,
      "learning_rate": 2.1707341548858205e-06,
      "loss": 1.6721,
      "step": 6359
    },
    {
      "epoch": 2.1676891615541924,
      "grad_norm": 15.880501905552856,
      "learning_rate": 2.1690993174370665e-06,
      "loss": 1.9122,
      "step": 6360
    },
    {
      "epoch": 2.1680299931833673,
      "grad_norm": 10.12951013549337,
      "learning_rate": 2.1674649253002593e-06,
      "loss": 1.2556,
      "step": 6361
    },
    {
      "epoch": 2.1683708248125426,
      "grad_norm": 19.040057214766268,
      "learning_rate": 2.1658309787324967e-06,
      "loss": 1.8297,
      "step": 6362
    },
    {
      "epoch": 2.168711656441718,
      "grad_norm": 14.913006274703895,
      "learning_rate": 2.1641974779907994e-06,
      "loss": 1.7444,
      "step": 6363
    },
    {
      "epoch": 2.169052488070893,
      "grad_norm": 13.30040764449652,
      "learning_rate": 2.1625644233321293e-06,
      "loss": 1.0295,
      "step": 6364
    },
    {
      "epoch": 2.169393319700068,
      "grad_norm": 19.081520281653173,
      "learning_rate": 2.1609318150133724e-06,
      "loss": 1.4962,
      "step": 6365
    },
    {
      "epoch": 2.1697341513292434,
      "grad_norm": 21.013875133402273,
      "learning_rate": 2.1592996532913397e-06,
      "loss": 1.7716,
      "step": 6366
    },
    {
      "epoch": 2.1700749829584187,
      "grad_norm": 15.380617361130238,
      "learning_rate": 2.1576679384227826e-06,
      "loss": 1.7069,
      "step": 6367
    },
    {
      "epoch": 2.1704158145875936,
      "grad_norm": 15.914795853630052,
      "learning_rate": 2.1560366706643716e-06,
      "loss": 1.9748,
      "step": 6368
    },
    {
      "epoch": 2.170756646216769,
      "grad_norm": 12.65551143407503,
      "learning_rate": 2.1544058502727123e-06,
      "loss": 1.3196,
      "step": 6369
    },
    {
      "epoch": 2.171097477845944,
      "grad_norm": 16.574190277886533,
      "learning_rate": 2.1527754775043436e-06,
      "loss": 1.3115,
      "step": 6370
    },
    {
      "epoch": 2.1714383094751195,
      "grad_norm": 17.60910812495016,
      "learning_rate": 2.1511455526157244e-06,
      "loss": 1.8509,
      "step": 6371
    },
    {
      "epoch": 2.1717791411042944,
      "grad_norm": 20.32275941425562,
      "learning_rate": 2.149516075863251e-06,
      "loss": 2.0637,
      "step": 6372
    },
    {
      "epoch": 2.1721199727334697,
      "grad_norm": 22.21748755382518,
      "learning_rate": 2.1478870475032455e-06,
      "loss": 1.2435,
      "step": 6373
    },
    {
      "epoch": 2.172460804362645,
      "grad_norm": 15.584038186028081,
      "learning_rate": 2.1462584677919613e-06,
      "loss": 1.2235,
      "step": 6374
    },
    {
      "epoch": 2.17280163599182,
      "grad_norm": 18.042211102614484,
      "learning_rate": 2.1446303369855797e-06,
      "loss": 1.6339,
      "step": 6375
    },
    {
      "epoch": 2.173142467620995,
      "grad_norm": 16.092096332836586,
      "learning_rate": 2.1430026553402134e-06,
      "loss": 1.1807,
      "step": 6376
    },
    {
      "epoch": 2.1734832992501705,
      "grad_norm": 16.679976149027365,
      "learning_rate": 2.141375423111902e-06,
      "loss": 2.1089,
      "step": 6377
    },
    {
      "epoch": 2.173824130879346,
      "grad_norm": 16.138266883323823,
      "learning_rate": 2.1397486405566166e-06,
      "loss": 1.6286,
      "step": 6378
    },
    {
      "epoch": 2.1741649625085206,
      "grad_norm": 12.251775688133684,
      "learning_rate": 2.1381223079302552e-06,
      "loss": 0.9213,
      "step": 6379
    },
    {
      "epoch": 2.174505794137696,
      "grad_norm": 15.13970053395009,
      "learning_rate": 2.1364964254886493e-06,
      "loss": 1.68,
      "step": 6380
    },
    {
      "epoch": 2.1748466257668713,
      "grad_norm": 12.311160746203376,
      "learning_rate": 2.1348709934875515e-06,
      "loss": 1.2173,
      "step": 6381
    },
    {
      "epoch": 2.1751874573960466,
      "grad_norm": 13.461168962281546,
      "learning_rate": 2.133246012182653e-06,
      "loss": 1.774,
      "step": 6382
    },
    {
      "epoch": 2.1755282890252214,
      "grad_norm": 28.50722205803135,
      "learning_rate": 2.131621481829571e-06,
      "loss": 1.6572,
      "step": 6383
    },
    {
      "epoch": 2.1758691206543967,
      "grad_norm": 16.231245992178028,
      "learning_rate": 2.1299974026838445e-06,
      "loss": 1.2849,
      "step": 6384
    },
    {
      "epoch": 2.176209952283572,
      "grad_norm": 12.086588333278401,
      "learning_rate": 2.128373775000955e-06,
      "loss": 1.4896,
      "step": 6385
    },
    {
      "epoch": 2.176550783912747,
      "grad_norm": 16.000258389633625,
      "learning_rate": 2.1267505990363e-06,
      "loss": 1.7448,
      "step": 6386
    },
    {
      "epoch": 2.1768916155419222,
      "grad_norm": 10.39648952375022,
      "learning_rate": 2.1251278750452136e-06,
      "loss": 1.2213,
      "step": 6387
    },
    {
      "epoch": 2.1772324471710975,
      "grad_norm": 16.641957976682935,
      "learning_rate": 2.123505603282957e-06,
      "loss": 2.1339,
      "step": 6388
    },
    {
      "epoch": 2.177573278800273,
      "grad_norm": 12.738859915713475,
      "learning_rate": 2.1218837840047193e-06,
      "loss": 1.1774,
      "step": 6389
    },
    {
      "epoch": 2.1779141104294477,
      "grad_norm": 19.918346723825316,
      "learning_rate": 2.1202624174656193e-06,
      "loss": 1.5249,
      "step": 6390
    },
    {
      "epoch": 2.178254942058623,
      "grad_norm": 17.152471415543843,
      "learning_rate": 2.118641503920705e-06,
      "loss": 1.5149,
      "step": 6391
    },
    {
      "epoch": 2.1785957736877983,
      "grad_norm": 16.199835621974575,
      "learning_rate": 2.1170210436249524e-06,
      "loss": 1.7361,
      "step": 6392
    },
    {
      "epoch": 2.1789366053169736,
      "grad_norm": 15.449693780672083,
      "learning_rate": 2.1154010368332658e-06,
      "loss": 1.3685,
      "step": 6393
    },
    {
      "epoch": 2.1792774369461485,
      "grad_norm": 15.51537089435563,
      "learning_rate": 2.113781483800479e-06,
      "loss": 1.5615,
      "step": 6394
    },
    {
      "epoch": 2.179618268575324,
      "grad_norm": 12.883989491980175,
      "learning_rate": 2.112162384781354e-06,
      "loss": 1.6957,
      "step": 6395
    },
    {
      "epoch": 2.179959100204499,
      "grad_norm": 17.756139577547703,
      "learning_rate": 2.110543740030583e-06,
      "loss": 1.1287,
      "step": 6396
    },
    {
      "epoch": 2.180299931833674,
      "grad_norm": 16.143448533018617,
      "learning_rate": 2.1089255498027815e-06,
      "loss": 1.4959,
      "step": 6397
    },
    {
      "epoch": 2.1806407634628493,
      "grad_norm": 16.432544628109877,
      "learning_rate": 2.107307814352501e-06,
      "loss": 0.9615,
      "step": 6398
    },
    {
      "epoch": 2.1809815950920246,
      "grad_norm": 17.095368777712203,
      "learning_rate": 2.1056905339342183e-06,
      "loss": 1.5621,
      "step": 6399
    },
    {
      "epoch": 2.1813224267212,
      "grad_norm": 39.38023358002484,
      "learning_rate": 2.1040737088023323e-06,
      "loss": 1.4668,
      "step": 6400
    },
    {
      "epoch": 2.181663258350375,
      "grad_norm": 13.391953166726896,
      "learning_rate": 2.102457339211184e-06,
      "loss": 1.544,
      "step": 6401
    },
    {
      "epoch": 2.18200408997955,
      "grad_norm": 19.3288018789814,
      "learning_rate": 2.1008414254150285e-06,
      "loss": 1.6265,
      "step": 6402
    },
    {
      "epoch": 2.1823449216087254,
      "grad_norm": 10.271870858974413,
      "learning_rate": 2.099225967668056e-06,
      "loss": 1.1742,
      "step": 6403
    },
    {
      "epoch": 2.1826857532379007,
      "grad_norm": 14.110353664663606,
      "learning_rate": 2.09761096622439e-06,
      "loss": 1.3137,
      "step": 6404
    },
    {
      "epoch": 2.1830265848670756,
      "grad_norm": 15.943071250867334,
      "learning_rate": 2.0959964213380707e-06,
      "loss": 1.2619,
      "step": 6405
    },
    {
      "epoch": 2.183367416496251,
      "grad_norm": 13.945995099717063,
      "learning_rate": 2.094382333263074e-06,
      "loss": 0.9038,
      "step": 6406
    },
    {
      "epoch": 2.183708248125426,
      "grad_norm": 13.748988116044995,
      "learning_rate": 2.0927687022533034e-06,
      "loss": 1.6231,
      "step": 6407
    },
    {
      "epoch": 2.184049079754601,
      "grad_norm": 17.17819409058399,
      "learning_rate": 2.091155528562589e-06,
      "loss": 1.7235,
      "step": 6408
    },
    {
      "epoch": 2.1843899113837764,
      "grad_norm": 17.71343140503629,
      "learning_rate": 2.0895428124446893e-06,
      "loss": 1.5126,
      "step": 6409
    },
    {
      "epoch": 2.1847307430129517,
      "grad_norm": 15.536612152165866,
      "learning_rate": 2.0879305541532914e-06,
      "loss": 1.3987,
      "step": 6410
    },
    {
      "epoch": 2.185071574642127,
      "grad_norm": 15.93570038352703,
      "learning_rate": 2.0863187539420086e-06,
      "loss": 1.4199,
      "step": 6411
    },
    {
      "epoch": 2.185412406271302,
      "grad_norm": 12.485677181701922,
      "learning_rate": 2.084707412064386e-06,
      "loss": 1.2827,
      "step": 6412
    },
    {
      "epoch": 2.185753237900477,
      "grad_norm": 16.684775772608106,
      "learning_rate": 2.0830965287738914e-06,
      "loss": 1.3299,
      "step": 6413
    },
    {
      "epoch": 2.1860940695296525,
      "grad_norm": 15.051046825273373,
      "learning_rate": 2.0814861043239265e-06,
      "loss": 1.7206,
      "step": 6414
    },
    {
      "epoch": 2.1864349011588278,
      "grad_norm": 15.458450275371668,
      "learning_rate": 2.079876138967811e-06,
      "loss": 1.3315,
      "step": 6415
    },
    {
      "epoch": 2.1867757327880026,
      "grad_norm": 15.367257372051684,
      "learning_rate": 2.0782666329588063e-06,
      "loss": 1.2881,
      "step": 6416
    },
    {
      "epoch": 2.187116564417178,
      "grad_norm": 15.12405283754705,
      "learning_rate": 2.076657586550092e-06,
      "loss": 1.6269,
      "step": 6417
    },
    {
      "epoch": 2.1874573960463533,
      "grad_norm": 11.620793368235901,
      "learning_rate": 2.075048999994775e-06,
      "loss": 1.296,
      "step": 6418
    },
    {
      "epoch": 2.187798227675528,
      "grad_norm": 12.452814330984637,
      "learning_rate": 2.073440873545894e-06,
      "loss": 1.5389,
      "step": 6419
    },
    {
      "epoch": 2.1881390593047034,
      "grad_norm": 13.026583857583542,
      "learning_rate": 2.0718332074564128e-06,
      "loss": 0.6218,
      "step": 6420
    },
    {
      "epoch": 2.1884798909338787,
      "grad_norm": 18.631756281772184,
      "learning_rate": 2.0702260019792254e-06,
      "loss": 1.0899,
      "step": 6421
    },
    {
      "epoch": 2.188820722563054,
      "grad_norm": 19.98006429638065,
      "learning_rate": 2.0686192573671497e-06,
      "loss": 1.7855,
      "step": 6422
    },
    {
      "epoch": 2.189161554192229,
      "grad_norm": 11.588220349307065,
      "learning_rate": 2.0670129738729346e-06,
      "loss": 1.1143,
      "step": 6423
    },
    {
      "epoch": 2.189502385821404,
      "grad_norm": 15.782831900499858,
      "learning_rate": 2.065407151749254e-06,
      "loss": 1.7666,
      "step": 6424
    },
    {
      "epoch": 2.1898432174505795,
      "grad_norm": 19.735154725975136,
      "learning_rate": 2.0638017912487103e-06,
      "loss": 2.3097,
      "step": 6425
    },
    {
      "epoch": 2.190184049079755,
      "grad_norm": 19.697308639032492,
      "learning_rate": 2.0621968926238335e-06,
      "loss": 1.6927,
      "step": 6426
    },
    {
      "epoch": 2.1905248807089297,
      "grad_norm": 15.292764729678327,
      "learning_rate": 2.0605924561270803e-06,
      "loss": 1.5597,
      "step": 6427
    },
    {
      "epoch": 2.190865712338105,
      "grad_norm": 18.87214337617264,
      "learning_rate": 2.058988482010834e-06,
      "loss": 1.9717,
      "step": 6428
    },
    {
      "epoch": 2.1912065439672803,
      "grad_norm": 13.311963100096081,
      "learning_rate": 2.057384970527408e-06,
      "loss": 1.3603,
      "step": 6429
    },
    {
      "epoch": 2.191547375596455,
      "grad_norm": 16.86011369358606,
      "learning_rate": 2.055781921929041e-06,
      "loss": 1.1594,
      "step": 6430
    },
    {
      "epoch": 2.1918882072256305,
      "grad_norm": 13.874629789662377,
      "learning_rate": 2.054179336467895e-06,
      "loss": 1.9579,
      "step": 6431
    },
    {
      "epoch": 2.192229038854806,
      "grad_norm": 11.897802095194654,
      "learning_rate": 2.052577214396067e-06,
      "loss": 1.1727,
      "step": 6432
    },
    {
      "epoch": 2.192569870483981,
      "grad_norm": 12.908782141407919,
      "learning_rate": 2.0509755559655785e-06,
      "loss": 1.4319,
      "step": 6433
    },
    {
      "epoch": 2.192910702113156,
      "grad_norm": 11.85418597997109,
      "learning_rate": 2.049374361428371e-06,
      "loss": 0.9313,
      "step": 6434
    },
    {
      "epoch": 2.1932515337423313,
      "grad_norm": 14.012202622661121,
      "learning_rate": 2.0477736310363257e-06,
      "loss": 1.4235,
      "step": 6435
    },
    {
      "epoch": 2.1935923653715066,
      "grad_norm": 12.573472144784116,
      "learning_rate": 2.0461733650412385e-06,
      "loss": 1.2747,
      "step": 6436
    },
    {
      "epoch": 2.1939331970006815,
      "grad_norm": 15.23020182682256,
      "learning_rate": 2.044573563694838e-06,
      "loss": 1.0825,
      "step": 6437
    },
    {
      "epoch": 2.1942740286298568,
      "grad_norm": 14.774945170242264,
      "learning_rate": 2.0429742272487833e-06,
      "loss": 1.0683,
      "step": 6438
    },
    {
      "epoch": 2.194614860259032,
      "grad_norm": 15.84066270419507,
      "learning_rate": 2.041375355954653e-06,
      "loss": 1.3354,
      "step": 6439
    },
    {
      "epoch": 2.1949556918882074,
      "grad_norm": 17.284715698514063,
      "learning_rate": 2.039776950063957e-06,
      "loss": 1.244,
      "step": 6440
    },
    {
      "epoch": 2.1952965235173822,
      "grad_norm": 13.345719247077106,
      "learning_rate": 2.03817900982813e-06,
      "loss": 1.6978,
      "step": 6441
    },
    {
      "epoch": 2.1956373551465576,
      "grad_norm": 14.180599566631459,
      "learning_rate": 2.036581535498536e-06,
      "loss": 1.4292,
      "step": 6442
    },
    {
      "epoch": 2.195978186775733,
      "grad_norm": 14.59112866912174,
      "learning_rate": 2.0349845273264625e-06,
      "loss": 1.4585,
      "step": 6443
    },
    {
      "epoch": 2.196319018404908,
      "grad_norm": 16.30304737634381,
      "learning_rate": 2.0333879855631255e-06,
      "loss": 1.901,
      "step": 6444
    },
    {
      "epoch": 2.196659850034083,
      "grad_norm": 17.04383123899611,
      "learning_rate": 2.0317919104596685e-06,
      "loss": 1.3474,
      "step": 6445
    },
    {
      "epoch": 2.1970006816632583,
      "grad_norm": 17.635220869074534,
      "learning_rate": 2.0301963022671593e-06,
      "loss": 1.1978,
      "step": 6446
    },
    {
      "epoch": 2.1973415132924337,
      "grad_norm": 18.44514853040743,
      "learning_rate": 2.0286011612365947e-06,
      "loss": 1.5502,
      "step": 6447
    },
    {
      "epoch": 2.1976823449216085,
      "grad_norm": 13.503229030792665,
      "learning_rate": 2.027006487618895e-06,
      "loss": 1.2718,
      "step": 6448
    },
    {
      "epoch": 2.198023176550784,
      "grad_norm": 22.536046635772795,
      "learning_rate": 2.0254122816649095e-06,
      "loss": 1.3576,
      "step": 6449
    },
    {
      "epoch": 2.198364008179959,
      "grad_norm": 16.67582392371029,
      "learning_rate": 2.0238185436254136e-06,
      "loss": 1.987,
      "step": 6450
    },
    {
      "epoch": 2.1987048398091344,
      "grad_norm": 21.345995398979394,
      "learning_rate": 2.0222252737511107e-06,
      "loss": 0.9905,
      "step": 6451
    },
    {
      "epoch": 2.1990456714383093,
      "grad_norm": 12.239400146121898,
      "learning_rate": 2.0206324722926234e-06,
      "loss": 1.9195,
      "step": 6452
    },
    {
      "epoch": 2.1993865030674846,
      "grad_norm": 13.65888724735172,
      "learning_rate": 2.019040139500507e-06,
      "loss": 1.4334,
      "step": 6453
    },
    {
      "epoch": 2.19972733469666,
      "grad_norm": 15.883565609858454,
      "learning_rate": 2.0174482756252465e-06,
      "loss": 1.6715,
      "step": 6454
    },
    {
      "epoch": 2.2000681663258352,
      "grad_norm": 17.25538765312383,
      "learning_rate": 2.0158568809172434e-06,
      "loss": 1.433,
      "step": 6455
    },
    {
      "epoch": 2.20040899795501,
      "grad_norm": 18.399164329610336,
      "learning_rate": 2.0142659556268325e-06,
      "loss": 0.9613,
      "step": 6456
    },
    {
      "epoch": 2.2007498295841854,
      "grad_norm": 14.627464525230518,
      "learning_rate": 2.012675500004272e-06,
      "loss": 1.2869,
      "step": 6457
    },
    {
      "epoch": 2.2010906612133607,
      "grad_norm": 19.11536646793278,
      "learning_rate": 2.011085514299747e-06,
      "loss": 1.736,
      "step": 6458
    },
    {
      "epoch": 2.2014314928425356,
      "grad_norm": 13.106918586112219,
      "learning_rate": 2.009495998763369e-06,
      "loss": 1.2206,
      "step": 6459
    },
    {
      "epoch": 2.201772324471711,
      "grad_norm": 19.36996854911595,
      "learning_rate": 2.0079069536451744e-06,
      "loss": 1.8391,
      "step": 6460
    },
    {
      "epoch": 2.202113156100886,
      "grad_norm": 18.09341913371272,
      "learning_rate": 2.006318379195127e-06,
      "loss": 1.6575,
      "step": 6461
    },
    {
      "epoch": 2.2024539877300615,
      "grad_norm": 11.865955872365587,
      "learning_rate": 2.0047302756631165e-06,
      "loss": 1.2597,
      "step": 6462
    },
    {
      "epoch": 2.2027948193592364,
      "grad_norm": 13.595200795880697,
      "learning_rate": 2.0031426432989566e-06,
      "loss": 1.753,
      "step": 6463
    },
    {
      "epoch": 2.2031356509884117,
      "grad_norm": 17.879354813875253,
      "learning_rate": 2.0015554823523904e-06,
      "loss": 1.9127,
      "step": 6464
    },
    {
      "epoch": 2.203476482617587,
      "grad_norm": 14.70358075999326,
      "learning_rate": 1.999968793073079e-06,
      "loss": 1.6472,
      "step": 6465
    },
    {
      "epoch": 2.2038173142467623,
      "grad_norm": 14.11642947962428,
      "learning_rate": 1.998382575710622e-06,
      "loss": 1.4666,
      "step": 6466
    },
    {
      "epoch": 2.204158145875937,
      "grad_norm": 10.780782530061988,
      "learning_rate": 1.996796830514535e-06,
      "loss": 1.1249,
      "step": 6467
    },
    {
      "epoch": 2.2044989775051125,
      "grad_norm": 14.13623041862937,
      "learning_rate": 1.9952115577342584e-06,
      "loss": 1.0541,
      "step": 6468
    },
    {
      "epoch": 2.204839809134288,
      "grad_norm": 23.483169914290396,
      "learning_rate": 1.993626757619169e-06,
      "loss": 1.8315,
      "step": 6469
    },
    {
      "epoch": 2.2051806407634627,
      "grad_norm": 13.238950035767267,
      "learning_rate": 1.9920424304185564e-06,
      "loss": 0.7205,
      "step": 6470
    },
    {
      "epoch": 2.205521472392638,
      "grad_norm": 12.203623783738196,
      "learning_rate": 1.9904585763816416e-06,
      "loss": 1.5955,
      "step": 6471
    },
    {
      "epoch": 2.2058623040218133,
      "grad_norm": 14.18030250162095,
      "learning_rate": 1.988875195757577e-06,
      "loss": 1.4373,
      "step": 6472
    },
    {
      "epoch": 2.2062031356509886,
      "grad_norm": 8.660776631691814,
      "learning_rate": 1.9872922887954283e-06,
      "loss": 1.0617,
      "step": 6473
    },
    {
      "epoch": 2.2065439672801634,
      "grad_norm": 10.212027473864431,
      "learning_rate": 1.9857098557441957e-06,
      "loss": 1.2248,
      "step": 6474
    },
    {
      "epoch": 2.2068847989093388,
      "grad_norm": 14.756276743058947,
      "learning_rate": 1.9841278968528023e-06,
      "loss": 1.6293,
      "step": 6475
    },
    {
      "epoch": 2.207225630538514,
      "grad_norm": 18.953617319206217,
      "learning_rate": 1.9825464123700957e-06,
      "loss": 1.7257,
      "step": 6476
    },
    {
      "epoch": 2.2075664621676894,
      "grad_norm": 40.5485771978725,
      "learning_rate": 1.9809654025448504e-06,
      "loss": 1.5301,
      "step": 6477
    },
    {
      "epoch": 2.2079072937968642,
      "grad_norm": 19.266794591612474,
      "learning_rate": 1.9793848676257654e-06,
      "loss": 1.037,
      "step": 6478
    },
    {
      "epoch": 2.2082481254260395,
      "grad_norm": 12.154146021072064,
      "learning_rate": 1.9778048078614652e-06,
      "loss": 1.3016,
      "step": 6479
    },
    {
      "epoch": 2.208588957055215,
      "grad_norm": 15.957899876355057,
      "learning_rate": 1.976225223500499e-06,
      "loss": 0.8797,
      "step": 6480
    },
    {
      "epoch": 2.2089297886843897,
      "grad_norm": 15.520924038882065,
      "learning_rate": 1.9746461147913425e-06,
      "loss": 1.3537,
      "step": 6481
    },
    {
      "epoch": 2.209270620313565,
      "grad_norm": 12.469371318419638,
      "learning_rate": 1.9730674819823955e-06,
      "loss": 1.3363,
      "step": 6482
    },
    {
      "epoch": 2.2096114519427403,
      "grad_norm": 16.067927289497664,
      "learning_rate": 1.9714893253219846e-06,
      "loss": 1.1516,
      "step": 6483
    },
    {
      "epoch": 2.2099522835719156,
      "grad_norm": 37.46801090564371,
      "learning_rate": 1.9699116450583557e-06,
      "loss": 1.944,
      "step": 6484
    },
    {
      "epoch": 2.2102931152010905,
      "grad_norm": 19.95398404261129,
      "learning_rate": 1.96833444143969e-06,
      "loss": 2.3612,
      "step": 6485
    },
    {
      "epoch": 2.210633946830266,
      "grad_norm": 12.940627792476441,
      "learning_rate": 1.966757714714083e-06,
      "loss": 1.3862,
      "step": 6486
    },
    {
      "epoch": 2.210974778459441,
      "grad_norm": 40.64430524180224,
      "learning_rate": 1.9651814651295607e-06,
      "loss": 1.5898,
      "step": 6487
    },
    {
      "epoch": 2.211315610088616,
      "grad_norm": 9.458303138233253,
      "learning_rate": 1.9636056929340787e-06,
      "loss": 1.263,
      "step": 6488
    },
    {
      "epoch": 2.2116564417177913,
      "grad_norm": 15.711740164787109,
      "learning_rate": 1.962030398375506e-06,
      "loss": 1.724,
      "step": 6489
    },
    {
      "epoch": 2.2119972733469666,
      "grad_norm": 13.718245982110663,
      "learning_rate": 1.960455581701646e-06,
      "loss": 1.4039,
      "step": 6490
    },
    {
      "epoch": 2.212338104976142,
      "grad_norm": 14.651702908462436,
      "learning_rate": 1.9588812431602216e-06,
      "loss": 1.7011,
      "step": 6491
    },
    {
      "epoch": 2.212678936605317,
      "grad_norm": 18.66852543012691,
      "learning_rate": 1.957307382998884e-06,
      "loss": 1.4043,
      "step": 6492
    },
    {
      "epoch": 2.213019768234492,
      "grad_norm": 15.478193013825067,
      "learning_rate": 1.9557340014652076e-06,
      "loss": 1.7288,
      "step": 6493
    },
    {
      "epoch": 2.2133605998636674,
      "grad_norm": 12.569206646060323,
      "learning_rate": 1.9541610988066913e-06,
      "loss": 1.249,
      "step": 6494
    },
    {
      "epoch": 2.2137014314928427,
      "grad_norm": 16.626633553507556,
      "learning_rate": 1.952588675270759e-06,
      "loss": 1.1383,
      "step": 6495
    },
    {
      "epoch": 2.2140422631220176,
      "grad_norm": 17.970293185044994,
      "learning_rate": 1.9510167311047588e-06,
      "loss": 1.6964,
      "step": 6496
    },
    {
      "epoch": 2.214383094751193,
      "grad_norm": 12.073703458347495,
      "learning_rate": 1.9494452665559645e-06,
      "loss": 1.1157,
      "step": 6497
    },
    {
      "epoch": 2.214723926380368,
      "grad_norm": 27.137770780891103,
      "learning_rate": 1.9478742818715747e-06,
      "loss": 1.045,
      "step": 6498
    },
    {
      "epoch": 2.215064758009543,
      "grad_norm": 14.669512390597491,
      "learning_rate": 1.9463037772987065e-06,
      "loss": 1.5316,
      "step": 6499
    },
    {
      "epoch": 2.2154055896387184,
      "grad_norm": 13.199730797803864,
      "learning_rate": 1.9447337530844125e-06,
      "loss": 1.221,
      "step": 6500
    },
    {
      "epoch": 2.2157464212678937,
      "grad_norm": 12.447744324948069,
      "learning_rate": 1.943164209475663e-06,
      "loss": 1.5117,
      "step": 6501
    },
    {
      "epoch": 2.216087252897069,
      "grad_norm": 12.97123653004605,
      "learning_rate": 1.9415951467193484e-06,
      "loss": 1.3554,
      "step": 6502
    },
    {
      "epoch": 2.216428084526244,
      "grad_norm": 12.567080263412016,
      "learning_rate": 1.9400265650622952e-06,
      "loss": 1.3284,
      "step": 6503
    },
    {
      "epoch": 2.216768916155419,
      "grad_norm": 12.790705954569662,
      "learning_rate": 1.9384584647512427e-06,
      "loss": 1.3197,
      "step": 6504
    },
    {
      "epoch": 2.2171097477845945,
      "grad_norm": 29.907757798630694,
      "learning_rate": 1.9368908460328595e-06,
      "loss": 1.6387,
      "step": 6505
    },
    {
      "epoch": 2.2174505794137698,
      "grad_norm": 13.035993807296924,
      "learning_rate": 1.9353237091537435e-06,
      "loss": 1.4484,
      "step": 6506
    },
    {
      "epoch": 2.2177914110429446,
      "grad_norm": 16.373808456331588,
      "learning_rate": 1.9337570543604056e-06,
      "loss": 1.5215,
      "step": 6507
    },
    {
      "epoch": 2.21813224267212,
      "grad_norm": 18.245489748991968,
      "learning_rate": 1.932190881899289e-06,
      "loss": 1.4155,
      "step": 6508
    },
    {
      "epoch": 2.2184730743012953,
      "grad_norm": 14.717471221205411,
      "learning_rate": 1.930625192016759e-06,
      "loss": 1.0236,
      "step": 6509
    },
    {
      "epoch": 2.21881390593047,
      "grad_norm": 11.515759628138111,
      "learning_rate": 1.929059984959105e-06,
      "loss": 1.2485,
      "step": 6510
    },
    {
      "epoch": 2.2191547375596454,
      "grad_norm": 15.360978071685393,
      "learning_rate": 1.92749526097254e-06,
      "loss": 1.2809,
      "step": 6511
    },
    {
      "epoch": 2.2194955691888207,
      "grad_norm": 11.409868419742354,
      "learning_rate": 1.9259310203032017e-06,
      "loss": 1.2296,
      "step": 6512
    },
    {
      "epoch": 2.219836400817996,
      "grad_norm": 10.987449250183968,
      "learning_rate": 1.9243672631971504e-06,
      "loss": 1.1509,
      "step": 6513
    },
    {
      "epoch": 2.220177232447171,
      "grad_norm": 17.431247476551846,
      "learning_rate": 1.9228039899003727e-06,
      "loss": 1.6348,
      "step": 6514
    },
    {
      "epoch": 2.220518064076346,
      "grad_norm": 12.338911222271015,
      "learning_rate": 1.921241200658776e-06,
      "loss": 1.5794,
      "step": 6515
    },
    {
      "epoch": 2.2208588957055215,
      "grad_norm": 14.532102731099304,
      "learning_rate": 1.919678895718195e-06,
      "loss": 1.1922,
      "step": 6516
    },
    {
      "epoch": 2.221199727334697,
      "grad_norm": 15.597162273870454,
      "learning_rate": 1.918117075324388e-06,
      "loss": 1.4518,
      "step": 6517
    },
    {
      "epoch": 2.2215405589638717,
      "grad_norm": 17.961680704497134,
      "learning_rate": 1.916555739723029e-06,
      "loss": 1.5674,
      "step": 6518
    },
    {
      "epoch": 2.221881390593047,
      "grad_norm": 16.55423249143276,
      "learning_rate": 1.91499488915973e-06,
      "loss": 1.2792,
      "step": 6519
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 15.23481547948054,
      "learning_rate": 1.913434523880014e-06,
      "loss": 1.3796,
      "step": 6520
    },
    {
      "epoch": 2.222563053851397,
      "grad_norm": 20.475637293451495,
      "learning_rate": 1.911874644129333e-06,
      "loss": 1.8474,
      "step": 6521
    },
    {
      "epoch": 2.2229038854805725,
      "grad_norm": 15.583183737449264,
      "learning_rate": 1.910315250153067e-06,
      "loss": 1.8963,
      "step": 6522
    },
    {
      "epoch": 2.223244717109748,
      "grad_norm": 14.575491941897035,
      "learning_rate": 1.9087563421965096e-06,
      "loss": 1.6655,
      "step": 6523
    },
    {
      "epoch": 2.223585548738923,
      "grad_norm": 11.047610225651493,
      "learning_rate": 1.9071979205048846e-06,
      "loss": 1.3768,
      "step": 6524
    },
    {
      "epoch": 2.223926380368098,
      "grad_norm": 14.601209344730876,
      "learning_rate": 1.9056399853233393e-06,
      "loss": 1.754,
      "step": 6525
    },
    {
      "epoch": 2.2242672119972733,
      "grad_norm": 22.96785851678997,
      "learning_rate": 1.9040825368969417e-06,
      "loss": 1.9469,
      "step": 6526
    },
    {
      "epoch": 2.2246080436264486,
      "grad_norm": 15.343330816281096,
      "learning_rate": 1.9025255754706851e-06,
      "loss": 1.8913,
      "step": 6527
    },
    {
      "epoch": 2.224948875255624,
      "grad_norm": 10.454000577564434,
      "learning_rate": 1.900969101289486e-06,
      "loss": 1.0823,
      "step": 6528
    },
    {
      "epoch": 2.2252897068847988,
      "grad_norm": 20.491338080746253,
      "learning_rate": 1.8994131145981843e-06,
      "loss": 1.3297,
      "step": 6529
    },
    {
      "epoch": 2.225630538513974,
      "grad_norm": 15.052290501737469,
      "learning_rate": 1.8978576156415419e-06,
      "loss": 1.3864,
      "step": 6530
    },
    {
      "epoch": 2.2259713701431494,
      "grad_norm": 40.095346596565456,
      "learning_rate": 1.8963026046642452e-06,
      "loss": 1.7226,
      "step": 6531
    },
    {
      "epoch": 2.2263122017723243,
      "grad_norm": 12.167137513195536,
      "learning_rate": 1.894748081910906e-06,
      "loss": 1.6773,
      "step": 6532
    },
    {
      "epoch": 2.2266530334014996,
      "grad_norm": 16.39952733244793,
      "learning_rate": 1.8931940476260512e-06,
      "loss": 1.7749,
      "step": 6533
    },
    {
      "epoch": 2.226993865030675,
      "grad_norm": 19.734943475469816,
      "learning_rate": 1.8916405020541417e-06,
      "loss": 0.9451,
      "step": 6534
    },
    {
      "epoch": 2.22733469665985,
      "grad_norm": 19.70249466153235,
      "learning_rate": 1.8900874454395574e-06,
      "loss": 1.63,
      "step": 6535
    },
    {
      "epoch": 2.227675528289025,
      "grad_norm": 14.881042185130394,
      "learning_rate": 1.8885348780265933e-06,
      "loss": 1.5483,
      "step": 6536
    },
    {
      "epoch": 2.2280163599182004,
      "grad_norm": 19.665293350715118,
      "learning_rate": 1.8869828000594814e-06,
      "loss": 1.6998,
      "step": 6537
    },
    {
      "epoch": 2.2283571915473757,
      "grad_norm": 14.381097723023993,
      "learning_rate": 1.8854312117823686e-06,
      "loss": 1.4899,
      "step": 6538
    },
    {
      "epoch": 2.228698023176551,
      "grad_norm": 19.217314749113985,
      "learning_rate": 1.8838801134393215e-06,
      "loss": 1.396,
      "step": 6539
    },
    {
      "epoch": 2.229038854805726,
      "grad_norm": 15.885545051289213,
      "learning_rate": 1.8823295052743402e-06,
      "loss": 1.9355,
      "step": 6540
    },
    {
      "epoch": 2.229379686434901,
      "grad_norm": 24.510706743564164,
      "learning_rate": 1.880779387531337e-06,
      "loss": 1.3089,
      "step": 6541
    },
    {
      "epoch": 2.2297205180640765,
      "grad_norm": 12.815977624665244,
      "learning_rate": 1.879229760454151e-06,
      "loss": 1.679,
      "step": 6542
    },
    {
      "epoch": 2.2300613496932513,
      "grad_norm": 13.14758397161614,
      "learning_rate": 1.8776806242865502e-06,
      "loss": 1.4149,
      "step": 6543
    },
    {
      "epoch": 2.2304021813224266,
      "grad_norm": 14.047469616234547,
      "learning_rate": 1.876131979272215e-06,
      "loss": 1.3565,
      "step": 6544
    },
    {
      "epoch": 2.230743012951602,
      "grad_norm": 14.068351636930561,
      "learning_rate": 1.8745838256547549e-06,
      "loss": 1.3718,
      "step": 6545
    },
    {
      "epoch": 2.2310838445807772,
      "grad_norm": 17.038843269308057,
      "learning_rate": 1.8730361636777e-06,
      "loss": 1.4993,
      "step": 6546
    },
    {
      "epoch": 2.231424676209952,
      "grad_norm": 13.579036108938267,
      "learning_rate": 1.8714889935845048e-06,
      "loss": 1.5445,
      "step": 6547
    },
    {
      "epoch": 2.2317655078391274,
      "grad_norm": 12.888503285827086,
      "learning_rate": 1.8699423156185464e-06,
      "loss": 1.2871,
      "step": 6548
    },
    {
      "epoch": 2.2321063394683027,
      "grad_norm": 18.128745650884326,
      "learning_rate": 1.8683961300231184e-06,
      "loss": 2.1477,
      "step": 6549
    },
    {
      "epoch": 2.232447171097478,
      "grad_norm": 20.25138243744824,
      "learning_rate": 1.866850437041447e-06,
      "loss": 1.5088,
      "step": 6550
    },
    {
      "epoch": 2.232788002726653,
      "grad_norm": 16.106754373829624,
      "learning_rate": 1.8653052369166763e-06,
      "loss": 1.6371,
      "step": 6551
    },
    {
      "epoch": 2.233128834355828,
      "grad_norm": 15.932182798514813,
      "learning_rate": 1.8637605298918665e-06,
      "loss": 1.7898,
      "step": 6552
    },
    {
      "epoch": 2.2334696659850035,
      "grad_norm": 17.33878609967074,
      "learning_rate": 1.8622163162100132e-06,
      "loss": 1.5606,
      "step": 6553
    },
    {
      "epoch": 2.2338104976141784,
      "grad_norm": 23.91904427278865,
      "learning_rate": 1.8606725961140226e-06,
      "loss": 1.1589,
      "step": 6554
    },
    {
      "epoch": 2.2341513292433537,
      "grad_norm": 15.179722079838037,
      "learning_rate": 1.8591293698467283e-06,
      "loss": 1.672,
      "step": 6555
    },
    {
      "epoch": 2.234492160872529,
      "grad_norm": 14.915975593618938,
      "learning_rate": 1.85758663765089e-06,
      "loss": 0.867,
      "step": 6556
    },
    {
      "epoch": 2.2348329925017043,
      "grad_norm": 25.860759379754995,
      "learning_rate": 1.8560443997691812e-06,
      "loss": 1.7142,
      "step": 6557
    },
    {
      "epoch": 2.235173824130879,
      "grad_norm": 16.98513394376174,
      "learning_rate": 1.8545026564442043e-06,
      "loss": 1.6397,
      "step": 6558
    },
    {
      "epoch": 2.2355146557600545,
      "grad_norm": 21.24468394230733,
      "learning_rate": 1.8529614079184803e-06,
      "loss": 1.7228,
      "step": 6559
    },
    {
      "epoch": 2.23585548738923,
      "grad_norm": 36.407818436233065,
      "learning_rate": 1.8514206544344548e-06,
      "loss": 1.3438,
      "step": 6560
    },
    {
      "epoch": 2.236196319018405,
      "grad_norm": 15.007186064432238,
      "learning_rate": 1.8498803962344942e-06,
      "loss": 1.9534,
      "step": 6561
    },
    {
      "epoch": 2.23653715064758,
      "grad_norm": 26.70383542113246,
      "learning_rate": 1.8483406335608873e-06,
      "loss": 1.6615,
      "step": 6562
    },
    {
      "epoch": 2.2368779822767553,
      "grad_norm": 19.325200850443554,
      "learning_rate": 1.8468013666558444e-06,
      "loss": 1.2694,
      "step": 6563
    },
    {
      "epoch": 2.2372188139059306,
      "grad_norm": 17.02716933084877,
      "learning_rate": 1.845262595761499e-06,
      "loss": 1.7997,
      "step": 6564
    },
    {
      "epoch": 2.2375596455351054,
      "grad_norm": 14.225012667572482,
      "learning_rate": 1.8437243211199052e-06,
      "loss": 1.4447,
      "step": 6565
    },
    {
      "epoch": 2.2379004771642808,
      "grad_norm": 7.047782923070132,
      "learning_rate": 1.8421865429730401e-06,
      "loss": 1.1337,
      "step": 6566
    },
    {
      "epoch": 2.238241308793456,
      "grad_norm": 17.625283390579007,
      "learning_rate": 1.840649261562803e-06,
      "loss": 1.9038,
      "step": 6567
    },
    {
      "epoch": 2.2385821404226314,
      "grad_norm": 13.983312899163542,
      "learning_rate": 1.839112477131013e-06,
      "loss": 1.2028,
      "step": 6568
    },
    {
      "epoch": 2.2389229720518062,
      "grad_norm": 24.86049864507031,
      "learning_rate": 1.837576189919415e-06,
      "loss": 1.3069,
      "step": 6569
    },
    {
      "epoch": 2.2392638036809815,
      "grad_norm": 12.261633560768544,
      "learning_rate": 1.836040400169668e-06,
      "loss": 1.4366,
      "step": 6570
    },
    {
      "epoch": 2.239604635310157,
      "grad_norm": 13.910338472745188,
      "learning_rate": 1.8345051081233622e-06,
      "loss": 0.9717,
      "step": 6571
    },
    {
      "epoch": 2.239945466939332,
      "grad_norm": 13.208122956180627,
      "learning_rate": 1.8329703140220063e-06,
      "loss": 0.9903,
      "step": 6572
    },
    {
      "epoch": 2.240286298568507,
      "grad_norm": 9.928422635037355,
      "learning_rate": 1.8314360181070234e-06,
      "loss": 1.3434,
      "step": 6573
    },
    {
      "epoch": 2.2406271301976823,
      "grad_norm": 19.232732848435422,
      "learning_rate": 1.829902220619772e-06,
      "loss": 1.4033,
      "step": 6574
    },
    {
      "epoch": 2.2409679618268576,
      "grad_norm": 11.01618517275507,
      "learning_rate": 1.8283689218015189e-06,
      "loss": 1.5363,
      "step": 6575
    },
    {
      "epoch": 2.2413087934560325,
      "grad_norm": 21.173436992937308,
      "learning_rate": 1.8268361218934583e-06,
      "loss": 1.3145,
      "step": 6576
    },
    {
      "epoch": 2.241649625085208,
      "grad_norm": 13.033733016733667,
      "learning_rate": 1.8253038211367102e-06,
      "loss": 1.4892,
      "step": 6577
    },
    {
      "epoch": 2.241990456714383,
      "grad_norm": 16.235827901469136,
      "learning_rate": 1.8237720197723075e-06,
      "loss": 1.9435,
      "step": 6578
    },
    {
      "epoch": 2.2423312883435584,
      "grad_norm": 16.28078392080204,
      "learning_rate": 1.8222407180412098e-06,
      "loss": 1.4583,
      "step": 6579
    },
    {
      "epoch": 2.2426721199727333,
      "grad_norm": 16.103211339345968,
      "learning_rate": 1.8207099161842972e-06,
      "loss": 1.5292,
      "step": 6580
    },
    {
      "epoch": 2.2430129516019086,
      "grad_norm": 17.64713186612963,
      "learning_rate": 1.8191796144423713e-06,
      "loss": 1.7809,
      "step": 6581
    },
    {
      "epoch": 2.243353783231084,
      "grad_norm": 14.506197797784152,
      "learning_rate": 1.8176498130561548e-06,
      "loss": 1.7332,
      "step": 6582
    },
    {
      "epoch": 2.2436946148602592,
      "grad_norm": 15.272077379944726,
      "learning_rate": 1.8161205122662885e-06,
      "loss": 1.476,
      "step": 6583
    },
    {
      "epoch": 2.244035446489434,
      "grad_norm": 15.665502285474176,
      "learning_rate": 1.8145917123133412e-06,
      "loss": 1.3227,
      "step": 6584
    },
    {
      "epoch": 2.2443762781186094,
      "grad_norm": 28.239980333793717,
      "learning_rate": 1.8130634134377994e-06,
      "loss": 1.5959,
      "step": 6585
    },
    {
      "epoch": 2.2447171097477847,
      "grad_norm": 14.517057664258646,
      "learning_rate": 1.8115356158800656e-06,
      "loss": 1.1364,
      "step": 6586
    },
    {
      "epoch": 2.2450579413769596,
      "grad_norm": 17.33830381677913,
      "learning_rate": 1.8100083198804757e-06,
      "loss": 1.3752,
      "step": 6587
    },
    {
      "epoch": 2.245398773006135,
      "grad_norm": 15.909895509419824,
      "learning_rate": 1.8084815256792742e-06,
      "loss": 1.6855,
      "step": 6588
    },
    {
      "epoch": 2.24573960463531,
      "grad_norm": 10.783483606770533,
      "learning_rate": 1.8069552335166318e-06,
      "loss": 1.1569,
      "step": 6589
    },
    {
      "epoch": 2.2460804362644855,
      "grad_norm": 14.713672542833457,
      "learning_rate": 1.8054294436326448e-06,
      "loss": 1.4026,
      "step": 6590
    },
    {
      "epoch": 2.2464212678936604,
      "grad_norm": 9.863831494760193,
      "learning_rate": 1.8039041562673215e-06,
      "loss": 0.8533,
      "step": 6591
    },
    {
      "epoch": 2.2467620995228357,
      "grad_norm": 22.227051784671584,
      "learning_rate": 1.8023793716605975e-06,
      "loss": 1.7393,
      "step": 6592
    },
    {
      "epoch": 2.247102931152011,
      "grad_norm": 16.01354109961644,
      "learning_rate": 1.8008550900523275e-06,
      "loss": 1.1814,
      "step": 6593
    },
    {
      "epoch": 2.2474437627811863,
      "grad_norm": 15.882306907548562,
      "learning_rate": 1.799331311682287e-06,
      "loss": 1.238,
      "step": 6594
    },
    {
      "epoch": 2.247784594410361,
      "grad_norm": 20.23711473714515,
      "learning_rate": 1.7978080367901717e-06,
      "loss": 2.1513,
      "step": 6595
    },
    {
      "epoch": 2.2481254260395365,
      "grad_norm": 14.220889626831703,
      "learning_rate": 1.7962852656156e-06,
      "loss": 1.8809,
      "step": 6596
    },
    {
      "epoch": 2.2484662576687118,
      "grad_norm": 14.412125884071777,
      "learning_rate": 1.794762998398109e-06,
      "loss": 1.4927,
      "step": 6597
    },
    {
      "epoch": 2.2488070892978866,
      "grad_norm": 17.06475866533461,
      "learning_rate": 1.7932412353771577e-06,
      "loss": 1.7292,
      "step": 6598
    },
    {
      "epoch": 2.249147920927062,
      "grad_norm": 19.509645583208286,
      "learning_rate": 1.7917199767921257e-06,
      "loss": 1.6938,
      "step": 6599
    },
    {
      "epoch": 2.2494887525562373,
      "grad_norm": 17.719341765589444,
      "learning_rate": 1.7901992228823123e-06,
      "loss": 2.2548,
      "step": 6600
    },
    {
      "epoch": 2.2498295841854126,
      "grad_norm": 14.991332218256371,
      "learning_rate": 1.7886789738869392e-06,
      "loss": 1.4401,
      "step": 6601
    },
    {
      "epoch": 2.2501704158145874,
      "grad_norm": 14.178386495300728,
      "learning_rate": 1.7871592300451462e-06,
      "loss": 1.5224,
      "step": 6602
    },
    {
      "epoch": 2.2505112474437627,
      "grad_norm": 10.95613254841417,
      "learning_rate": 1.7856399915959982e-06,
      "loss": 1.0864,
      "step": 6603
    },
    {
      "epoch": 2.250852079072938,
      "grad_norm": 10.394440360046358,
      "learning_rate": 1.7841212587784718e-06,
      "loss": 1.0963,
      "step": 6604
    },
    {
      "epoch": 2.2511929107021134,
      "grad_norm": 16.544083143953596,
      "learning_rate": 1.7826030318314746e-06,
      "loss": 1.4097,
      "step": 6605
    },
    {
      "epoch": 2.2515337423312882,
      "grad_norm": 10.825776113319959,
      "learning_rate": 1.7810853109938297e-06,
      "loss": 1.1794,
      "step": 6606
    },
    {
      "epoch": 2.2518745739604635,
      "grad_norm": 12.60370604914609,
      "learning_rate": 1.7795680965042762e-06,
      "loss": 1.2495,
      "step": 6607
    },
    {
      "epoch": 2.252215405589639,
      "grad_norm": 14.325452739419799,
      "learning_rate": 1.7780513886014838e-06,
      "loss": 1.5262,
      "step": 6608
    },
    {
      "epoch": 2.2525562372188137,
      "grad_norm": 17.778945256073854,
      "learning_rate": 1.7765351875240323e-06,
      "loss": 1.266,
      "step": 6609
    },
    {
      "epoch": 2.252897068847989,
      "grad_norm": 14.059063508773713,
      "learning_rate": 1.7750194935104265e-06,
      "loss": 1.8387,
      "step": 6610
    },
    {
      "epoch": 2.2532379004771643,
      "grad_norm": 19.49041722117298,
      "learning_rate": 1.773504306799095e-06,
      "loss": 1.6624,
      "step": 6611
    },
    {
      "epoch": 2.2535787321063396,
      "grad_norm": 13.38073921183939,
      "learning_rate": 1.7719896276283783e-06,
      "loss": 1.7142,
      "step": 6612
    },
    {
      "epoch": 2.2539195637355145,
      "grad_norm": 25.388975490755232,
      "learning_rate": 1.7704754562365434e-06,
      "loss": 1.4283,
      "step": 6613
    },
    {
      "epoch": 2.25426039536469,
      "grad_norm": 9.888007882207221,
      "learning_rate": 1.7689617928617754e-06,
      "loss": 1.1461,
      "step": 6614
    },
    {
      "epoch": 2.254601226993865,
      "grad_norm": 17.36124321175439,
      "learning_rate": 1.767448637742179e-06,
      "loss": 1.4603,
      "step": 6615
    },
    {
      "epoch": 2.2549420586230404,
      "grad_norm": 18.59219004309801,
      "learning_rate": 1.765935991115782e-06,
      "loss": 1.3044,
      "step": 6616
    },
    {
      "epoch": 2.2552828902522153,
      "grad_norm": 12.923952579194598,
      "learning_rate": 1.7644238532205238e-06,
      "loss": 1.3854,
      "step": 6617
    },
    {
      "epoch": 2.2556237218813906,
      "grad_norm": 12.670191336817188,
      "learning_rate": 1.7629122242942753e-06,
      "loss": 1.0385,
      "step": 6618
    },
    {
      "epoch": 2.255964553510566,
      "grad_norm": 18.34483319293623,
      "learning_rate": 1.761401104574822e-06,
      "loss": 1.9168,
      "step": 6619
    },
    {
      "epoch": 2.2563053851397408,
      "grad_norm": 15.499594422280586,
      "learning_rate": 1.7598904942998635e-06,
      "loss": 1.9945,
      "step": 6620
    },
    {
      "epoch": 2.256646216768916,
      "grad_norm": 20.29388934460904,
      "learning_rate": 1.758380393707032e-06,
      "loss": 1.4382,
      "step": 6621
    },
    {
      "epoch": 2.2569870483980914,
      "grad_norm": 19.0871657280359,
      "learning_rate": 1.756870803033867e-06,
      "loss": 1.4828,
      "step": 6622
    },
    {
      "epoch": 2.2573278800272667,
      "grad_norm": 11.83987335304635,
      "learning_rate": 1.7553617225178332e-06,
      "loss": 1.3436,
      "step": 6623
    },
    {
      "epoch": 2.2576687116564416,
      "grad_norm": 12.069524963247996,
      "learning_rate": 1.7538531523963199e-06,
      "loss": 1.4179,
      "step": 6624
    },
    {
      "epoch": 2.258009543285617,
      "grad_norm": 19.396392579003372,
      "learning_rate": 1.752345092906626e-06,
      "loss": 1.3676,
      "step": 6625
    },
    {
      "epoch": 2.258350374914792,
      "grad_norm": 13.908115126923699,
      "learning_rate": 1.750837544285977e-06,
      "loss": 1.2024,
      "step": 6626
    },
    {
      "epoch": 2.2586912065439675,
      "grad_norm": 13.093756116760877,
      "learning_rate": 1.7493305067715166e-06,
      "loss": 1.0771,
      "step": 6627
    },
    {
      "epoch": 2.2590320381731424,
      "grad_norm": 31.810091407268335,
      "learning_rate": 1.7478239806003077e-06,
      "loss": 1.5335,
      "step": 6628
    },
    {
      "epoch": 2.2593728698023177,
      "grad_norm": 17.66247879927456,
      "learning_rate": 1.7463179660093333e-06,
      "loss": 1.3096,
      "step": 6629
    },
    {
      "epoch": 2.259713701431493,
      "grad_norm": 13.407469672024108,
      "learning_rate": 1.7448124632354946e-06,
      "loss": 1.648,
      "step": 6630
    },
    {
      "epoch": 2.260054533060668,
      "grad_norm": 18.933582044430942,
      "learning_rate": 1.7433074725156134e-06,
      "loss": 1.493,
      "step": 6631
    },
    {
      "epoch": 2.260395364689843,
      "grad_norm": 20.2546094639788,
      "learning_rate": 1.741802994086431e-06,
      "loss": 1.3441,
      "step": 6632
    },
    {
      "epoch": 2.2607361963190185,
      "grad_norm": 17.738129204316206,
      "learning_rate": 1.7402990281846082e-06,
      "loss": 1.4795,
      "step": 6633
    },
    {
      "epoch": 2.2610770279481938,
      "grad_norm": 18.28411105453398,
      "learning_rate": 1.738795575046724e-06,
      "loss": 1.6102,
      "step": 6634
    },
    {
      "epoch": 2.2614178595773686,
      "grad_norm": 15.90048643971666,
      "learning_rate": 1.7372926349092778e-06,
      "loss": 1.187,
      "step": 6635
    },
    {
      "epoch": 2.261758691206544,
      "grad_norm": 14.730672523376057,
      "learning_rate": 1.7357902080086887e-06,
      "loss": 1.727,
      "step": 6636
    },
    {
      "epoch": 2.2620995228357192,
      "grad_norm": 16.707669545418618,
      "learning_rate": 1.734288294581295e-06,
      "loss": 1.5973,
      "step": 6637
    },
    {
      "epoch": 2.2624403544648946,
      "grad_norm": 12.976342657717828,
      "learning_rate": 1.7327868948633498e-06,
      "loss": 1.9282,
      "step": 6638
    },
    {
      "epoch": 2.2627811860940694,
      "grad_norm": 14.922873273383766,
      "learning_rate": 1.7312860090910332e-06,
      "loss": 1.2642,
      "step": 6639
    },
    {
      "epoch": 2.2631220177232447,
      "grad_norm": 18.322585656307428,
      "learning_rate": 1.7297856375004417e-06,
      "loss": 1.7845,
      "step": 6640
    },
    {
      "epoch": 2.26346284935242,
      "grad_norm": 17.14213471619805,
      "learning_rate": 1.7282857803275838e-06,
      "loss": 1.1955,
      "step": 6641
    },
    {
      "epoch": 2.263803680981595,
      "grad_norm": 15.280505872823365,
      "learning_rate": 1.7267864378083997e-06,
      "loss": 1.6735,
      "step": 6642
    },
    {
      "epoch": 2.26414451261077,
      "grad_norm": 20.159906151953376,
      "learning_rate": 1.7252876101787375e-06,
      "loss": 1.2082,
      "step": 6643
    },
    {
      "epoch": 2.2644853442399455,
      "grad_norm": 17.355334166888507,
      "learning_rate": 1.7237892976743682e-06,
      "loss": 1.2618,
      "step": 6644
    },
    {
      "epoch": 2.264826175869121,
      "grad_norm": 19.509458069037358,
      "learning_rate": 1.7222915005309877e-06,
      "loss": 1.3894,
      "step": 6645
    },
    {
      "epoch": 2.2651670074982957,
      "grad_norm": 15.296974855845528,
      "learning_rate": 1.7207942189841997e-06,
      "loss": 1.7524,
      "step": 6646
    },
    {
      "epoch": 2.265507839127471,
      "grad_norm": 15.330207631276974,
      "learning_rate": 1.7192974532695356e-06,
      "loss": 1.7839,
      "step": 6647
    },
    {
      "epoch": 2.2658486707566463,
      "grad_norm": 17.290479792637278,
      "learning_rate": 1.7178012036224416e-06,
      "loss": 1.1248,
      "step": 6648
    },
    {
      "epoch": 2.2661895023858216,
      "grad_norm": 13.488514316487159,
      "learning_rate": 1.716305470278284e-06,
      "loss": 1.0104,
      "step": 6649
    },
    {
      "epoch": 2.2665303340149965,
      "grad_norm": 33.379068569676114,
      "learning_rate": 1.7148102534723472e-06,
      "loss": 1.375,
      "step": 6650
    },
    {
      "epoch": 2.266871165644172,
      "grad_norm": 12.027128486920105,
      "learning_rate": 1.7133155534398354e-06,
      "loss": 1.1939,
      "step": 6651
    },
    {
      "epoch": 2.267211997273347,
      "grad_norm": 20.085954482362204,
      "learning_rate": 1.7118213704158704e-06,
      "loss": 1.758,
      "step": 6652
    },
    {
      "epoch": 2.267552828902522,
      "grad_norm": 13.018821713303696,
      "learning_rate": 1.7103277046354954e-06,
      "loss": 1.3956,
      "step": 6653
    },
    {
      "epoch": 2.2678936605316973,
      "grad_norm": 17.10777902636449,
      "learning_rate": 1.708834556333664e-06,
      "loss": 1.8285,
      "step": 6654
    },
    {
      "epoch": 2.2682344921608726,
      "grad_norm": 19.25681890852201,
      "learning_rate": 1.7073419257452605e-06,
      "loss": 1.5933,
      "step": 6655
    },
    {
      "epoch": 2.268575323790048,
      "grad_norm": 11.56301599524918,
      "learning_rate": 1.7058498131050804e-06,
      "loss": 1.462,
      "step": 6656
    },
    {
      "epoch": 2.2689161554192228,
      "grad_norm": 14.982223901618772,
      "learning_rate": 1.7043582186478351e-06,
      "loss": 1.486,
      "step": 6657
    },
    {
      "epoch": 2.269256987048398,
      "grad_norm": 11.56712984215238,
      "learning_rate": 1.7028671426081644e-06,
      "loss": 1.4272,
      "step": 6658
    },
    {
      "epoch": 2.2695978186775734,
      "grad_norm": 34.295907249115274,
      "learning_rate": 1.701376585220616e-06,
      "loss": 1.8885,
      "step": 6659
    },
    {
      "epoch": 2.2699386503067487,
      "grad_norm": 17.652741349883513,
      "learning_rate": 1.6998865467196597e-06,
      "loss": 1.3708,
      "step": 6660
    },
    {
      "epoch": 2.2702794819359235,
      "grad_norm": 15.440533496657599,
      "learning_rate": 1.6983970273396899e-06,
      "loss": 1.2729,
      "step": 6661
    },
    {
      "epoch": 2.270620313565099,
      "grad_norm": 11.901069723071158,
      "learning_rate": 1.696908027315009e-06,
      "loss": 1.215,
      "step": 6662
    },
    {
      "epoch": 2.270961145194274,
      "grad_norm": 16.8461188692679,
      "learning_rate": 1.6954195468798445e-06,
      "loss": 2.243,
      "step": 6663
    },
    {
      "epoch": 2.271301976823449,
      "grad_norm": 13.292778116944755,
      "learning_rate": 1.6939315862683398e-06,
      "loss": 1.1515,
      "step": 6664
    },
    {
      "epoch": 2.2716428084526243,
      "grad_norm": 12.616848259758049,
      "learning_rate": 1.6924441457145575e-06,
      "loss": 1.6608,
      "step": 6665
    },
    {
      "epoch": 2.2719836400817996,
      "grad_norm": 14.479716321438318,
      "learning_rate": 1.6909572254524769e-06,
      "loss": 1.7715,
      "step": 6666
    },
    {
      "epoch": 2.272324471710975,
      "grad_norm": 15.442574184630516,
      "learning_rate": 1.689470825715998e-06,
      "loss": 1.406,
      "step": 6667
    },
    {
      "epoch": 2.27266530334015,
      "grad_norm": 15.697665983084818,
      "learning_rate": 1.687984946738937e-06,
      "loss": 1.5777,
      "step": 6668
    },
    {
      "epoch": 2.273006134969325,
      "grad_norm": 12.647653878023837,
      "learning_rate": 1.6864995887550279e-06,
      "loss": 1.4677,
      "step": 6669
    },
    {
      "epoch": 2.2733469665985004,
      "grad_norm": 15.59843716557701,
      "learning_rate": 1.6850147519979232e-06,
      "loss": 1.1386,
      "step": 6670
    },
    {
      "epoch": 2.2736877982276757,
      "grad_norm": 14.42697782532658,
      "learning_rate": 1.6835304367011967e-06,
      "loss": 1.8743,
      "step": 6671
    },
    {
      "epoch": 2.2740286298568506,
      "grad_norm": 11.104829899307529,
      "learning_rate": 1.6820466430983306e-06,
      "loss": 0.9995,
      "step": 6672
    },
    {
      "epoch": 2.274369461486026,
      "grad_norm": 12.512092358326173,
      "learning_rate": 1.6805633714227377e-06,
      "loss": 1.4562,
      "step": 6673
    },
    {
      "epoch": 2.2747102931152012,
      "grad_norm": 16.099335910393112,
      "learning_rate": 1.6790806219077422e-06,
      "loss": 1.1228,
      "step": 6674
    },
    {
      "epoch": 2.275051124744376,
      "grad_norm": 21.8918745121986,
      "learning_rate": 1.677598394786581e-06,
      "loss": 1.2609,
      "step": 6675
    },
    {
      "epoch": 2.2753919563735514,
      "grad_norm": 11.207601408741517,
      "learning_rate": 1.6761166902924215e-06,
      "loss": 1.0375,
      "step": 6676
    },
    {
      "epoch": 2.2757327880027267,
      "grad_norm": 8.926238408198289,
      "learning_rate": 1.6746355086583361e-06,
      "loss": 0.8715,
      "step": 6677
    },
    {
      "epoch": 2.276073619631902,
      "grad_norm": 10.252891477848543,
      "learning_rate": 1.6731548501173234e-06,
      "loss": 1.2149,
      "step": 6678
    },
    {
      "epoch": 2.276414451261077,
      "grad_norm": 13.268835709083957,
      "learning_rate": 1.6716747149022955e-06,
      "loss": 1.359,
      "step": 6679
    },
    {
      "epoch": 2.276755282890252,
      "grad_norm": 16.22521325882433,
      "learning_rate": 1.670195103246085e-06,
      "loss": 1.9143,
      "step": 6680
    },
    {
      "epoch": 2.2770961145194275,
      "grad_norm": 15.420220716224126,
      "learning_rate": 1.668716015381439e-06,
      "loss": 1.5475,
      "step": 6681
    },
    {
      "epoch": 2.277436946148603,
      "grad_norm": 26.415790764445475,
      "learning_rate": 1.6672374515410255e-06,
      "loss": 2.1361,
      "step": 6682
    },
    {
      "epoch": 2.2777777777777777,
      "grad_norm": 16.244785446506842,
      "learning_rate": 1.6657594119574276e-06,
      "loss": 1.4185,
      "step": 6683
    },
    {
      "epoch": 2.278118609406953,
      "grad_norm": 18.315055447400223,
      "learning_rate": 1.6642818968631474e-06,
      "loss": 1.5227,
      "step": 6684
    },
    {
      "epoch": 2.2784594410361283,
      "grad_norm": 16.218923301598036,
      "learning_rate": 1.6628049064906038e-06,
      "loss": 1.5621,
      "step": 6685
    },
    {
      "epoch": 2.278800272665303,
      "grad_norm": 17.978990255624762,
      "learning_rate": 1.661328441072132e-06,
      "loss": 1.4604,
      "step": 6686
    },
    {
      "epoch": 2.2791411042944785,
      "grad_norm": 20.290659411502354,
      "learning_rate": 1.659852500839989e-06,
      "loss": 1.356,
      "step": 6687
    },
    {
      "epoch": 2.279481935923654,
      "grad_norm": 13.260005176860616,
      "learning_rate": 1.658377086026341e-06,
      "loss": 1.3363,
      "step": 6688
    },
    {
      "epoch": 2.279822767552829,
      "grad_norm": 43.39425193294985,
      "learning_rate": 1.6569021968632808e-06,
      "loss": 1.3586,
      "step": 6689
    },
    {
      "epoch": 2.280163599182004,
      "grad_norm": 13.335758018778543,
      "learning_rate": 1.6554278335828155e-06,
      "loss": 1.0115,
      "step": 6690
    },
    {
      "epoch": 2.2805044308111793,
      "grad_norm": 19.138463071820496,
      "learning_rate": 1.6539539964168621e-06,
      "loss": 1.7568,
      "step": 6691
    },
    {
      "epoch": 2.2808452624403546,
      "grad_norm": 11.056214692515844,
      "learning_rate": 1.6524806855972681e-06,
      "loss": 1.3107,
      "step": 6692
    },
    {
      "epoch": 2.28118609406953,
      "grad_norm": 15.739803043005855,
      "learning_rate": 1.6510079013557868e-06,
      "loss": 1.277,
      "step": 6693
    },
    {
      "epoch": 2.2815269256987047,
      "grad_norm": 21.35635990038058,
      "learning_rate": 1.6495356439240916e-06,
      "loss": 1.4902,
      "step": 6694
    },
    {
      "epoch": 2.28186775732788,
      "grad_norm": 11.295348047793405,
      "learning_rate": 1.6480639135337811e-06,
      "loss": 1.3739,
      "step": 6695
    },
    {
      "epoch": 2.2822085889570554,
      "grad_norm": 13.877860193643915,
      "learning_rate": 1.6465927104163587e-06,
      "loss": 1.2055,
      "step": 6696
    },
    {
      "epoch": 2.2825494205862302,
      "grad_norm": 24.619563090479133,
      "learning_rate": 1.6451220348032515e-06,
      "loss": 0.9322,
      "step": 6697
    },
    {
      "epoch": 2.2828902522154055,
      "grad_norm": 14.034511974241067,
      "learning_rate": 1.6436518869258038e-06,
      "loss": 1.1877,
      "step": 6698
    },
    {
      "epoch": 2.283231083844581,
      "grad_norm": 17.104662749577862,
      "learning_rate": 1.6421822670152748e-06,
      "loss": 1.1943,
      "step": 6699
    },
    {
      "epoch": 2.283571915473756,
      "grad_norm": 12.98046030020724,
      "learning_rate": 1.640713175302842e-06,
      "loss": 1.4216,
      "step": 6700
    },
    {
      "epoch": 2.283912747102931,
      "grad_norm": 18.621795132695226,
      "learning_rate": 1.6392446120195992e-06,
      "loss": 2.0926,
      "step": 6701
    },
    {
      "epoch": 2.2842535787321063,
      "grad_norm": 15.643808761429078,
      "learning_rate": 1.6377765773965576e-06,
      "loss": 1.3443,
      "step": 6702
    },
    {
      "epoch": 2.2845944103612816,
      "grad_norm": 21.75848646805825,
      "learning_rate": 1.6363090716646446e-06,
      "loss": 1.7728,
      "step": 6703
    },
    {
      "epoch": 2.284935241990457,
      "grad_norm": 13.740036653053219,
      "learning_rate": 1.6348420950547045e-06,
      "loss": 1.2649,
      "step": 6704
    },
    {
      "epoch": 2.285276073619632,
      "grad_norm": 19.76421990587547,
      "learning_rate": 1.6333756477975005e-06,
      "loss": 1.7383,
      "step": 6705
    },
    {
      "epoch": 2.285616905248807,
      "grad_norm": 8.556965950946745,
      "learning_rate": 1.6319097301237053e-06,
      "loss": 1.0391,
      "step": 6706
    },
    {
      "epoch": 2.2859577368779824,
      "grad_norm": 20.169521973952364,
      "learning_rate": 1.6304443422639194e-06,
      "loss": 1.6594,
      "step": 6707
    },
    {
      "epoch": 2.2862985685071573,
      "grad_norm": 17.50467015880147,
      "learning_rate": 1.628979484448653e-06,
      "loss": 1.9248,
      "step": 6708
    },
    {
      "epoch": 2.2866394001363326,
      "grad_norm": 16.226646018426138,
      "learning_rate": 1.6275151569083298e-06,
      "loss": 1.1571,
      "step": 6709
    },
    {
      "epoch": 2.286980231765508,
      "grad_norm": 10.890602050018646,
      "learning_rate": 1.6260513598733007e-06,
      "loss": 1.3413,
      "step": 6710
    },
    {
      "epoch": 2.287321063394683,
      "grad_norm": 11.506208589985594,
      "learning_rate": 1.6245880935738218e-06,
      "loss": 0.9149,
      "step": 6711
    },
    {
      "epoch": 2.287661895023858,
      "grad_norm": 18.531380972145875,
      "learning_rate": 1.623125358240073e-06,
      "loss": 1.6056,
      "step": 6712
    },
    {
      "epoch": 2.2880027266530334,
      "grad_norm": 16.953052687159012,
      "learning_rate": 1.6216631541021478e-06,
      "loss": 1.2174,
      "step": 6713
    },
    {
      "epoch": 2.2883435582822087,
      "grad_norm": 11.078583096432386,
      "learning_rate": 1.6202014813900563e-06,
      "loss": 1.3785,
      "step": 6714
    },
    {
      "epoch": 2.288684389911384,
      "grad_norm": 15.156781262739901,
      "learning_rate": 1.6187403403337266e-06,
      "loss": 1.1882,
      "step": 6715
    },
    {
      "epoch": 2.289025221540559,
      "grad_norm": 9.172399671324946,
      "learning_rate": 1.6172797311630011e-06,
      "loss": 1.2359,
      "step": 6716
    },
    {
      "epoch": 2.289366053169734,
      "grad_norm": 14.491246737446339,
      "learning_rate": 1.6158196541076399e-06,
      "loss": 1.3682,
      "step": 6717
    },
    {
      "epoch": 2.2897068847989095,
      "grad_norm": 17.297751447598113,
      "learning_rate": 1.6143601093973188e-06,
      "loss": 1.5846,
      "step": 6718
    },
    {
      "epoch": 2.2900477164280844,
      "grad_norm": 15.08607700487343,
      "learning_rate": 1.61290109726163e-06,
      "loss": 1.2764,
      "step": 6719
    },
    {
      "epoch": 2.2903885480572597,
      "grad_norm": 18.83537165812134,
      "learning_rate": 1.6114426179300812e-06,
      "loss": 1.6266,
      "step": 6720
    },
    {
      "epoch": 2.290729379686435,
      "grad_norm": 15.558708202172674,
      "learning_rate": 1.6099846716320998e-06,
      "loss": 1.4862,
      "step": 6721
    },
    {
      "epoch": 2.29107021131561,
      "grad_norm": 15.715240984777392,
      "learning_rate": 1.6085272585970214e-06,
      "loss": 1.8158,
      "step": 6722
    },
    {
      "epoch": 2.291411042944785,
      "grad_norm": 16.909837827683425,
      "learning_rate": 1.607070379054107e-06,
      "loss": 1.0237,
      "step": 6723
    },
    {
      "epoch": 2.2917518745739605,
      "grad_norm": 14.641389965145565,
      "learning_rate": 1.6056140332325304e-06,
      "loss": 1.133,
      "step": 6724
    },
    {
      "epoch": 2.2920927062031358,
      "grad_norm": 11.336944624453302,
      "learning_rate": 1.6041582213613754e-06,
      "loss": 0.8207,
      "step": 6725
    },
    {
      "epoch": 2.292433537832311,
      "grad_norm": 17.98109384359319,
      "learning_rate": 1.6027029436696533e-06,
      "loss": 1.4761,
      "step": 6726
    },
    {
      "epoch": 2.292774369461486,
      "grad_norm": 13.767278833664237,
      "learning_rate": 1.6012482003862806e-06,
      "loss": 1.2306,
      "step": 6727
    },
    {
      "epoch": 2.2931152010906612,
      "grad_norm": 27.88181186676902,
      "learning_rate": 1.5997939917400934e-06,
      "loss": 1.7846,
      "step": 6728
    },
    {
      "epoch": 2.2934560327198366,
      "grad_norm": 14.68877615988441,
      "learning_rate": 1.5983403179598506e-06,
      "loss": 1.3233,
      "step": 6729
    },
    {
      "epoch": 2.2937968643490114,
      "grad_norm": 19.182574325203056,
      "learning_rate": 1.5968871792742151e-06,
      "loss": 1.5726,
      "step": 6730
    },
    {
      "epoch": 2.2941376959781867,
      "grad_norm": 13.594741447136839,
      "learning_rate": 1.595434575911774e-06,
      "loss": 1.4803,
      "step": 6731
    },
    {
      "epoch": 2.294478527607362,
      "grad_norm": 12.609263266448384,
      "learning_rate": 1.5939825081010262e-06,
      "loss": 1.5238,
      "step": 6732
    },
    {
      "epoch": 2.294819359236537,
      "grad_norm": 18.55880494169227,
      "learning_rate": 1.5925309760703884e-06,
      "loss": 1.9712,
      "step": 6733
    },
    {
      "epoch": 2.295160190865712,
      "grad_norm": 26.152546812458528,
      "learning_rate": 1.591079980048193e-06,
      "loss": 1.4101,
      "step": 6734
    },
    {
      "epoch": 2.2955010224948875,
      "grad_norm": 15.197016375398364,
      "learning_rate": 1.589629520262686e-06,
      "loss": 1.584,
      "step": 6735
    },
    {
      "epoch": 2.295841854124063,
      "grad_norm": 22.764280939933794,
      "learning_rate": 1.5881795969420321e-06,
      "loss": 1.0934,
      "step": 6736
    },
    {
      "epoch": 2.296182685753238,
      "grad_norm": 12.709266781976341,
      "learning_rate": 1.586730210314309e-06,
      "loss": 0.9672,
      "step": 6737
    },
    {
      "epoch": 2.296523517382413,
      "grad_norm": 17.70363499991965,
      "learning_rate": 1.5852813606075113e-06,
      "loss": 1.6462,
      "step": 6738
    },
    {
      "epoch": 2.2968643490115883,
      "grad_norm": 17.029598280211598,
      "learning_rate": 1.5838330480495506e-06,
      "loss": 1.5977,
      "step": 6739
    },
    {
      "epoch": 2.2972051806407636,
      "grad_norm": 14.362172763840741,
      "learning_rate": 1.5823852728682471e-06,
      "loss": 1.4167,
      "step": 6740
    },
    {
      "epoch": 2.2975460122699385,
      "grad_norm": 13.102356617283949,
      "learning_rate": 1.580938035291346e-06,
      "loss": 1.3382,
      "step": 6741
    },
    {
      "epoch": 2.297886843899114,
      "grad_norm": 13.961558625289065,
      "learning_rate": 1.5794913355465047e-06,
      "loss": 1.2359,
      "step": 6742
    },
    {
      "epoch": 2.298227675528289,
      "grad_norm": 15.458421689406428,
      "learning_rate": 1.5780451738612907e-06,
      "loss": 1.2786,
      "step": 6743
    },
    {
      "epoch": 2.298568507157464,
      "grad_norm": 13.10221699302413,
      "learning_rate": 1.576599550463191e-06,
      "loss": 1.0488,
      "step": 6744
    },
    {
      "epoch": 2.2989093387866393,
      "grad_norm": 13.143928694482318,
      "learning_rate": 1.5751544655796137e-06,
      "loss": 1.3512,
      "step": 6745
    },
    {
      "epoch": 2.2992501704158146,
      "grad_norm": 18.928704738872668,
      "learning_rate": 1.5737099194378702e-06,
      "loss": 1.116,
      "step": 6746
    },
    {
      "epoch": 2.29959100204499,
      "grad_norm": 14.24044128584685,
      "learning_rate": 1.5722659122651956e-06,
      "loss": 1.2011,
      "step": 6747
    },
    {
      "epoch": 2.299931833674165,
      "grad_norm": 10.557616471691759,
      "learning_rate": 1.5708224442887387e-06,
      "loss": 1.2577,
      "step": 6748
    },
    {
      "epoch": 2.30027266530334,
      "grad_norm": 16.001950385839248,
      "learning_rate": 1.5693795157355618e-06,
      "loss": 0.9955,
      "step": 6749
    },
    {
      "epoch": 2.3006134969325154,
      "grad_norm": 19.909435760417786,
      "learning_rate": 1.5679371268326432e-06,
      "loss": 1.6249,
      "step": 6750
    },
    {
      "epoch": 2.3009543285616907,
      "grad_norm": 13.000418764495407,
      "learning_rate": 1.566495277806877e-06,
      "loss": 1.4044,
      "step": 6751
    },
    {
      "epoch": 2.3012951601908656,
      "grad_norm": 25.75223467673309,
      "learning_rate": 1.5650539688850719e-06,
      "loss": 1.4686,
      "step": 6752
    },
    {
      "epoch": 2.301635991820041,
      "grad_norm": 14.421689608071082,
      "learning_rate": 1.563613200293951e-06,
      "loss": 1.5243,
      "step": 6753
    },
    {
      "epoch": 2.301976823449216,
      "grad_norm": 15.204563206067506,
      "learning_rate": 1.562172972260153e-06,
      "loss": 1.0579,
      "step": 6754
    },
    {
      "epoch": 2.302317655078391,
      "grad_norm": 12.473521589071993,
      "learning_rate": 1.5607332850102335e-06,
      "loss": 1.4003,
      "step": 6755
    },
    {
      "epoch": 2.3026584867075663,
      "grad_norm": 13.19118104319589,
      "learning_rate": 1.5592941387706562e-06,
      "loss": 1.1751,
      "step": 6756
    },
    {
      "epoch": 2.3029993183367417,
      "grad_norm": 12.571881670563181,
      "learning_rate": 1.5578555337678086e-06,
      "loss": 1.8917,
      "step": 6757
    },
    {
      "epoch": 2.303340149965917,
      "grad_norm": 22.57273094090433,
      "learning_rate": 1.55641747022799e-06,
      "loss": 1.6363,
      "step": 6758
    },
    {
      "epoch": 2.3036809815950923,
      "grad_norm": 16.290690123574187,
      "learning_rate": 1.554979948377408e-06,
      "loss": 1.7881,
      "step": 6759
    },
    {
      "epoch": 2.304021813224267,
      "grad_norm": 13.912896648605884,
      "learning_rate": 1.5535429684421972e-06,
      "loss": 1.4826,
      "step": 6760
    },
    {
      "epoch": 2.3043626448534424,
      "grad_norm": 15.339064915848784,
      "learning_rate": 1.5521065306483952e-06,
      "loss": 1.5438,
      "step": 6761
    },
    {
      "epoch": 2.3047034764826178,
      "grad_norm": 13.545603113541175,
      "learning_rate": 1.5506706352219596e-06,
      "loss": 1.3037,
      "step": 6762
    },
    {
      "epoch": 2.3050443081117926,
      "grad_norm": 16.712171561680368,
      "learning_rate": 1.5492352823887674e-06,
      "loss": 1.7138,
      "step": 6763
    },
    {
      "epoch": 2.305385139740968,
      "grad_norm": 14.68446960784123,
      "learning_rate": 1.5478004723746004e-06,
      "loss": 1.7869,
      "step": 6764
    },
    {
      "epoch": 2.3057259713701432,
      "grad_norm": 12.806497219527673,
      "learning_rate": 1.5463662054051615e-06,
      "loss": 1.6202,
      "step": 6765
    },
    {
      "epoch": 2.306066802999318,
      "grad_norm": 15.549733330456567,
      "learning_rate": 1.5449324817060668e-06,
      "loss": 1.0624,
      "step": 6766
    },
    {
      "epoch": 2.3064076346284934,
      "grad_norm": 19.36211847631616,
      "learning_rate": 1.5434993015028466e-06,
      "loss": 2.01,
      "step": 6767
    },
    {
      "epoch": 2.3067484662576687,
      "grad_norm": 12.523392246983564,
      "learning_rate": 1.5420666650209453e-06,
      "loss": 1.3782,
      "step": 6768
    },
    {
      "epoch": 2.307089297886844,
      "grad_norm": 16.163878744747123,
      "learning_rate": 1.5406345724857236e-06,
      "loss": 1.8176,
      "step": 6769
    },
    {
      "epoch": 2.3074301295160193,
      "grad_norm": 12.447565047430416,
      "learning_rate": 1.5392030241224536e-06,
      "loss": 1.2012,
      "step": 6770
    },
    {
      "epoch": 2.307770961145194,
      "grad_norm": 11.324011124473,
      "learning_rate": 1.537772020156325e-06,
      "loss": 1.366,
      "step": 6771
    },
    {
      "epoch": 2.3081117927743695,
      "grad_norm": 12.227371762724612,
      "learning_rate": 1.5363415608124405e-06,
      "loss": 1.3692,
      "step": 6772
    },
    {
      "epoch": 2.308452624403545,
      "grad_norm": 17.598700283900854,
      "learning_rate": 1.5349116463158158e-06,
      "loss": 1.2154,
      "step": 6773
    },
    {
      "epoch": 2.3087934560327197,
      "grad_norm": 11.21271415022091,
      "learning_rate": 1.5334822768913833e-06,
      "loss": 1.3843,
      "step": 6774
    },
    {
      "epoch": 2.309134287661895,
      "grad_norm": 17.25332266413702,
      "learning_rate": 1.5320534527639873e-06,
      "loss": 1.0401,
      "step": 6775
    },
    {
      "epoch": 2.3094751192910703,
      "grad_norm": 13.692125824880344,
      "learning_rate": 1.5306251741583906e-06,
      "loss": 0.9937,
      "step": 6776
    },
    {
      "epoch": 2.309815950920245,
      "grad_norm": 13.800168781399183,
      "learning_rate": 1.5291974412992633e-06,
      "loss": 1.1928,
      "step": 6777
    },
    {
      "epoch": 2.3101567825494205,
      "grad_norm": 29.11774387943874,
      "learning_rate": 1.5277702544111927e-06,
      "loss": 1.9215,
      "step": 6778
    },
    {
      "epoch": 2.310497614178596,
      "grad_norm": 18.52897973682685,
      "learning_rate": 1.5263436137186872e-06,
      "loss": 1.2699,
      "step": 6779
    },
    {
      "epoch": 2.310838445807771,
      "grad_norm": 24.542621840845126,
      "learning_rate": 1.524917519446157e-06,
      "loss": 1.8465,
      "step": 6780
    },
    {
      "epoch": 2.311179277436946,
      "grad_norm": 12.848609245222075,
      "learning_rate": 1.523491971817934e-06,
      "loss": 1.1834,
      "step": 6781
    },
    {
      "epoch": 2.3115201090661213,
      "grad_norm": 11.995451511725484,
      "learning_rate": 1.5220669710582642e-06,
      "loss": 1.6044,
      "step": 6782
    },
    {
      "epoch": 2.3118609406952966,
      "grad_norm": 55.26716984816829,
      "learning_rate": 1.5206425173913037e-06,
      "loss": 1.5489,
      "step": 6783
    },
    {
      "epoch": 2.312201772324472,
      "grad_norm": 18.84281376308303,
      "learning_rate": 1.519218611041126e-06,
      "loss": 2.3916,
      "step": 6784
    },
    {
      "epoch": 2.3125426039536467,
      "grad_norm": 22.526907953532003,
      "learning_rate": 1.5177952522317173e-06,
      "loss": 1.6027,
      "step": 6785
    },
    {
      "epoch": 2.312883435582822,
      "grad_norm": 12.210801080284314,
      "learning_rate": 1.5163724411869773e-06,
      "loss": 1.1535,
      "step": 6786
    },
    {
      "epoch": 2.3132242672119974,
      "grad_norm": 11.03956773306301,
      "learning_rate": 1.5149501781307196e-06,
      "loss": 1.1996,
      "step": 6787
    },
    {
      "epoch": 2.3135650988411722,
      "grad_norm": 12.280068973584374,
      "learning_rate": 1.513528463286672e-06,
      "loss": 0.9397,
      "step": 6788
    },
    {
      "epoch": 2.3139059304703475,
      "grad_norm": 12.184354773024825,
      "learning_rate": 1.5121072968784772e-06,
      "loss": 1.4346,
      "step": 6789
    },
    {
      "epoch": 2.314246762099523,
      "grad_norm": 10.482354269405946,
      "learning_rate": 1.5106866791296865e-06,
      "loss": 1.1159,
      "step": 6790
    },
    {
      "epoch": 2.314587593728698,
      "grad_norm": 14.365329746289467,
      "learning_rate": 1.5092666102637727e-06,
      "loss": 1.218,
      "step": 6791
    },
    {
      "epoch": 2.314928425357873,
      "grad_norm": 15.194421549103048,
      "learning_rate": 1.5078470905041188e-06,
      "loss": 1.4854,
      "step": 6792
    },
    {
      "epoch": 2.3152692569870483,
      "grad_norm": 11.277752994270068,
      "learning_rate": 1.5064281200740155e-06,
      "loss": 1.3574,
      "step": 6793
    },
    {
      "epoch": 2.3156100886162236,
      "grad_norm": 14.793994018875994,
      "learning_rate": 1.5050096991966788e-06,
      "loss": 1.542,
      "step": 6794
    },
    {
      "epoch": 2.315950920245399,
      "grad_norm": 20.977041337046984,
      "learning_rate": 1.5035918280952282e-06,
      "loss": 1.4081,
      "step": 6795
    },
    {
      "epoch": 2.316291751874574,
      "grad_norm": 15.262028527572815,
      "learning_rate": 1.5021745069927002e-06,
      "loss": 1.5739,
      "step": 6796
    },
    {
      "epoch": 2.316632583503749,
      "grad_norm": 20.095456587303214,
      "learning_rate": 1.5007577361120484e-06,
      "loss": 1.3203,
      "step": 6797
    },
    {
      "epoch": 2.3169734151329244,
      "grad_norm": 14.5962745453524,
      "learning_rate": 1.4993415156761338e-06,
      "loss": 1.2429,
      "step": 6798
    },
    {
      "epoch": 2.3173142467620993,
      "grad_norm": 20.0191007897799,
      "learning_rate": 1.4979258459077334e-06,
      "loss": 1.3649,
      "step": 6799
    },
    {
      "epoch": 2.3176550783912746,
      "grad_norm": 13.12857755394873,
      "learning_rate": 1.4965107270295386e-06,
      "loss": 1.3127,
      "step": 6800
    },
    {
      "epoch": 2.31799591002045,
      "grad_norm": 13.012164624762967,
      "learning_rate": 1.4950961592641534e-06,
      "loss": 1.1499,
      "step": 6801
    },
    {
      "epoch": 2.318336741649625,
      "grad_norm": 15.393046291799273,
      "learning_rate": 1.493682142834094e-06,
      "loss": 1.6572,
      "step": 6802
    },
    {
      "epoch": 2.3186775732788,
      "grad_norm": 13.372591011868023,
      "learning_rate": 1.4922686779617918e-06,
      "loss": 1.5224,
      "step": 6803
    },
    {
      "epoch": 2.3190184049079754,
      "grad_norm": 17.757662585223102,
      "learning_rate": 1.49085576486959e-06,
      "loss": 1.6141,
      "step": 6804
    },
    {
      "epoch": 2.3193592365371507,
      "grad_norm": 12.772824393122471,
      "learning_rate": 1.489443403779745e-06,
      "loss": 0.8032,
      "step": 6805
    },
    {
      "epoch": 2.319700068166326,
      "grad_norm": 33.988997575519754,
      "learning_rate": 1.4880315949144276e-06,
      "loss": 1.5481,
      "step": 6806
    },
    {
      "epoch": 2.320040899795501,
      "grad_norm": 14.875636352062903,
      "learning_rate": 1.4866203384957207e-06,
      "loss": 1.5062,
      "step": 6807
    },
    {
      "epoch": 2.320381731424676,
      "grad_norm": 13.390818562628388,
      "learning_rate": 1.485209634745622e-06,
      "loss": 1.6192,
      "step": 6808
    },
    {
      "epoch": 2.3207225630538515,
      "grad_norm": 41.30107680320918,
      "learning_rate": 1.4837994838860365e-06,
      "loss": 1.6463,
      "step": 6809
    },
    {
      "epoch": 2.3210633946830264,
      "grad_norm": 22.068219967284286,
      "learning_rate": 1.482389886138792e-06,
      "loss": 1.0337,
      "step": 6810
    },
    {
      "epoch": 2.3214042263122017,
      "grad_norm": 13.982614672816474,
      "learning_rate": 1.48098084172562e-06,
      "loss": 1.8987,
      "step": 6811
    },
    {
      "epoch": 2.321745057941377,
      "grad_norm": 10.0623864375454,
      "learning_rate": 1.4795723508681682e-06,
      "loss": 1.2738,
      "step": 6812
    },
    {
      "epoch": 2.3220858895705523,
      "grad_norm": 11.38963554332603,
      "learning_rate": 1.478164413788003e-06,
      "loss": 1.2018,
      "step": 6813
    },
    {
      "epoch": 2.322426721199727,
      "grad_norm": 11.930530353265775,
      "learning_rate": 1.4767570307065938e-06,
      "loss": 1.3074,
      "step": 6814
    },
    {
      "epoch": 2.3227675528289025,
      "grad_norm": 23.125528513683182,
      "learning_rate": 1.475350201845328e-06,
      "loss": 1.7024,
      "step": 6815
    },
    {
      "epoch": 2.3231083844580778,
      "grad_norm": 26.238006223764096,
      "learning_rate": 1.4739439274255068e-06,
      "loss": 1.8091,
      "step": 6816
    },
    {
      "epoch": 2.323449216087253,
      "grad_norm": 17.033226606559104,
      "learning_rate": 1.4725382076683425e-06,
      "loss": 0.9755,
      "step": 6817
    },
    {
      "epoch": 2.323790047716428,
      "grad_norm": 13.983427053771837,
      "learning_rate": 1.4711330427949599e-06,
      "loss": 1.4239,
      "step": 6818
    },
    {
      "epoch": 2.3241308793456033,
      "grad_norm": 15.581671620791772,
      "learning_rate": 1.4697284330263971e-06,
      "loss": 1.6632,
      "step": 6819
    },
    {
      "epoch": 2.3244717109747786,
      "grad_norm": 15.045577889819684,
      "learning_rate": 1.4683243785836054e-06,
      "loss": 1.2086,
      "step": 6820
    },
    {
      "epoch": 2.3248125426039534,
      "grad_norm": 17.243412289814625,
      "learning_rate": 1.4669208796874474e-06,
      "loss": 1.3519,
      "step": 6821
    },
    {
      "epoch": 2.3251533742331287,
      "grad_norm": 8.609527442118134,
      "learning_rate": 1.4655179365586992e-06,
      "loss": 0.6026,
      "step": 6822
    },
    {
      "epoch": 2.325494205862304,
      "grad_norm": 15.066509517337296,
      "learning_rate": 1.4641155494180515e-06,
      "loss": 1.527,
      "step": 6823
    },
    {
      "epoch": 2.3258350374914794,
      "grad_norm": 21.777522569912048,
      "learning_rate": 1.4627137184861001e-06,
      "loss": 1.4248,
      "step": 6824
    },
    {
      "epoch": 2.326175869120654,
      "grad_norm": 19.70366987392369,
      "learning_rate": 1.4613124439833638e-06,
      "loss": 1.2756,
      "step": 6825
    },
    {
      "epoch": 2.3265167007498295,
      "grad_norm": 17.246912507807966,
      "learning_rate": 1.4599117261302681e-06,
      "loss": 1.3245,
      "step": 6826
    },
    {
      "epoch": 2.326857532379005,
      "grad_norm": 21.28860945652932,
      "learning_rate": 1.4585115651471466e-06,
      "loss": 1.3544,
      "step": 6827
    },
    {
      "epoch": 2.32719836400818,
      "grad_norm": 15.516567097820449,
      "learning_rate": 1.4571119612542573e-06,
      "loss": 1.3145,
      "step": 6828
    },
    {
      "epoch": 2.327539195637355,
      "grad_norm": 16.59127823934273,
      "learning_rate": 1.4557129146717585e-06,
      "loss": 2.0661,
      "step": 6829
    },
    {
      "epoch": 2.3278800272665303,
      "grad_norm": 18.914424273424576,
      "learning_rate": 1.4543144256197256e-06,
      "loss": 1.627,
      "step": 6830
    },
    {
      "epoch": 2.3282208588957056,
      "grad_norm": 11.169128774583875,
      "learning_rate": 1.4529164943181507e-06,
      "loss": 1.1925,
      "step": 6831
    },
    {
      "epoch": 2.3285616905248805,
      "grad_norm": 20.187675431801246,
      "learning_rate": 1.4515191209869305e-06,
      "loss": 1.3517,
      "step": 6832
    },
    {
      "epoch": 2.328902522154056,
      "grad_norm": 15.001786464576563,
      "learning_rate": 1.450122305845878e-06,
      "loss": 1.4915,
      "step": 6833
    },
    {
      "epoch": 2.329243353783231,
      "grad_norm": 15.276839479291493,
      "learning_rate": 1.4487260491147176e-06,
      "loss": 1.9139,
      "step": 6834
    },
    {
      "epoch": 2.3295841854124064,
      "grad_norm": 11.347145850389566,
      "learning_rate": 1.447330351013087e-06,
      "loss": 1.1315,
      "step": 6835
    },
    {
      "epoch": 2.3299250170415813,
      "grad_norm": 11.696811142221314,
      "learning_rate": 1.4459352117605347e-06,
      "loss": 1.4671,
      "step": 6836
    },
    {
      "epoch": 2.3302658486707566,
      "grad_norm": 15.091661564004434,
      "learning_rate": 1.444540631576521e-06,
      "loss": 1.8422,
      "step": 6837
    },
    {
      "epoch": 2.330606680299932,
      "grad_norm": 15.46144754860721,
      "learning_rate": 1.4431466106804198e-06,
      "loss": 1.3101,
      "step": 6838
    },
    {
      "epoch": 2.330947511929107,
      "grad_norm": 16.2638448486805,
      "learning_rate": 1.441753149291517e-06,
      "loss": 1.7912,
      "step": 6839
    },
    {
      "epoch": 2.331288343558282,
      "grad_norm": 13.842879800782557,
      "learning_rate": 1.4403602476290057e-06,
      "loss": 1.7586,
      "step": 6840
    },
    {
      "epoch": 2.3316291751874574,
      "grad_norm": 19.063205168570175,
      "learning_rate": 1.4389679059119987e-06,
      "loss": 1.267,
      "step": 6841
    },
    {
      "epoch": 2.3319700068166327,
      "grad_norm": 13.108203981589552,
      "learning_rate": 1.437576124359517e-06,
      "loss": 1.5203,
      "step": 6842
    },
    {
      "epoch": 2.3323108384458076,
      "grad_norm": 16.580254721682167,
      "learning_rate": 1.4361849031904895e-06,
      "loss": 1.1453,
      "step": 6843
    },
    {
      "epoch": 2.332651670074983,
      "grad_norm": 7.981124102526594,
      "learning_rate": 1.4347942426237666e-06,
      "loss": 0.8505,
      "step": 6844
    },
    {
      "epoch": 2.332992501704158,
      "grad_norm": 15.907146028807505,
      "learning_rate": 1.4334041428781003e-06,
      "loss": 0.8349,
      "step": 6845
    },
    {
      "epoch": 2.3333333333333335,
      "grad_norm": 22.981256128552914,
      "learning_rate": 1.4320146041721582e-06,
      "loss": 1.7752,
      "step": 6846
    },
    {
      "epoch": 2.3336741649625083,
      "grad_norm": 12.705866619568669,
      "learning_rate": 1.430625626724525e-06,
      "loss": 1.4825,
      "step": 6847
    },
    {
      "epoch": 2.3340149965916837,
      "grad_norm": 27.287052703104372,
      "learning_rate": 1.4292372107536883e-06,
      "loss": 1.9613,
      "step": 6848
    },
    {
      "epoch": 2.334355828220859,
      "grad_norm": 13.575089740991263,
      "learning_rate": 1.4278493564780531e-06,
      "loss": 1.648,
      "step": 6849
    },
    {
      "epoch": 2.3346966598500343,
      "grad_norm": 11.25891730638185,
      "learning_rate": 1.4264620641159339e-06,
      "loss": 1.0767,
      "step": 6850
    },
    {
      "epoch": 2.335037491479209,
      "grad_norm": 21.078499383769493,
      "learning_rate": 1.4250753338855572e-06,
      "loss": 1.4757,
      "step": 6851
    },
    {
      "epoch": 2.3353783231083844,
      "grad_norm": 23.174498772220517,
      "learning_rate": 1.4236891660050611e-06,
      "loss": 1.1858,
      "step": 6852
    },
    {
      "epoch": 2.3357191547375598,
      "grad_norm": 13.833487829167005,
      "learning_rate": 1.4223035606924961e-06,
      "loss": 1.1578,
      "step": 6853
    },
    {
      "epoch": 2.3360599863667346,
      "grad_norm": 10.331434959499516,
      "learning_rate": 1.420918518165823e-06,
      "loss": 1.025,
      "step": 6854
    },
    {
      "epoch": 2.33640081799591,
      "grad_norm": 13.149668915446572,
      "learning_rate": 1.4195340386429147e-06,
      "loss": 1.3889,
      "step": 6855
    },
    {
      "epoch": 2.3367416496250852,
      "grad_norm": 16.515303595578484,
      "learning_rate": 1.4181501223415545e-06,
      "loss": 1.4766,
      "step": 6856
    },
    {
      "epoch": 2.3370824812542605,
      "grad_norm": 12.757883692990825,
      "learning_rate": 1.416766769479439e-06,
      "loss": 1.8699,
      "step": 6857
    },
    {
      "epoch": 2.3374233128834354,
      "grad_norm": 14.480414212771501,
      "learning_rate": 1.4153839802741748e-06,
      "loss": 1.1457,
      "step": 6858
    },
    {
      "epoch": 2.3377641445126107,
      "grad_norm": 14.463197201583133,
      "learning_rate": 1.41400175494328e-06,
      "loss": 1.2605,
      "step": 6859
    },
    {
      "epoch": 2.338104976141786,
      "grad_norm": 21.79265923707415,
      "learning_rate": 1.4126200937041857e-06,
      "loss": 1.7573,
      "step": 6860
    },
    {
      "epoch": 2.3384458077709613,
      "grad_norm": 15.01356567375351,
      "learning_rate": 1.4112389967742274e-06,
      "loss": 1.2835,
      "step": 6861
    },
    {
      "epoch": 2.338786639400136,
      "grad_norm": 16.61836092747709,
      "learning_rate": 1.409858464370663e-06,
      "loss": 1.6126,
      "step": 6862
    },
    {
      "epoch": 2.3391274710293115,
      "grad_norm": 24.894490874315537,
      "learning_rate": 1.4084784967106556e-06,
      "loss": 1.6772,
      "step": 6863
    },
    {
      "epoch": 2.339468302658487,
      "grad_norm": 19.882754407374478,
      "learning_rate": 1.4070990940112744e-06,
      "loss": 1.3097,
      "step": 6864
    },
    {
      "epoch": 2.3398091342876617,
      "grad_norm": 21.459429267608133,
      "learning_rate": 1.405720256489511e-06,
      "loss": 1.9409,
      "step": 6865
    },
    {
      "epoch": 2.340149965916837,
      "grad_norm": 35.071493629777265,
      "learning_rate": 1.404341984362258e-06,
      "loss": 1.2986,
      "step": 6866
    },
    {
      "epoch": 2.3404907975460123,
      "grad_norm": 24.971665664980037,
      "learning_rate": 1.4029642778463226e-06,
      "loss": 1.5061,
      "step": 6867
    },
    {
      "epoch": 2.3408316291751876,
      "grad_norm": 14.22368069076007,
      "learning_rate": 1.401587137158429e-06,
      "loss": 1.6439,
      "step": 6868
    },
    {
      "epoch": 2.3411724608043625,
      "grad_norm": 18.328209287451564,
      "learning_rate": 1.4002105625152007e-06,
      "loss": 0.9759,
      "step": 6869
    },
    {
      "epoch": 2.341513292433538,
      "grad_norm": 11.256594558437866,
      "learning_rate": 1.398834554133181e-06,
      "loss": 0.8927,
      "step": 6870
    },
    {
      "epoch": 2.341854124062713,
      "grad_norm": 17.59205085944068,
      "learning_rate": 1.3974591122288216e-06,
      "loss": 1.197,
      "step": 6871
    },
    {
      "epoch": 2.3421949556918884,
      "grad_norm": 14.984163057042698,
      "learning_rate": 1.3960842370184852e-06,
      "loss": 1.3412,
      "step": 6872
    },
    {
      "epoch": 2.3425357873210633,
      "grad_norm": 16.1130629922975,
      "learning_rate": 1.3947099287184457e-06,
      "loss": 1.6555,
      "step": 6873
    },
    {
      "epoch": 2.3428766189502386,
      "grad_norm": 24.37885468712867,
      "learning_rate": 1.3933361875448831e-06,
      "loss": 1.5797,
      "step": 6874
    },
    {
      "epoch": 2.343217450579414,
      "grad_norm": 16.021953324647395,
      "learning_rate": 1.3919630137138973e-06,
      "loss": 1.4541,
      "step": 6875
    },
    {
      "epoch": 2.3435582822085887,
      "grad_norm": 16.017609606047746,
      "learning_rate": 1.3905904074414934e-06,
      "loss": 1.6313,
      "step": 6876
    },
    {
      "epoch": 2.343899113837764,
      "grad_norm": 14.886763044167857,
      "learning_rate": 1.389218368943584e-06,
      "loss": 1.7891,
      "step": 6877
    },
    {
      "epoch": 2.3442399454669394,
      "grad_norm": 15.01082457642746,
      "learning_rate": 1.3878468984360017e-06,
      "loss": 1.7901,
      "step": 6878
    },
    {
      "epoch": 2.3445807770961147,
      "grad_norm": 19.096149937832507,
      "learning_rate": 1.3864759961344798e-06,
      "loss": 1.7732,
      "step": 6879
    },
    {
      "epoch": 2.3449216087252895,
      "grad_norm": 18.099792545189313,
      "learning_rate": 1.3851056622546667e-06,
      "loss": 1.5425,
      "step": 6880
    },
    {
      "epoch": 2.345262440354465,
      "grad_norm": 12.435393393843423,
      "learning_rate": 1.3837358970121262e-06,
      "loss": 1.2139,
      "step": 6881
    },
    {
      "epoch": 2.34560327198364,
      "grad_norm": 13.861659826171381,
      "learning_rate": 1.382366700622323e-06,
      "loss": 1.698,
      "step": 6882
    },
    {
      "epoch": 2.3459441036128155,
      "grad_norm": 14.530774351046299,
      "learning_rate": 1.380998073300639e-06,
      "loss": 1.7631,
      "step": 6883
    },
    {
      "epoch": 2.3462849352419903,
      "grad_norm": 21.055915541753045,
      "learning_rate": 1.3796300152623642e-06,
      "loss": 1.5848,
      "step": 6884
    },
    {
      "epoch": 2.3466257668711656,
      "grad_norm": 12.896581944035924,
      "learning_rate": 1.3782625267227002e-06,
      "loss": 1.2976,
      "step": 6885
    },
    {
      "epoch": 2.346966598500341,
      "grad_norm": 20.80932865404713,
      "learning_rate": 1.3768956078967578e-06,
      "loss": 2.0026,
      "step": 6886
    },
    {
      "epoch": 2.347307430129516,
      "grad_norm": 14.942914694967834,
      "learning_rate": 1.3755292589995588e-06,
      "loss": 1.0096,
      "step": 6887
    },
    {
      "epoch": 2.347648261758691,
      "grad_norm": 24.530144583947045,
      "learning_rate": 1.3741634802460357e-06,
      "loss": 1.312,
      "step": 6888
    },
    {
      "epoch": 2.3479890933878664,
      "grad_norm": 16.11045777838077,
      "learning_rate": 1.3727982718510314e-06,
      "loss": 0.6572,
      "step": 6889
    },
    {
      "epoch": 2.3483299250170417,
      "grad_norm": 8.90477570617953,
      "learning_rate": 1.3714336340292971e-06,
      "loss": 1.0004,
      "step": 6890
    },
    {
      "epoch": 2.3486707566462166,
      "grad_norm": 14.092983748847075,
      "learning_rate": 1.3700695669954972e-06,
      "loss": 1.2074,
      "step": 6891
    },
    {
      "epoch": 2.349011588275392,
      "grad_norm": 53.44140803960102,
      "learning_rate": 1.3687060709642042e-06,
      "loss": 1.5928,
      "step": 6892
    },
    {
      "epoch": 2.3493524199045672,
      "grad_norm": 11.9097323296817,
      "learning_rate": 1.3673431461499016e-06,
      "loss": 1.1095,
      "step": 6893
    },
    {
      "epoch": 2.3496932515337425,
      "grad_norm": 11.938018284936215,
      "learning_rate": 1.3659807927669844e-06,
      "loss": 1.3894,
      "step": 6894
    },
    {
      "epoch": 2.3500340831629174,
      "grad_norm": 24.153141521958293,
      "learning_rate": 1.3646190110297515e-06,
      "loss": 1.9008,
      "step": 6895
    },
    {
      "epoch": 2.3503749147920927,
      "grad_norm": 17.17008918367805,
      "learning_rate": 1.363257801152421e-06,
      "loss": 1.6471,
      "step": 6896
    },
    {
      "epoch": 2.350715746421268,
      "grad_norm": 14.705491386694023,
      "learning_rate": 1.3618971633491174e-06,
      "loss": 1.6144,
      "step": 6897
    },
    {
      "epoch": 2.351056578050443,
      "grad_norm": 14.231279017050605,
      "learning_rate": 1.3605370978338694e-06,
      "loss": 1.7055,
      "step": 6898
    },
    {
      "epoch": 2.351397409679618,
      "grad_norm": 16.671025614757088,
      "learning_rate": 1.3591776048206268e-06,
      "loss": 1.1526,
      "step": 6899
    },
    {
      "epoch": 2.3517382413087935,
      "grad_norm": 18.972160213212696,
      "learning_rate": 1.3578186845232389e-06,
      "loss": 1.108,
      "step": 6900
    },
    {
      "epoch": 2.352079072937969,
      "grad_norm": 14.416138047836414,
      "learning_rate": 1.3564603371554685e-06,
      "loss": 1.4124,
      "step": 6901
    },
    {
      "epoch": 2.3524199045671437,
      "grad_norm": 14.761643797602227,
      "learning_rate": 1.355102562930995e-06,
      "loss": 1.2873,
      "step": 6902
    },
    {
      "epoch": 2.352760736196319,
      "grad_norm": 11.749790063297029,
      "learning_rate": 1.3537453620633956e-06,
      "loss": 1.3165,
      "step": 6903
    },
    {
      "epoch": 2.3531015678254943,
      "grad_norm": 19.048418609872265,
      "learning_rate": 1.3523887347661658e-06,
      "loss": 1.5522,
      "step": 6904
    },
    {
      "epoch": 2.3534423994546696,
      "grad_norm": 11.492549773683917,
      "learning_rate": 1.3510326812527086e-06,
      "loss": 1.5647,
      "step": 6905
    },
    {
      "epoch": 2.3537832310838445,
      "grad_norm": 19.986125792525126,
      "learning_rate": 1.3496772017363363e-06,
      "loss": 1.3087,
      "step": 6906
    },
    {
      "epoch": 2.3541240627130198,
      "grad_norm": 17.389724514822323,
      "learning_rate": 1.348322296430273e-06,
      "loss": 1.2393,
      "step": 6907
    },
    {
      "epoch": 2.354464894342195,
      "grad_norm": 33.74753881527748,
      "learning_rate": 1.3469679655476453e-06,
      "loss": 1.747,
      "step": 6908
    },
    {
      "epoch": 2.35480572597137,
      "grad_norm": 14.371487733404559,
      "learning_rate": 1.3456142093015002e-06,
      "loss": 1.6892,
      "step": 6909
    },
    {
      "epoch": 2.3551465576005453,
      "grad_norm": 15.23288159062672,
      "learning_rate": 1.3442610279047885e-06,
      "loss": 1.577,
      "step": 6910
    },
    {
      "epoch": 2.3554873892297206,
      "grad_norm": 25.69648389333611,
      "learning_rate": 1.3429084215703675e-06,
      "loss": 1.7081,
      "step": 6911
    },
    {
      "epoch": 2.355828220858896,
      "grad_norm": 13.59330374349659,
      "learning_rate": 1.3415563905110124e-06,
      "loss": 0.9884,
      "step": 6912
    },
    {
      "epoch": 2.3561690524880707,
      "grad_norm": 15.772229693850846,
      "learning_rate": 1.3402049349393991e-06,
      "loss": 1.4417,
      "step": 6913
    },
    {
      "epoch": 2.356509884117246,
      "grad_norm": 20.072034205179303,
      "learning_rate": 1.3388540550681168e-06,
      "loss": 2.0185,
      "step": 6914
    },
    {
      "epoch": 2.3568507157464214,
      "grad_norm": 16.021080381212556,
      "learning_rate": 1.3375037511096688e-06,
      "loss": 1.9639,
      "step": 6915
    },
    {
      "epoch": 2.3571915473755967,
      "grad_norm": 13.234228969885658,
      "learning_rate": 1.3361540232764586e-06,
      "loss": 1.5235,
      "step": 6916
    },
    {
      "epoch": 2.3575323790047715,
      "grad_norm": 15.877523661295369,
      "learning_rate": 1.3348048717808059e-06,
      "loss": 1.6906,
      "step": 6917
    },
    {
      "epoch": 2.357873210633947,
      "grad_norm": 18.596339406322844,
      "learning_rate": 1.3334562968349368e-06,
      "loss": 1.0381,
      "step": 6918
    },
    {
      "epoch": 2.358214042263122,
      "grad_norm": 13.328790137484765,
      "learning_rate": 1.3321082986509882e-06,
      "loss": 1.2631,
      "step": 6919
    },
    {
      "epoch": 2.358554873892297,
      "grad_norm": 14.007035819843269,
      "learning_rate": 1.3307608774410046e-06,
      "loss": 0.991,
      "step": 6920
    },
    {
      "epoch": 2.3588957055214723,
      "grad_norm": 12.071419151327644,
      "learning_rate": 1.329414033416942e-06,
      "loss": 0.6723,
      "step": 6921
    },
    {
      "epoch": 2.3592365371506476,
      "grad_norm": 17.699157078120784,
      "learning_rate": 1.3280677667906638e-06,
      "loss": 1.6153,
      "step": 6922
    },
    {
      "epoch": 2.359577368779823,
      "grad_norm": 15.913191643159758,
      "learning_rate": 1.3267220777739426e-06,
      "loss": 2.0157,
      "step": 6923
    },
    {
      "epoch": 2.359918200408998,
      "grad_norm": 13.194259317807461,
      "learning_rate": 1.325376966578461e-06,
      "loss": 1.1622,
      "step": 6924
    },
    {
      "epoch": 2.360259032038173,
      "grad_norm": 13.70402081110196,
      "learning_rate": 1.3240324334158106e-06,
      "loss": 1.136,
      "step": 6925
    },
    {
      "epoch": 2.3605998636673484,
      "grad_norm": 14.022400916412966,
      "learning_rate": 1.3226884784974914e-06,
      "loss": 1.3944,
      "step": 6926
    },
    {
      "epoch": 2.3609406952965237,
      "grad_norm": 17.392967959771063,
      "learning_rate": 1.3213451020349133e-06,
      "loss": 1.573,
      "step": 6927
    },
    {
      "epoch": 2.3612815269256986,
      "grad_norm": 14.045184073568171,
      "learning_rate": 1.320002304239396e-06,
      "loss": 1.1975,
      "step": 6928
    },
    {
      "epoch": 2.361622358554874,
      "grad_norm": 45.913764625646394,
      "learning_rate": 1.3186600853221625e-06,
      "loss": 1.1174,
      "step": 6929
    },
    {
      "epoch": 2.361963190184049,
      "grad_norm": 19.003968199387394,
      "learning_rate": 1.3173184454943538e-06,
      "loss": 2.2929,
      "step": 6930
    },
    {
      "epoch": 2.362304021813224,
      "grad_norm": 11.844114511921473,
      "learning_rate": 1.3159773849670144e-06,
      "loss": 1.1568,
      "step": 6931
    },
    {
      "epoch": 2.3626448534423994,
      "grad_norm": 12.627343876647407,
      "learning_rate": 1.3146369039510954e-06,
      "loss": 1.6683,
      "step": 6932
    },
    {
      "epoch": 2.3629856850715747,
      "grad_norm": 12.537147552743384,
      "learning_rate": 1.3132970026574643e-06,
      "loss": 1.1234,
      "step": 6933
    },
    {
      "epoch": 2.36332651670075,
      "grad_norm": 14.636367449672935,
      "learning_rate": 1.3119576812968893e-06,
      "loss": 1.751,
      "step": 6934
    },
    {
      "epoch": 2.363667348329925,
      "grad_norm": 13.389112364964891,
      "learning_rate": 1.3106189400800513e-06,
      "loss": 1.8461,
      "step": 6935
    },
    {
      "epoch": 2.3640081799591,
      "grad_norm": 12.84100538058378,
      "learning_rate": 1.3092807792175428e-06,
      "loss": 1.4986,
      "step": 6936
    },
    {
      "epoch": 2.3643490115882755,
      "grad_norm": 20.692470366974813,
      "learning_rate": 1.3079431989198587e-06,
      "loss": 1.1694,
      "step": 6937
    },
    {
      "epoch": 2.364689843217451,
      "grad_norm": 14.212761331634802,
      "learning_rate": 1.306606199397406e-06,
      "loss": 1.3339,
      "step": 6938
    },
    {
      "epoch": 2.3650306748466257,
      "grad_norm": 13.668625130002185,
      "learning_rate": 1.3052697808605013e-06,
      "loss": 1.5452,
      "step": 6939
    },
    {
      "epoch": 2.365371506475801,
      "grad_norm": 21.165116788917064,
      "learning_rate": 1.3039339435193677e-06,
      "loss": 1.7968,
      "step": 6940
    },
    {
      "epoch": 2.3657123381049763,
      "grad_norm": 15.287749170368901,
      "learning_rate": 1.3025986875841396e-06,
      "loss": 1.7229,
      "step": 6941
    },
    {
      "epoch": 2.366053169734151,
      "grad_norm": 12.732317703626853,
      "learning_rate": 1.3012640132648536e-06,
      "loss": 1.1455,
      "step": 6942
    },
    {
      "epoch": 2.3663940013633264,
      "grad_norm": 9.918826392676523,
      "learning_rate": 1.2999299207714634e-06,
      "loss": 1.2792,
      "step": 6943
    },
    {
      "epoch": 2.3667348329925018,
      "grad_norm": 19.882135596921223,
      "learning_rate": 1.298596410313827e-06,
      "loss": 1.9929,
      "step": 6944
    },
    {
      "epoch": 2.367075664621677,
      "grad_norm": 13.967213903222222,
      "learning_rate": 1.2972634821017072e-06,
      "loss": 1.477,
      "step": 6945
    },
    {
      "epoch": 2.367416496250852,
      "grad_norm": 12.471087721652314,
      "learning_rate": 1.2959311363447842e-06,
      "loss": 1.4324,
      "step": 6946
    },
    {
      "epoch": 2.3677573278800272,
      "grad_norm": 17.876704152579418,
      "learning_rate": 1.294599373252637e-06,
      "loss": 1.0392,
      "step": 6947
    },
    {
      "epoch": 2.3680981595092025,
      "grad_norm": 15.482069803795651,
      "learning_rate": 1.2932681930347568e-06,
      "loss": 1.4305,
      "step": 6948
    },
    {
      "epoch": 2.368438991138378,
      "grad_norm": 15.20943249398943,
      "learning_rate": 1.2919375959005493e-06,
      "loss": 1.2714,
      "step": 6949
    },
    {
      "epoch": 2.3687798227675527,
      "grad_norm": 13.114094429128906,
      "learning_rate": 1.2906075820593166e-06,
      "loss": 1.9379,
      "step": 6950
    },
    {
      "epoch": 2.369120654396728,
      "grad_norm": 42.61191701155305,
      "learning_rate": 1.2892781517202767e-06,
      "loss": 1.8965,
      "step": 6951
    },
    {
      "epoch": 2.3694614860259033,
      "grad_norm": 16.42140635537054,
      "learning_rate": 1.2879493050925578e-06,
      "loss": 1.6126,
      "step": 6952
    },
    {
      "epoch": 2.369802317655078,
      "grad_norm": 15.549077429892614,
      "learning_rate": 1.2866210423851882e-06,
      "loss": 1.3019,
      "step": 6953
    },
    {
      "epoch": 2.3701431492842535,
      "grad_norm": 14.236481603428127,
      "learning_rate": 1.2852933638071114e-06,
      "loss": 1.6961,
      "step": 6954
    },
    {
      "epoch": 2.370483980913429,
      "grad_norm": 18.638722865657183,
      "learning_rate": 1.283966269567175e-06,
      "loss": 1.6279,
      "step": 6955
    },
    {
      "epoch": 2.370824812542604,
      "grad_norm": 25.659903455213257,
      "learning_rate": 1.282639759874138e-06,
      "loss": 1.6431,
      "step": 6956
    },
    {
      "epoch": 2.371165644171779,
      "grad_norm": 20.62248242023754,
      "learning_rate": 1.2813138349366644e-06,
      "loss": 1.5438,
      "step": 6957
    },
    {
      "epoch": 2.3715064758009543,
      "grad_norm": 12.303292072548448,
      "learning_rate": 1.2799884949633274e-06,
      "loss": 1.2436,
      "step": 6958
    },
    {
      "epoch": 2.3718473074301296,
      "grad_norm": 18.13752907850377,
      "learning_rate": 1.278663740162609e-06,
      "loss": 1.4326,
      "step": 6959
    },
    {
      "epoch": 2.372188139059305,
      "grad_norm": 13.409820786821584,
      "learning_rate": 1.277339570742898e-06,
      "loss": 1.2325,
      "step": 6960
    },
    {
      "epoch": 2.37252897068848,
      "grad_norm": 13.281682487278548,
      "learning_rate": 1.2760159869124916e-06,
      "loss": 1.2503,
      "step": 6961
    },
    {
      "epoch": 2.372869802317655,
      "grad_norm": 14.568492007209182,
      "learning_rate": 1.274692988879596e-06,
      "loss": 1.4656,
      "step": 6962
    },
    {
      "epoch": 2.3732106339468304,
      "grad_norm": 30.162897538181554,
      "learning_rate": 1.2733705768523203e-06,
      "loss": 1.3545,
      "step": 6963
    },
    {
      "epoch": 2.3735514655760053,
      "grad_norm": 14.305693062838511,
      "learning_rate": 1.2720487510386893e-06,
      "loss": 1.6591,
      "step": 6964
    },
    {
      "epoch": 2.3738922972051806,
      "grad_norm": 11.199114838420437,
      "learning_rate": 1.2707275116466306e-06,
      "loss": 1.5526,
      "step": 6965
    },
    {
      "epoch": 2.374233128834356,
      "grad_norm": 13.649920900142094,
      "learning_rate": 1.2694068588839775e-06,
      "loss": 1.3413,
      "step": 6966
    },
    {
      "epoch": 2.374573960463531,
      "grad_norm": 16.79702901190038,
      "learning_rate": 1.2680867929584783e-06,
      "loss": 1.4527,
      "step": 6967
    },
    {
      "epoch": 2.374914792092706,
      "grad_norm": 24.781500394787756,
      "learning_rate": 1.2667673140777819e-06,
      "loss": 0.9448,
      "step": 6968
    },
    {
      "epoch": 2.3752556237218814,
      "grad_norm": 14.88290357633069,
      "learning_rate": 1.2654484224494474e-06,
      "loss": 1.4149,
      "step": 6969
    },
    {
      "epoch": 2.3755964553510567,
      "grad_norm": 17.641558986949004,
      "learning_rate": 1.264130118280943e-06,
      "loss": 1.9408,
      "step": 6970
    },
    {
      "epoch": 2.375937286980232,
      "grad_norm": 15.102681884510146,
      "learning_rate": 1.2628124017796424e-06,
      "loss": 1.692,
      "step": 6971
    },
    {
      "epoch": 2.376278118609407,
      "grad_norm": 23.66544277253339,
      "learning_rate": 1.2614952731528274e-06,
      "loss": 1.0848,
      "step": 6972
    },
    {
      "epoch": 2.376618950238582,
      "grad_norm": 14.410010122101026,
      "learning_rate": 1.2601787326076887e-06,
      "loss": 1.0601,
      "step": 6973
    },
    {
      "epoch": 2.3769597818677575,
      "grad_norm": 12.567294484094354,
      "learning_rate": 1.2588627803513226e-06,
      "loss": 1.6451,
      "step": 6974
    },
    {
      "epoch": 2.3773006134969323,
      "grad_norm": 13.624149937757027,
      "learning_rate": 1.2575474165907336e-06,
      "loss": 1.0157,
      "step": 6975
    },
    {
      "epoch": 2.3776414451261076,
      "grad_norm": 14.220848256214689,
      "learning_rate": 1.256232641532834e-06,
      "loss": 1.2987,
      "step": 6976
    },
    {
      "epoch": 2.377982276755283,
      "grad_norm": 18.935339730392744,
      "learning_rate": 1.2549184553844428e-06,
      "loss": 1.7898,
      "step": 6977
    },
    {
      "epoch": 2.3783231083844583,
      "grad_norm": 43.79873430644483,
      "learning_rate": 1.2536048583522886e-06,
      "loss": 1.3228,
      "step": 6978
    },
    {
      "epoch": 2.378663940013633,
      "grad_norm": 15.320431104738805,
      "learning_rate": 1.2522918506430004e-06,
      "loss": 1.3611,
      "step": 6979
    },
    {
      "epoch": 2.3790047716428084,
      "grad_norm": 12.97989069748599,
      "learning_rate": 1.2509794324631247e-06,
      "loss": 1.115,
      "step": 6980
    },
    {
      "epoch": 2.3793456032719837,
      "grad_norm": 12.795421545903066,
      "learning_rate": 1.2496676040191092e-06,
      "loss": 1.508,
      "step": 6981
    },
    {
      "epoch": 2.379686434901159,
      "grad_norm": 16.40834559349729,
      "learning_rate": 1.2483563655173065e-06,
      "loss": 1.0984,
      "step": 6982
    },
    {
      "epoch": 2.380027266530334,
      "grad_norm": 14.797290783533818,
      "learning_rate": 1.2470457171639843e-06,
      "loss": 1.7167,
      "step": 6983
    },
    {
      "epoch": 2.3803680981595092,
      "grad_norm": 18.55770344201877,
      "learning_rate": 1.2457356591653091e-06,
      "loss": 1.7436,
      "step": 6984
    },
    {
      "epoch": 2.3807089297886845,
      "grad_norm": 19.524658419273354,
      "learning_rate": 1.2444261917273588e-06,
      "loss": 1.1198,
      "step": 6985
    },
    {
      "epoch": 2.3810497614178594,
      "grad_norm": 16.68527495223767,
      "learning_rate": 1.2431173150561204e-06,
      "loss": 1.5301,
      "step": 6986
    },
    {
      "epoch": 2.3813905930470347,
      "grad_norm": 18.59392890461778,
      "learning_rate": 1.241809029357483e-06,
      "loss": 1.1901,
      "step": 6987
    },
    {
      "epoch": 2.38173142467621,
      "grad_norm": 14.134157098471109,
      "learning_rate": 1.2405013348372457e-06,
      "loss": 1.4953,
      "step": 6988
    },
    {
      "epoch": 2.3820722563053853,
      "grad_norm": 21.302106797698798,
      "learning_rate": 1.239194231701114e-06,
      "loss": 1.9773,
      "step": 6989
    },
    {
      "epoch": 2.38241308793456,
      "grad_norm": 20.25191222623521,
      "learning_rate": 1.2378877201547002e-06,
      "loss": 1.8932,
      "step": 6990
    },
    {
      "epoch": 2.3827539195637355,
      "grad_norm": 31.499990150423645,
      "learning_rate": 1.2365818004035246e-06,
      "loss": 1.4572,
      "step": 6991
    },
    {
      "epoch": 2.383094751192911,
      "grad_norm": 30.788647490476766,
      "learning_rate": 1.2352764726530131e-06,
      "loss": 1.4803,
      "step": 6992
    },
    {
      "epoch": 2.383435582822086,
      "grad_norm": 15.858119375872036,
      "learning_rate": 1.2339717371084986e-06,
      "loss": 1.3608,
      "step": 6993
    },
    {
      "epoch": 2.383776414451261,
      "grad_norm": 13.46585075939964,
      "learning_rate": 1.232667593975222e-06,
      "loss": 1.1625,
      "step": 6994
    },
    {
      "epoch": 2.3841172460804363,
      "grad_norm": 16.078581332487328,
      "learning_rate": 1.231364043458329e-06,
      "loss": 1.6507,
      "step": 6995
    },
    {
      "epoch": 2.3844580777096116,
      "grad_norm": 15.967388421152492,
      "learning_rate": 1.2300610857628765e-06,
      "loss": 1.5809,
      "step": 6996
    },
    {
      "epoch": 2.3847989093387865,
      "grad_norm": 15.843909261561034,
      "learning_rate": 1.228758721093819e-06,
      "loss": 1.8243,
      "step": 6997
    },
    {
      "epoch": 2.3851397409679618,
      "grad_norm": 17.85406237921348,
      "learning_rate": 1.2274569496560284e-06,
      "loss": 1.8169,
      "step": 6998
    },
    {
      "epoch": 2.385480572597137,
      "grad_norm": 16.47686166843419,
      "learning_rate": 1.2261557716542787e-06,
      "loss": 1.6329,
      "step": 6999
    },
    {
      "epoch": 2.3858214042263124,
      "grad_norm": 14.343838635940568,
      "learning_rate": 1.2248551872932463e-06,
      "loss": 1.1968,
      "step": 7000
    },
    {
      "epoch": 2.3861622358554873,
      "grad_norm": 15.49213221433796,
      "learning_rate": 1.2235551967775239e-06,
      "loss": 1.4102,
      "step": 7001
    },
    {
      "epoch": 2.3865030674846626,
      "grad_norm": 13.112898138185567,
      "learning_rate": 1.2222558003116002e-06,
      "loss": 1.1654,
      "step": 7002
    },
    {
      "epoch": 2.386843899113838,
      "grad_norm": 17.310010536255177,
      "learning_rate": 1.220956998099878e-06,
      "loss": 1.2332,
      "step": 7003
    },
    {
      "epoch": 2.387184730743013,
      "grad_norm": 18.22467520859551,
      "learning_rate": 1.2196587903466634e-06,
      "loss": 1.5653,
      "step": 7004
    },
    {
      "epoch": 2.387525562372188,
      "grad_norm": 11.67953496598649,
      "learning_rate": 1.2183611772561698e-06,
      "loss": 0.9619,
      "step": 7005
    },
    {
      "epoch": 2.3878663940013634,
      "grad_norm": 13.77191605584328,
      "learning_rate": 1.2170641590325166e-06,
      "loss": 1.6336,
      "step": 7006
    },
    {
      "epoch": 2.3882072256305387,
      "grad_norm": 12.79057897959964,
      "learning_rate": 1.2157677358797304e-06,
      "loss": 1.7458,
      "step": 7007
    },
    {
      "epoch": 2.3885480572597135,
      "grad_norm": 21.35310057102306,
      "learning_rate": 1.2144719080017431e-06,
      "loss": 1.5974,
      "step": 7008
    },
    {
      "epoch": 2.388888888888889,
      "grad_norm": 17.365410457559342,
      "learning_rate": 1.2131766756023945e-06,
      "loss": 1.3649,
      "step": 7009
    },
    {
      "epoch": 2.389229720518064,
      "grad_norm": 13.578519499943866,
      "learning_rate": 1.211882038885429e-06,
      "loss": 1.4866,
      "step": 7010
    },
    {
      "epoch": 2.3895705521472395,
      "grad_norm": 15.41621458078481,
      "learning_rate": 1.2105879980544987e-06,
      "loss": 1.7434,
      "step": 7011
    },
    {
      "epoch": 2.3899113837764143,
      "grad_norm": 16.195667332040912,
      "learning_rate": 1.2092945533131623e-06,
      "loss": 1.1849,
      "step": 7012
    },
    {
      "epoch": 2.3902522154055896,
      "grad_norm": 19.150970663866296,
      "learning_rate": 1.2080017048648795e-06,
      "loss": 1.2507,
      "step": 7013
    },
    {
      "epoch": 2.390593047034765,
      "grad_norm": 15.195159840159308,
      "learning_rate": 1.2067094529130253e-06,
      "loss": 2.0224,
      "step": 7014
    },
    {
      "epoch": 2.3909338786639402,
      "grad_norm": 14.557186974164281,
      "learning_rate": 1.205417797660876e-06,
      "loss": 1.506,
      "step": 7015
    },
    {
      "epoch": 2.391274710293115,
      "grad_norm": 15.812773772825977,
      "learning_rate": 1.2041267393116096e-06,
      "loss": 2.0729,
      "step": 7016
    },
    {
      "epoch": 2.3916155419222904,
      "grad_norm": 17.10048913519934,
      "learning_rate": 1.20283627806832e-06,
      "loss": 1.868,
      "step": 7017
    },
    {
      "epoch": 2.3919563735514657,
      "grad_norm": 14.78916733663254,
      "learning_rate": 1.2015464141339984e-06,
      "loss": 2.0143,
      "step": 7018
    },
    {
      "epoch": 2.3922972051806406,
      "grad_norm": 13.96069017004883,
      "learning_rate": 1.2002571477115448e-06,
      "loss": 1.3359,
      "step": 7019
    },
    {
      "epoch": 2.392638036809816,
      "grad_norm": 13.63921874886805,
      "learning_rate": 1.1989684790037714e-06,
      "loss": 0.9749,
      "step": 7020
    },
    {
      "epoch": 2.392978868438991,
      "grad_norm": 17.69042882804265,
      "learning_rate": 1.1976804082133852e-06,
      "loss": 1.5044,
      "step": 7021
    },
    {
      "epoch": 2.393319700068166,
      "grad_norm": 17.030141038439524,
      "learning_rate": 1.196392935543007e-06,
      "loss": 2.072,
      "step": 7022
    },
    {
      "epoch": 2.3936605316973414,
      "grad_norm": 15.014342300595306,
      "learning_rate": 1.1951060611951615e-06,
      "loss": 1.4243,
      "step": 7023
    },
    {
      "epoch": 2.3940013633265167,
      "grad_norm": 24.584826393154177,
      "learning_rate": 1.1938197853722794e-06,
      "loss": 1.3802,
      "step": 7024
    },
    {
      "epoch": 2.394342194955692,
      "grad_norm": 14.564857927531339,
      "learning_rate": 1.192534108276696e-06,
      "loss": 1.2822,
      "step": 7025
    },
    {
      "epoch": 2.3946830265848673,
      "grad_norm": 17.230959177496132,
      "learning_rate": 1.1912490301106537e-06,
      "loss": 1.6805,
      "step": 7026
    },
    {
      "epoch": 2.395023858214042,
      "grad_norm": 11.509296208180213,
      "learning_rate": 1.1899645510763015e-06,
      "loss": 1.569,
      "step": 7027
    },
    {
      "epoch": 2.3953646898432175,
      "grad_norm": 19.862169354633593,
      "learning_rate": 1.1886806713756922e-06,
      "loss": 1.6138,
      "step": 7028
    },
    {
      "epoch": 2.395705521472393,
      "grad_norm": 13.801228281538613,
      "learning_rate": 1.187397391210785e-06,
      "loss": 1.4264,
      "step": 7029
    },
    {
      "epoch": 2.3960463531015677,
      "grad_norm": 14.717161038247282,
      "learning_rate": 1.1861147107834464e-06,
      "loss": 1.6696,
      "step": 7030
    },
    {
      "epoch": 2.396387184730743,
      "grad_norm": 12.330478534988172,
      "learning_rate": 1.1848326302954439e-06,
      "loss": 1.5595,
      "step": 7031
    },
    {
      "epoch": 2.3967280163599183,
      "grad_norm": 14.026062989258138,
      "learning_rate": 1.1835511499484564e-06,
      "loss": 1.1489,
      "step": 7032
    },
    {
      "epoch": 2.397068847989093,
      "grad_norm": 15.358377457675845,
      "learning_rate": 1.1822702699440675e-06,
      "loss": 1.4529,
      "step": 7033
    },
    {
      "epoch": 2.3974096796182685,
      "grad_norm": 15.276961939596886,
      "learning_rate": 1.1809899904837614e-06,
      "loss": 1.7157,
      "step": 7034
    },
    {
      "epoch": 2.3977505112474438,
      "grad_norm": 19.645765438564013,
      "learning_rate": 1.1797103117689318e-06,
      "loss": 1.556,
      "step": 7035
    },
    {
      "epoch": 2.398091342876619,
      "grad_norm": 11.088985495144144,
      "learning_rate": 1.1784312340008786e-06,
      "loss": 1.0321,
      "step": 7036
    },
    {
      "epoch": 2.3984321745057944,
      "grad_norm": 14.034685284548413,
      "learning_rate": 1.1771527573808044e-06,
      "loss": 1.233,
      "step": 7037
    },
    {
      "epoch": 2.3987730061349692,
      "grad_norm": 17.875418813063465,
      "learning_rate": 1.17587488210982e-06,
      "loss": 1.8297,
      "step": 7038
    },
    {
      "epoch": 2.3991138377641446,
      "grad_norm": 20.201180391839706,
      "learning_rate": 1.1745976083889388e-06,
      "loss": 1.6173,
      "step": 7039
    },
    {
      "epoch": 2.39945466939332,
      "grad_norm": 12.916569460187533,
      "learning_rate": 1.1733209364190822e-06,
      "loss": 1.4033,
      "step": 7040
    },
    {
      "epoch": 2.3997955010224947,
      "grad_norm": 8.889416545427792,
      "learning_rate": 1.1720448664010753e-06,
      "loss": 1.0879,
      "step": 7041
    },
    {
      "epoch": 2.40013633265167,
      "grad_norm": 20.795211346122944,
      "learning_rate": 1.1707693985356495e-06,
      "loss": 1.6864,
      "step": 7042
    },
    {
      "epoch": 2.4004771642808453,
      "grad_norm": 15.847959829100834,
      "learning_rate": 1.1694945330234403e-06,
      "loss": 1.614,
      "step": 7043
    },
    {
      "epoch": 2.40081799591002,
      "grad_norm": 18.788473198468118,
      "learning_rate": 1.1682202700649898e-06,
      "loss": 1.7367,
      "step": 7044
    },
    {
      "epoch": 2.4011588275391955,
      "grad_norm": 63.12189574126776,
      "learning_rate": 1.1669466098607434e-06,
      "loss": 1.7914,
      "step": 7045
    },
    {
      "epoch": 2.401499659168371,
      "grad_norm": 8.497290437069513,
      "learning_rate": 1.1656735526110558e-06,
      "loss": 1.1216,
      "step": 7046
    },
    {
      "epoch": 2.401840490797546,
      "grad_norm": 15.747156633187076,
      "learning_rate": 1.1644010985161791e-06,
      "loss": 1.4531,
      "step": 7047
    },
    {
      "epoch": 2.4021813224267214,
      "grad_norm": 18.8081522040647,
      "learning_rate": 1.1631292477762795e-06,
      "loss": 1.457,
      "step": 7048
    },
    {
      "epoch": 2.4025221540558963,
      "grad_norm": 15.290144931578189,
      "learning_rate": 1.1618580005914248e-06,
      "loss": 1.6402,
      "step": 7049
    },
    {
      "epoch": 2.4028629856850716,
      "grad_norm": 23.713016634879033,
      "learning_rate": 1.1605873571615823e-06,
      "loss": 1.7239,
      "step": 7050
    },
    {
      "epoch": 2.403203817314247,
      "grad_norm": 17.057990509780307,
      "learning_rate": 1.1593173176866346e-06,
      "loss": 1.6484,
      "step": 7051
    },
    {
      "epoch": 2.403544648943422,
      "grad_norm": 14.38782401749088,
      "learning_rate": 1.158047882366361e-06,
      "loss": 1.2655,
      "step": 7052
    },
    {
      "epoch": 2.403885480572597,
      "grad_norm": 18.56286285929236,
      "learning_rate": 1.156779051400448e-06,
      "loss": 1.5789,
      "step": 7053
    },
    {
      "epoch": 2.4042263122017724,
      "grad_norm": 11.570175888639799,
      "learning_rate": 1.1555108249884916e-06,
      "loss": 1.392,
      "step": 7054
    },
    {
      "epoch": 2.4045671438309473,
      "grad_norm": 12.2242050048677,
      "learning_rate": 1.154243203329985e-06,
      "loss": 1.2173,
      "step": 7055
    },
    {
      "epoch": 2.4049079754601226,
      "grad_norm": 13.579864752454561,
      "learning_rate": 1.1529761866243317e-06,
      "loss": 1.6638,
      "step": 7056
    },
    {
      "epoch": 2.405248807089298,
      "grad_norm": 24.98382267790553,
      "learning_rate": 1.1517097750708383e-06,
      "loss": 1.1107,
      "step": 7057
    },
    {
      "epoch": 2.405589638718473,
      "grad_norm": 9.391527762550352,
      "learning_rate": 1.1504439688687168e-06,
      "loss": 1.2071,
      "step": 7058
    },
    {
      "epoch": 2.4059304703476485,
      "grad_norm": 15.793792111747186,
      "learning_rate": 1.149178768217083e-06,
      "loss": 1.3567,
      "step": 7059
    },
    {
      "epoch": 2.4062713019768234,
      "grad_norm": 15.739614569008056,
      "learning_rate": 1.1479141733149585e-06,
      "loss": 1.519,
      "step": 7060
    },
    {
      "epoch": 2.4066121336059987,
      "grad_norm": 17.589995472551717,
      "learning_rate": 1.1466501843612681e-06,
      "loss": 1.4502,
      "step": 7061
    },
    {
      "epoch": 2.406952965235174,
      "grad_norm": 14.323967917246092,
      "learning_rate": 1.1453868015548436e-06,
      "loss": 1.341,
      "step": 7062
    },
    {
      "epoch": 2.407293796864349,
      "grad_norm": 9.085143381605834,
      "learning_rate": 1.1441240250944192e-06,
      "loss": 0.7196,
      "step": 7063
    },
    {
      "epoch": 2.407634628493524,
      "grad_norm": 14.520647349747339,
      "learning_rate": 1.1428618551786364e-06,
      "loss": 1.6496,
      "step": 7064
    },
    {
      "epoch": 2.4079754601226995,
      "grad_norm": 18.48700234152163,
      "learning_rate": 1.1416002920060353e-06,
      "loss": 1.322,
      "step": 7065
    },
    {
      "epoch": 2.4083162917518743,
      "grad_norm": 22.31584688320049,
      "learning_rate": 1.1403393357750696e-06,
      "loss": 1.8924,
      "step": 7066
    },
    {
      "epoch": 2.4086571233810496,
      "grad_norm": 13.78481037065328,
      "learning_rate": 1.1390789866840918e-06,
      "loss": 1.6013,
      "step": 7067
    },
    {
      "epoch": 2.408997955010225,
      "grad_norm": 11.089031513598648,
      "learning_rate": 1.1378192449313575e-06,
      "loss": 1.0885,
      "step": 7068
    },
    {
      "epoch": 2.4093387866394003,
      "grad_norm": 37.25570369565831,
      "learning_rate": 1.1365601107150292e-06,
      "loss": 1.4194,
      "step": 7069
    },
    {
      "epoch": 2.409679618268575,
      "grad_norm": 13.085105617965022,
      "learning_rate": 1.1353015842331771e-06,
      "loss": 1.7585,
      "step": 7070
    },
    {
      "epoch": 2.4100204498977504,
      "grad_norm": 15.046023251949068,
      "learning_rate": 1.1340436656837694e-06,
      "loss": 1.6423,
      "step": 7071
    },
    {
      "epoch": 2.4103612815269257,
      "grad_norm": 18.62919705442938,
      "learning_rate": 1.132786355264683e-06,
      "loss": 1.5826,
      "step": 7072
    },
    {
      "epoch": 2.410702113156101,
      "grad_norm": 17.350113854187743,
      "learning_rate": 1.1315296531736968e-06,
      "loss": 1.8454,
      "step": 7073
    },
    {
      "epoch": 2.411042944785276,
      "grad_norm": 14.847478515867026,
      "learning_rate": 1.1302735596084957e-06,
      "loss": 1.7459,
      "step": 7074
    },
    {
      "epoch": 2.4113837764144512,
      "grad_norm": 7.786197563691469,
      "learning_rate": 1.129018074766668e-06,
      "loss": 0.8365,
      "step": 7075
    },
    {
      "epoch": 2.4117246080436265,
      "grad_norm": 20.24825254797009,
      "learning_rate": 1.1277631988457073e-06,
      "loss": 1.7471,
      "step": 7076
    },
    {
      "epoch": 2.4120654396728014,
      "grad_norm": 11.225915806106249,
      "learning_rate": 1.1265089320430089e-06,
      "loss": 1.4868,
      "step": 7077
    },
    {
      "epoch": 2.4124062713019767,
      "grad_norm": 15.56525215425584,
      "learning_rate": 1.125255274555876e-06,
      "loss": 1.1467,
      "step": 7078
    },
    {
      "epoch": 2.412747102931152,
      "grad_norm": 17.30941104132712,
      "learning_rate": 1.124002226581512e-06,
      "loss": 1.1405,
      "step": 7079
    },
    {
      "epoch": 2.4130879345603273,
      "grad_norm": 13.910186971757367,
      "learning_rate": 1.122749788317029e-06,
      "loss": 1.5828,
      "step": 7080
    },
    {
      "epoch": 2.413428766189502,
      "grad_norm": 17.125558478055297,
      "learning_rate": 1.1214979599594362e-06,
      "loss": 1.3406,
      "step": 7081
    },
    {
      "epoch": 2.4137695978186775,
      "grad_norm": 14.12667435320065,
      "learning_rate": 1.1202467417056546e-06,
      "loss": 1.2574,
      "step": 7082
    },
    {
      "epoch": 2.414110429447853,
      "grad_norm": 16.022605856091594,
      "learning_rate": 1.1189961337525069e-06,
      "loss": 1.6862,
      "step": 7083
    },
    {
      "epoch": 2.414451261077028,
      "grad_norm": 10.824339665033548,
      "learning_rate": 1.1177461362967135e-06,
      "loss": 1.6799,
      "step": 7084
    },
    {
      "epoch": 2.414792092706203,
      "grad_norm": 15.970867917947709,
      "learning_rate": 1.11649674953491e-06,
      "loss": 1.4583,
      "step": 7085
    },
    {
      "epoch": 2.4151329243353783,
      "grad_norm": 19.938767425699176,
      "learning_rate": 1.1152479736636262e-06,
      "loss": 1.7508,
      "step": 7086
    },
    {
      "epoch": 2.4154737559645536,
      "grad_norm": 17.271671002541062,
      "learning_rate": 1.1139998088792985e-06,
      "loss": 1.3655,
      "step": 7087
    },
    {
      "epoch": 2.4158145875937285,
      "grad_norm": 17.86557953872723,
      "learning_rate": 1.1127522553782731e-06,
      "loss": 0.9778,
      "step": 7088
    },
    {
      "epoch": 2.4161554192229038,
      "grad_norm": 15.401788537597023,
      "learning_rate": 1.111505313356791e-06,
      "loss": 1.7464,
      "step": 7089
    },
    {
      "epoch": 2.416496250852079,
      "grad_norm": 18.805370963452326,
      "learning_rate": 1.1102589830110023e-06,
      "loss": 1.619,
      "step": 7090
    },
    {
      "epoch": 2.4168370824812544,
      "grad_norm": 13.730701263628776,
      "learning_rate": 1.1090132645369595e-06,
      "loss": 1.1127,
      "step": 7091
    },
    {
      "epoch": 2.4171779141104293,
      "grad_norm": 12.13506384702224,
      "learning_rate": 1.1077681581306194e-06,
      "loss": 1.5546,
      "step": 7092
    },
    {
      "epoch": 2.4175187457396046,
      "grad_norm": 10.510919811423076,
      "learning_rate": 1.1065236639878419e-06,
      "loss": 1.2782,
      "step": 7093
    },
    {
      "epoch": 2.41785957736878,
      "grad_norm": 16.059786072933125,
      "learning_rate": 1.1052797823043914e-06,
      "loss": 1.9169,
      "step": 7094
    },
    {
      "epoch": 2.418200408997955,
      "grad_norm": 11.847389105352597,
      "learning_rate": 1.104036513275935e-06,
      "loss": 1.0675,
      "step": 7095
    },
    {
      "epoch": 2.41854124062713,
      "grad_norm": 12.73303147839179,
      "learning_rate": 1.102793857098044e-06,
      "loss": 1.3634,
      "step": 7096
    },
    {
      "epoch": 2.4188820722563054,
      "grad_norm": 16.456801749318732,
      "learning_rate": 1.1015518139661935e-06,
      "loss": 1.4058,
      "step": 7097
    },
    {
      "epoch": 2.4192229038854807,
      "grad_norm": 11.909038669358392,
      "learning_rate": 1.1003103840757612e-06,
      "loss": 1.5088,
      "step": 7098
    },
    {
      "epoch": 2.4195637355146555,
      "grad_norm": 12.759009162211667,
      "learning_rate": 1.0990695676220303e-06,
      "loss": 0.9713,
      "step": 7099
    },
    {
      "epoch": 2.419904567143831,
      "grad_norm": 14.324306080480063,
      "learning_rate": 1.0978293648001826e-06,
      "loss": 1.4712,
      "step": 7100
    },
    {
      "epoch": 2.420245398773006,
      "grad_norm": 13.803605554053629,
      "learning_rate": 1.0965897758053124e-06,
      "loss": 1.1397,
      "step": 7101
    },
    {
      "epoch": 2.4205862304021815,
      "grad_norm": 25.614722978476266,
      "learning_rate": 1.0953508008324076e-06,
      "loss": 1.5058,
      "step": 7102
    },
    {
      "epoch": 2.4209270620313563,
      "grad_norm": 14.634662708050781,
      "learning_rate": 1.0941124400763637e-06,
      "loss": 1.3194,
      "step": 7103
    },
    {
      "epoch": 2.4212678936605316,
      "grad_norm": 11.520007653990081,
      "learning_rate": 1.0928746937319844e-06,
      "loss": 1.2202,
      "step": 7104
    },
    {
      "epoch": 2.421608725289707,
      "grad_norm": 11.622557579274886,
      "learning_rate": 1.0916375619939678e-06,
      "loss": 1.6477,
      "step": 7105
    },
    {
      "epoch": 2.4219495569188823,
      "grad_norm": 13.81427588432943,
      "learning_rate": 1.0904010450569209e-06,
      "loss": 1.5044,
      "step": 7106
    },
    {
      "epoch": 2.422290388548057,
      "grad_norm": 18.115273386028875,
      "learning_rate": 1.0891651431153527e-06,
      "loss": 1.1851,
      "step": 7107
    },
    {
      "epoch": 2.4226312201772324,
      "grad_norm": 17.207037061878033,
      "learning_rate": 1.0879298563636758e-06,
      "loss": 1.286,
      "step": 7108
    },
    {
      "epoch": 2.4229720518064077,
      "grad_norm": 29.793197083970952,
      "learning_rate": 1.0866951849962054e-06,
      "loss": 1.7181,
      "step": 7109
    },
    {
      "epoch": 2.4233128834355826,
      "grad_norm": 12.491347235778273,
      "learning_rate": 1.085461129207161e-06,
      "loss": 0.9687,
      "step": 7110
    },
    {
      "epoch": 2.423653715064758,
      "grad_norm": 16.73720297463059,
      "learning_rate": 1.0842276891906628e-06,
      "loss": 1.1401,
      "step": 7111
    },
    {
      "epoch": 2.423994546693933,
      "grad_norm": 12.66518582946452,
      "learning_rate": 1.0829948651407374e-06,
      "loss": 1.6341,
      "step": 7112
    },
    {
      "epoch": 2.4243353783231085,
      "grad_norm": 15.091405838639584,
      "learning_rate": 1.081762657251313e-06,
      "loss": 1.3658,
      "step": 7113
    },
    {
      "epoch": 2.4246762099522834,
      "grad_norm": 15.341949726275107,
      "learning_rate": 1.0805310657162206e-06,
      "loss": 1.3334,
      "step": 7114
    },
    {
      "epoch": 2.4250170415814587,
      "grad_norm": 10.315824382751103,
      "learning_rate": 1.0793000907291917e-06,
      "loss": 1.4225,
      "step": 7115
    },
    {
      "epoch": 2.425357873210634,
      "grad_norm": 13.301266991679588,
      "learning_rate": 1.0780697324838674e-06,
      "loss": 1.2304,
      "step": 7116
    },
    {
      "epoch": 2.4256987048398093,
      "grad_norm": 13.906059791791161,
      "learning_rate": 1.0768399911737876e-06,
      "loss": 1.2848,
      "step": 7117
    },
    {
      "epoch": 2.426039536468984,
      "grad_norm": 18.619211554307576,
      "learning_rate": 1.075610866992392e-06,
      "loss": 1.4381,
      "step": 7118
    },
    {
      "epoch": 2.4263803680981595,
      "grad_norm": 23.22007309504667,
      "learning_rate": 1.0743823601330316e-06,
      "loss": 1.6922,
      "step": 7119
    },
    {
      "epoch": 2.426721199727335,
      "grad_norm": 12.430442367438534,
      "learning_rate": 1.0731544707889519e-06,
      "loss": 1.6499,
      "step": 7120
    },
    {
      "epoch": 2.4270620313565097,
      "grad_norm": 16.66284595993648,
      "learning_rate": 1.0719271991533042e-06,
      "loss": 1.5648,
      "step": 7121
    },
    {
      "epoch": 2.427402862985685,
      "grad_norm": 23.150045428144303,
      "learning_rate": 1.0707005454191472e-06,
      "loss": 1.3762,
      "step": 7122
    },
    {
      "epoch": 2.4277436946148603,
      "grad_norm": 21.476017038566606,
      "learning_rate": 1.0694745097794351e-06,
      "loss": 1.7551,
      "step": 7123
    },
    {
      "epoch": 2.4280845262440356,
      "grad_norm": 12.667598173731351,
      "learning_rate": 1.068249092427029e-06,
      "loss": 1.1269,
      "step": 7124
    },
    {
      "epoch": 2.4284253578732105,
      "grad_norm": 18.423747588292517,
      "learning_rate": 1.067024293554692e-06,
      "loss": 1.1292,
      "step": 7125
    },
    {
      "epoch": 2.4287661895023858,
      "grad_norm": 22.846097846321975,
      "learning_rate": 1.0658001133550899e-06,
      "loss": 1.122,
      "step": 7126
    },
    {
      "epoch": 2.429107021131561,
      "grad_norm": 18.86687595292585,
      "learning_rate": 1.0645765520207913e-06,
      "loss": 1.3992,
      "step": 7127
    },
    {
      "epoch": 2.4294478527607364,
      "grad_norm": 20.34203407561231,
      "learning_rate": 1.0633536097442671e-06,
      "loss": 1.2087,
      "step": 7128
    },
    {
      "epoch": 2.4297886843899112,
      "grad_norm": 16.39243063289598,
      "learning_rate": 1.0621312867178913e-06,
      "loss": 1.3791,
      "step": 7129
    },
    {
      "epoch": 2.4301295160190866,
      "grad_norm": 21.37298828794129,
      "learning_rate": 1.0609095831339388e-06,
      "loss": 1.5701,
      "step": 7130
    },
    {
      "epoch": 2.430470347648262,
      "grad_norm": 25.069697877037836,
      "learning_rate": 1.0596884991845902e-06,
      "loss": 1.8068,
      "step": 7131
    },
    {
      "epoch": 2.4308111792774367,
      "grad_norm": 19.21318709342924,
      "learning_rate": 1.0584680350619265e-06,
      "loss": 1.1256,
      "step": 7132
    },
    {
      "epoch": 2.431152010906612,
      "grad_norm": 10.760179588178525,
      "learning_rate": 1.057248190957932e-06,
      "loss": 1.1747,
      "step": 7133
    },
    {
      "epoch": 2.4314928425357873,
      "grad_norm": 16.24756980000676,
      "learning_rate": 1.0560289670644896e-06,
      "loss": 1.4922,
      "step": 7134
    },
    {
      "epoch": 2.4318336741649627,
      "grad_norm": 20.911650069422834,
      "learning_rate": 1.0548103635733937e-06,
      "loss": 1.4788,
      "step": 7135
    },
    {
      "epoch": 2.4321745057941375,
      "grad_norm": 11.035724783387087,
      "learning_rate": 1.0535923806763305e-06,
      "loss": 1.1321,
      "step": 7136
    },
    {
      "epoch": 2.432515337423313,
      "grad_norm": 20.64871033623647,
      "learning_rate": 1.0523750185648946e-06,
      "loss": 1.3925,
      "step": 7137
    },
    {
      "epoch": 2.432856169052488,
      "grad_norm": 10.643667266020392,
      "learning_rate": 1.0511582774305855e-06,
      "loss": 1.3709,
      "step": 7138
    },
    {
      "epoch": 2.4331970006816634,
      "grad_norm": 12.26696273258922,
      "learning_rate": 1.0499421574647977e-06,
      "loss": 1.2436,
      "step": 7139
    },
    {
      "epoch": 2.4335378323108383,
      "grad_norm": 12.444754393903482,
      "learning_rate": 1.048726658858833e-06,
      "loss": 1.4068,
      "step": 7140
    },
    {
      "epoch": 2.4338786639400136,
      "grad_norm": 20.011379153717154,
      "learning_rate": 1.0475117818038933e-06,
      "loss": 1.1694,
      "step": 7141
    },
    {
      "epoch": 2.434219495569189,
      "grad_norm": 13.970963287917797,
      "learning_rate": 1.0462975264910847e-06,
      "loss": 1.5911,
      "step": 7142
    },
    {
      "epoch": 2.434560327198364,
      "grad_norm": 13.205531546370915,
      "learning_rate": 1.0450838931114137e-06,
      "loss": 1.4693,
      "step": 7143
    },
    {
      "epoch": 2.434901158827539,
      "grad_norm": 16.45159288316795,
      "learning_rate": 1.0438708818557902e-06,
      "loss": 1.2645,
      "step": 7144
    },
    {
      "epoch": 2.4352419904567144,
      "grad_norm": 12.710235156389514,
      "learning_rate": 1.042658492915025e-06,
      "loss": 1.4887,
      "step": 7145
    },
    {
      "epoch": 2.4355828220858897,
      "grad_norm": 18.07307968072705,
      "learning_rate": 1.0414467264798327e-06,
      "loss": 1.2976,
      "step": 7146
    },
    {
      "epoch": 2.4359236537150646,
      "grad_norm": 14.129865158001456,
      "learning_rate": 1.0402355827408283e-06,
      "loss": 1.3888,
      "step": 7147
    },
    {
      "epoch": 2.43626448534424,
      "grad_norm": 13.081556518413857,
      "learning_rate": 1.0390250618885307e-06,
      "loss": 1.5739,
      "step": 7148
    },
    {
      "epoch": 2.436605316973415,
      "grad_norm": 13.598146721714945,
      "learning_rate": 1.037815164113356e-06,
      "loss": 1.3265,
      "step": 7149
    },
    {
      "epoch": 2.4369461486025905,
      "grad_norm": 20.160380677801303,
      "learning_rate": 1.0366058896056296e-06,
      "loss": 1.8446,
      "step": 7150
    },
    {
      "epoch": 2.4372869802317654,
      "grad_norm": 13.221189023748131,
      "learning_rate": 1.0353972385555761e-06,
      "loss": 1.3325,
      "step": 7151
    },
    {
      "epoch": 2.4376278118609407,
      "grad_norm": 12.192226605175833,
      "learning_rate": 1.0341892111533165e-06,
      "loss": 1.3386,
      "step": 7152
    },
    {
      "epoch": 2.437968643490116,
      "grad_norm": 12.732308628508301,
      "learning_rate": 1.0329818075888832e-06,
      "loss": 1.3764,
      "step": 7153
    },
    {
      "epoch": 2.438309475119291,
      "grad_norm": 25.28563881877948,
      "learning_rate": 1.031775028052202e-06,
      "loss": 1.6427,
      "step": 7154
    },
    {
      "epoch": 2.438650306748466,
      "grad_norm": 20.87275224703967,
      "learning_rate": 1.0305688727331041e-06,
      "loss": 0.8538,
      "step": 7155
    },
    {
      "epoch": 2.4389911383776415,
      "grad_norm": 12.529244170059357,
      "learning_rate": 1.0293633418213267e-06,
      "loss": 1.4225,
      "step": 7156
    },
    {
      "epoch": 2.439331970006817,
      "grad_norm": 13.205856168960803,
      "learning_rate": 1.0281584355065006e-06,
      "loss": 1.6709,
      "step": 7157
    },
    {
      "epoch": 2.4396728016359917,
      "grad_norm": 11.592059313546304,
      "learning_rate": 1.0269541539781618e-06,
      "loss": 1.586,
      "step": 7158
    },
    {
      "epoch": 2.440013633265167,
      "grad_norm": 24.545225792126,
      "learning_rate": 1.0257504974257538e-06,
      "loss": 1.4787,
      "step": 7159
    },
    {
      "epoch": 2.4403544648943423,
      "grad_norm": 21.472539464160924,
      "learning_rate": 1.0245474660386113e-06,
      "loss": 1.5991,
      "step": 7160
    },
    {
      "epoch": 2.4406952965235176,
      "grad_norm": 16.59670312576498,
      "learning_rate": 1.023345060005978e-06,
      "loss": 1.6018,
      "step": 7161
    },
    {
      "epoch": 2.4410361281526924,
      "grad_norm": 17.462023887859964,
      "learning_rate": 1.0221432795169966e-06,
      "loss": 1.7785,
      "step": 7162
    },
    {
      "epoch": 2.4413769597818678,
      "grad_norm": 16.351531910906346,
      "learning_rate": 1.020942124760712e-06,
      "loss": 1.3093,
      "step": 7163
    },
    {
      "epoch": 2.441717791411043,
      "grad_norm": 25.742755621875826,
      "learning_rate": 1.0197415959260732e-06,
      "loss": 1.2187,
      "step": 7164
    },
    {
      "epoch": 2.442058623040218,
      "grad_norm": 22.58556373268965,
      "learning_rate": 1.018541693201922e-06,
      "loss": 1.745,
      "step": 7165
    },
    {
      "epoch": 2.4423994546693932,
      "grad_norm": 15.285258438497669,
      "learning_rate": 1.0173424167770141e-06,
      "loss": 1.4242,
      "step": 7166
    },
    {
      "epoch": 2.4427402862985685,
      "grad_norm": 18.75762409549442,
      "learning_rate": 1.0161437668399988e-06,
      "loss": 0.903,
      "step": 7167
    },
    {
      "epoch": 2.443081117927744,
      "grad_norm": 11.439159631846241,
      "learning_rate": 1.0149457435794252e-06,
      "loss": 1.3394,
      "step": 7168
    },
    {
      "epoch": 2.4434219495569187,
      "grad_norm": 13.052281208399728,
      "learning_rate": 1.0137483471837529e-06,
      "loss": 1.5572,
      "step": 7169
    },
    {
      "epoch": 2.443762781186094,
      "grad_norm": 15.38057135098321,
      "learning_rate": 1.012551577841332e-06,
      "loss": 1.9112,
      "step": 7170
    },
    {
      "epoch": 2.4441036128152693,
      "grad_norm": 17.540768765286426,
      "learning_rate": 1.0113554357404199e-06,
      "loss": 1.3663,
      "step": 7171
    },
    {
      "epoch": 2.4444444444444446,
      "grad_norm": 12.534619906945778,
      "learning_rate": 1.010159921069178e-06,
      "loss": 0.883,
      "step": 7172
    },
    {
      "epoch": 2.4447852760736195,
      "grad_norm": 15.527856441101953,
      "learning_rate": 1.008965034015662e-06,
      "loss": 1.8254,
      "step": 7173
    },
    {
      "epoch": 2.445126107702795,
      "grad_norm": 15.21566996958172,
      "learning_rate": 1.0077707747678339e-06,
      "loss": 1.8015,
      "step": 7174
    },
    {
      "epoch": 2.44546693933197,
      "grad_norm": 17.578904511838907,
      "learning_rate": 1.0065771435135546e-06,
      "loss": 1.8197,
      "step": 7175
    },
    {
      "epoch": 2.445807770961145,
      "grad_norm": 10.8992401969301,
      "learning_rate": 1.0053841404405878e-06,
      "loss": 1.1596,
      "step": 7176
    },
    {
      "epoch": 2.4461486025903203,
      "grad_norm": 13.519823886982053,
      "learning_rate": 1.0041917657365974e-06,
      "loss": 1.5636,
      "step": 7177
    },
    {
      "epoch": 2.4464894342194956,
      "grad_norm": 19.49895569267218,
      "learning_rate": 1.0030000195891486e-06,
      "loss": 1.9788,
      "step": 7178
    },
    {
      "epoch": 2.446830265848671,
      "grad_norm": 15.164066070945205,
      "learning_rate": 1.001808902185708e-06,
      "loss": 1.6152,
      "step": 7179
    },
    {
      "epoch": 2.447171097477846,
      "grad_norm": 18.365426992121996,
      "learning_rate": 1.0006184137136426e-06,
      "loss": 1.5637,
      "step": 7180
    },
    {
      "epoch": 2.447511929107021,
      "grad_norm": 26.74460709948205,
      "learning_rate": 9.994285543602217e-07,
      "loss": 1.3699,
      "step": 7181
    },
    {
      "epoch": 2.4478527607361964,
      "grad_norm": 14.410508492109862,
      "learning_rate": 9.982393243126142e-07,
      "loss": 1.5025,
      "step": 7182
    },
    {
      "epoch": 2.4481935923653717,
      "grad_norm": 14.498190116549527,
      "learning_rate": 9.970507237578914e-07,
      "loss": 1.3214,
      "step": 7183
    },
    {
      "epoch": 2.4485344239945466,
      "grad_norm": 10.78599835856865,
      "learning_rate": 9.958627528830239e-07,
      "loss": 1.7572,
      "step": 7184
    },
    {
      "epoch": 2.448875255623722,
      "grad_norm": 16.72751419940287,
      "learning_rate": 9.946754118748868e-07,
      "loss": 1.161,
      "step": 7185
    },
    {
      "epoch": 2.449216087252897,
      "grad_norm": 16.8445693738929,
      "learning_rate": 9.934887009202494e-07,
      "loss": 1.2553,
      "step": 7186
    },
    {
      "epoch": 2.449556918882072,
      "grad_norm": 11.24946671174485,
      "learning_rate": 9.92302620205789e-07,
      "loss": 1.3664,
      "step": 7187
    },
    {
      "epoch": 2.4498977505112474,
      "grad_norm": 14.818794960105349,
      "learning_rate": 9.911171699180817e-07,
      "loss": 1.5927,
      "step": 7188
    },
    {
      "epoch": 2.4502385821404227,
      "grad_norm": 14.069749257709743,
      "learning_rate": 9.899323502435993e-07,
      "loss": 1.7592,
      "step": 7189
    },
    {
      "epoch": 2.450579413769598,
      "grad_norm": 15.629690629614114,
      "learning_rate": 9.887481613687245e-07,
      "loss": 0.8777,
      "step": 7190
    },
    {
      "epoch": 2.450920245398773,
      "grad_norm": 13.39520631686912,
      "learning_rate": 9.8756460347973e-07,
      "loss": 0.9351,
      "step": 7191
    },
    {
      "epoch": 2.451261077027948,
      "grad_norm": 18.803148215768317,
      "learning_rate": 9.863816767627943e-07,
      "loss": 0.8648,
      "step": 7192
    },
    {
      "epoch": 2.4516019086571235,
      "grad_norm": 14.254662614855748,
      "learning_rate": 9.851993814040011e-07,
      "loss": 1.4748,
      "step": 7193
    },
    {
      "epoch": 2.4519427402862988,
      "grad_norm": 11.363620518404744,
      "learning_rate": 9.840177175893262e-07,
      "loss": 1.3806,
      "step": 7194
    },
    {
      "epoch": 2.4522835719154736,
      "grad_norm": 10.961294660406253,
      "learning_rate": 9.828366855046507e-07,
      "loss": 1.3043,
      "step": 7195
    },
    {
      "epoch": 2.452624403544649,
      "grad_norm": 11.410141045633805,
      "learning_rate": 9.816562853357564e-07,
      "loss": 0.9093,
      "step": 7196
    },
    {
      "epoch": 2.4529652351738243,
      "grad_norm": 13.927484231166975,
      "learning_rate": 9.80476517268324e-07,
      "loss": 1.4542,
      "step": 7197
    },
    {
      "epoch": 2.453306066802999,
      "grad_norm": 15.148515229663053,
      "learning_rate": 9.792973814879376e-07,
      "loss": 1.3128,
      "step": 7198
    },
    {
      "epoch": 2.4536468984321744,
      "grad_norm": 10.960153867689325,
      "learning_rate": 9.78118878180076e-07,
      "loss": 1.502,
      "step": 7199
    },
    {
      "epoch": 2.4539877300613497,
      "grad_norm": 12.964453482058124,
      "learning_rate": 9.769410075301267e-07,
      "loss": 1.6452,
      "step": 7200
    },
    {
      "epoch": 2.454328561690525,
      "grad_norm": 16.182466326357858,
      "learning_rate": 9.757637697233723e-07,
      "loss": 1.1855,
      "step": 7201
    },
    {
      "epoch": 2.4546693933197,
      "grad_norm": 16.187040222625527,
      "learning_rate": 9.745871649449944e-07,
      "loss": 1.397,
      "step": 7202
    },
    {
      "epoch": 2.455010224948875,
      "grad_norm": 19.58717837756131,
      "learning_rate": 9.734111933800827e-07,
      "loss": 1.2609,
      "step": 7203
    },
    {
      "epoch": 2.4553510565780505,
      "grad_norm": 10.168480713735102,
      "learning_rate": 9.722358552136174e-07,
      "loss": 1.361,
      "step": 7204
    },
    {
      "epoch": 2.455691888207226,
      "grad_norm": 11.838139631796846,
      "learning_rate": 9.710611506304845e-07,
      "loss": 1.3299,
      "step": 7205
    },
    {
      "epoch": 2.4560327198364007,
      "grad_norm": 11.677211987502739,
      "learning_rate": 9.698870798154736e-07,
      "loss": 0.9799,
      "step": 7206
    },
    {
      "epoch": 2.456373551465576,
      "grad_norm": 22.201120175248597,
      "learning_rate": 9.687136429532662e-07,
      "loss": 1.5797,
      "step": 7207
    },
    {
      "epoch": 2.4567143830947513,
      "grad_norm": 14.812152062407037,
      "learning_rate": 9.67540840228451e-07,
      "loss": 1.4067,
      "step": 7208
    },
    {
      "epoch": 2.457055214723926,
      "grad_norm": 12.824720784414266,
      "learning_rate": 9.663686718255138e-07,
      "loss": 1.3922,
      "step": 7209
    },
    {
      "epoch": 2.4573960463531015,
      "grad_norm": 18.849793068130023,
      "learning_rate": 9.65197137928841e-07,
      "loss": 0.8182,
      "step": 7210
    },
    {
      "epoch": 2.457736877982277,
      "grad_norm": 14.17527324126507,
      "learning_rate": 9.640262387227206e-07,
      "loss": 1.3101,
      "step": 7211
    },
    {
      "epoch": 2.458077709611452,
      "grad_norm": 13.98788571627043,
      "learning_rate": 9.62855974391339e-07,
      "loss": 1.5673,
      "step": 7212
    },
    {
      "epoch": 2.458418541240627,
      "grad_norm": 17.364585038630988,
      "learning_rate": 9.616863451187836e-07,
      "loss": 1.4092,
      "step": 7213
    },
    {
      "epoch": 2.4587593728698023,
      "grad_norm": 14.277996737852556,
      "learning_rate": 9.605173510890415e-07,
      "loss": 1.303,
      "step": 7214
    },
    {
      "epoch": 2.4591002044989776,
      "grad_norm": 16.870737996666172,
      "learning_rate": 9.593489924859999e-07,
      "loss": 1.3883,
      "step": 7215
    },
    {
      "epoch": 2.459441036128153,
      "grad_norm": 23.57177672096655,
      "learning_rate": 9.581812694934472e-07,
      "loss": 1.3017,
      "step": 7216
    },
    {
      "epoch": 2.4597818677573278,
      "grad_norm": 12.461238986142392,
      "learning_rate": 9.570141822950708e-07,
      "loss": 1.5356,
      "step": 7217
    },
    {
      "epoch": 2.460122699386503,
      "grad_norm": 22.562103751534757,
      "learning_rate": 9.558477310744568e-07,
      "loss": 0.9012,
      "step": 7218
    },
    {
      "epoch": 2.4604635310156784,
      "grad_norm": 18.127004769759765,
      "learning_rate": 9.546819160150956e-07,
      "loss": 1.4209,
      "step": 7219
    },
    {
      "epoch": 2.4608043626448532,
      "grad_norm": 22.8403729083543,
      "learning_rate": 9.535167373003695e-07,
      "loss": 1.4932,
      "step": 7220
    },
    {
      "epoch": 2.4611451942740286,
      "grad_norm": 9.270554983289106,
      "learning_rate": 9.523521951135701e-07,
      "loss": 1.3037,
      "step": 7221
    },
    {
      "epoch": 2.461486025903204,
      "grad_norm": 11.645806817482242,
      "learning_rate": 9.511882896378844e-07,
      "loss": 1.4348,
      "step": 7222
    },
    {
      "epoch": 2.461826857532379,
      "grad_norm": 17.87513809929892,
      "learning_rate": 9.50025021056396e-07,
      "loss": 1.3196,
      "step": 7223
    },
    {
      "epoch": 2.462167689161554,
      "grad_norm": 16.13476683348674,
      "learning_rate": 9.488623895520959e-07,
      "loss": 1.7471,
      "step": 7224
    },
    {
      "epoch": 2.4625085207907293,
      "grad_norm": 15.083576141262654,
      "learning_rate": 9.477003953078673e-07,
      "loss": 1.0385,
      "step": 7225
    },
    {
      "epoch": 2.4628493524199047,
      "grad_norm": 14.368290074912666,
      "learning_rate": 9.465390385064965e-07,
      "loss": 1.4466,
      "step": 7226
    },
    {
      "epoch": 2.46319018404908,
      "grad_norm": 13.444666024993124,
      "learning_rate": 9.453783193306737e-07,
      "loss": 1.5229,
      "step": 7227
    },
    {
      "epoch": 2.463531015678255,
      "grad_norm": 15.667801033702055,
      "learning_rate": 9.442182379629805e-07,
      "loss": 1.6101,
      "step": 7228
    },
    {
      "epoch": 2.46387184730743,
      "grad_norm": 17.875506923700254,
      "learning_rate": 9.430587945859032e-07,
      "loss": 1.9103,
      "step": 7229
    },
    {
      "epoch": 2.4642126789366054,
      "grad_norm": 17.49330484459176,
      "learning_rate": 9.418999893818276e-07,
      "loss": 1.3707,
      "step": 7230
    },
    {
      "epoch": 2.4645535105657803,
      "grad_norm": 17.955943203993076,
      "learning_rate": 9.407418225330378e-07,
      "loss": 1.9069,
      "step": 7231
    },
    {
      "epoch": 2.4648943421949556,
      "grad_norm": 13.636319585377386,
      "learning_rate": 9.395842942217198e-07,
      "loss": 1.584,
      "step": 7232
    },
    {
      "epoch": 2.465235173824131,
      "grad_norm": 16.839059590235827,
      "learning_rate": 9.38427404629953e-07,
      "loss": 1.2318,
      "step": 7233
    },
    {
      "epoch": 2.4655760054533062,
      "grad_norm": 10.330959930581962,
      "learning_rate": 9.37271153939725e-07,
      "loss": 1.1717,
      "step": 7234
    },
    {
      "epoch": 2.465916837082481,
      "grad_norm": 20.433025698107343,
      "learning_rate": 9.361155423329188e-07,
      "loss": 1.8687,
      "step": 7235
    },
    {
      "epoch": 2.4662576687116564,
      "grad_norm": 40.03858249058752,
      "learning_rate": 9.349605699913128e-07,
      "loss": 1.2256,
      "step": 7236
    },
    {
      "epoch": 2.4665985003408317,
      "grad_norm": 18.82611296232626,
      "learning_rate": 9.338062370965933e-07,
      "loss": 1.6781,
      "step": 7237
    },
    {
      "epoch": 2.466939331970007,
      "grad_norm": 13.268400931992273,
      "learning_rate": 9.32652543830338e-07,
      "loss": 1.3044,
      "step": 7238
    },
    {
      "epoch": 2.467280163599182,
      "grad_norm": 15.7931899415614,
      "learning_rate": 9.314994903740277e-07,
      "loss": 1.2555,
      "step": 7239
    },
    {
      "epoch": 2.467620995228357,
      "grad_norm": 13.14380409053015,
      "learning_rate": 9.303470769090456e-07,
      "loss": 1.4949,
      "step": 7240
    },
    {
      "epoch": 2.4679618268575325,
      "grad_norm": 17.12241143231086,
      "learning_rate": 9.291953036166673e-07,
      "loss": 1.5554,
      "step": 7241
    },
    {
      "epoch": 2.4683026584867074,
      "grad_norm": 15.620158303830387,
      "learning_rate": 9.280441706780729e-07,
      "loss": 1.4395,
      "step": 7242
    },
    {
      "epoch": 2.4686434901158827,
      "grad_norm": 15.41240465227015,
      "learning_rate": 9.268936782743398e-07,
      "loss": 1.5205,
      "step": 7243
    },
    {
      "epoch": 2.468984321745058,
      "grad_norm": 18.41532110964911,
      "learning_rate": 9.257438265864454e-07,
      "loss": 1.7013,
      "step": 7244
    },
    {
      "epoch": 2.4693251533742333,
      "grad_norm": 12.144089250184471,
      "learning_rate": 9.245946157952656e-07,
      "loss": 1.4905,
      "step": 7245
    },
    {
      "epoch": 2.469665985003408,
      "grad_norm": 15.658000063337223,
      "learning_rate": 9.234460460815763e-07,
      "loss": 1.5081,
      "step": 7246
    },
    {
      "epoch": 2.4700068166325835,
      "grad_norm": 12.59686655548456,
      "learning_rate": 9.222981176260515e-07,
      "loss": 1.6243,
      "step": 7247
    },
    {
      "epoch": 2.470347648261759,
      "grad_norm": 16.043615750645948,
      "learning_rate": 9.211508306092654e-07,
      "loss": 1.9371,
      "step": 7248
    },
    {
      "epoch": 2.470688479890934,
      "grad_norm": 14.9662544793883,
      "learning_rate": 9.200041852116903e-07,
      "loss": 1.414,
      "step": 7249
    },
    {
      "epoch": 2.471029311520109,
      "grad_norm": 15.597322042713373,
      "learning_rate": 9.188581816136988e-07,
      "loss": 1.8718,
      "step": 7250
    },
    {
      "epoch": 2.4713701431492843,
      "grad_norm": 14.982816326494522,
      "learning_rate": 9.17712819995561e-07,
      "loss": 1.8928,
      "step": 7251
    },
    {
      "epoch": 2.4717109747784596,
      "grad_norm": 14.963856948213275,
      "learning_rate": 9.165681005374477e-07,
      "loss": 1.504,
      "step": 7252
    },
    {
      "epoch": 2.4720518064076344,
      "grad_norm": 13.693706777317677,
      "learning_rate": 9.154240234194283e-07,
      "loss": 1.5746,
      "step": 7253
    },
    {
      "epoch": 2.4723926380368098,
      "grad_norm": 27.363992450065297,
      "learning_rate": 9.142805888214673e-07,
      "loss": 1.2542,
      "step": 7254
    },
    {
      "epoch": 2.472733469665985,
      "grad_norm": 19.697659165278434,
      "learning_rate": 9.131377969234351e-07,
      "loss": 1.3685,
      "step": 7255
    },
    {
      "epoch": 2.4730743012951604,
      "grad_norm": 19.579023001738474,
      "learning_rate": 9.119956479050979e-07,
      "loss": 1.7573,
      "step": 7256
    },
    {
      "epoch": 2.4734151329243352,
      "grad_norm": 12.931122520869199,
      "learning_rate": 9.108541419461159e-07,
      "loss": 1.5519,
      "step": 7257
    },
    {
      "epoch": 2.4737559645535105,
      "grad_norm": 15.47420862218098,
      "learning_rate": 9.097132792260582e-07,
      "loss": 1.2184,
      "step": 7258
    },
    {
      "epoch": 2.474096796182686,
      "grad_norm": 14.74881546475577,
      "learning_rate": 9.085730599243825e-07,
      "loss": 1.403,
      "step": 7259
    },
    {
      "epoch": 2.474437627811861,
      "grad_norm": 17.484296609969448,
      "learning_rate": 9.074334842204502e-07,
      "loss": 1.4673,
      "step": 7260
    },
    {
      "epoch": 2.474778459441036,
      "grad_norm": 10.078901244125214,
      "learning_rate": 9.062945522935251e-07,
      "loss": 0.8473,
      "step": 7261
    },
    {
      "epoch": 2.4751192910702113,
      "grad_norm": 16.249880127589734,
      "learning_rate": 9.051562643227618e-07,
      "loss": 1.4238,
      "step": 7262
    },
    {
      "epoch": 2.4754601226993866,
      "grad_norm": 23.8161549515034,
      "learning_rate": 9.040186204872187e-07,
      "loss": 1.0332,
      "step": 7263
    },
    {
      "epoch": 2.4758009543285615,
      "grad_norm": 13.237333071598837,
      "learning_rate": 9.02881620965852e-07,
      "loss": 0.7131,
      "step": 7264
    },
    {
      "epoch": 2.476141785957737,
      "grad_norm": 11.337234602730579,
      "learning_rate": 9.017452659375158e-07,
      "loss": 1.0034,
      "step": 7265
    },
    {
      "epoch": 2.476482617586912,
      "grad_norm": 17.4055310616967,
      "learning_rate": 9.006095555809646e-07,
      "loss": 1.6727,
      "step": 7266
    },
    {
      "epoch": 2.4768234492160874,
      "grad_norm": 14.005628033926971,
      "learning_rate": 8.994744900748464e-07,
      "loss": 1.6518,
      "step": 7267
    },
    {
      "epoch": 2.4771642808452623,
      "grad_norm": 12.918073777405686,
      "learning_rate": 8.983400695977157e-07,
      "loss": 1.9534,
      "step": 7268
    },
    {
      "epoch": 2.4775051124744376,
      "grad_norm": 11.671468300043681,
      "learning_rate": 8.972062943280213e-07,
      "loss": 1.5628,
      "step": 7269
    },
    {
      "epoch": 2.477845944103613,
      "grad_norm": 21.773698671504246,
      "learning_rate": 8.960731644441062e-07,
      "loss": 1.9202,
      "step": 7270
    },
    {
      "epoch": 2.4781867757327882,
      "grad_norm": 12.016252813064868,
      "learning_rate": 8.949406801242216e-07,
      "loss": 1.536,
      "step": 7271
    },
    {
      "epoch": 2.478527607361963,
      "grad_norm": 18.603103597652378,
      "learning_rate": 8.938088415465085e-07,
      "loss": 1.3815,
      "step": 7272
    },
    {
      "epoch": 2.4788684389911384,
      "grad_norm": 14.937071268401132,
      "learning_rate": 8.926776488890088e-07,
      "loss": 1.4647,
      "step": 7273
    },
    {
      "epoch": 2.4792092706203137,
      "grad_norm": 13.748867576204576,
      "learning_rate": 8.915471023296679e-07,
      "loss": 1.2132,
      "step": 7274
    },
    {
      "epoch": 2.4795501022494886,
      "grad_norm": 12.158054227426947,
      "learning_rate": 8.904172020463209e-07,
      "loss": 1.3479,
      "step": 7275
    },
    {
      "epoch": 2.479890933878664,
      "grad_norm": 10.940859701880347,
      "learning_rate": 8.892879482167066e-07,
      "loss": 1.5001,
      "step": 7276
    },
    {
      "epoch": 2.480231765507839,
      "grad_norm": 13.391777854471858,
      "learning_rate": 8.881593410184642e-07,
      "loss": 1.6364,
      "step": 7277
    },
    {
      "epoch": 2.4805725971370145,
      "grad_norm": 14.97460204398591,
      "learning_rate": 8.870313806291248e-07,
      "loss": 1.6099,
      "step": 7278
    },
    {
      "epoch": 2.4809134287661894,
      "grad_norm": 14.648752701461312,
      "learning_rate": 8.859040672261216e-07,
      "loss": 1.7127,
      "step": 7279
    },
    {
      "epoch": 2.4812542603953647,
      "grad_norm": 17.185113957049612,
      "learning_rate": 8.84777400986786e-07,
      "loss": 2.1919,
      "step": 7280
    },
    {
      "epoch": 2.48159509202454,
      "grad_norm": 21.035684668432104,
      "learning_rate": 8.836513820883468e-07,
      "loss": 1.7349,
      "step": 7281
    },
    {
      "epoch": 2.4819359236537153,
      "grad_norm": 19.291868187192378,
      "learning_rate": 8.825260107079314e-07,
      "loss": 1.673,
      "step": 7282
    },
    {
      "epoch": 2.48227675528289,
      "grad_norm": 17.969087104437563,
      "learning_rate": 8.814012870225652e-07,
      "loss": 1.5835,
      "step": 7283
    },
    {
      "epoch": 2.4826175869120655,
      "grad_norm": 11.518077340679586,
      "learning_rate": 8.802772112091723e-07,
      "loss": 0.9113,
      "step": 7284
    },
    {
      "epoch": 2.4829584185412408,
      "grad_norm": 14.937137887495476,
      "learning_rate": 8.791537834445734e-07,
      "loss": 1.1796,
      "step": 7285
    },
    {
      "epoch": 2.4832992501704156,
      "grad_norm": 13.05421455830903,
      "learning_rate": 8.780310039054884e-07,
      "loss": 1.6152,
      "step": 7286
    },
    {
      "epoch": 2.483640081799591,
      "grad_norm": 17.3514101966365,
      "learning_rate": 8.769088727685365e-07,
      "loss": 1.1963,
      "step": 7287
    },
    {
      "epoch": 2.4839809134287663,
      "grad_norm": 18.493868641136718,
      "learning_rate": 8.757873902102293e-07,
      "loss": 1.6259,
      "step": 7288
    },
    {
      "epoch": 2.4843217450579416,
      "grad_norm": 35.77229918087143,
      "learning_rate": 8.746665564069845e-07,
      "loss": 1.6115,
      "step": 7289
    },
    {
      "epoch": 2.4846625766871164,
      "grad_norm": 17.232779489242258,
      "learning_rate": 8.735463715351139e-07,
      "loss": 1.6328,
      "step": 7290
    },
    {
      "epoch": 2.4850034083162917,
      "grad_norm": 18.611301167372854,
      "learning_rate": 8.724268357708227e-07,
      "loss": 1.8876,
      "step": 7291
    },
    {
      "epoch": 2.485344239945467,
      "grad_norm": 14.739756991505775,
      "learning_rate": 8.713079492902237e-07,
      "loss": 1.1354,
      "step": 7292
    },
    {
      "epoch": 2.4856850715746424,
      "grad_norm": 14.518684781049565,
      "learning_rate": 8.701897122693187e-07,
      "loss": 1.5207,
      "step": 7293
    },
    {
      "epoch": 2.4860259032038172,
      "grad_norm": 12.103499737354694,
      "learning_rate": 8.690721248840117e-07,
      "loss": 1.5234,
      "step": 7294
    },
    {
      "epoch": 2.4863667348329925,
      "grad_norm": 13.28495703757433,
      "learning_rate": 8.679551873101033e-07,
      "loss": 1.5038,
      "step": 7295
    },
    {
      "epoch": 2.486707566462168,
      "grad_norm": 23.62572993506726,
      "learning_rate": 8.66838899723293e-07,
      "loss": 1.2484,
      "step": 7296
    },
    {
      "epoch": 2.4870483980913427,
      "grad_norm": 12.882316914884425,
      "learning_rate": 8.657232622991768e-07,
      "loss": 1.6313,
      "step": 7297
    },
    {
      "epoch": 2.487389229720518,
      "grad_norm": 14.612987567502403,
      "learning_rate": 8.64608275213249e-07,
      "loss": 1.2636,
      "step": 7298
    },
    {
      "epoch": 2.4877300613496933,
      "grad_norm": 17.744060436582874,
      "learning_rate": 8.634939386409019e-07,
      "loss": 2.0273,
      "step": 7299
    },
    {
      "epoch": 2.4880708929788686,
      "grad_norm": 16.36339956859085,
      "learning_rate": 8.623802527574243e-07,
      "loss": 1.2677,
      "step": 7300
    },
    {
      "epoch": 2.4884117246080435,
      "grad_norm": 23.123910063041055,
      "learning_rate": 8.612672177380038e-07,
      "loss": 1.2786,
      "step": 7301
    },
    {
      "epoch": 2.488752556237219,
      "grad_norm": 19.265320379996886,
      "learning_rate": 8.601548337577259e-07,
      "loss": 1.8027,
      "step": 7302
    },
    {
      "epoch": 2.489093387866394,
      "grad_norm": 16.374129960695843,
      "learning_rate": 8.590431009915734e-07,
      "loss": 1.6699,
      "step": 7303
    },
    {
      "epoch": 2.4894342194955694,
      "grad_norm": 12.830818275697034,
      "learning_rate": 8.579320196144225e-07,
      "loss": 1.0333,
      "step": 7304
    },
    {
      "epoch": 2.4897750511247443,
      "grad_norm": 19.649175584007732,
      "learning_rate": 8.568215898010551e-07,
      "loss": 1.9452,
      "step": 7305
    },
    {
      "epoch": 2.4901158827539196,
      "grad_norm": 15.097972362608665,
      "learning_rate": 8.55711811726146e-07,
      "loss": 1.4861,
      "step": 7306
    },
    {
      "epoch": 2.490456714383095,
      "grad_norm": 14.40721914423309,
      "learning_rate": 8.546026855642631e-07,
      "loss": 1.4387,
      "step": 7307
    },
    {
      "epoch": 2.4907975460122698,
      "grad_norm": 18.818559534259357,
      "learning_rate": 8.534942114898831e-07,
      "loss": 1.6418,
      "step": 7308
    },
    {
      "epoch": 2.491138377641445,
      "grad_norm": 15.355395683992542,
      "learning_rate": 8.523863896773682e-07,
      "loss": 1.0962,
      "step": 7309
    },
    {
      "epoch": 2.4914792092706204,
      "grad_norm": 13.672089660936196,
      "learning_rate": 8.512792203009834e-07,
      "loss": 1.44,
      "step": 7310
    },
    {
      "epoch": 2.4918200408997953,
      "grad_norm": 14.639548549627127,
      "learning_rate": 8.501727035348956e-07,
      "loss": 1.1745,
      "step": 7311
    },
    {
      "epoch": 2.4921608725289706,
      "grad_norm": 11.987301842415631,
      "learning_rate": 8.490668395531592e-07,
      "loss": 1.0465,
      "step": 7312
    },
    {
      "epoch": 2.492501704158146,
      "grad_norm": 16.509511061867904,
      "learning_rate": 8.479616285297331e-07,
      "loss": 1.4955,
      "step": 7313
    },
    {
      "epoch": 2.492842535787321,
      "grad_norm": 13.789730819556569,
      "learning_rate": 8.46857070638471e-07,
      "loss": 1.327,
      "step": 7314
    },
    {
      "epoch": 2.4931833674164965,
      "grad_norm": 25.47945493986472,
      "learning_rate": 8.457531660531243e-07,
      "loss": 1.17,
      "step": 7315
    },
    {
      "epoch": 2.4935241990456714,
      "grad_norm": 12.086911735336566,
      "learning_rate": 8.446499149473419e-07,
      "loss": 1.3855,
      "step": 7316
    },
    {
      "epoch": 2.4938650306748467,
      "grad_norm": 14.750849672338111,
      "learning_rate": 8.435473174946695e-07,
      "loss": 1.5196,
      "step": 7317
    },
    {
      "epoch": 2.494205862304022,
      "grad_norm": 15.129561697219751,
      "learning_rate": 8.424453738685495e-07,
      "loss": 1.8154,
      "step": 7318
    },
    {
      "epoch": 2.494546693933197,
      "grad_norm": 16.576557855776038,
      "learning_rate": 8.413440842423226e-07,
      "loss": 1.2594,
      "step": 7319
    },
    {
      "epoch": 2.494887525562372,
      "grad_norm": 17.581886249802572,
      "learning_rate": 8.402434487892264e-07,
      "loss": 1.3065,
      "step": 7320
    },
    {
      "epoch": 2.4952283571915475,
      "grad_norm": 16.815338922423148,
      "learning_rate": 8.391434676823951e-07,
      "loss": 1.7271,
      "step": 7321
    },
    {
      "epoch": 2.4955691888207223,
      "grad_norm": 16.600701314810742,
      "learning_rate": 8.38044141094858e-07,
      "loss": 1.6915,
      "step": 7322
    },
    {
      "epoch": 2.4959100204498976,
      "grad_norm": 11.988104697693768,
      "learning_rate": 8.36945469199546e-07,
      "loss": 0.9004,
      "step": 7323
    },
    {
      "epoch": 2.496250852079073,
      "grad_norm": 10.399512918248465,
      "learning_rate": 8.358474521692855e-07,
      "loss": 0.6607,
      "step": 7324
    },
    {
      "epoch": 2.4965916837082482,
      "grad_norm": 19.616397109801234,
      "learning_rate": 8.347500901767947e-07,
      "loss": 2.0397,
      "step": 7325
    },
    {
      "epoch": 2.4969325153374236,
      "grad_norm": 16.053113989114443,
      "learning_rate": 8.33653383394698e-07,
      "loss": 1.5719,
      "step": 7326
    },
    {
      "epoch": 2.4972733469665984,
      "grad_norm": 14.887494043534236,
      "learning_rate": 8.325573319955083e-07,
      "loss": 1.4806,
      "step": 7327
    },
    {
      "epoch": 2.4976141785957737,
      "grad_norm": 19.450787495956874,
      "learning_rate": 8.314619361516396e-07,
      "loss": 1.5588,
      "step": 7328
    },
    {
      "epoch": 2.497955010224949,
      "grad_norm": 10.859380862098183,
      "learning_rate": 8.303671960354026e-07,
      "loss": 0.7832,
      "step": 7329
    },
    {
      "epoch": 2.498295841854124,
      "grad_norm": 16.284460310556423,
      "learning_rate": 8.292731118190039e-07,
      "loss": 1.767,
      "step": 7330
    },
    {
      "epoch": 2.498636673483299,
      "grad_norm": 13.651347310241343,
      "learning_rate": 8.281796836745476e-07,
      "loss": 1.5013,
      "step": 7331
    },
    {
      "epoch": 2.4989775051124745,
      "grad_norm": 13.557289607670956,
      "learning_rate": 8.270869117740343e-07,
      "loss": 1.2004,
      "step": 7332
    },
    {
      "epoch": 2.4993183367416494,
      "grad_norm": 18.826562863831565,
      "learning_rate": 8.259947962893605e-07,
      "loss": 1.1302,
      "step": 7333
    },
    {
      "epoch": 2.4996591683708247,
      "grad_norm": 22.89463103124782,
      "learning_rate": 8.249033373923215e-07,
      "loss": 1.6083,
      "step": 7334
    },
    {
      "epoch": 2.5,
      "grad_norm": 21.10486774315084,
      "learning_rate": 8.238125352546078e-07,
      "loss": 1.4667,
      "step": 7335
    },
    {
      "epoch": 2.5003408316291753,
      "grad_norm": 8.977924145551727,
      "learning_rate": 8.22722390047806e-07,
      "loss": 1.0005,
      "step": 7336
    },
    {
      "epoch": 2.5006816632583506,
      "grad_norm": 13.69659112590352,
      "learning_rate": 8.216329019434028e-07,
      "loss": 1.9537,
      "step": 7337
    },
    {
      "epoch": 2.5010224948875255,
      "grad_norm": 10.178160168628994,
      "learning_rate": 8.205440711127749e-07,
      "loss": 0.7666,
      "step": 7338
    },
    {
      "epoch": 2.501363326516701,
      "grad_norm": 13.792566914346132,
      "learning_rate": 8.194558977272027e-07,
      "loss": 1.6983,
      "step": 7339
    },
    {
      "epoch": 2.501704158145876,
      "grad_norm": 14.99074874198622,
      "learning_rate": 8.183683819578614e-07,
      "loss": 1.3055,
      "step": 7340
    },
    {
      "epoch": 2.502044989775051,
      "grad_norm": 8.28071515922257,
      "learning_rate": 8.17281523975817e-07,
      "loss": 0.8247,
      "step": 7341
    },
    {
      "epoch": 2.5023858214042263,
      "grad_norm": 14.880819963916805,
      "learning_rate": 8.16195323952042e-07,
      "loss": 1.4722,
      "step": 7342
    },
    {
      "epoch": 2.5027266530334016,
      "grad_norm": 15.455770904442437,
      "learning_rate": 8.151097820573955e-07,
      "loss": 1.5712,
      "step": 7343
    },
    {
      "epoch": 2.5030674846625764,
      "grad_norm": 19.713317092032014,
      "learning_rate": 8.140248984626387e-07,
      "loss": 1.6699,
      "step": 7344
    },
    {
      "epoch": 2.5034083162917518,
      "grad_norm": 15.659107460063895,
      "learning_rate": 8.12940673338431e-07,
      "loss": 1.9309,
      "step": 7345
    },
    {
      "epoch": 2.503749147920927,
      "grad_norm": 17.783972067668234,
      "learning_rate": 8.118571068553222e-07,
      "loss": 1.9212,
      "step": 7346
    },
    {
      "epoch": 2.5040899795501024,
      "grad_norm": 14.48541665206304,
      "learning_rate": 8.10774199183762e-07,
      "loss": 1.6459,
      "step": 7347
    },
    {
      "epoch": 2.5044308111792777,
      "grad_norm": 17.664023263507936,
      "learning_rate": 8.09691950494097e-07,
      "loss": 1.5613,
      "step": 7348
    },
    {
      "epoch": 2.5047716428084525,
      "grad_norm": 21.373022931698106,
      "learning_rate": 8.086103609565688e-07,
      "loss": 1.4619,
      "step": 7349
    },
    {
      "epoch": 2.505112474437628,
      "grad_norm": 19.584549353159748,
      "learning_rate": 8.075294307413156e-07,
      "loss": 1.4934,
      "step": 7350
    },
    {
      "epoch": 2.505453306066803,
      "grad_norm": 13.232044562144287,
      "learning_rate": 8.064491600183722e-07,
      "loss": 1.685,
      "step": 7351
    },
    {
      "epoch": 2.505794137695978,
      "grad_norm": 11.054763609237785,
      "learning_rate": 8.053695489576697e-07,
      "loss": 0.7657,
      "step": 7352
    },
    {
      "epoch": 2.5061349693251533,
      "grad_norm": 20.24879136025583,
      "learning_rate": 8.04290597729035e-07,
      "loss": 1.589,
      "step": 7353
    },
    {
      "epoch": 2.5064758009543286,
      "grad_norm": 12.049023010676693,
      "learning_rate": 8.032123065021913e-07,
      "loss": 1.045,
      "step": 7354
    },
    {
      "epoch": 2.5068166325835035,
      "grad_norm": 14.375461201391476,
      "learning_rate": 8.021346754467596e-07,
      "loss": 2.1813,
      "step": 7355
    },
    {
      "epoch": 2.507157464212679,
      "grad_norm": 14.998850711403684,
      "learning_rate": 8.010577047322515e-07,
      "loss": 1.3718,
      "step": 7356
    },
    {
      "epoch": 2.507498295841854,
      "grad_norm": 15.34967352374433,
      "learning_rate": 7.999813945280826e-07,
      "loss": 1.8656,
      "step": 7357
    },
    {
      "epoch": 2.5078391274710294,
      "grad_norm": 13.602744070199778,
      "learning_rate": 7.989057450035614e-07,
      "loss": 1.7002,
      "step": 7358
    },
    {
      "epoch": 2.5081799591002047,
      "grad_norm": 14.276615019224245,
      "learning_rate": 7.978307563278881e-07,
      "loss": 1.1421,
      "step": 7359
    },
    {
      "epoch": 2.5085207907293796,
      "grad_norm": 14.23944293588689,
      "learning_rate": 7.967564286701656e-07,
      "loss": 1.5142,
      "step": 7360
    },
    {
      "epoch": 2.508861622358555,
      "grad_norm": 13.996526179131441,
      "learning_rate": 7.956827621993885e-07,
      "loss": 1.2263,
      "step": 7361
    },
    {
      "epoch": 2.5092024539877302,
      "grad_norm": 14.233305213942131,
      "learning_rate": 7.946097570844491e-07,
      "loss": 1.8298,
      "step": 7362
    },
    {
      "epoch": 2.509543285616905,
      "grad_norm": 16.784348897198672,
      "learning_rate": 7.935374134941359e-07,
      "loss": 1.3478,
      "step": 7363
    },
    {
      "epoch": 2.5098841172460804,
      "grad_norm": 15.572479093873591,
      "learning_rate": 7.92465731597133e-07,
      "loss": 2.0343,
      "step": 7364
    },
    {
      "epoch": 2.5102249488752557,
      "grad_norm": 12.384354815029436,
      "learning_rate": 7.913947115620191e-07,
      "loss": 1.5259,
      "step": 7365
    },
    {
      "epoch": 2.5105657805044306,
      "grad_norm": 12.359896086592785,
      "learning_rate": 7.903243535572708e-07,
      "loss": 1.1324,
      "step": 7366
    },
    {
      "epoch": 2.510906612133606,
      "grad_norm": 15.344179204428357,
      "learning_rate": 7.892546577512589e-07,
      "loss": 1.515,
      "step": 7367
    },
    {
      "epoch": 2.511247443762781,
      "grad_norm": 13.647111740340657,
      "learning_rate": 7.881856243122521e-07,
      "loss": 1.363,
      "step": 7368
    },
    {
      "epoch": 2.5115882753919565,
      "grad_norm": 17.415695786671527,
      "learning_rate": 7.871172534084126e-07,
      "loss": 1.4744,
      "step": 7369
    },
    {
      "epoch": 2.511929107021132,
      "grad_norm": 15.35150354076101,
      "learning_rate": 7.860495452077993e-07,
      "loss": 1.5351,
      "step": 7370
    },
    {
      "epoch": 2.5122699386503067,
      "grad_norm": 15.78615190601747,
      "learning_rate": 7.84982499878369e-07,
      "loss": 1.2808,
      "step": 7371
    },
    {
      "epoch": 2.512610770279482,
      "grad_norm": 18.857883777610134,
      "learning_rate": 7.839161175879679e-07,
      "loss": 1.873,
      "step": 7372
    },
    {
      "epoch": 2.5129516019086573,
      "grad_norm": 12.960363357502438,
      "learning_rate": 7.828503985043462e-07,
      "loss": 1.1541,
      "step": 7373
    },
    {
      "epoch": 2.513292433537832,
      "grad_norm": 16.170782737568736,
      "learning_rate": 7.817853427951449e-07,
      "loss": 1.3803,
      "step": 7374
    },
    {
      "epoch": 2.5136332651670075,
      "grad_norm": 15.160987559636919,
      "learning_rate": 7.80720950627899e-07,
      "loss": 1.2531,
      "step": 7375
    },
    {
      "epoch": 2.5139740967961828,
      "grad_norm": 14.528107256723471,
      "learning_rate": 7.796572221700461e-07,
      "loss": 1.7039,
      "step": 7376
    },
    {
      "epoch": 2.5143149284253576,
      "grad_norm": 12.377666602027306,
      "learning_rate": 7.785941575889106e-07,
      "loss": 1.3853,
      "step": 7377
    },
    {
      "epoch": 2.514655760054533,
      "grad_norm": 16.618353786038536,
      "learning_rate": 7.775317570517177e-07,
      "loss": 1.7623,
      "step": 7378
    },
    {
      "epoch": 2.5149965916837083,
      "grad_norm": 23.430186925953976,
      "learning_rate": 7.764700207255904e-07,
      "loss": 1.8534,
      "step": 7379
    },
    {
      "epoch": 2.5153374233128836,
      "grad_norm": 17.915662250660127,
      "learning_rate": 7.754089487775406e-07,
      "loss": 1.2594,
      "step": 7380
    },
    {
      "epoch": 2.515678254942059,
      "grad_norm": 15.397664422691138,
      "learning_rate": 7.743485413744805e-07,
      "loss": 1.8082,
      "step": 7381
    },
    {
      "epoch": 2.5160190865712337,
      "grad_norm": 16.448836143967586,
      "learning_rate": 7.73288798683216e-07,
      "loss": 1.5979,
      "step": 7382
    },
    {
      "epoch": 2.516359918200409,
      "grad_norm": 11.23949734230631,
      "learning_rate": 7.722297208704488e-07,
      "loss": 1.1814,
      "step": 7383
    },
    {
      "epoch": 2.5167007498295844,
      "grad_norm": 15.010891263826833,
      "learning_rate": 7.711713081027767e-07,
      "loss": 1.2757,
      "step": 7384
    },
    {
      "epoch": 2.5170415814587592,
      "grad_norm": 14.074609454795562,
      "learning_rate": 7.701135605466914e-07,
      "loss": 1.261,
      "step": 7385
    },
    {
      "epoch": 2.5173824130879345,
      "grad_norm": 13.764414447460863,
      "learning_rate": 7.690564783685817e-07,
      "loss": 1.0638,
      "step": 7386
    },
    {
      "epoch": 2.51772324471711,
      "grad_norm": 16.022714723594095,
      "learning_rate": 7.680000617347299e-07,
      "loss": 1.2868,
      "step": 7387
    },
    {
      "epoch": 2.5180640763462847,
      "grad_norm": 11.799480559366899,
      "learning_rate": 7.669443108113151e-07,
      "loss": 1.6154,
      "step": 7388
    },
    {
      "epoch": 2.51840490797546,
      "grad_norm": 11.768644236199794,
      "learning_rate": 7.658892257644113e-07,
      "loss": 0.939,
      "step": 7389
    },
    {
      "epoch": 2.5187457396046353,
      "grad_norm": 11.465430660173231,
      "learning_rate": 7.64834806759987e-07,
      "loss": 1.4763,
      "step": 7390
    },
    {
      "epoch": 2.5190865712338106,
      "grad_norm": 10.422517501772361,
      "learning_rate": 7.637810539639067e-07,
      "loss": 1.3888,
      "step": 7391
    },
    {
      "epoch": 2.519427402862986,
      "grad_norm": 15.43662398509854,
      "learning_rate": 7.627279675419319e-07,
      "loss": 1.5599,
      "step": 7392
    },
    {
      "epoch": 2.519768234492161,
      "grad_norm": 12.820704781004274,
      "learning_rate": 7.616755476597137e-07,
      "loss": 1.0902,
      "step": 7393
    },
    {
      "epoch": 2.520109066121336,
      "grad_norm": 10.917417252559016,
      "learning_rate": 7.606237944828026e-07,
      "loss": 0.925,
      "step": 7394
    },
    {
      "epoch": 2.5204498977505114,
      "grad_norm": 14.347677486464546,
      "learning_rate": 7.595727081766468e-07,
      "loss": 1.2378,
      "step": 7395
    },
    {
      "epoch": 2.5207907293796863,
      "grad_norm": 13.640542659126242,
      "learning_rate": 7.585222889065829e-07,
      "loss": 0.7724,
      "step": 7396
    },
    {
      "epoch": 2.5211315610088616,
      "grad_norm": 13.187901524998502,
      "learning_rate": 7.574725368378477e-07,
      "loss": 1.2352,
      "step": 7397
    },
    {
      "epoch": 2.521472392638037,
      "grad_norm": 30.348865617751507,
      "learning_rate": 7.564234521355707e-07,
      "loss": 1.2088,
      "step": 7398
    },
    {
      "epoch": 2.5218132242672118,
      "grad_norm": 19.194658111405282,
      "learning_rate": 7.553750349647776e-07,
      "loss": 1.2974,
      "step": 7399
    },
    {
      "epoch": 2.522154055896387,
      "grad_norm": 16.331683618798603,
      "learning_rate": 7.543272854903882e-07,
      "loss": 2.1875,
      "step": 7400
    },
    {
      "epoch": 2.5224948875255624,
      "grad_norm": 11.124908017083714,
      "learning_rate": 7.53280203877218e-07,
      "loss": 1.1949,
      "step": 7401
    },
    {
      "epoch": 2.5228357191547377,
      "grad_norm": 14.127976827762394,
      "learning_rate": 7.522337902899768e-07,
      "loss": 1.6357,
      "step": 7402
    },
    {
      "epoch": 2.523176550783913,
      "grad_norm": 18.998008078900295,
      "learning_rate": 7.511880448932696e-07,
      "loss": 1.4481,
      "step": 7403
    },
    {
      "epoch": 2.523517382413088,
      "grad_norm": 13.49533295551422,
      "learning_rate": 7.50142967851597e-07,
      "loss": 1.4648,
      "step": 7404
    },
    {
      "epoch": 2.523858214042263,
      "grad_norm": 15.337134188455785,
      "learning_rate": 7.490985593293537e-07,
      "loss": 2.1124,
      "step": 7405
    },
    {
      "epoch": 2.5241990456714385,
      "grad_norm": 14.116031960572537,
      "learning_rate": 7.480548194908271e-07,
      "loss": 1.272,
      "step": 7406
    },
    {
      "epoch": 2.5245398773006134,
      "grad_norm": 9.900935165223613,
      "learning_rate": 7.470117485002044e-07,
      "loss": 1.1918,
      "step": 7407
    },
    {
      "epoch": 2.5248807089297887,
      "grad_norm": 14.37814763734437,
      "learning_rate": 7.459693465215651e-07,
      "loss": 1.4858,
      "step": 7408
    },
    {
      "epoch": 2.525221540558964,
      "grad_norm": 12.582151709465794,
      "learning_rate": 7.449276137188794e-07,
      "loss": 1.0836,
      "step": 7409
    },
    {
      "epoch": 2.525562372188139,
      "grad_norm": 16.335214110110382,
      "learning_rate": 7.438865502560211e-07,
      "loss": 2.134,
      "step": 7410
    },
    {
      "epoch": 2.525903203817314,
      "grad_norm": 26.137384615545418,
      "learning_rate": 7.428461562967499e-07,
      "loss": 1.6227,
      "step": 7411
    },
    {
      "epoch": 2.5262440354464895,
      "grad_norm": 14.979597600658169,
      "learning_rate": 7.418064320047236e-07,
      "loss": 1.2952,
      "step": 7412
    },
    {
      "epoch": 2.5265848670756648,
      "grad_norm": 15.814414130009325,
      "learning_rate": 7.407673775434987e-07,
      "loss": 1.715,
      "step": 7413
    },
    {
      "epoch": 2.52692569870484,
      "grad_norm": 32.81427698791769,
      "learning_rate": 7.397289930765195e-07,
      "loss": 1.718,
      "step": 7414
    },
    {
      "epoch": 2.527266530334015,
      "grad_norm": 13.842699621292997,
      "learning_rate": 7.386912787671286e-07,
      "loss": 1.7704,
      "step": 7415
    },
    {
      "epoch": 2.5276073619631902,
      "grad_norm": 15.666791842269564,
      "learning_rate": 7.376542347785631e-07,
      "loss": 1.3374,
      "step": 7416
    },
    {
      "epoch": 2.5279481935923656,
      "grad_norm": 12.902460042865902,
      "learning_rate": 7.366178612739538e-07,
      "loss": 1.0773,
      "step": 7417
    },
    {
      "epoch": 2.5282890252215404,
      "grad_norm": 12.424832870006833,
      "learning_rate": 7.355821584163264e-07,
      "loss": 1.3416,
      "step": 7418
    },
    {
      "epoch": 2.5286298568507157,
      "grad_norm": 13.691257950976492,
      "learning_rate": 7.345471263686011e-07,
      "loss": 0.901,
      "step": 7419
    },
    {
      "epoch": 2.528970688479891,
      "grad_norm": 14.745461203725633,
      "learning_rate": 7.335127652935925e-07,
      "loss": 1.3336,
      "step": 7420
    },
    {
      "epoch": 2.529311520109066,
      "grad_norm": 17.01075597041676,
      "learning_rate": 7.324790753540101e-07,
      "loss": 1.6479,
      "step": 7421
    },
    {
      "epoch": 2.529652351738241,
      "grad_norm": 22.0880869909676,
      "learning_rate": 7.314460567124571e-07,
      "loss": 1.3919,
      "step": 7422
    },
    {
      "epoch": 2.5299931833674165,
      "grad_norm": 14.840927920129124,
      "learning_rate": 7.304137095314312e-07,
      "loss": 1.9994,
      "step": 7423
    },
    {
      "epoch": 2.530334014996592,
      "grad_norm": 12.385464311332294,
      "learning_rate": 7.293820339733265e-07,
      "loss": 1.6584,
      "step": 7424
    },
    {
      "epoch": 2.530674846625767,
      "grad_norm": 14.182593846513335,
      "learning_rate": 7.28351030200426e-07,
      "loss": 1.5844,
      "step": 7425
    },
    {
      "epoch": 2.531015678254942,
      "grad_norm": 13.54925437503414,
      "learning_rate": 7.273206983749148e-07,
      "loss": 1.129,
      "step": 7426
    },
    {
      "epoch": 2.5313565098841173,
      "grad_norm": 13.618172048926102,
      "learning_rate": 7.262910386588657e-07,
      "loss": 1.3588,
      "step": 7427
    },
    {
      "epoch": 2.5316973415132926,
      "grad_norm": 15.975769236053301,
      "learning_rate": 7.252620512142472e-07,
      "loss": 1.8363,
      "step": 7428
    },
    {
      "epoch": 2.5320381731424675,
      "grad_norm": 24.210289699663964,
      "learning_rate": 7.242337362029273e-07,
      "loss": 0.6429,
      "step": 7429
    },
    {
      "epoch": 2.532379004771643,
      "grad_norm": 21.201744509798647,
      "learning_rate": 7.232060937866602e-07,
      "loss": 1.6521,
      "step": 7430
    },
    {
      "epoch": 2.532719836400818,
      "grad_norm": 17.0037989097734,
      "learning_rate": 7.221791241270998e-07,
      "loss": 1.5373,
      "step": 7431
    },
    {
      "epoch": 2.533060668029993,
      "grad_norm": 11.681451731553782,
      "learning_rate": 7.21152827385792e-07,
      "loss": 0.9405,
      "step": 7432
    },
    {
      "epoch": 2.5334014996591683,
      "grad_norm": 22.021592518746917,
      "learning_rate": 7.20127203724178e-07,
      "loss": 0.8153,
      "step": 7433
    },
    {
      "epoch": 2.5337423312883436,
      "grad_norm": 12.896057103080716,
      "learning_rate": 7.191022533035913e-07,
      "loss": 1.6441,
      "step": 7434
    },
    {
      "epoch": 2.534083162917519,
      "grad_norm": 18.524799128037106,
      "learning_rate": 7.180779762852618e-07,
      "loss": 2.0216,
      "step": 7435
    },
    {
      "epoch": 2.534423994546694,
      "grad_norm": 16.2142834908373,
      "learning_rate": 7.170543728303119e-07,
      "loss": 1.2325,
      "step": 7436
    },
    {
      "epoch": 2.534764826175869,
      "grad_norm": 15.470132078984223,
      "learning_rate": 7.160314430997589e-07,
      "loss": 1.4961,
      "step": 7437
    },
    {
      "epoch": 2.5351056578050444,
      "grad_norm": 15.859849735897038,
      "learning_rate": 7.150091872545129e-07,
      "loss": 1.6399,
      "step": 7438
    },
    {
      "epoch": 2.5354464894342197,
      "grad_norm": 23.29977911765609,
      "learning_rate": 7.139876054553807e-07,
      "loss": 1.8268,
      "step": 7439
    },
    {
      "epoch": 2.5357873210633946,
      "grad_norm": 15.585284149163423,
      "learning_rate": 7.129666978630579e-07,
      "loss": 1.5916,
      "step": 7440
    },
    {
      "epoch": 2.53612815269257,
      "grad_norm": 16.425190801716614,
      "learning_rate": 7.119464646381397e-07,
      "loss": 1.2795,
      "step": 7441
    },
    {
      "epoch": 2.536468984321745,
      "grad_norm": 23.377126824157838,
      "learning_rate": 7.109269059411139e-07,
      "loss": 1.3624,
      "step": 7442
    },
    {
      "epoch": 2.53680981595092,
      "grad_norm": 14.67346501935475,
      "learning_rate": 7.099080219323567e-07,
      "loss": 1.8565,
      "step": 7443
    },
    {
      "epoch": 2.5371506475800953,
      "grad_norm": 14.010663884799238,
      "learning_rate": 7.088898127721489e-07,
      "loss": 0.7907,
      "step": 7444
    },
    {
      "epoch": 2.5374914792092707,
      "grad_norm": 16.79252131885413,
      "learning_rate": 7.078722786206532e-07,
      "loss": 1.2841,
      "step": 7445
    },
    {
      "epoch": 2.537832310838446,
      "grad_norm": 15.453622519245345,
      "learning_rate": 7.068554196379335e-07,
      "loss": 1.8678,
      "step": 7446
    },
    {
      "epoch": 2.5381731424676213,
      "grad_norm": 13.690597144248576,
      "learning_rate": 7.058392359839483e-07,
      "loss": 1.574,
      "step": 7447
    },
    {
      "epoch": 2.538513974096796,
      "grad_norm": 11.088629490097476,
      "learning_rate": 7.048237278185449e-07,
      "loss": 0.8778,
      "step": 7448
    },
    {
      "epoch": 2.5388548057259714,
      "grad_norm": 40.03221779026003,
      "learning_rate": 7.038088953014666e-07,
      "loss": 1.79,
      "step": 7449
    },
    {
      "epoch": 2.5391956373551468,
      "grad_norm": 10.48398376501175,
      "learning_rate": 7.027947385923511e-07,
      "loss": 0.938,
      "step": 7450
    },
    {
      "epoch": 2.5395364689843216,
      "grad_norm": 15.536694302111991,
      "learning_rate": 7.017812578507299e-07,
      "loss": 1.3474,
      "step": 7451
    },
    {
      "epoch": 2.539877300613497,
      "grad_norm": 16.2709075221783,
      "learning_rate": 7.007684532360265e-07,
      "loss": 1.5791,
      "step": 7452
    },
    {
      "epoch": 2.5402181322426722,
      "grad_norm": 15.60145672252382,
      "learning_rate": 6.9975632490756e-07,
      "loss": 1.194,
      "step": 7453
    },
    {
      "epoch": 2.540558963871847,
      "grad_norm": 12.002147655828264,
      "learning_rate": 6.987448730245422e-07,
      "loss": 1.3491,
      "step": 7454
    },
    {
      "epoch": 2.5408997955010224,
      "grad_norm": 9.868236115850912,
      "learning_rate": 6.977340977460778e-07,
      "loss": 1.0245,
      "step": 7455
    },
    {
      "epoch": 2.5412406271301977,
      "grad_norm": 17.341225885734325,
      "learning_rate": 6.967239992311658e-07,
      "loss": 1.748,
      "step": 7456
    },
    {
      "epoch": 2.541581458759373,
      "grad_norm": 15.842698171608319,
      "learning_rate": 6.957145776386998e-07,
      "loss": 1.3395,
      "step": 7457
    },
    {
      "epoch": 2.5419222903885483,
      "grad_norm": 15.35613491589239,
      "learning_rate": 6.947058331274658e-07,
      "loss": 1.5127,
      "step": 7458
    },
    {
      "epoch": 2.542263122017723,
      "grad_norm": 16.994190976461557,
      "learning_rate": 6.936977658561405e-07,
      "loss": 1.2297,
      "step": 7459
    },
    {
      "epoch": 2.5426039536468985,
      "grad_norm": 19.367637297241817,
      "learning_rate": 6.926903759833009e-07,
      "loss": 2.0924,
      "step": 7460
    },
    {
      "epoch": 2.542944785276074,
      "grad_norm": 10.610100308042385,
      "learning_rate": 6.916836636674113e-07,
      "loss": 1.1283,
      "step": 7461
    },
    {
      "epoch": 2.5432856169052487,
      "grad_norm": 17.881806431530663,
      "learning_rate": 6.906776290668299e-07,
      "loss": 1.2584,
      "step": 7462
    },
    {
      "epoch": 2.543626448534424,
      "grad_norm": 16.439164700097205,
      "learning_rate": 6.896722723398142e-07,
      "loss": 1.1934,
      "step": 7463
    },
    {
      "epoch": 2.5439672801635993,
      "grad_norm": 11.833445767299883,
      "learning_rate": 6.886675936445075e-07,
      "loss": 1.4468,
      "step": 7464
    },
    {
      "epoch": 2.544308111792774,
      "grad_norm": 10.311150731931095,
      "learning_rate": 6.876635931389508e-07,
      "loss": 0.7819,
      "step": 7465
    },
    {
      "epoch": 2.5446489434219495,
      "grad_norm": 13.69926120122674,
      "learning_rate": 6.866602709810777e-07,
      "loss": 1.1414,
      "step": 7466
    },
    {
      "epoch": 2.544989775051125,
      "grad_norm": 15.790945826421298,
      "learning_rate": 6.856576273287136e-07,
      "loss": 1.5101,
      "step": 7467
    },
    {
      "epoch": 2.5453306066803,
      "grad_norm": 13.357162892711615,
      "learning_rate": 6.846556623395795e-07,
      "loss": 1.3342,
      "step": 7468
    },
    {
      "epoch": 2.545671438309475,
      "grad_norm": 9.999257473381297,
      "learning_rate": 6.836543761712883e-07,
      "loss": 1.0993,
      "step": 7469
    },
    {
      "epoch": 2.5460122699386503,
      "grad_norm": 15.456092316377015,
      "learning_rate": 6.826537689813462e-07,
      "loss": 1.8562,
      "step": 7470
    },
    {
      "epoch": 2.5463531015678256,
      "grad_norm": 13.350889786251932,
      "learning_rate": 6.816538409271522e-07,
      "loss": 1.3233,
      "step": 7471
    },
    {
      "epoch": 2.546693933197001,
      "grad_norm": 16.00767677219847,
      "learning_rate": 6.806545921659996e-07,
      "loss": 0.9706,
      "step": 7472
    },
    {
      "epoch": 2.5470347648261757,
      "grad_norm": 12.916934163471954,
      "learning_rate": 6.796560228550747e-07,
      "loss": 1.3898,
      "step": 7473
    },
    {
      "epoch": 2.547375596455351,
      "grad_norm": 23.521405366526686,
      "learning_rate": 6.786581331514535e-07,
      "loss": 1.7989,
      "step": 7474
    },
    {
      "epoch": 2.5477164280845264,
      "grad_norm": 11.309291526506726,
      "learning_rate": 6.776609232121118e-07,
      "loss": 1.2695,
      "step": 7475
    },
    {
      "epoch": 2.5480572597137012,
      "grad_norm": 21.53296969970772,
      "learning_rate": 6.766643931939143e-07,
      "loss": 1.2414,
      "step": 7476
    },
    {
      "epoch": 2.5483980913428765,
      "grad_norm": 19.91683270165153,
      "learning_rate": 6.756685432536153e-07,
      "loss": 1.3956,
      "step": 7477
    },
    {
      "epoch": 2.548738922972052,
      "grad_norm": 12.69360505543974,
      "learning_rate": 6.746733735478711e-07,
      "loss": 1.4268,
      "step": 7478
    },
    {
      "epoch": 2.549079754601227,
      "grad_norm": 19.634314026974394,
      "learning_rate": 6.73678884233222e-07,
      "loss": 1.353,
      "step": 7479
    },
    {
      "epoch": 2.549420586230402,
      "grad_norm": 17.358765437518457,
      "learning_rate": 6.726850754661057e-07,
      "loss": 1.7538,
      "step": 7480
    },
    {
      "epoch": 2.5497614178595773,
      "grad_norm": 26.144711291586887,
      "learning_rate": 6.716919474028549e-07,
      "loss": 1.1928,
      "step": 7481
    },
    {
      "epoch": 2.5501022494887526,
      "grad_norm": 24.522609041085005,
      "learning_rate": 6.706995001996897e-07,
      "loss": 1.266,
      "step": 7482
    },
    {
      "epoch": 2.550443081117928,
      "grad_norm": 11.39086458431068,
      "learning_rate": 6.697077340127262e-07,
      "loss": 1.5944,
      "step": 7483
    },
    {
      "epoch": 2.550783912747103,
      "grad_norm": 23.077119385971162,
      "learning_rate": 6.687166489979763e-07,
      "loss": 1.281,
      "step": 7484
    },
    {
      "epoch": 2.551124744376278,
      "grad_norm": 17.94143664758822,
      "learning_rate": 6.677262453113381e-07,
      "loss": 1.0701,
      "step": 7485
    },
    {
      "epoch": 2.5514655760054534,
      "grad_norm": 19.24041315783454,
      "learning_rate": 6.667365231086076e-07,
      "loss": 1.6118,
      "step": 7486
    },
    {
      "epoch": 2.5518064076346283,
      "grad_norm": 14.80668599994533,
      "learning_rate": 6.657474825454712e-07,
      "loss": 1.1814,
      "step": 7487
    },
    {
      "epoch": 2.5521472392638036,
      "grad_norm": 16.992535209583476,
      "learning_rate": 6.647591237775098e-07,
      "loss": 1.4056,
      "step": 7488
    },
    {
      "epoch": 2.552488070892979,
      "grad_norm": 22.3806552825105,
      "learning_rate": 6.637714469601969e-07,
      "loss": 2.0174,
      "step": 7489
    },
    {
      "epoch": 2.552828902522154,
      "grad_norm": 14.698556921374086,
      "learning_rate": 6.627844522488947e-07,
      "loss": 1.4303,
      "step": 7490
    },
    {
      "epoch": 2.553169734151329,
      "grad_norm": 20.35596216705866,
      "learning_rate": 6.617981397988649e-07,
      "loss": 1.013,
      "step": 7491
    },
    {
      "epoch": 2.5535105657805044,
      "grad_norm": 12.385620679160192,
      "learning_rate": 6.608125097652579e-07,
      "loss": 1.1662,
      "step": 7492
    },
    {
      "epoch": 2.5538513974096797,
      "grad_norm": 15.436551592414942,
      "learning_rate": 6.598275623031142e-07,
      "loss": 1.5772,
      "step": 7493
    },
    {
      "epoch": 2.554192229038855,
      "grad_norm": 14.34127917401642,
      "learning_rate": 6.588432975673742e-07,
      "loss": 1.2449,
      "step": 7494
    },
    {
      "epoch": 2.55453306066803,
      "grad_norm": 15.642030454053655,
      "learning_rate": 6.578597157128642e-07,
      "loss": 1.6719,
      "step": 7495
    },
    {
      "epoch": 2.554873892297205,
      "grad_norm": 8.782445128229245,
      "learning_rate": 6.56876816894304e-07,
      "loss": 1.1693,
      "step": 7496
    },
    {
      "epoch": 2.5552147239263805,
      "grad_norm": 18.058506836493294,
      "learning_rate": 6.558946012663125e-07,
      "loss": 1.2447,
      "step": 7497
    },
    {
      "epoch": 2.5555555555555554,
      "grad_norm": 14.340363798815176,
      "learning_rate": 6.549130689833915e-07,
      "loss": 1.7319,
      "step": 7498
    },
    {
      "epoch": 2.5558963871847307,
      "grad_norm": 13.727178991245914,
      "learning_rate": 6.539322201999421e-07,
      "loss": 1.6583,
      "step": 7499
    },
    {
      "epoch": 2.556237218813906,
      "grad_norm": 13.534538491793398,
      "learning_rate": 6.529520550702545e-07,
      "loss": 1.7296,
      "step": 7500
    },
    {
      "epoch": 2.556578050443081,
      "grad_norm": 18.519079930487678,
      "learning_rate": 6.519725737485139e-07,
      "loss": 1.4234,
      "step": 7501
    },
    {
      "epoch": 2.556918882072256,
      "grad_norm": 11.827437430681869,
      "learning_rate": 6.509937763887953e-07,
      "loss": 0.9797,
      "step": 7502
    },
    {
      "epoch": 2.5572597137014315,
      "grad_norm": 15.94786915910308,
      "learning_rate": 6.500156631450682e-07,
      "loss": 1.6703,
      "step": 7503
    },
    {
      "epoch": 2.5576005453306068,
      "grad_norm": 15.871260150276552,
      "learning_rate": 6.490382341711937e-07,
      "loss": 1.7732,
      "step": 7504
    },
    {
      "epoch": 2.557941376959782,
      "grad_norm": 15.146228473401834,
      "learning_rate": 6.480614896209253e-07,
      "loss": 1.8148,
      "step": 7505
    },
    {
      "epoch": 2.558282208588957,
      "grad_norm": 34.5192280080981,
      "learning_rate": 6.470854296479084e-07,
      "loss": 1.7932,
      "step": 7506
    },
    {
      "epoch": 2.5586230402181322,
      "grad_norm": 8.937878576555123,
      "learning_rate": 6.461100544056814e-07,
      "loss": 1.0607,
      "step": 7507
    },
    {
      "epoch": 2.5589638718473076,
      "grad_norm": 16.755890482089264,
      "learning_rate": 6.451353640476749e-07,
      "loss": 1.6062,
      "step": 7508
    },
    {
      "epoch": 2.5593047034764824,
      "grad_norm": 14.683883607711413,
      "learning_rate": 6.44161358727211e-07,
      "loss": 1.4719,
      "step": 7509
    },
    {
      "epoch": 2.5596455351056577,
      "grad_norm": 13.72906187230467,
      "learning_rate": 6.431880385975064e-07,
      "loss": 1.1821,
      "step": 7510
    },
    {
      "epoch": 2.559986366734833,
      "grad_norm": 14.657126653716805,
      "learning_rate": 6.422154038116646e-07,
      "loss": 1.5824,
      "step": 7511
    },
    {
      "epoch": 2.560327198364008,
      "grad_norm": 10.756076837967816,
      "learning_rate": 6.412434545226876e-07,
      "loss": 1.51,
      "step": 7512
    },
    {
      "epoch": 2.560668029993183,
      "grad_norm": 14.490943693257083,
      "learning_rate": 6.40272190883468e-07,
      "loss": 1.2049,
      "step": 7513
    },
    {
      "epoch": 2.5610088616223585,
      "grad_norm": 18.056694077025742,
      "learning_rate": 6.393016130467855e-07,
      "loss": 1.7334,
      "step": 7514
    },
    {
      "epoch": 2.561349693251534,
      "grad_norm": 18.468549011510394,
      "learning_rate": 6.383317211653206e-07,
      "loss": 1.7978,
      "step": 7515
    },
    {
      "epoch": 2.561690524880709,
      "grad_norm": 13.802332533766752,
      "learning_rate": 6.37362515391638e-07,
      "loss": 1.2166,
      "step": 7516
    },
    {
      "epoch": 2.562031356509884,
      "grad_norm": 13.975517711085459,
      "learning_rate": 6.363939958781968e-07,
      "loss": 1.5202,
      "step": 7517
    },
    {
      "epoch": 2.5623721881390593,
      "grad_norm": 28.932858433824084,
      "learning_rate": 6.354261627773528e-07,
      "loss": 1.3765,
      "step": 7518
    },
    {
      "epoch": 2.5627130197682346,
      "grad_norm": 17.50363424532597,
      "learning_rate": 6.344590162413467e-07,
      "loss": 1.363,
      "step": 7519
    },
    {
      "epoch": 2.5630538513974095,
      "grad_norm": 16.243152362631704,
      "learning_rate": 6.334925564223154e-07,
      "loss": 0.8087,
      "step": 7520
    },
    {
      "epoch": 2.563394683026585,
      "grad_norm": 13.676637689773194,
      "learning_rate": 6.32526783472287e-07,
      "loss": 1.1545,
      "step": 7521
    },
    {
      "epoch": 2.56373551465576,
      "grad_norm": 18.626554144425764,
      "learning_rate": 6.315616975431816e-07,
      "loss": 1.6043,
      "step": 7522
    },
    {
      "epoch": 2.564076346284935,
      "grad_norm": 14.68966546310584,
      "learning_rate": 6.30597298786812e-07,
      "loss": 1.3608,
      "step": 7523
    },
    {
      "epoch": 2.5644171779141103,
      "grad_norm": 21.410316737982576,
      "learning_rate": 6.296335873548787e-07,
      "loss": 1.4842,
      "step": 7524
    },
    {
      "epoch": 2.5647580095432856,
      "grad_norm": 16.585501366665145,
      "learning_rate": 6.28670563398981e-07,
      "loss": 0.7889,
      "step": 7525
    },
    {
      "epoch": 2.565098841172461,
      "grad_norm": 15.756272525125036,
      "learning_rate": 6.277082270706053e-07,
      "loss": 1.3538,
      "step": 7526
    },
    {
      "epoch": 2.565439672801636,
      "grad_norm": 17.386642965041144,
      "learning_rate": 6.267465785211291e-07,
      "loss": 1.3714,
      "step": 7527
    },
    {
      "epoch": 2.565780504430811,
      "grad_norm": 15.174932697990073,
      "learning_rate": 6.257856179018268e-07,
      "loss": 1.8268,
      "step": 7528
    },
    {
      "epoch": 2.5661213360599864,
      "grad_norm": 12.935610519893192,
      "learning_rate": 6.248253453638586e-07,
      "loss": 1.0392,
      "step": 7529
    },
    {
      "epoch": 2.5664621676891617,
      "grad_norm": 15.93124030674824,
      "learning_rate": 6.238657610582788e-07,
      "loss": 1.3081,
      "step": 7530
    },
    {
      "epoch": 2.5668029993183366,
      "grad_norm": 12.40525804926018,
      "learning_rate": 6.229068651360376e-07,
      "loss": 1.1104,
      "step": 7531
    },
    {
      "epoch": 2.567143830947512,
      "grad_norm": 12.795507694260392,
      "learning_rate": 6.219486577479694e-07,
      "loss": 1.2855,
      "step": 7532
    },
    {
      "epoch": 2.567484662576687,
      "grad_norm": 10.745389066872951,
      "learning_rate": 6.209911390448053e-07,
      "loss": 1.1567,
      "step": 7533
    },
    {
      "epoch": 2.567825494205862,
      "grad_norm": 14.301298562606224,
      "learning_rate": 6.200343091771671e-07,
      "loss": 1.2952,
      "step": 7534
    },
    {
      "epoch": 2.5681663258350373,
      "grad_norm": 14.34397698997906,
      "learning_rate": 6.190781682955676e-07,
      "loss": 1.2093,
      "step": 7535
    },
    {
      "epoch": 2.5685071574642127,
      "grad_norm": 12.579256822064343,
      "learning_rate": 6.181227165504123e-07,
      "loss": 1.3373,
      "step": 7536
    },
    {
      "epoch": 2.568847989093388,
      "grad_norm": 24.252253435035048,
      "learning_rate": 6.171679540919961e-07,
      "loss": 1.2051,
      "step": 7537
    },
    {
      "epoch": 2.5691888207225633,
      "grad_norm": 13.089668003072703,
      "learning_rate": 6.162138810705082e-07,
      "loss": 1.7413,
      "step": 7538
    },
    {
      "epoch": 2.569529652351738,
      "grad_norm": 21.068002866637766,
      "learning_rate": 6.152604976360271e-07,
      "loss": 1.1302,
      "step": 7539
    },
    {
      "epoch": 2.5698704839809134,
      "grad_norm": 19.496122597180182,
      "learning_rate": 6.143078039385247e-07,
      "loss": 1.7931,
      "step": 7540
    },
    {
      "epoch": 2.5702113156100888,
      "grad_norm": 13.628838270755262,
      "learning_rate": 6.133558001278633e-07,
      "loss": 1.0128,
      "step": 7541
    },
    {
      "epoch": 2.5705521472392636,
      "grad_norm": 19.55965554832037,
      "learning_rate": 6.12404486353797e-07,
      "loss": 1.6733,
      "step": 7542
    },
    {
      "epoch": 2.570892978868439,
      "grad_norm": 16.170665940253034,
      "learning_rate": 6.114538627659705e-07,
      "loss": 1.3652,
      "step": 7543
    },
    {
      "epoch": 2.5712338104976142,
      "grad_norm": 24.859163801150487,
      "learning_rate": 6.105039295139226e-07,
      "loss": 1.34,
      "step": 7544
    },
    {
      "epoch": 2.571574642126789,
      "grad_norm": 13.570133449185306,
      "learning_rate": 6.095546867470776e-07,
      "loss": 1.4805,
      "step": 7545
    },
    {
      "epoch": 2.5719154737559644,
      "grad_norm": 26.230587614994175,
      "learning_rate": 6.086061346147593e-07,
      "loss": 1.3337,
      "step": 7546
    },
    {
      "epoch": 2.5722563053851397,
      "grad_norm": 15.032702086385363,
      "learning_rate": 6.076582732661778e-07,
      "loss": 1.2779,
      "step": 7547
    },
    {
      "epoch": 2.572597137014315,
      "grad_norm": 19.000299418117,
      "learning_rate": 6.067111028504324e-07,
      "loss": 1.7698,
      "step": 7548
    },
    {
      "epoch": 2.5729379686434903,
      "grad_norm": 11.023054838700272,
      "learning_rate": 6.05764623516522e-07,
      "loss": 1.162,
      "step": 7549
    },
    {
      "epoch": 2.573278800272665,
      "grad_norm": 11.372707914925734,
      "learning_rate": 6.048188354133272e-07,
      "loss": 1.2493,
      "step": 7550
    },
    {
      "epoch": 2.5736196319018405,
      "grad_norm": 19.74772530419478,
      "learning_rate": 6.038737386896254e-07,
      "loss": 1.4789,
      "step": 7551
    },
    {
      "epoch": 2.573960463531016,
      "grad_norm": 18.413126976107833,
      "learning_rate": 6.029293334940862e-07,
      "loss": 1.2351,
      "step": 7552
    },
    {
      "epoch": 2.5743012951601907,
      "grad_norm": 11.407322138984718,
      "learning_rate": 6.019856199752655e-07,
      "loss": 1.1112,
      "step": 7553
    },
    {
      "epoch": 2.574642126789366,
      "grad_norm": 13.522349350624626,
      "learning_rate": 6.010425982816143e-07,
      "loss": 0.7853,
      "step": 7554
    },
    {
      "epoch": 2.5749829584185413,
      "grad_norm": 19.00770798126287,
      "learning_rate": 6.001002685614732e-07,
      "loss": 1.8336,
      "step": 7555
    },
    {
      "epoch": 2.575323790047716,
      "grad_norm": 21.289596753311248,
      "learning_rate": 5.991586309630754e-07,
      "loss": 0.8119,
      "step": 7556
    },
    {
      "epoch": 2.5756646216768915,
      "grad_norm": 14.838626665225064,
      "learning_rate": 5.982176856345445e-07,
      "loss": 1.5879,
      "step": 7557
    },
    {
      "epoch": 2.576005453306067,
      "grad_norm": 12.274013150311129,
      "learning_rate": 5.972774327238923e-07,
      "loss": 1.5406,
      "step": 7558
    },
    {
      "epoch": 2.576346284935242,
      "grad_norm": 24.81919635482738,
      "learning_rate": 5.96337872379027e-07,
      "loss": 1.49,
      "step": 7559
    },
    {
      "epoch": 2.5766871165644174,
      "grad_norm": 18.391527502845687,
      "learning_rate": 5.953990047477454e-07,
      "loss": 1.3091,
      "step": 7560
    },
    {
      "epoch": 2.5770279481935923,
      "grad_norm": 12.64798912522365,
      "learning_rate": 5.944608299777316e-07,
      "loss": 1.0762,
      "step": 7561
    },
    {
      "epoch": 2.5773687798227676,
      "grad_norm": 18.649494734354118,
      "learning_rate": 5.935233482165692e-07,
      "loss": 1.446,
      "step": 7562
    },
    {
      "epoch": 2.577709611451943,
      "grad_norm": 17.608749283860497,
      "learning_rate": 5.925865596117242e-07,
      "loss": 1.6205,
      "step": 7563
    },
    {
      "epoch": 2.5780504430811177,
      "grad_norm": 14.482754492052926,
      "learning_rate": 5.91650464310557e-07,
      "loss": 1.786,
      "step": 7564
    },
    {
      "epoch": 2.578391274710293,
      "grad_norm": 14.44253574409249,
      "learning_rate": 5.907150624603225e-07,
      "loss": 1.581,
      "step": 7565
    },
    {
      "epoch": 2.5787321063394684,
      "grad_norm": 15.332856588864901,
      "learning_rate": 5.897803542081604e-07,
      "loss": 1.8783,
      "step": 7566
    },
    {
      "epoch": 2.5790729379686432,
      "grad_norm": 24.083653045672996,
      "learning_rate": 5.888463397011046e-07,
      "loss": 1.2323,
      "step": 7567
    },
    {
      "epoch": 2.5794137695978185,
      "grad_norm": 18.079461208672765,
      "learning_rate": 5.879130190860788e-07,
      "loss": 1.4314,
      "step": 7568
    },
    {
      "epoch": 2.579754601226994,
      "grad_norm": 18.250169969097662,
      "learning_rate": 5.869803925098994e-07,
      "loss": 1.0131,
      "step": 7569
    },
    {
      "epoch": 2.580095432856169,
      "grad_norm": 12.78784255549784,
      "learning_rate": 5.860484601192706e-07,
      "loss": 1.0866,
      "step": 7570
    },
    {
      "epoch": 2.5804362644853445,
      "grad_norm": 15.016282839827316,
      "learning_rate": 5.851172220607903e-07,
      "loss": 1.2379,
      "step": 7571
    },
    {
      "epoch": 2.5807770961145193,
      "grad_norm": 11.108655055531205,
      "learning_rate": 5.841866784809457e-07,
      "loss": 1.1546,
      "step": 7572
    },
    {
      "epoch": 2.5811179277436946,
      "grad_norm": 15.480034018147299,
      "learning_rate": 5.832568295261148e-07,
      "loss": 1.5815,
      "step": 7573
    },
    {
      "epoch": 2.58145875937287,
      "grad_norm": 10.10583016966508,
      "learning_rate": 5.823276753425672e-07,
      "loss": 1.1795,
      "step": 7574
    },
    {
      "epoch": 2.581799591002045,
      "grad_norm": 23.109608434869788,
      "learning_rate": 5.813992160764614e-07,
      "loss": 1.1411,
      "step": 7575
    },
    {
      "epoch": 2.58214042263122,
      "grad_norm": 14.274630908388552,
      "learning_rate": 5.804714518738486e-07,
      "loss": 1.3388,
      "step": 7576
    },
    {
      "epoch": 2.5824812542603954,
      "grad_norm": 12.34232514961632,
      "learning_rate": 5.79544382880669e-07,
      "loss": 1.5614,
      "step": 7577
    },
    {
      "epoch": 2.5828220858895703,
      "grad_norm": 13.85441329552113,
      "learning_rate": 5.786180092427562e-07,
      "loss": 1.4195,
      "step": 7578
    },
    {
      "epoch": 2.5831629175187456,
      "grad_norm": 14.712760473473,
      "learning_rate": 5.776923311058292e-07,
      "loss": 1.2241,
      "step": 7579
    },
    {
      "epoch": 2.583503749147921,
      "grad_norm": 14.84973221142647,
      "learning_rate": 5.767673486155034e-07,
      "loss": 2.0231,
      "step": 7580
    },
    {
      "epoch": 2.5838445807770962,
      "grad_norm": 12.813271959484416,
      "learning_rate": 5.758430619172828e-07,
      "loss": 1.0574,
      "step": 7581
    },
    {
      "epoch": 2.5841854124062715,
      "grad_norm": 15.873306573370773,
      "learning_rate": 5.749194711565581e-07,
      "loss": 1.719,
      "step": 7582
    },
    {
      "epoch": 2.5845262440354464,
      "grad_norm": 17.77459886164835,
      "learning_rate": 5.739965764786177e-07,
      "loss": 1.8418,
      "step": 7583
    },
    {
      "epoch": 2.5848670756646217,
      "grad_norm": 15.26556484009428,
      "learning_rate": 5.73074378028634e-07,
      "loss": 1.2007,
      "step": 7584
    },
    {
      "epoch": 2.585207907293797,
      "grad_norm": 23.560816372382835,
      "learning_rate": 5.721528759516721e-07,
      "loss": 0.9166,
      "step": 7585
    },
    {
      "epoch": 2.585548738922972,
      "grad_norm": 16.23121556706933,
      "learning_rate": 5.712320703926905e-07,
      "loss": 1.4422,
      "step": 7586
    },
    {
      "epoch": 2.585889570552147,
      "grad_norm": 16.78378858772311,
      "learning_rate": 5.703119614965335e-07,
      "loss": 1.4507,
      "step": 7587
    },
    {
      "epoch": 2.5862304021813225,
      "grad_norm": 13.21942248440853,
      "learning_rate": 5.693925494079389e-07,
      "loss": 1.529,
      "step": 7588
    },
    {
      "epoch": 2.5865712338104974,
      "grad_norm": 15.924732289009594,
      "learning_rate": 5.684738342715329e-07,
      "loss": 1.0063,
      "step": 7589
    },
    {
      "epoch": 2.5869120654396727,
      "grad_norm": 12.923813563974988,
      "learning_rate": 5.67555816231834e-07,
      "loss": 1.2073,
      "step": 7590
    },
    {
      "epoch": 2.587252897068848,
      "grad_norm": 11.880310242561448,
      "learning_rate": 5.666384954332505e-07,
      "loss": 1.4026,
      "step": 7591
    },
    {
      "epoch": 2.5875937286980233,
      "grad_norm": 17.610140269477185,
      "learning_rate": 5.65721872020078e-07,
      "loss": 1.3519,
      "step": 7592
    },
    {
      "epoch": 2.5879345603271986,
      "grad_norm": 14.023329970917652,
      "learning_rate": 5.64805946136508e-07,
      "loss": 1.616,
      "step": 7593
    },
    {
      "epoch": 2.5882753919563735,
      "grad_norm": 14.68695762808496,
      "learning_rate": 5.63890717926619e-07,
      "loss": 1.1191,
      "step": 7594
    },
    {
      "epoch": 2.5886162235855488,
      "grad_norm": 10.044929801869007,
      "learning_rate": 5.629761875343776e-07,
      "loss": 0.9482,
      "step": 7595
    },
    {
      "epoch": 2.588957055214724,
      "grad_norm": 14.585230726878313,
      "learning_rate": 5.620623551036453e-07,
      "loss": 1.7916,
      "step": 7596
    },
    {
      "epoch": 2.589297886843899,
      "grad_norm": 21.632658396129237,
      "learning_rate": 5.611492207781715e-07,
      "loss": 1.2568,
      "step": 7597
    },
    {
      "epoch": 2.5896387184730743,
      "grad_norm": 25.547891239589216,
      "learning_rate": 5.602367847015938e-07,
      "loss": 1.3806,
      "step": 7598
    },
    {
      "epoch": 2.5899795501022496,
      "grad_norm": 15.360920041375469,
      "learning_rate": 5.59325047017446e-07,
      "loss": 1.2547,
      "step": 7599
    },
    {
      "epoch": 2.5903203817314244,
      "grad_norm": 16.886976862212187,
      "learning_rate": 5.584140078691441e-07,
      "loss": 1.084,
      "step": 7600
    },
    {
      "epoch": 2.5906612133605997,
      "grad_norm": 31.08548770146703,
      "learning_rate": 5.575036673999984e-07,
      "loss": 1.4638,
      "step": 7601
    },
    {
      "epoch": 2.591002044989775,
      "grad_norm": 12.376511342749444,
      "learning_rate": 5.565940257532132e-07,
      "loss": 1.09,
      "step": 7602
    },
    {
      "epoch": 2.5913428766189504,
      "grad_norm": 12.099344339808326,
      "learning_rate": 5.556850830718741e-07,
      "loss": 1.1049,
      "step": 7603
    },
    {
      "epoch": 2.5916837082481257,
      "grad_norm": 10.665697961255859,
      "learning_rate": 5.547768394989638e-07,
      "loss": 1.3053,
      "step": 7604
    },
    {
      "epoch": 2.5920245398773005,
      "grad_norm": 12.657019809649363,
      "learning_rate": 5.538692951773517e-07,
      "loss": 1.1978,
      "step": 7605
    },
    {
      "epoch": 2.592365371506476,
      "grad_norm": 18.163908797838573,
      "learning_rate": 5.529624502497988e-07,
      "loss": 1.6117,
      "step": 7606
    },
    {
      "epoch": 2.592706203135651,
      "grad_norm": 14.95999624579253,
      "learning_rate": 5.520563048589545e-07,
      "loss": 1.3498,
      "step": 7607
    },
    {
      "epoch": 2.593047034764826,
      "grad_norm": 20.506759943613577,
      "learning_rate": 5.511508591473602e-07,
      "loss": 1.1816,
      "step": 7608
    },
    {
      "epoch": 2.5933878663940013,
      "grad_norm": 15.223233374474844,
      "learning_rate": 5.502461132574455e-07,
      "loss": 1.3085,
      "step": 7609
    },
    {
      "epoch": 2.5937286980231766,
      "grad_norm": 13.969090668177662,
      "learning_rate": 5.493420673315308e-07,
      "loss": 0.9485,
      "step": 7610
    },
    {
      "epoch": 2.5940695296523515,
      "grad_norm": 9.883939125038136,
      "learning_rate": 5.484387215118259e-07,
      "loss": 1.0597,
      "step": 7611
    },
    {
      "epoch": 2.594410361281527,
      "grad_norm": 12.758281531446567,
      "learning_rate": 5.475360759404313e-07,
      "loss": 1.0226,
      "step": 7612
    },
    {
      "epoch": 2.594751192910702,
      "grad_norm": 19.21004808077746,
      "learning_rate": 5.466341307593348e-07,
      "loss": 1.7968,
      "step": 7613
    },
    {
      "epoch": 2.5950920245398774,
      "grad_norm": 14.862162960826184,
      "learning_rate": 5.457328861104177e-07,
      "loss": 1.6059,
      "step": 7614
    },
    {
      "epoch": 2.5954328561690527,
      "grad_norm": 19.632711903054332,
      "learning_rate": 5.448323421354507e-07,
      "loss": 1.963,
      "step": 7615
    },
    {
      "epoch": 2.5957736877982276,
      "grad_norm": 13.2227837257219,
      "learning_rate": 5.439324989760891e-07,
      "loss": 1.1763,
      "step": 7616
    },
    {
      "epoch": 2.596114519427403,
      "grad_norm": 16.278360233951233,
      "learning_rate": 5.430333567738854e-07,
      "loss": 1.5994,
      "step": 7617
    },
    {
      "epoch": 2.596455351056578,
      "grad_norm": 17.91166564730727,
      "learning_rate": 5.421349156702765e-07,
      "loss": 1.1361,
      "step": 7618
    },
    {
      "epoch": 2.596796182685753,
      "grad_norm": 20.717650706314274,
      "learning_rate": 5.41237175806591e-07,
      "loss": 1.2107,
      "step": 7619
    },
    {
      "epoch": 2.5971370143149284,
      "grad_norm": 12.3617290221532,
      "learning_rate": 5.403401373240469e-07,
      "loss": 1.7145,
      "step": 7620
    },
    {
      "epoch": 2.5974778459441037,
      "grad_norm": 11.417697086418292,
      "learning_rate": 5.39443800363752e-07,
      "loss": 1.0997,
      "step": 7621
    },
    {
      "epoch": 2.5978186775732786,
      "grad_norm": 13.692385585454653,
      "learning_rate": 5.385481650667041e-07,
      "loss": 1.5414,
      "step": 7622
    },
    {
      "epoch": 2.598159509202454,
      "grad_norm": 13.265281185028426,
      "learning_rate": 5.376532315737898e-07,
      "loss": 1.3023,
      "step": 7623
    },
    {
      "epoch": 2.598500340831629,
      "grad_norm": 14.708475582283823,
      "learning_rate": 5.367590000257855e-07,
      "loss": 1.7442,
      "step": 7624
    },
    {
      "epoch": 2.5988411724608045,
      "grad_norm": 12.855259601572147,
      "learning_rate": 5.358654705633576e-07,
      "loss": 1.3278,
      "step": 7625
    },
    {
      "epoch": 2.59918200408998,
      "grad_norm": 13.344797115151028,
      "learning_rate": 5.349726433270619e-07,
      "loss": 1.3636,
      "step": 7626
    },
    {
      "epoch": 2.5995228357191547,
      "grad_norm": 13.081721869459685,
      "learning_rate": 5.340805184573438e-07,
      "loss": 1.5649,
      "step": 7627
    },
    {
      "epoch": 2.59986366734833,
      "grad_norm": 16.298469232670715,
      "learning_rate": 5.33189096094539e-07,
      "loss": 1.2182,
      "step": 7628
    },
    {
      "epoch": 2.6002044989775053,
      "grad_norm": 13.839660768432886,
      "learning_rate": 5.322983763788686e-07,
      "loss": 1.5229,
      "step": 7629
    },
    {
      "epoch": 2.60054533060668,
      "grad_norm": 18.870286932878596,
      "learning_rate": 5.314083594504493e-07,
      "loss": 1.2202,
      "step": 7630
    },
    {
      "epoch": 2.6008861622358554,
      "grad_norm": 17.040723310454997,
      "learning_rate": 5.305190454492848e-07,
      "loss": 1.5158,
      "step": 7631
    },
    {
      "epoch": 2.6012269938650308,
      "grad_norm": 17.869614736498,
      "learning_rate": 5.296304345152642e-07,
      "loss": 1.6303,
      "step": 7632
    },
    {
      "epoch": 2.6015678254942056,
      "grad_norm": 13.229862487674161,
      "learning_rate": 5.28742526788174e-07,
      "loss": 0.6866,
      "step": 7633
    },
    {
      "epoch": 2.601908657123381,
      "grad_norm": 10.334670869950893,
      "learning_rate": 5.278553224076821e-07,
      "loss": 1.0687,
      "step": 7634
    },
    {
      "epoch": 2.6022494887525562,
      "grad_norm": 13.435897346647335,
      "learning_rate": 5.2696882151335e-07,
      "loss": 1.4004,
      "step": 7635
    },
    {
      "epoch": 2.6025903203817315,
      "grad_norm": 9.700081488956696,
      "learning_rate": 5.260830242446302e-07,
      "loss": 1.0694,
      "step": 7636
    },
    {
      "epoch": 2.602931152010907,
      "grad_norm": 13.0599868615492,
      "learning_rate": 5.251979307408589e-07,
      "loss": 1.4749,
      "step": 7637
    },
    {
      "epoch": 2.6032719836400817,
      "grad_norm": 11.395234467712974,
      "learning_rate": 5.243135411412665e-07,
      "loss": 1.4845,
      "step": 7638
    },
    {
      "epoch": 2.603612815269257,
      "grad_norm": 10.771842203211738,
      "learning_rate": 5.234298555849709e-07,
      "loss": 0.8352,
      "step": 7639
    },
    {
      "epoch": 2.6039536468984323,
      "grad_norm": 18.678632086138453,
      "learning_rate": 5.22546874210979e-07,
      "loss": 1.7726,
      "step": 7640
    },
    {
      "epoch": 2.604294478527607,
      "grad_norm": 16.53284863547154,
      "learning_rate": 5.21664597158188e-07,
      "loss": 0.9549,
      "step": 7641
    },
    {
      "epoch": 2.6046353101567825,
      "grad_norm": 13.379710825594456,
      "learning_rate": 5.207830245653827e-07,
      "loss": 1.4244,
      "step": 7642
    },
    {
      "epoch": 2.604976141785958,
      "grad_norm": 13.192893711602583,
      "learning_rate": 5.199021565712381e-07,
      "loss": 1.5713,
      "step": 7643
    },
    {
      "epoch": 2.6053169734151327,
      "grad_norm": 14.416678380847602,
      "learning_rate": 5.190219933143193e-07,
      "loss": 1.7034,
      "step": 7644
    },
    {
      "epoch": 2.605657805044308,
      "grad_norm": 18.743131930181526,
      "learning_rate": 5.181425349330776e-07,
      "loss": 1.4534,
      "step": 7645
    },
    {
      "epoch": 2.6059986366734833,
      "grad_norm": 15.085510731503705,
      "learning_rate": 5.172637815658583e-07,
      "loss": 1.4469,
      "step": 7646
    },
    {
      "epoch": 2.6063394683026586,
      "grad_norm": 14.241416708138514,
      "learning_rate": 5.163857333508888e-07,
      "loss": 1.4272,
      "step": 7647
    },
    {
      "epoch": 2.606680299931834,
      "grad_norm": 20.9475130667021,
      "learning_rate": 5.155083904262925e-07,
      "loss": 1.3887,
      "step": 7648
    },
    {
      "epoch": 2.607021131561009,
      "grad_norm": 9.65156605279421,
      "learning_rate": 5.146317529300787e-07,
      "loss": 1.3087,
      "step": 7649
    },
    {
      "epoch": 2.607361963190184,
      "grad_norm": 15.572629204454698,
      "learning_rate": 5.137558210001437e-07,
      "loss": 1.9056,
      "step": 7650
    },
    {
      "epoch": 2.6077027948193594,
      "grad_norm": 14.545494644640566,
      "learning_rate": 5.128805947742788e-07,
      "loss": 1.4606,
      "step": 7651
    },
    {
      "epoch": 2.6080436264485343,
      "grad_norm": 12.764362570214423,
      "learning_rate": 5.120060743901578e-07,
      "loss": 1.5002,
      "step": 7652
    },
    {
      "epoch": 2.6083844580777096,
      "grad_norm": 25.68624695452573,
      "learning_rate": 5.111322599853469e-07,
      "loss": 1.2751,
      "step": 7653
    },
    {
      "epoch": 2.608725289706885,
      "grad_norm": 18.782897951184097,
      "learning_rate": 5.102591516973005e-07,
      "loss": 1.2298,
      "step": 7654
    },
    {
      "epoch": 2.6090661213360598,
      "grad_norm": 18.376331997446762,
      "learning_rate": 5.09386749663362e-07,
      "loss": 1.5143,
      "step": 7655
    },
    {
      "epoch": 2.609406952965235,
      "grad_norm": 16.00591132101434,
      "learning_rate": 5.085150540207639e-07,
      "loss": 1.1487,
      "step": 7656
    },
    {
      "epoch": 2.6097477845944104,
      "grad_norm": 12.911772052321503,
      "learning_rate": 5.076440649066277e-07,
      "loss": 1.0996,
      "step": 7657
    },
    {
      "epoch": 2.6100886162235857,
      "grad_norm": 15.583305808746854,
      "learning_rate": 5.067737824579627e-07,
      "loss": 1.392,
      "step": 7658
    },
    {
      "epoch": 2.610429447852761,
      "grad_norm": 14.556019346639575,
      "learning_rate": 5.059042068116687e-07,
      "loss": 1.711,
      "step": 7659
    },
    {
      "epoch": 2.610770279481936,
      "grad_norm": 22.784794184067323,
      "learning_rate": 5.050353381045326e-07,
      "loss": 2.064,
      "step": 7660
    },
    {
      "epoch": 2.611111111111111,
      "grad_norm": 14.885913992818663,
      "learning_rate": 5.041671764732314e-07,
      "loss": 1.4587,
      "step": 7661
    },
    {
      "epoch": 2.6114519427402865,
      "grad_norm": 9.712231079307701,
      "learning_rate": 5.032997220543307e-07,
      "loss": 1.1508,
      "step": 7662
    },
    {
      "epoch": 2.6117927743694613,
      "grad_norm": 9.554327363543404,
      "learning_rate": 5.024329749842827e-07,
      "loss": 1.1248,
      "step": 7663
    },
    {
      "epoch": 2.6121336059986366,
      "grad_norm": 10.053131744416408,
      "learning_rate": 5.015669353994324e-07,
      "loss": 1.1432,
      "step": 7664
    },
    {
      "epoch": 2.612474437627812,
      "grad_norm": 14.611327011695462,
      "learning_rate": 5.007016034360113e-07,
      "loss": 1.3088,
      "step": 7665
    },
    {
      "epoch": 2.612815269256987,
      "grad_norm": 14.888229469918,
      "learning_rate": 4.998369792301366e-07,
      "loss": 1.7243,
      "step": 7666
    },
    {
      "epoch": 2.613156100886162,
      "grad_norm": 13.12397123450047,
      "learning_rate": 4.989730629178208e-07,
      "loss": 1.2319,
      "step": 7667
    },
    {
      "epoch": 2.6134969325153374,
      "grad_norm": 15.232165850450656,
      "learning_rate": 4.981098546349594e-07,
      "loss": 1.4397,
      "step": 7668
    },
    {
      "epoch": 2.6138377641445127,
      "grad_norm": 13.940008269027095,
      "learning_rate": 4.972473545173379e-07,
      "loss": 1.2343,
      "step": 7669
    },
    {
      "epoch": 2.614178595773688,
      "grad_norm": 18.99113390079362,
      "learning_rate": 4.963855627006331e-07,
      "loss": 1.1965,
      "step": 7670
    },
    {
      "epoch": 2.614519427402863,
      "grad_norm": 11.514383419468558,
      "learning_rate": 4.955244793204067e-07,
      "loss": 1.0754,
      "step": 7671
    },
    {
      "epoch": 2.6148602590320382,
      "grad_norm": 11.392428016814703,
      "learning_rate": 4.946641045121115e-07,
      "loss": 1.0203,
      "step": 7672
    },
    {
      "epoch": 2.6152010906612135,
      "grad_norm": 11.587284746616552,
      "learning_rate": 4.938044384110868e-07,
      "loss": 1.4241,
      "step": 7673
    },
    {
      "epoch": 2.6155419222903884,
      "grad_norm": 12.859058843385037,
      "learning_rate": 4.92945481152562e-07,
      "loss": 1.4092,
      "step": 7674
    },
    {
      "epoch": 2.6158827539195637,
      "grad_norm": 17.833400560899033,
      "learning_rate": 4.920872328716547e-07,
      "loss": 1.281,
      "step": 7675
    },
    {
      "epoch": 2.616223585548739,
      "grad_norm": 19.14000646312008,
      "learning_rate": 4.912296937033706e-07,
      "loss": 1.6787,
      "step": 7676
    },
    {
      "epoch": 2.616564417177914,
      "grad_norm": 12.036128791014066,
      "learning_rate": 4.903728637826038e-07,
      "loss": 1.3302,
      "step": 7677
    },
    {
      "epoch": 2.616905248807089,
      "grad_norm": 13.68930587287639,
      "learning_rate": 4.895167432441378e-07,
      "loss": 1.3586,
      "step": 7678
    },
    {
      "epoch": 2.6172460804362645,
      "grad_norm": 16.089817213294467,
      "learning_rate": 4.88661332222643e-07,
      "loss": 1.611,
      "step": 7679
    },
    {
      "epoch": 2.61758691206544,
      "grad_norm": 13.663495264062934,
      "learning_rate": 4.878066308526803e-07,
      "loss": 0.7885,
      "step": 7680
    },
    {
      "epoch": 2.617927743694615,
      "grad_norm": 36.899342208629406,
      "learning_rate": 4.869526392686946e-07,
      "loss": 1.1679,
      "step": 7681
    },
    {
      "epoch": 2.61826857532379,
      "grad_norm": 12.788538713030725,
      "learning_rate": 4.860993576050249e-07,
      "loss": 1.4739,
      "step": 7682
    },
    {
      "epoch": 2.6186094069529653,
      "grad_norm": 14.601751201818121,
      "learning_rate": 4.852467859958965e-07,
      "loss": 1.0906,
      "step": 7683
    },
    {
      "epoch": 2.6189502385821406,
      "grad_norm": 12.753006114731432,
      "learning_rate": 4.843949245754192e-07,
      "loss": 1.5752,
      "step": 7684
    },
    {
      "epoch": 2.6192910702113155,
      "grad_norm": 18.55938115571647,
      "learning_rate": 4.835437734775966e-07,
      "loss": 1.706,
      "step": 7685
    },
    {
      "epoch": 2.6196319018404908,
      "grad_norm": 15.076659195420685,
      "learning_rate": 4.82693332836317e-07,
      "loss": 1.7997,
      "step": 7686
    },
    {
      "epoch": 2.619972733469666,
      "grad_norm": 20.023439794461428,
      "learning_rate": 4.818436027853584e-07,
      "loss": 1.1161,
      "step": 7687
    },
    {
      "epoch": 2.620313565098841,
      "grad_norm": 14.260296081364507,
      "learning_rate": 4.809945834583873e-07,
      "loss": 1.0781,
      "step": 7688
    },
    {
      "epoch": 2.6206543967280163,
      "grad_norm": 12.233580707751281,
      "learning_rate": 4.801462749889568e-07,
      "loss": 1.0002,
      "step": 7689
    },
    {
      "epoch": 2.6209952283571916,
      "grad_norm": 13.589209495368374,
      "learning_rate": 4.792986775105102e-07,
      "loss": 1.4921,
      "step": 7690
    },
    {
      "epoch": 2.621336059986367,
      "grad_norm": 16.023154162511414,
      "learning_rate": 4.784517911563774e-07,
      "loss": 1.3418,
      "step": 7691
    },
    {
      "epoch": 2.621676891615542,
      "grad_norm": 15.52327331704132,
      "learning_rate": 4.776056160597769e-07,
      "loss": 1.2877,
      "step": 7692
    },
    {
      "epoch": 2.622017723244717,
      "grad_norm": 16.047076298300865,
      "learning_rate": 4.767601523538157e-07,
      "loss": 2.0197,
      "step": 7693
    },
    {
      "epoch": 2.6223585548738924,
      "grad_norm": 17.571215943529168,
      "learning_rate": 4.759154001714883e-07,
      "loss": 1.0487,
      "step": 7694
    },
    {
      "epoch": 2.6226993865030677,
      "grad_norm": 16.04375037655816,
      "learning_rate": 4.7507135964567795e-07,
      "loss": 1.634,
      "step": 7695
    },
    {
      "epoch": 2.6230402181322425,
      "grad_norm": 51.62853440939501,
      "learning_rate": 4.742280309091563e-07,
      "loss": 1.7843,
      "step": 7696
    },
    {
      "epoch": 2.623381049761418,
      "grad_norm": 11.578911506686502,
      "learning_rate": 4.733854140945798e-07,
      "loss": 1.1521,
      "step": 7697
    },
    {
      "epoch": 2.623721881390593,
      "grad_norm": 20.13857142351193,
      "learning_rate": 4.7254350933449764e-07,
      "loss": 1.5795,
      "step": 7698
    },
    {
      "epoch": 2.624062713019768,
      "grad_norm": 21.52270531774509,
      "learning_rate": 4.7170231676134546e-07,
      "loss": 1.1399,
      "step": 7699
    },
    {
      "epoch": 2.6244035446489433,
      "grad_norm": 13.44238971855142,
      "learning_rate": 4.70861836507443e-07,
      "loss": 1.4992,
      "step": 7700
    },
    {
      "epoch": 2.6247443762781186,
      "grad_norm": 16.980077529800855,
      "learning_rate": 4.70022068705005e-07,
      "loss": 0.9568,
      "step": 7701
    },
    {
      "epoch": 2.625085207907294,
      "grad_norm": 14.49006163371389,
      "learning_rate": 4.691830134861275e-07,
      "loss": 1.7436,
      "step": 7702
    },
    {
      "epoch": 2.6254260395364692,
      "grad_norm": 15.115783045182448,
      "learning_rate": 4.683446709827971e-07,
      "loss": 1.8825,
      "step": 7703
    },
    {
      "epoch": 2.625766871165644,
      "grad_norm": 14.6068027082302,
      "learning_rate": 4.675070413268917e-07,
      "loss": 1.4441,
      "step": 7704
    },
    {
      "epoch": 2.6261077027948194,
      "grad_norm": 15.518059222146553,
      "learning_rate": 4.666701246501698e-07,
      "loss": 1.4723,
      "step": 7705
    },
    {
      "epoch": 2.6264485344239947,
      "grad_norm": 12.298871315562094,
      "learning_rate": 4.658339210842844e-07,
      "loss": 1.2641,
      "step": 7706
    },
    {
      "epoch": 2.6267893660531696,
      "grad_norm": 13.995629812915185,
      "learning_rate": 4.6499843076077187e-07,
      "loss": 1.6411,
      "step": 7707
    },
    {
      "epoch": 2.627130197682345,
      "grad_norm": 15.050877844870971,
      "learning_rate": 4.641636538110594e-07,
      "loss": 1.5603,
      "step": 7708
    },
    {
      "epoch": 2.62747102931152,
      "grad_norm": 17.158471121418742,
      "learning_rate": 4.6332959036645975e-07,
      "loss": 1.4349,
      "step": 7709
    },
    {
      "epoch": 2.627811860940695,
      "grad_norm": 13.825309760861645,
      "learning_rate": 4.6249624055817564e-07,
      "loss": 0.994,
      "step": 7710
    },
    {
      "epoch": 2.6281526925698704,
      "grad_norm": 16.09532300929592,
      "learning_rate": 4.6166360451729564e-07,
      "loss": 1.4629,
      "step": 7711
    },
    {
      "epoch": 2.6284935241990457,
      "grad_norm": 15.650926629425596,
      "learning_rate": 4.6083168237479603e-07,
      "loss": 1.6271,
      "step": 7712
    },
    {
      "epoch": 2.628834355828221,
      "grad_norm": 15.283637381330518,
      "learning_rate": 4.600004742615427e-07,
      "loss": 1.3527,
      "step": 7713
    },
    {
      "epoch": 2.6291751874573963,
      "grad_norm": 15.25932191302661,
      "learning_rate": 4.591699803082872e-07,
      "loss": 1.1761,
      "step": 7714
    },
    {
      "epoch": 2.629516019086571,
      "grad_norm": 18.608322730171544,
      "learning_rate": 4.5834020064566997e-07,
      "loss": 1.7409,
      "step": 7715
    },
    {
      "epoch": 2.6298568507157465,
      "grad_norm": 16.808885830580586,
      "learning_rate": 4.5751113540421885e-07,
      "loss": 1.4939,
      "step": 7716
    },
    {
      "epoch": 2.630197682344922,
      "grad_norm": 14.039345793201036,
      "learning_rate": 4.56682784714349e-07,
      "loss": 1.1231,
      "step": 7717
    },
    {
      "epoch": 2.6305385139740967,
      "grad_norm": 12.786467094496615,
      "learning_rate": 4.5585514870636225e-07,
      "loss": 1.2704,
      "step": 7718
    },
    {
      "epoch": 2.630879345603272,
      "grad_norm": 16.694828217918154,
      "learning_rate": 4.550282275104495e-07,
      "loss": 1.4531,
      "step": 7719
    },
    {
      "epoch": 2.6312201772324473,
      "grad_norm": 11.508467338842266,
      "learning_rate": 4.5420202125669e-07,
      "loss": 1.7934,
      "step": 7720
    },
    {
      "epoch": 2.631561008861622,
      "grad_norm": 23.171279724577914,
      "learning_rate": 4.5337653007504814e-07,
      "loss": 1.7467,
      "step": 7721
    },
    {
      "epoch": 2.6319018404907975,
      "grad_norm": 29.499970643466533,
      "learning_rate": 4.5255175409537667e-07,
      "loss": 1.5627,
      "step": 7722
    },
    {
      "epoch": 2.6322426721199728,
      "grad_norm": 13.71483448524475,
      "learning_rate": 4.5172769344741686e-07,
      "loss": 1.5466,
      "step": 7723
    },
    {
      "epoch": 2.632583503749148,
      "grad_norm": 13.509711699035044,
      "learning_rate": 4.509043482607961e-07,
      "loss": 1.1888,
      "step": 7724
    },
    {
      "epoch": 2.6329243353783234,
      "grad_norm": 8.431942467158406,
      "learning_rate": 4.500817186650308e-07,
      "loss": 1.2252,
      "step": 7725
    },
    {
      "epoch": 2.6332651670074982,
      "grad_norm": 18.64211826246726,
      "learning_rate": 4.4925980478952303e-07,
      "loss": 1.449,
      "step": 7726
    },
    {
      "epoch": 2.6336059986366736,
      "grad_norm": 12.110938237420717,
      "learning_rate": 4.4843860676356376e-07,
      "loss": 0.9156,
      "step": 7727
    },
    {
      "epoch": 2.633946830265849,
      "grad_norm": 17.135542634957233,
      "learning_rate": 4.476181247163297e-07,
      "loss": 1.6148,
      "step": 7728
    },
    {
      "epoch": 2.6342876618950237,
      "grad_norm": 21.49570533243647,
      "learning_rate": 4.4679835877688706e-07,
      "loss": 1.1998,
      "step": 7729
    },
    {
      "epoch": 2.634628493524199,
      "grad_norm": 14.509506421986927,
      "learning_rate": 4.459793090741887e-07,
      "loss": 1.404,
      "step": 7730
    },
    {
      "epoch": 2.6349693251533743,
      "grad_norm": 15.991652792735968,
      "learning_rate": 4.4516097573707175e-07,
      "loss": 1.917,
      "step": 7731
    },
    {
      "epoch": 2.635310156782549,
      "grad_norm": 14.60052849195606,
      "learning_rate": 4.443433588942658e-07,
      "loss": 1.4789,
      "step": 7732
    },
    {
      "epoch": 2.6356509884117245,
      "grad_norm": 15.637998593313293,
      "learning_rate": 4.4352645867438594e-07,
      "loss": 0.7317,
      "step": 7733
    },
    {
      "epoch": 2.6359918200409,
      "grad_norm": 18.216432614139986,
      "learning_rate": 4.427102752059298e-07,
      "loss": 1.5986,
      "step": 7734
    },
    {
      "epoch": 2.636332651670075,
      "grad_norm": 17.1453145850343,
      "learning_rate": 4.4189480861729137e-07,
      "loss": 1.0067,
      "step": 7735
    },
    {
      "epoch": 2.6366734832992504,
      "grad_norm": 12.598979418878164,
      "learning_rate": 4.410800590367431e-07,
      "loss": 0.8576,
      "step": 7736
    },
    {
      "epoch": 2.6370143149284253,
      "grad_norm": 15.395934148449621,
      "learning_rate": 4.4026602659244856e-07,
      "loss": 1.6912,
      "step": 7737
    },
    {
      "epoch": 2.6373551465576006,
      "grad_norm": 14.633686757949757,
      "learning_rate": 4.394527114124614e-07,
      "loss": 1.7482,
      "step": 7738
    },
    {
      "epoch": 2.637695978186776,
      "grad_norm": 13.152639176432928,
      "learning_rate": 4.386401136247159e-07,
      "loss": 1.5672,
      "step": 7739
    },
    {
      "epoch": 2.638036809815951,
      "grad_norm": 12.985020453735148,
      "learning_rate": 4.3782823335703883e-07,
      "loss": 1.5336,
      "step": 7740
    },
    {
      "epoch": 2.638377641445126,
      "grad_norm": 16.638008359243816,
      "learning_rate": 4.3701707073714127e-07,
      "loss": 1.519,
      "step": 7741
    },
    {
      "epoch": 2.6387184730743014,
      "grad_norm": 12.525150719213373,
      "learning_rate": 4.3620662589262285e-07,
      "loss": 0.9935,
      "step": 7742
    },
    {
      "epoch": 2.6390593047034763,
      "grad_norm": 19.96362900125915,
      "learning_rate": 4.3539689895096936e-07,
      "loss": 1.2397,
      "step": 7743
    },
    {
      "epoch": 2.6394001363326516,
      "grad_norm": 17.465068023336027,
      "learning_rate": 4.3458789003955395e-07,
      "loss": 1.4404,
      "step": 7744
    },
    {
      "epoch": 2.639740967961827,
      "grad_norm": 17.24616384308173,
      "learning_rate": 4.337795992856381e-07,
      "loss": 1.345,
      "step": 7745
    },
    {
      "epoch": 2.640081799591002,
      "grad_norm": 16.333202135337864,
      "learning_rate": 4.3297202681636785e-07,
      "loss": 1.5025,
      "step": 7746
    },
    {
      "epoch": 2.6404226312201775,
      "grad_norm": 14.51267999763345,
      "learning_rate": 4.3216517275877777e-07,
      "loss": 1.5114,
      "step": 7747
    },
    {
      "epoch": 2.6407634628493524,
      "grad_norm": 16.04824115965645,
      "learning_rate": 4.3135903723979023e-07,
      "loss": 1.3824,
      "step": 7748
    },
    {
      "epoch": 2.6411042944785277,
      "grad_norm": 20.461962073131346,
      "learning_rate": 4.305536203862132e-07,
      "loss": 1.3215,
      "step": 7749
    },
    {
      "epoch": 2.641445126107703,
      "grad_norm": 11.330043259420776,
      "learning_rate": 4.297489223247403e-07,
      "loss": 0.954,
      "step": 7750
    },
    {
      "epoch": 2.641785957736878,
      "grad_norm": 20.67936990850048,
      "learning_rate": 4.2894494318195645e-07,
      "loss": 0.9849,
      "step": 7751
    },
    {
      "epoch": 2.642126789366053,
      "grad_norm": 10.46168015670137,
      "learning_rate": 4.281416830843288e-07,
      "loss": 0.9257,
      "step": 7752
    },
    {
      "epoch": 2.6424676209952285,
      "grad_norm": 10.221074681429823,
      "learning_rate": 4.273391421582124e-07,
      "loss": 0.9859,
      "step": 7753
    },
    {
      "epoch": 2.6428084526244033,
      "grad_norm": 13.053603268781615,
      "learning_rate": 4.2653732052985397e-07,
      "loss": 1.2832,
      "step": 7754
    },
    {
      "epoch": 2.6431492842535786,
      "grad_norm": 10.917827568348867,
      "learning_rate": 4.257362183253799e-07,
      "loss": 1.347,
      "step": 7755
    },
    {
      "epoch": 2.643490115882754,
      "grad_norm": 11.989655909652017,
      "learning_rate": 4.2493583567080777e-07,
      "loss": 1.3557,
      "step": 7756
    },
    {
      "epoch": 2.6438309475119293,
      "grad_norm": 13.81648315144623,
      "learning_rate": 4.2413617269204066e-07,
      "loss": 1.3846,
      "step": 7757
    },
    {
      "epoch": 2.644171779141104,
      "grad_norm": 27.28153799691946,
      "learning_rate": 4.2333722951486965e-07,
      "loss": 0.7134,
      "step": 7758
    },
    {
      "epoch": 2.6445126107702794,
      "grad_norm": 20.072140412242078,
      "learning_rate": 4.2253900626497093e-07,
      "loss": 1.3419,
      "step": 7759
    },
    {
      "epoch": 2.6448534423994547,
      "grad_norm": 16.57675300469527,
      "learning_rate": 4.2174150306790786e-07,
      "loss": 1.7682,
      "step": 7760
    },
    {
      "epoch": 2.64519427402863,
      "grad_norm": 21.105021708550275,
      "learning_rate": 4.209447200491318e-07,
      "loss": 1.6267,
      "step": 7761
    },
    {
      "epoch": 2.645535105657805,
      "grad_norm": 19.328896153712027,
      "learning_rate": 4.201486573339791e-07,
      "loss": 1.351,
      "step": 7762
    },
    {
      "epoch": 2.6458759372869802,
      "grad_norm": 14.717174975104685,
      "learning_rate": 4.193533150476742e-07,
      "loss": 1.5083,
      "step": 7763
    },
    {
      "epoch": 2.6462167689161555,
      "grad_norm": 17.186452842631493,
      "learning_rate": 4.185586933153285e-07,
      "loss": 1.9581,
      "step": 7764
    },
    {
      "epoch": 2.6465576005453304,
      "grad_norm": 14.262743758511634,
      "learning_rate": 4.1776479226193543e-07,
      "loss": 1.3581,
      "step": 7765
    },
    {
      "epoch": 2.6468984321745057,
      "grad_norm": 16.236783291800926,
      "learning_rate": 4.169716120123829e-07,
      "loss": 1.5944,
      "step": 7766
    },
    {
      "epoch": 2.647239263803681,
      "grad_norm": 16.559460227348495,
      "learning_rate": 4.1617915269144047e-07,
      "loss": 1.5905,
      "step": 7767
    },
    {
      "epoch": 2.6475800954328563,
      "grad_norm": 15.907168742272804,
      "learning_rate": 4.1538741442376297e-07,
      "loss": 1.9036,
      "step": 7768
    },
    {
      "epoch": 2.647920927062031,
      "grad_norm": 16.328384364697417,
      "learning_rate": 4.1459639733389676e-07,
      "loss": 1.7929,
      "step": 7769
    },
    {
      "epoch": 2.6482617586912065,
      "grad_norm": 17.94846134149216,
      "learning_rate": 4.138061015462702e-07,
      "loss": 1.464,
      "step": 7770
    },
    {
      "epoch": 2.648602590320382,
      "grad_norm": 14.183304025461293,
      "learning_rate": 4.1301652718519934e-07,
      "loss": 1.1229,
      "step": 7771
    },
    {
      "epoch": 2.648943421949557,
      "grad_norm": 18.297017736508003,
      "learning_rate": 4.1222767437488986e-07,
      "loss": 1.2185,
      "step": 7772
    },
    {
      "epoch": 2.649284253578732,
      "grad_norm": 46.35249206972199,
      "learning_rate": 4.114395432394297e-07,
      "loss": 1.5993,
      "step": 7773
    },
    {
      "epoch": 2.6496250852079073,
      "grad_norm": 15.417262165983074,
      "learning_rate": 4.1065213390279524e-07,
      "loss": 1.26,
      "step": 7774
    },
    {
      "epoch": 2.6499659168370826,
      "grad_norm": 14.684580946569552,
      "learning_rate": 4.098654464888496e-07,
      "loss": 1.9025,
      "step": 7775
    },
    {
      "epoch": 2.6503067484662575,
      "grad_norm": 15.708029001238208,
      "learning_rate": 4.090794811213411e-07,
      "loss": 1.7399,
      "step": 7776
    },
    {
      "epoch": 2.6506475800954328,
      "grad_norm": 22.032778170819377,
      "learning_rate": 4.0829423792390577e-07,
      "loss": 0.7559,
      "step": 7777
    },
    {
      "epoch": 2.650988411724608,
      "grad_norm": 26.869847806976644,
      "learning_rate": 4.0750971702006537e-07,
      "loss": 1.4664,
      "step": 7778
    },
    {
      "epoch": 2.6513292433537834,
      "grad_norm": 19.109611233766728,
      "learning_rate": 4.0672591853322784e-07,
      "loss": 2.1366,
      "step": 7779
    },
    {
      "epoch": 2.6516700749829583,
      "grad_norm": 13.40718289200999,
      "learning_rate": 4.0594284258668846e-07,
      "loss": 1.6611,
      "step": 7780
    },
    {
      "epoch": 2.6520109066121336,
      "grad_norm": 16.422820364696385,
      "learning_rate": 4.051604893036276e-07,
      "loss": 1.5282,
      "step": 7781
    },
    {
      "epoch": 2.652351738241309,
      "grad_norm": 18.930119501659625,
      "learning_rate": 4.0437885880711293e-07,
      "loss": 2.0103,
      "step": 7782
    },
    {
      "epoch": 2.652692569870484,
      "grad_norm": 22.682415420500924,
      "learning_rate": 4.0359795122009825e-07,
      "loss": 1.2153,
      "step": 7783
    },
    {
      "epoch": 2.653033401499659,
      "grad_norm": 11.140848012062857,
      "learning_rate": 4.0281776666542204e-07,
      "loss": 1.3739,
      "step": 7784
    },
    {
      "epoch": 2.6533742331288344,
      "grad_norm": 26.024361997643695,
      "learning_rate": 4.0203830526581277e-07,
      "loss": 1.6619,
      "step": 7785
    },
    {
      "epoch": 2.6537150647580097,
      "grad_norm": 13.725842586865713,
      "learning_rate": 4.012595671438807e-07,
      "loss": 1.6374,
      "step": 7786
    },
    {
      "epoch": 2.6540558963871845,
      "grad_norm": 20.365610856290534,
      "learning_rate": 4.004815524221245e-07,
      "loss": 1.5722,
      "step": 7787
    },
    {
      "epoch": 2.65439672801636,
      "grad_norm": 17.8274138520206,
      "learning_rate": 3.9970426122293125e-07,
      "loss": 1.9641,
      "step": 7788
    },
    {
      "epoch": 2.654737559645535,
      "grad_norm": 23.74683616599662,
      "learning_rate": 3.9892769366856986e-07,
      "loss": 1.1312,
      "step": 7789
    },
    {
      "epoch": 2.6550783912747105,
      "grad_norm": 13.613434100441507,
      "learning_rate": 3.9815184988119817e-07,
      "loss": 1.5715,
      "step": 7790
    },
    {
      "epoch": 2.6554192229038853,
      "grad_norm": 14.645207458140527,
      "learning_rate": 3.9737672998285904e-07,
      "loss": 1.4185,
      "step": 7791
    },
    {
      "epoch": 2.6557600545330606,
      "grad_norm": 15.980630532727378,
      "learning_rate": 3.966023340954822e-07,
      "loss": 1.7762,
      "step": 7792
    },
    {
      "epoch": 2.656100886162236,
      "grad_norm": 9.367205504333942,
      "learning_rate": 3.958286623408836e-07,
      "loss": 1.4074,
      "step": 7793
    },
    {
      "epoch": 2.6564417177914113,
      "grad_norm": 19.113259508477366,
      "learning_rate": 3.950557148407641e-07,
      "loss": 1.6684,
      "step": 7794
    },
    {
      "epoch": 2.656782549420586,
      "grad_norm": 17.52479240233182,
      "learning_rate": 3.942834917167121e-07,
      "loss": 1.1084,
      "step": 7795
    },
    {
      "epoch": 2.6571233810497614,
      "grad_norm": 12.180743250538848,
      "learning_rate": 3.935119930902015e-07,
      "loss": 0.9065,
      "step": 7796
    },
    {
      "epoch": 2.6574642126789367,
      "grad_norm": 41.32326177623418,
      "learning_rate": 3.927412190825913e-07,
      "loss": 1.128,
      "step": 7797
    },
    {
      "epoch": 2.6578050443081116,
      "grad_norm": 8.903573193658897,
      "learning_rate": 3.9197116981512907e-07,
      "loss": 1.1785,
      "step": 7798
    },
    {
      "epoch": 2.658145875937287,
      "grad_norm": 8.798601425975658,
      "learning_rate": 3.9120184540894336e-07,
      "loss": 0.9837,
      "step": 7799
    },
    {
      "epoch": 2.658486707566462,
      "grad_norm": 18.069434661624744,
      "learning_rate": 3.904332459850546e-07,
      "loss": 1.6623,
      "step": 7800
    },
    {
      "epoch": 2.658827539195637,
      "grad_norm": 19.07043219490256,
      "learning_rate": 3.896653716643667e-07,
      "loss": 1.5415,
      "step": 7801
    },
    {
      "epoch": 2.6591683708248124,
      "grad_norm": 13.351870086632644,
      "learning_rate": 3.888982225676669e-07,
      "loss": 1.4736,
      "step": 7802
    },
    {
      "epoch": 2.6595092024539877,
      "grad_norm": 18.545711315894586,
      "learning_rate": 3.881317988156336e-07,
      "loss": 1.5606,
      "step": 7803
    },
    {
      "epoch": 2.659850034083163,
      "grad_norm": 13.533396922778447,
      "learning_rate": 3.873661005288276e-07,
      "loss": 1.2747,
      "step": 7804
    },
    {
      "epoch": 2.6601908657123383,
      "grad_norm": 21.70281722198345,
      "learning_rate": 3.8660112782769433e-07,
      "loss": 1.2614,
      "step": 7805
    },
    {
      "epoch": 2.660531697341513,
      "grad_norm": 17.938850830498293,
      "learning_rate": 3.858368808325702e-07,
      "loss": 2.0465,
      "step": 7806
    },
    {
      "epoch": 2.6608725289706885,
      "grad_norm": 15.635854036135262,
      "learning_rate": 3.8507335966367186e-07,
      "loss": 1.4135,
      "step": 7807
    },
    {
      "epoch": 2.661213360599864,
      "grad_norm": 18.569405228942006,
      "learning_rate": 3.843105644411038e-07,
      "loss": 2.2657,
      "step": 7808
    },
    {
      "epoch": 2.6615541922290387,
      "grad_norm": 18.404488904246644,
      "learning_rate": 3.8354849528486005e-07,
      "loss": 1.8543,
      "step": 7809
    },
    {
      "epoch": 2.661895023858214,
      "grad_norm": 15.229405513604094,
      "learning_rate": 3.8278715231481414e-07,
      "loss": 1.5476,
      "step": 7810
    },
    {
      "epoch": 2.6622358554873893,
      "grad_norm": 14.319818486101033,
      "learning_rate": 3.820265356507291e-07,
      "loss": 1.4331,
      "step": 7811
    },
    {
      "epoch": 2.662576687116564,
      "grad_norm": 12.738526045985424,
      "learning_rate": 3.8126664541225324e-07,
      "loss": 1.4386,
      "step": 7812
    },
    {
      "epoch": 2.6629175187457395,
      "grad_norm": 13.107749889435853,
      "learning_rate": 3.8050748171892025e-07,
      "loss": 1.2599,
      "step": 7813
    },
    {
      "epoch": 2.6632583503749148,
      "grad_norm": 12.383125411684835,
      "learning_rate": 3.797490446901497e-07,
      "loss": 1.5627,
      "step": 7814
    },
    {
      "epoch": 2.66359918200409,
      "grad_norm": 19.99085334021286,
      "learning_rate": 3.789913344452456e-07,
      "loss": 1.2343,
      "step": 7815
    },
    {
      "epoch": 2.6639400136332654,
      "grad_norm": 14.00483702119686,
      "learning_rate": 3.7823435110340036e-07,
      "loss": 1.2823,
      "step": 7816
    },
    {
      "epoch": 2.6642808452624402,
      "grad_norm": 16.681705190051034,
      "learning_rate": 3.774780947836909e-07,
      "loss": 1.5107,
      "step": 7817
    },
    {
      "epoch": 2.6646216768916156,
      "grad_norm": 11.261882973420477,
      "learning_rate": 3.76722565605076e-07,
      "loss": 1.3255,
      "step": 7818
    },
    {
      "epoch": 2.664962508520791,
      "grad_norm": 10.590859641799282,
      "learning_rate": 3.7596776368640843e-07,
      "loss": 1.3015,
      "step": 7819
    },
    {
      "epoch": 2.6653033401499657,
      "grad_norm": 28.145325359830053,
      "learning_rate": 3.75213689146417e-07,
      "loss": 1.5126,
      "step": 7820
    },
    {
      "epoch": 2.665644171779141,
      "grad_norm": 24.691701265521274,
      "learning_rate": 3.744603421037224e-07,
      "loss": 1.5785,
      "step": 7821
    },
    {
      "epoch": 2.6659850034083163,
      "grad_norm": 20.44950377751187,
      "learning_rate": 3.7370772267683044e-07,
      "loss": 1.0925,
      "step": 7822
    },
    {
      "epoch": 2.666325835037491,
      "grad_norm": 13.235536036525458,
      "learning_rate": 3.7295583098412913e-07,
      "loss": 1.3877,
      "step": 7823
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 12.359995018635434,
      "learning_rate": 3.72204667143895e-07,
      "loss": 1.1274,
      "step": 7824
    },
    {
      "epoch": 2.667007498295842,
      "grad_norm": 10.275599584228384,
      "learning_rate": 3.714542312742886e-07,
      "loss": 0.8469,
      "step": 7825
    },
    {
      "epoch": 2.667348329925017,
      "grad_norm": 16.022969414278013,
      "learning_rate": 3.7070452349335704e-07,
      "loss": 1.3223,
      "step": 7826
    },
    {
      "epoch": 2.6676891615541924,
      "grad_norm": 16.03249179843436,
      "learning_rate": 3.699555439190322e-07,
      "loss": 1.8954,
      "step": 7827
    },
    {
      "epoch": 2.6680299931833673,
      "grad_norm": 17.429179154201318,
      "learning_rate": 3.692072926691309e-07,
      "loss": 1.9761,
      "step": 7828
    },
    {
      "epoch": 2.6683708248125426,
      "grad_norm": 18.63018811027642,
      "learning_rate": 3.6845976986135735e-07,
      "loss": 0.9547,
      "step": 7829
    },
    {
      "epoch": 2.668711656441718,
      "grad_norm": 25.03063135990437,
      "learning_rate": 3.6771297561329854e-07,
      "loss": 1.0864,
      "step": 7830
    },
    {
      "epoch": 2.669052488070893,
      "grad_norm": 9.91755067254692,
      "learning_rate": 3.669669100424289e-07,
      "loss": 1.3366,
      "step": 7831
    },
    {
      "epoch": 2.669393319700068,
      "grad_norm": 15.160286899973935,
      "learning_rate": 3.662215732661073e-07,
      "loss": 1.3539,
      "step": 7832
    },
    {
      "epoch": 2.6697341513292434,
      "grad_norm": 9.742803357653077,
      "learning_rate": 3.654769654015783e-07,
      "loss": 1.0024,
      "step": 7833
    },
    {
      "epoch": 2.6700749829584183,
      "grad_norm": 25.31883700598765,
      "learning_rate": 3.647330865659715e-07,
      "loss": 1.0595,
      "step": 7834
    },
    {
      "epoch": 2.6704158145875936,
      "grad_norm": 19.988728134099347,
      "learning_rate": 3.639899368763028e-07,
      "loss": 1.1669,
      "step": 7835
    },
    {
      "epoch": 2.670756646216769,
      "grad_norm": 11.081578133819924,
      "learning_rate": 3.6324751644947085e-07,
      "loss": 1.2867,
      "step": 7836
    },
    {
      "epoch": 2.671097477845944,
      "grad_norm": 15.329988560235401,
      "learning_rate": 3.6250582540226286e-07,
      "loss": 1.4313,
      "step": 7837
    },
    {
      "epoch": 2.6714383094751195,
      "grad_norm": 16.46541053977173,
      "learning_rate": 3.617648638513499e-07,
      "loss": 1.8117,
      "step": 7838
    },
    {
      "epoch": 2.6717791411042944,
      "grad_norm": 21.795895582666517,
      "learning_rate": 3.610246319132865e-07,
      "loss": 1.7401,
      "step": 7839
    },
    {
      "epoch": 2.6721199727334697,
      "grad_norm": 20.201877888176863,
      "learning_rate": 3.6028512970451623e-07,
      "loss": 1.8716,
      "step": 7840
    },
    {
      "epoch": 2.672460804362645,
      "grad_norm": 19.298641304048488,
      "learning_rate": 3.595463573413643e-07,
      "loss": 1.0439,
      "step": 7841
    },
    {
      "epoch": 2.67280163599182,
      "grad_norm": 23.134455226320387,
      "learning_rate": 3.588083149400423e-07,
      "loss": 1.4072,
      "step": 7842
    },
    {
      "epoch": 2.673142467620995,
      "grad_norm": 18.017739350635498,
      "learning_rate": 3.5807100261664894e-07,
      "loss": 0.557,
      "step": 7843
    },
    {
      "epoch": 2.6734832992501705,
      "grad_norm": 15.374244083213435,
      "learning_rate": 3.5733442048716484e-07,
      "loss": 1.637,
      "step": 7844
    },
    {
      "epoch": 2.6738241308793453,
      "grad_norm": 14.201387814855059,
      "learning_rate": 3.565985686674578e-07,
      "loss": 1.7156,
      "step": 7845
    },
    {
      "epoch": 2.6741649625085206,
      "grad_norm": 11.24183847801657,
      "learning_rate": 3.5586344727328027e-07,
      "loss": 0.9711,
      "step": 7846
    },
    {
      "epoch": 2.674505794137696,
      "grad_norm": 13.860981074638048,
      "learning_rate": 3.5512905642026974e-07,
      "loss": 1.9548,
      "step": 7847
    },
    {
      "epoch": 2.6748466257668713,
      "grad_norm": 20.671953982241124,
      "learning_rate": 3.5439539622394934e-07,
      "loss": 1.0235,
      "step": 7848
    },
    {
      "epoch": 2.6751874573960466,
      "grad_norm": 10.19923186322359,
      "learning_rate": 3.536624667997246e-07,
      "loss": 1.2436,
      "step": 7849
    },
    {
      "epoch": 2.6755282890252214,
      "grad_norm": 15.353681050362521,
      "learning_rate": 3.5293026826289043e-07,
      "loss": 1.6182,
      "step": 7850
    },
    {
      "epoch": 2.6758691206543967,
      "grad_norm": 15.05699545961098,
      "learning_rate": 3.5219880072862534e-07,
      "loss": 1.5452,
      "step": 7851
    },
    {
      "epoch": 2.676209952283572,
      "grad_norm": 12.31068488350357,
      "learning_rate": 3.514680643119889e-07,
      "loss": 1.5833,
      "step": 7852
    },
    {
      "epoch": 2.676550783912747,
      "grad_norm": 9.697543257832464,
      "learning_rate": 3.5073805912793244e-07,
      "loss": 0.951,
      "step": 7853
    },
    {
      "epoch": 2.6768916155419222,
      "grad_norm": 15.39533400541716,
      "learning_rate": 3.500087852912859e-07,
      "loss": 1.0527,
      "step": 7854
    },
    {
      "epoch": 2.6772324471710975,
      "grad_norm": 15.053015019663684,
      "learning_rate": 3.492802429167669e-07,
      "loss": 1.8663,
      "step": 7855
    },
    {
      "epoch": 2.6775732788002724,
      "grad_norm": 16.293585749483363,
      "learning_rate": 3.4855243211898146e-07,
      "loss": 1.6889,
      "step": 7856
    },
    {
      "epoch": 2.6779141104294477,
      "grad_norm": 15.373781206926635,
      "learning_rate": 3.4782535301241315e-07,
      "loss": 1.6508,
      "step": 7857
    },
    {
      "epoch": 2.678254942058623,
      "grad_norm": 15.923568925037225,
      "learning_rate": 3.470990057114365e-07,
      "loss": 1.2919,
      "step": 7858
    },
    {
      "epoch": 2.6785957736877983,
      "grad_norm": 16.262113744569717,
      "learning_rate": 3.463733903303085e-07,
      "loss": 1.1861,
      "step": 7859
    },
    {
      "epoch": 2.6789366053169736,
      "grad_norm": 14.747995633682201,
      "learning_rate": 3.456485069831711e-07,
      "loss": 1.3213,
      "step": 7860
    },
    {
      "epoch": 2.6792774369461485,
      "grad_norm": 13.601744151893882,
      "learning_rate": 3.4492435578405146e-07,
      "loss": 0.8925,
      "step": 7861
    },
    {
      "epoch": 2.679618268575324,
      "grad_norm": 16.22083915076897,
      "learning_rate": 3.442009368468613e-07,
      "loss": 1.5687,
      "step": 7862
    },
    {
      "epoch": 2.679959100204499,
      "grad_norm": 15.617567296264024,
      "learning_rate": 3.434782502853978e-07,
      "loss": 1.6859,
      "step": 7863
    },
    {
      "epoch": 2.680299931833674,
      "grad_norm": 26.838900136746254,
      "learning_rate": 3.427562962133418e-07,
      "loss": 1.8461,
      "step": 7864
    },
    {
      "epoch": 2.6806407634628493,
      "grad_norm": 14.214350050477996,
      "learning_rate": 3.4203507474425967e-07,
      "loss": 1.7913,
      "step": 7865
    },
    {
      "epoch": 2.6809815950920246,
      "grad_norm": 14.90565586369843,
      "learning_rate": 3.413145859916028e-07,
      "loss": 1.3606,
      "step": 7866
    },
    {
      "epoch": 2.6813224267211995,
      "grad_norm": 30.978719757063537,
      "learning_rate": 3.4059483006870677e-07,
      "loss": 1.6558,
      "step": 7867
    },
    {
      "epoch": 2.681663258350375,
      "grad_norm": 12.630233057229475,
      "learning_rate": 3.3987580708879143e-07,
      "loss": 1.5652,
      "step": 7868
    },
    {
      "epoch": 2.68200408997955,
      "grad_norm": 16.709220674371704,
      "learning_rate": 3.3915751716496413e-07,
      "loss": 1.3903,
      "step": 7869
    },
    {
      "epoch": 2.6823449216087254,
      "grad_norm": 13.243827878807327,
      "learning_rate": 3.3843996041021066e-07,
      "loss": 1.7876,
      "step": 7870
    },
    {
      "epoch": 2.6826857532379007,
      "grad_norm": 17.012702660471586,
      "learning_rate": 3.3772313693740954e-07,
      "loss": 1.118,
      "step": 7871
    },
    {
      "epoch": 2.6830265848670756,
      "grad_norm": 13.706615296183385,
      "learning_rate": 3.3700704685931895e-07,
      "loss": 1.4805,
      "step": 7872
    },
    {
      "epoch": 2.683367416496251,
      "grad_norm": 17.610271940098563,
      "learning_rate": 3.3629169028858046e-07,
      "loss": 1.171,
      "step": 7873
    },
    {
      "epoch": 2.683708248125426,
      "grad_norm": 9.984825469416746,
      "learning_rate": 3.355770673377262e-07,
      "loss": 1.0858,
      "step": 7874
    },
    {
      "epoch": 2.684049079754601,
      "grad_norm": 10.83568414876771,
      "learning_rate": 3.348631781191664e-07,
      "loss": 0.9809,
      "step": 7875
    },
    {
      "epoch": 2.6843899113837764,
      "grad_norm": 21.32135702335243,
      "learning_rate": 3.3415002274519824e-07,
      "loss": 1.3589,
      "step": 7876
    },
    {
      "epoch": 2.6847307430129517,
      "grad_norm": 14.380075492172123,
      "learning_rate": 3.334376013280072e-07,
      "loss": 1.4442,
      "step": 7877
    },
    {
      "epoch": 2.6850715746421265,
      "grad_norm": 12.693135501946157,
      "learning_rate": 3.327259139796568e-07,
      "loss": 1.1404,
      "step": 7878
    },
    {
      "epoch": 2.685412406271302,
      "grad_norm": 22.036208220965033,
      "learning_rate": 3.3201496081209984e-07,
      "loss": 1.2565,
      "step": 7879
    },
    {
      "epoch": 2.685753237900477,
      "grad_norm": 25.901666855532397,
      "learning_rate": 3.3130474193717176e-07,
      "loss": 1.1245,
      "step": 7880
    },
    {
      "epoch": 2.6860940695296525,
      "grad_norm": 18.67299819235336,
      "learning_rate": 3.305952574665927e-07,
      "loss": 1.892,
      "step": 7881
    },
    {
      "epoch": 2.6864349011588278,
      "grad_norm": 22.957280924843023,
      "learning_rate": 3.298865075119684e-07,
      "loss": 1.6049,
      "step": 7882
    },
    {
      "epoch": 2.6867757327880026,
      "grad_norm": 9.93875014266664,
      "learning_rate": 3.2917849218478513e-07,
      "loss": 0.7872,
      "step": 7883
    },
    {
      "epoch": 2.687116564417178,
      "grad_norm": 13.153510219454484,
      "learning_rate": 3.284712115964195e-07,
      "loss": 1.2284,
      "step": 7884
    },
    {
      "epoch": 2.6874573960463533,
      "grad_norm": 13.998178577135569,
      "learning_rate": 3.27764665858129e-07,
      "loss": 1.0878,
      "step": 7885
    },
    {
      "epoch": 2.687798227675528,
      "grad_norm": 33.372843210446796,
      "learning_rate": 3.270588550810544e-07,
      "loss": 1.5453,
      "step": 7886
    },
    {
      "epoch": 2.6881390593047034,
      "grad_norm": 11.695270833569362,
      "learning_rate": 3.2635377937622557e-07,
      "loss": 1.2095,
      "step": 7887
    },
    {
      "epoch": 2.6884798909338787,
      "grad_norm": 15.081144001207846,
      "learning_rate": 3.2564943885455114e-07,
      "loss": 0.9385,
      "step": 7888
    },
    {
      "epoch": 2.6888207225630536,
      "grad_norm": 12.97981022892759,
      "learning_rate": 3.249458336268263e-07,
      "loss": 1.2923,
      "step": 7889
    },
    {
      "epoch": 2.689161554192229,
      "grad_norm": 12.271765563489652,
      "learning_rate": 3.242429638037342e-07,
      "loss": 1.1517,
      "step": 7890
    },
    {
      "epoch": 2.689502385821404,
      "grad_norm": 16.043021741690577,
      "learning_rate": 3.2354082949583587e-07,
      "loss": 1.4855,
      "step": 7891
    },
    {
      "epoch": 2.6898432174505795,
      "grad_norm": 10.777005923528108,
      "learning_rate": 3.228394308135813e-07,
      "loss": 0.8444,
      "step": 7892
    },
    {
      "epoch": 2.690184049079755,
      "grad_norm": 13.144777943409839,
      "learning_rate": 3.221387678673027e-07,
      "loss": 1.2258,
      "step": 7893
    },
    {
      "epoch": 2.6905248807089297,
      "grad_norm": 23.698076661797646,
      "learning_rate": 3.2143884076721753e-07,
      "loss": 1.407,
      "step": 7894
    },
    {
      "epoch": 2.690865712338105,
      "grad_norm": 12.27075959737697,
      "learning_rate": 3.207396496234266e-07,
      "loss": 0.7508,
      "step": 7895
    },
    {
      "epoch": 2.6912065439672803,
      "grad_norm": 14.639541514138164,
      "learning_rate": 3.2004119454591643e-07,
      "loss": 1.44,
      "step": 7896
    },
    {
      "epoch": 2.691547375596455,
      "grad_norm": 14.625336550304194,
      "learning_rate": 3.193434756445557e-07,
      "loss": 1.3885,
      "step": 7897
    },
    {
      "epoch": 2.6918882072256305,
      "grad_norm": 11.18334700566323,
      "learning_rate": 3.186464930290989e-07,
      "loss": 1.0583,
      "step": 7898
    },
    {
      "epoch": 2.692229038854806,
      "grad_norm": 9.413567707880546,
      "learning_rate": 3.179502468091844e-07,
      "loss": 1.1087,
      "step": 7899
    },
    {
      "epoch": 2.6925698704839807,
      "grad_norm": 11.717920562710704,
      "learning_rate": 3.1725473709433405e-07,
      "loss": 1.2788,
      "step": 7900
    },
    {
      "epoch": 2.692910702113156,
      "grad_norm": 49.289291540221626,
      "learning_rate": 3.1655996399395483e-07,
      "loss": 1.4423,
      "step": 7901
    },
    {
      "epoch": 2.6932515337423313,
      "grad_norm": 13.531759992399126,
      "learning_rate": 3.158659276173365e-07,
      "loss": 1.2649,
      "step": 7902
    },
    {
      "epoch": 2.6935923653715066,
      "grad_norm": 34.61880627059901,
      "learning_rate": 3.1517262807365554e-07,
      "loss": 1.4064,
      "step": 7903
    },
    {
      "epoch": 2.693933197000682,
      "grad_norm": 18.16156681330489,
      "learning_rate": 3.144800654719676e-07,
      "loss": 1.8184,
      "step": 7904
    },
    {
      "epoch": 2.6942740286298568,
      "grad_norm": 18.31046686179254,
      "learning_rate": 3.1378823992121776e-07,
      "loss": 1.5392,
      "step": 7905
    },
    {
      "epoch": 2.694614860259032,
      "grad_norm": 29.328940182679954,
      "learning_rate": 3.130971515302339e-07,
      "loss": 1.3962,
      "step": 7906
    },
    {
      "epoch": 2.6949556918882074,
      "grad_norm": 20.036750014921676,
      "learning_rate": 3.124068004077235e-07,
      "loss": 1.1446,
      "step": 7907
    },
    {
      "epoch": 2.6952965235173822,
      "grad_norm": 12.672395051674501,
      "learning_rate": 3.117171866622853e-07,
      "loss": 1.3195,
      "step": 7908
    },
    {
      "epoch": 2.6956373551465576,
      "grad_norm": 17.197056183392977,
      "learning_rate": 3.110283104023959e-07,
      "loss": 1.3408,
      "step": 7909
    },
    {
      "epoch": 2.695978186775733,
      "grad_norm": 13.148957632326098,
      "learning_rate": 3.103401717364185e-07,
      "loss": 1.314,
      "step": 7910
    },
    {
      "epoch": 2.6963190184049077,
      "grad_norm": 14.401642031317264,
      "learning_rate": 3.096527707726016e-07,
      "loss": 1.2919,
      "step": 7911
    },
    {
      "epoch": 2.696659850034083,
      "grad_norm": 23.791054390101714,
      "learning_rate": 3.0896610761907474e-07,
      "loss": 1.4195,
      "step": 7912
    },
    {
      "epoch": 2.6970006816632583,
      "grad_norm": 21.685901644878495,
      "learning_rate": 3.082801823838527e-07,
      "loss": 1.0859,
      "step": 7913
    },
    {
      "epoch": 2.6973415132924337,
      "grad_norm": 16.67645659565729,
      "learning_rate": 3.075949951748347e-07,
      "loss": 1.7195,
      "step": 7914
    },
    {
      "epoch": 2.697682344921609,
      "grad_norm": 11.198207929606362,
      "learning_rate": 3.069105460998034e-07,
      "loss": 1.0759,
      "step": 7915
    },
    {
      "epoch": 2.698023176550784,
      "grad_norm": 13.747794339084923,
      "learning_rate": 3.062268352664255e-07,
      "loss": 1.5475,
      "step": 7916
    },
    {
      "epoch": 2.698364008179959,
      "grad_norm": 22.208438034093966,
      "learning_rate": 3.05543862782251e-07,
      "loss": 1.6429,
      "step": 7917
    },
    {
      "epoch": 2.6987048398091344,
      "grad_norm": 20.617760878037288,
      "learning_rate": 3.048616287547146e-07,
      "loss": 1.9196,
      "step": 7918
    },
    {
      "epoch": 2.6990456714383093,
      "grad_norm": 15.862016538980958,
      "learning_rate": 3.0418013329113483e-07,
      "loss": 1.3976,
      "step": 7919
    },
    {
      "epoch": 2.6993865030674846,
      "grad_norm": 16.443246158018482,
      "learning_rate": 3.034993764987115e-07,
      "loss": 1.4784,
      "step": 7920
    },
    {
      "epoch": 2.69972733469666,
      "grad_norm": 17.972093647965423,
      "learning_rate": 3.028193584845329e-07,
      "loss": 2.0999,
      "step": 7921
    },
    {
      "epoch": 2.700068166325835,
      "grad_norm": 8.330028479386932,
      "learning_rate": 3.0214007935556844e-07,
      "loss": 1.0816,
      "step": 7922
    },
    {
      "epoch": 2.70040899795501,
      "grad_norm": 17.025962566902162,
      "learning_rate": 3.0146153921866873e-07,
      "loss": 1.4721,
      "step": 7923
    },
    {
      "epoch": 2.7007498295841854,
      "grad_norm": 27.543883639883468,
      "learning_rate": 3.007837381805745e-07,
      "loss": 1.1798,
      "step": 7924
    },
    {
      "epoch": 2.7010906612133607,
      "grad_norm": 20.45026199532222,
      "learning_rate": 3.001066763479038e-07,
      "loss": 1.4235,
      "step": 7925
    },
    {
      "epoch": 2.701431492842536,
      "grad_norm": 13.460894753004986,
      "learning_rate": 2.9943035382716143e-07,
      "loss": 1.6709,
      "step": 7926
    },
    {
      "epoch": 2.701772324471711,
      "grad_norm": 16.02992666212317,
      "learning_rate": 2.987547707247379e-07,
      "loss": 1.6812,
      "step": 7927
    },
    {
      "epoch": 2.702113156100886,
      "grad_norm": 21.604582800458903,
      "learning_rate": 2.980799271469026e-07,
      "loss": 0.7154,
      "step": 7928
    },
    {
      "epoch": 2.7024539877300615,
      "grad_norm": 17.941121156708135,
      "learning_rate": 2.9740582319981227e-07,
      "loss": 1.6154,
      "step": 7929
    },
    {
      "epoch": 2.7027948193592364,
      "grad_norm": 18.675835699468696,
      "learning_rate": 2.9673245898950543e-07,
      "loss": 1.7565,
      "step": 7930
    },
    {
      "epoch": 2.7031356509884117,
      "grad_norm": 17.76911188752681,
      "learning_rate": 2.9605983462190567e-07,
      "loss": 1.6894,
      "step": 7931
    },
    {
      "epoch": 2.703476482617587,
      "grad_norm": 15.041053693535646,
      "learning_rate": 2.953879502028195e-07,
      "loss": 0.9798,
      "step": 7932
    },
    {
      "epoch": 2.703817314246762,
      "grad_norm": 19.091477130465055,
      "learning_rate": 2.947168058379363e-07,
      "loss": 2.2291,
      "step": 7933
    },
    {
      "epoch": 2.704158145875937,
      "grad_norm": 16.444759844741558,
      "learning_rate": 2.9404640163283037e-07,
      "loss": 1.8707,
      "step": 7934
    },
    {
      "epoch": 2.7044989775051125,
      "grad_norm": 13.226558385509746,
      "learning_rate": 2.9337673769295916e-07,
      "loss": 0.7609,
      "step": 7935
    },
    {
      "epoch": 2.704839809134288,
      "grad_norm": 13.554132798897864,
      "learning_rate": 2.927078141236628e-07,
      "loss": 1.3824,
      "step": 7936
    },
    {
      "epoch": 2.705180640763463,
      "grad_norm": 15.31613950929831,
      "learning_rate": 2.9203963103016655e-07,
      "loss": 1.5386,
      "step": 7937
    },
    {
      "epoch": 2.705521472392638,
      "grad_norm": 11.384348445800972,
      "learning_rate": 2.913721885175763e-07,
      "loss": 1.048,
      "step": 7938
    },
    {
      "epoch": 2.7058623040218133,
      "grad_norm": 19.080066690077146,
      "learning_rate": 2.907054866908854e-07,
      "loss": 1.7981,
      "step": 7939
    },
    {
      "epoch": 2.7062031356509886,
      "grad_norm": 21.55670071652056,
      "learning_rate": 2.9003952565496875e-07,
      "loss": 1.2614,
      "step": 7940
    },
    {
      "epoch": 2.7065439672801634,
      "grad_norm": 11.001998137699625,
      "learning_rate": 2.893743055145826e-07,
      "loss": 1.1502,
      "step": 7941
    },
    {
      "epoch": 2.7068847989093388,
      "grad_norm": 15.000966879405219,
      "learning_rate": 2.887098263743715e-07,
      "loss": 1.2024,
      "step": 7942
    },
    {
      "epoch": 2.707225630538514,
      "grad_norm": 15.273645974282932,
      "learning_rate": 2.880460883388586e-07,
      "loss": 1.2917,
      "step": 7943
    },
    {
      "epoch": 2.707566462167689,
      "grad_norm": 20.660426498154,
      "learning_rate": 2.8738309151245315e-07,
      "loss": 1.7341,
      "step": 7944
    },
    {
      "epoch": 2.7079072937968642,
      "grad_norm": 19.556665131861717,
      "learning_rate": 2.8672083599944724e-07,
      "loss": 1.8222,
      "step": 7945
    },
    {
      "epoch": 2.7082481254260395,
      "grad_norm": 13.659075378376922,
      "learning_rate": 2.8605932190401643e-07,
      "loss": 0.8071,
      "step": 7946
    },
    {
      "epoch": 2.708588957055215,
      "grad_norm": 197.61682057999684,
      "learning_rate": 2.8539854933021916e-07,
      "loss": 1.4817,
      "step": 7947
    },
    {
      "epoch": 2.70892978868439,
      "grad_norm": 17.5158688269508,
      "learning_rate": 2.847385183819984e-07,
      "loss": 1.038,
      "step": 7948
    },
    {
      "epoch": 2.709270620313565,
      "grad_norm": 13.580501244059793,
      "learning_rate": 2.840792291631783e-07,
      "loss": 1.5004,
      "step": 7949
    },
    {
      "epoch": 2.7096114519427403,
      "grad_norm": 14.597328595328262,
      "learning_rate": 2.8342068177746874e-07,
      "loss": 1.3535,
      "step": 7950
    },
    {
      "epoch": 2.7099522835719156,
      "grad_norm": 18.293684932746842,
      "learning_rate": 2.827628763284618e-07,
      "loss": 1.4249,
      "step": 7951
    },
    {
      "epoch": 2.7102931152010905,
      "grad_norm": 14.248178669128482,
      "learning_rate": 2.8210581291963244e-07,
      "loss": 1.8212,
      "step": 7952
    },
    {
      "epoch": 2.710633946830266,
      "grad_norm": 15.36798637784822,
      "learning_rate": 2.8144949165434034e-07,
      "loss": 1.2004,
      "step": 7953
    },
    {
      "epoch": 2.710974778459441,
      "grad_norm": 8.547773084224296,
      "learning_rate": 2.80793912635825e-07,
      "loss": 0.8923,
      "step": 7954
    },
    {
      "epoch": 2.711315610088616,
      "grad_norm": 28.013229893766844,
      "learning_rate": 2.801390759672146e-07,
      "loss": 1.6774,
      "step": 7955
    },
    {
      "epoch": 2.7116564417177913,
      "grad_norm": 12.748486342859875,
      "learning_rate": 2.794849817515166e-07,
      "loss": 1.4802,
      "step": 7956
    },
    {
      "epoch": 2.7119972733469666,
      "grad_norm": 19.36399524164748,
      "learning_rate": 2.788316300916205e-07,
      "loss": 2.0698,
      "step": 7957
    },
    {
      "epoch": 2.712338104976142,
      "grad_norm": 13.754776991418014,
      "learning_rate": 2.781790210903046e-07,
      "loss": 1.5114,
      "step": 7958
    },
    {
      "epoch": 2.7126789366053172,
      "grad_norm": 24.89201462627269,
      "learning_rate": 2.7752715485022454e-07,
      "loss": 1.2071,
      "step": 7959
    },
    {
      "epoch": 2.713019768234492,
      "grad_norm": 15.878326604973246,
      "learning_rate": 2.7687603147392105e-07,
      "loss": 1.5331,
      "step": 7960
    },
    {
      "epoch": 2.7133605998636674,
      "grad_norm": 9.216239558475298,
      "learning_rate": 2.7622565106382116e-07,
      "loss": 0.987,
      "step": 7961
    },
    {
      "epoch": 2.7137014314928427,
      "grad_norm": 17.055067880531148,
      "learning_rate": 2.7557601372222966e-07,
      "loss": 1.3391,
      "step": 7962
    },
    {
      "epoch": 2.7140422631220176,
      "grad_norm": 13.728079820074564,
      "learning_rate": 2.749271195513387e-07,
      "loss": 1.0585,
      "step": 7963
    },
    {
      "epoch": 2.714383094751193,
      "grad_norm": 12.936745243100821,
      "learning_rate": 2.7427896865322043e-07,
      "loss": 1.2226,
      "step": 7964
    },
    {
      "epoch": 2.714723926380368,
      "grad_norm": 17.02869242598727,
      "learning_rate": 2.736315611298329e-07,
      "loss": 1.834,
      "step": 7965
    },
    {
      "epoch": 2.715064758009543,
      "grad_norm": 15.713471282809051,
      "learning_rate": 2.7298489708301513e-07,
      "loss": 1.3336,
      "step": 7966
    },
    {
      "epoch": 2.7154055896387184,
      "grad_norm": 23.292500614783695,
      "learning_rate": 2.723389766144907e-07,
      "loss": 0.8076,
      "step": 7967
    },
    {
      "epoch": 2.7157464212678937,
      "grad_norm": 52.772783225283014,
      "learning_rate": 2.716937998258645e-07,
      "loss": 1.1896,
      "step": 7968
    },
    {
      "epoch": 2.716087252897069,
      "grad_norm": 17.030356438238478,
      "learning_rate": 2.710493668186259e-07,
      "loss": 1.4926,
      "step": 7969
    },
    {
      "epoch": 2.7164280845262443,
      "grad_norm": 12.703133432744371,
      "learning_rate": 2.704056776941466e-07,
      "loss": 1.5019,
      "step": 7970
    },
    {
      "epoch": 2.716768916155419,
      "grad_norm": 12.15774823127215,
      "learning_rate": 2.697627325536828e-07,
      "loss": 1.5022,
      "step": 7971
    },
    {
      "epoch": 2.7171097477845945,
      "grad_norm": 16.903253828203827,
      "learning_rate": 2.691205314983697e-07,
      "loss": 1.3218,
      "step": 7972
    },
    {
      "epoch": 2.7174505794137698,
      "grad_norm": 16.43967837499582,
      "learning_rate": 2.684790746292298e-07,
      "loss": 1.7903,
      "step": 7973
    },
    {
      "epoch": 2.7177914110429446,
      "grad_norm": 17.94141438178002,
      "learning_rate": 2.678383620471675e-07,
      "loss": 1.7723,
      "step": 7974
    },
    {
      "epoch": 2.71813224267212,
      "grad_norm": 15.147567769591271,
      "learning_rate": 2.671983938529671e-07,
      "loss": 1.1167,
      "step": 7975
    },
    {
      "epoch": 2.7184730743012953,
      "grad_norm": 19.23553444642319,
      "learning_rate": 2.665591701473014e-07,
      "loss": 1.3393,
      "step": 7976
    },
    {
      "epoch": 2.71881390593047,
      "grad_norm": 14.29565455848557,
      "learning_rate": 2.6592069103071993e-07,
      "loss": 1.7271,
      "step": 7977
    },
    {
      "epoch": 2.7191547375596454,
      "grad_norm": 23.834010306685858,
      "learning_rate": 2.6528295660365855e-07,
      "loss": 1.22,
      "step": 7978
    },
    {
      "epoch": 2.7194955691888207,
      "grad_norm": 21.6623954553323,
      "learning_rate": 2.646459669664364e-07,
      "loss": 1.6547,
      "step": 7979
    },
    {
      "epoch": 2.719836400817996,
      "grad_norm": 12.221742410829695,
      "learning_rate": 2.6400972221925394e-07,
      "loss": 1.2655,
      "step": 7980
    },
    {
      "epoch": 2.7201772324471714,
      "grad_norm": 10.577518404491006,
      "learning_rate": 2.633742224621949e-07,
      "loss": 1.2598,
      "step": 7981
    },
    {
      "epoch": 2.720518064076346,
      "grad_norm": 19.048002977817436,
      "learning_rate": 2.627394677952266e-07,
      "loss": 1.8956,
      "step": 7982
    },
    {
      "epoch": 2.7208588957055215,
      "grad_norm": 18.86522345361547,
      "learning_rate": 2.6210545831819756e-07,
      "loss": 1.4989,
      "step": 7983
    },
    {
      "epoch": 2.721199727334697,
      "grad_norm": 15.591515666256416,
      "learning_rate": 2.614721941308401e-07,
      "loss": 1.531,
      "step": 7984
    },
    {
      "epoch": 2.7215405589638717,
      "grad_norm": 14.383243902356828,
      "learning_rate": 2.608396753327702e-07,
      "loss": 1.3646,
      "step": 7985
    },
    {
      "epoch": 2.721881390593047,
      "grad_norm": 14.444769449514615,
      "learning_rate": 2.6020790202348434e-07,
      "loss": 1.6323,
      "step": 7986
    },
    {
      "epoch": 2.7222222222222223,
      "grad_norm": 14.580964161532115,
      "learning_rate": 2.595768743023641e-07,
      "loss": 1.4098,
      "step": 7987
    },
    {
      "epoch": 2.722563053851397,
      "grad_norm": 15.429590082523145,
      "learning_rate": 2.589465922686712e-07,
      "loss": 1.6463,
      "step": 7988
    },
    {
      "epoch": 2.7229038854805725,
      "grad_norm": 15.750203964276686,
      "learning_rate": 2.5831705602155256e-07,
      "loss": 1.4012,
      "step": 7989
    },
    {
      "epoch": 2.723244717109748,
      "grad_norm": 21.601554094027126,
      "learning_rate": 2.576882656600377e-07,
      "loss": 1.7735,
      "step": 7990
    },
    {
      "epoch": 2.723585548738923,
      "grad_norm": 12.454560739481266,
      "learning_rate": 2.5706022128303474e-07,
      "loss": 1.4361,
      "step": 7991
    },
    {
      "epoch": 2.7239263803680984,
      "grad_norm": 18.011516499393014,
      "learning_rate": 2.564329229893414e-07,
      "loss": 1.6007,
      "step": 7992
    },
    {
      "epoch": 2.7242672119972733,
      "grad_norm": 12.108549131983066,
      "learning_rate": 2.558063708776315e-07,
      "loss": 1.2894,
      "step": 7993
    },
    {
      "epoch": 2.7246080436264486,
      "grad_norm": 15.134777705919934,
      "learning_rate": 2.551805650464645e-07,
      "loss": 1.997,
      "step": 7994
    },
    {
      "epoch": 2.724948875255624,
      "grad_norm": 26.99305205880644,
      "learning_rate": 2.545555055942839e-07,
      "loss": 1.6634,
      "step": 7995
    },
    {
      "epoch": 2.7252897068847988,
      "grad_norm": 11.946238786918919,
      "learning_rate": 2.5393119261941215e-07,
      "loss": 1.5398,
      "step": 7996
    },
    {
      "epoch": 2.725630538513974,
      "grad_norm": 15.423249571773761,
      "learning_rate": 2.533076262200568e-07,
      "loss": 1.1774,
      "step": 7997
    },
    {
      "epoch": 2.7259713701431494,
      "grad_norm": 23.729368309896937,
      "learning_rate": 2.5268480649430717e-07,
      "loss": 1.2914,
      "step": 7998
    },
    {
      "epoch": 2.7263122017723243,
      "grad_norm": 16.98704957041396,
      "learning_rate": 2.5206273354013545e-07,
      "loss": 1.8676,
      "step": 7999
    },
    {
      "epoch": 2.7266530334014996,
      "grad_norm": 16.927803959374334,
      "learning_rate": 2.5144140745539615e-07,
      "loss": 1.701,
      "step": 8000
    },
    {
      "epoch": 2.726993865030675,
      "grad_norm": 14.542656185517494,
      "learning_rate": 2.508208283378266e-07,
      "loss": 0.941,
      "step": 8001
    },
    {
      "epoch": 2.72733469665985,
      "grad_norm": 14.926994949286993,
      "learning_rate": 2.5020099628504603e-07,
      "loss": 1.2465,
      "step": 8002
    },
    {
      "epoch": 2.7276755282890255,
      "grad_norm": 20.274184177559007,
      "learning_rate": 2.4958191139455635e-07,
      "loss": 1.9504,
      "step": 8003
    },
    {
      "epoch": 2.7280163599182004,
      "grad_norm": 18.976217290048623,
      "learning_rate": 2.4896357376374247e-07,
      "loss": 1.1565,
      "step": 8004
    },
    {
      "epoch": 2.7283571915473757,
      "grad_norm": 18.09198532526831,
      "learning_rate": 2.4834598348987215e-07,
      "loss": 1.4387,
      "step": 8005
    },
    {
      "epoch": 2.728698023176551,
      "grad_norm": 19.989631264383593,
      "learning_rate": 2.477291406700927e-07,
      "loss": 1.7646,
      "step": 8006
    },
    {
      "epoch": 2.729038854805726,
      "grad_norm": 14.428165954028136,
      "learning_rate": 2.47113045401437e-07,
      "loss": 1.1664,
      "step": 8007
    },
    {
      "epoch": 2.729379686434901,
      "grad_norm": 13.06914767547314,
      "learning_rate": 2.464976977808209e-07,
      "loss": 1.5235,
      "step": 8008
    },
    {
      "epoch": 2.7297205180640765,
      "grad_norm": 16.385229766847026,
      "learning_rate": 2.4588309790503807e-07,
      "loss": 1.6467,
      "step": 8009
    },
    {
      "epoch": 2.7300613496932513,
      "grad_norm": 17.252306136240176,
      "learning_rate": 2.4526924587076897e-07,
      "loss": 1.7769,
      "step": 8010
    },
    {
      "epoch": 2.7304021813224266,
      "grad_norm": 27.53897805848033,
      "learning_rate": 2.446561417745763e-07,
      "loss": 1.0877,
      "step": 8011
    },
    {
      "epoch": 2.730743012951602,
      "grad_norm": 17.979534407460974,
      "learning_rate": 2.4404378571290134e-07,
      "loss": 1.1576,
      "step": 8012
    },
    {
      "epoch": 2.7310838445807772,
      "grad_norm": 18.01615740087379,
      "learning_rate": 2.434321777820714e-07,
      "loss": 1.5651,
      "step": 8013
    },
    {
      "epoch": 2.7314246762099526,
      "grad_norm": 17.268431744829886,
      "learning_rate": 2.4282131807829514e-07,
      "loss": 1.6115,
      "step": 8014
    },
    {
      "epoch": 2.7317655078391274,
      "grad_norm": 33.01078195286599,
      "learning_rate": 2.422112066976623e-07,
      "loss": 1.4326,
      "step": 8015
    },
    {
      "epoch": 2.7321063394683027,
      "grad_norm": 21.726212750313106,
      "learning_rate": 2.416018437361461e-07,
      "loss": 1.669,
      "step": 8016
    },
    {
      "epoch": 2.732447171097478,
      "grad_norm": 15.88764291068917,
      "learning_rate": 2.409932292896028e-07,
      "loss": 1.5688,
      "step": 8017
    },
    {
      "epoch": 2.732788002726653,
      "grad_norm": 12.60672310828375,
      "learning_rate": 2.4038536345376837e-07,
      "loss": 0.922,
      "step": 8018
    },
    {
      "epoch": 2.733128834355828,
      "grad_norm": 15.17108067484828,
      "learning_rate": 2.3977824632426315e-07,
      "loss": 1.8778,
      "step": 8019
    },
    {
      "epoch": 2.7334696659850035,
      "grad_norm": 14.832164972405886,
      "learning_rate": 2.391718779965896e-07,
      "loss": 1.4675,
      "step": 8020
    },
    {
      "epoch": 2.7338104976141784,
      "grad_norm": 18.067165350662524,
      "learning_rate": 2.385662585661319e-07,
      "loss": 0.8629,
      "step": 8021
    },
    {
      "epoch": 2.7341513292433537,
      "grad_norm": 17.49616322115636,
      "learning_rate": 2.379613881281545e-07,
      "loss": 1.3502,
      "step": 8022
    },
    {
      "epoch": 2.734492160872529,
      "grad_norm": 18.568013787983052,
      "learning_rate": 2.373572667778079e-07,
      "loss": 1.485,
      "step": 8023
    },
    {
      "epoch": 2.7348329925017043,
      "grad_norm": 16.08490939312666,
      "learning_rate": 2.3675389461012277e-07,
      "loss": 0.8774,
      "step": 8024
    },
    {
      "epoch": 2.7351738241308796,
      "grad_norm": 16.184170556922993,
      "learning_rate": 2.361512717200104e-07,
      "loss": 1.0303,
      "step": 8025
    },
    {
      "epoch": 2.7355146557600545,
      "grad_norm": 12.298412031776735,
      "learning_rate": 2.355493982022683e-07,
      "loss": 1.498,
      "step": 8026
    },
    {
      "epoch": 2.73585548738923,
      "grad_norm": 10.659676435373422,
      "learning_rate": 2.3494827415157073e-07,
      "loss": 1.3037,
      "step": 8027
    },
    {
      "epoch": 2.736196319018405,
      "grad_norm": 25.500503130277124,
      "learning_rate": 2.3434789966247819e-07,
      "loss": 1.6319,
      "step": 8028
    },
    {
      "epoch": 2.73653715064758,
      "grad_norm": 15.841047405476026,
      "learning_rate": 2.3374827482943284e-07,
      "loss": 1.8296,
      "step": 8029
    },
    {
      "epoch": 2.7368779822767553,
      "grad_norm": 11.466089992123193,
      "learning_rate": 2.3314939974675654e-07,
      "loss": 0.7136,
      "step": 8030
    },
    {
      "epoch": 2.7372188139059306,
      "grad_norm": 19.605687241656792,
      "learning_rate": 2.3255127450865556e-07,
      "loss": 2.2206,
      "step": 8031
    },
    {
      "epoch": 2.7375596455351054,
      "grad_norm": 20.037781253174384,
      "learning_rate": 2.319538992092174e-07,
      "loss": 1.5214,
      "step": 8032
    },
    {
      "epoch": 2.7379004771642808,
      "grad_norm": 12.185185582976224,
      "learning_rate": 2.3135727394241135e-07,
      "loss": 0.933,
      "step": 8033
    },
    {
      "epoch": 2.738241308793456,
      "grad_norm": 15.932446870108082,
      "learning_rate": 2.3076139880208848e-07,
      "loss": 1.413,
      "step": 8034
    },
    {
      "epoch": 2.7385821404226314,
      "grad_norm": 11.456850308827228,
      "learning_rate": 2.301662738819832e-07,
      "loss": 0.9231,
      "step": 8035
    },
    {
      "epoch": 2.7389229720518067,
      "grad_norm": 14.221084584620275,
      "learning_rate": 2.2957189927571067e-07,
      "loss": 1.3425,
      "step": 8036
    },
    {
      "epoch": 2.7392638036809815,
      "grad_norm": 10.547114177096399,
      "learning_rate": 2.2897827507676828e-07,
      "loss": 1.0822,
      "step": 8037
    },
    {
      "epoch": 2.739604635310157,
      "grad_norm": 35.503599147557075,
      "learning_rate": 2.283854013785358e-07,
      "loss": 1.684,
      "step": 8038
    },
    {
      "epoch": 2.739945466939332,
      "grad_norm": 12.314456891851659,
      "learning_rate": 2.2779327827427423e-07,
      "loss": 1.4911,
      "step": 8039
    },
    {
      "epoch": 2.740286298568507,
      "grad_norm": 19.215651314397363,
      "learning_rate": 2.2720190585712675e-07,
      "loss": 1.7581,
      "step": 8040
    },
    {
      "epoch": 2.7406271301976823,
      "grad_norm": 13.762081904524175,
      "learning_rate": 2.2661128422011902e-07,
      "loss": 1.5541,
      "step": 8041
    },
    {
      "epoch": 2.7409679618268576,
      "grad_norm": 23.544006338046398,
      "learning_rate": 2.260214134561589e-07,
      "loss": 1.5338,
      "step": 8042
    },
    {
      "epoch": 2.7413087934560325,
      "grad_norm": 13.891976178921853,
      "learning_rate": 2.2543229365803388e-07,
      "loss": 1.2814,
      "step": 8043
    },
    {
      "epoch": 2.741649625085208,
      "grad_norm": 15.630492025749321,
      "learning_rate": 2.2484392491841478e-07,
      "loss": 1.5571,
      "step": 8044
    },
    {
      "epoch": 2.741990456714383,
      "grad_norm": 17.747622878697886,
      "learning_rate": 2.242563073298565e-07,
      "loss": 1.6969,
      "step": 8045
    },
    {
      "epoch": 2.7423312883435584,
      "grad_norm": 12.676554466246637,
      "learning_rate": 2.236694409847917e-07,
      "loss": 1.2652,
      "step": 8046
    },
    {
      "epoch": 2.7426721199727337,
      "grad_norm": 15.438378465563972,
      "learning_rate": 2.230833259755366e-07,
      "loss": 1.2396,
      "step": 8047
    },
    {
      "epoch": 2.7430129516019086,
      "grad_norm": 12.05975451604001,
      "learning_rate": 2.2249796239429077e-07,
      "loss": 1.755,
      "step": 8048
    },
    {
      "epoch": 2.743353783231084,
      "grad_norm": 16.654247204596377,
      "learning_rate": 2.2191335033313333e-07,
      "loss": 0.792,
      "step": 8049
    },
    {
      "epoch": 2.7436946148602592,
      "grad_norm": 18.227265482141664,
      "learning_rate": 2.213294898840268e-07,
      "loss": 1.0457,
      "step": 8050
    },
    {
      "epoch": 2.744035446489434,
      "grad_norm": 13.590446656273961,
      "learning_rate": 2.2074638113881387e-07,
      "loss": 1.1177,
      "step": 8051
    },
    {
      "epoch": 2.7443762781186094,
      "grad_norm": 11.798789466556903,
      "learning_rate": 2.2016402418922e-07,
      "loss": 0.6563,
      "step": 8052
    },
    {
      "epoch": 2.7447171097477847,
      "grad_norm": 21.662529484682764,
      "learning_rate": 2.1958241912685308e-07,
      "loss": 1.272,
      "step": 8053
    },
    {
      "epoch": 2.7450579413769596,
      "grad_norm": 17.469318493486988,
      "learning_rate": 2.1900156604320154e-07,
      "loss": 1.3762,
      "step": 8054
    },
    {
      "epoch": 2.745398773006135,
      "grad_norm": 15.153564160173236,
      "learning_rate": 2.1842146502963568e-07,
      "loss": 0.5443,
      "step": 8055
    },
    {
      "epoch": 2.74573960463531,
      "grad_norm": 17.427679750970054,
      "learning_rate": 2.1784211617740692e-07,
      "loss": 1.685,
      "step": 8056
    },
    {
      "epoch": 2.7460804362644855,
      "grad_norm": 12.666362059019631,
      "learning_rate": 2.172635195776507e-07,
      "loss": 1.245,
      "step": 8057
    },
    {
      "epoch": 2.7464212678936604,
      "grad_norm": 11.384723431949269,
      "learning_rate": 2.1668567532138195e-07,
      "loss": 1.179,
      "step": 8058
    },
    {
      "epoch": 2.7467620995228357,
      "grad_norm": 13.837598869528144,
      "learning_rate": 2.1610858349949693e-07,
      "loss": 1.7304,
      "step": 8059
    },
    {
      "epoch": 2.747102931152011,
      "grad_norm": 40.22327788582566,
      "learning_rate": 2.1553224420277629e-07,
      "loss": 1.6781,
      "step": 8060
    },
    {
      "epoch": 2.7474437627811863,
      "grad_norm": 13.87101584810105,
      "learning_rate": 2.1495665752187866e-07,
      "loss": 1.1848,
      "step": 8061
    },
    {
      "epoch": 2.747784594410361,
      "grad_norm": 15.465899971504937,
      "learning_rate": 2.143818235473466e-07,
      "loss": 1.1726,
      "step": 8062
    },
    {
      "epoch": 2.7481254260395365,
      "grad_norm": 12.883909498844002,
      "learning_rate": 2.1380774236960555e-07,
      "loss": 0.8505,
      "step": 8063
    },
    {
      "epoch": 2.7484662576687118,
      "grad_norm": 25.576814505003547,
      "learning_rate": 2.1323441407895883e-07,
      "loss": 1.0972,
      "step": 8064
    },
    {
      "epoch": 2.7488070892978866,
      "grad_norm": 11.323084713368194,
      "learning_rate": 2.1266183876559377e-07,
      "loss": 1.3365,
      "step": 8065
    },
    {
      "epoch": 2.749147920927062,
      "grad_norm": 13.121440056508133,
      "learning_rate": 2.120900165195783e-07,
      "loss": 1.2805,
      "step": 8066
    },
    {
      "epoch": 2.7494887525562373,
      "grad_norm": 13.27551493476925,
      "learning_rate": 2.115189474308632e-07,
      "loss": 1.0859,
      "step": 8067
    },
    {
      "epoch": 2.7498295841854126,
      "grad_norm": 15.871500629656257,
      "learning_rate": 2.1094863158927947e-07,
      "loss": 1.148,
      "step": 8068
    },
    {
      "epoch": 2.7501704158145874,
      "grad_norm": 14.570448611242304,
      "learning_rate": 2.1037906908453976e-07,
      "loss": 1.4385,
      "step": 8069
    },
    {
      "epoch": 2.7505112474437627,
      "grad_norm": 11.878778618718865,
      "learning_rate": 2.0981026000623905e-07,
      "loss": 0.9275,
      "step": 8070
    },
    {
      "epoch": 2.750852079072938,
      "grad_norm": 13.572795450176923,
      "learning_rate": 2.09242204443853e-07,
      "loss": 1.1234,
      "step": 8071
    },
    {
      "epoch": 2.7511929107021134,
      "grad_norm": 17.78385295705943,
      "learning_rate": 2.0867490248673962e-07,
      "loss": 1.6557,
      "step": 8072
    },
    {
      "epoch": 2.7515337423312882,
      "grad_norm": 15.342226935848622,
      "learning_rate": 2.0810835422413634e-07,
      "loss": 1.9695,
      "step": 8073
    },
    {
      "epoch": 2.7518745739604635,
      "grad_norm": 16.120582437686735,
      "learning_rate": 2.0754255974516578e-07,
      "loss": 1.4478,
      "step": 8074
    },
    {
      "epoch": 2.752215405589639,
      "grad_norm": 20.0350989903574,
      "learning_rate": 2.069775191388268e-07,
      "loss": 1.0904,
      "step": 8075
    },
    {
      "epoch": 2.7525562372188137,
      "grad_norm": 14.254041110146149,
      "learning_rate": 2.0641323249400492e-07,
      "loss": 1.2898,
      "step": 8076
    },
    {
      "epoch": 2.752897068847989,
      "grad_norm": 15.085266736779047,
      "learning_rate": 2.0584969989946357e-07,
      "loss": 1.9499,
      "step": 8077
    },
    {
      "epoch": 2.7532379004771643,
      "grad_norm": 19.4120950492033,
      "learning_rate": 2.05286921443848e-07,
      "loss": 1.5242,
      "step": 8078
    },
    {
      "epoch": 2.7535787321063396,
      "grad_norm": 13.963152037505315,
      "learning_rate": 2.0472489721568733e-07,
      "loss": 1.4645,
      "step": 8079
    },
    {
      "epoch": 2.7539195637355145,
      "grad_norm": 13.966458299296342,
      "learning_rate": 2.0416362730338867e-07,
      "loss": 1.5248,
      "step": 8080
    },
    {
      "epoch": 2.75426039536469,
      "grad_norm": 14.45055689143638,
      "learning_rate": 2.0360311179524306e-07,
      "loss": 1.8254,
      "step": 8081
    },
    {
      "epoch": 2.754601226993865,
      "grad_norm": 13.967849206101805,
      "learning_rate": 2.030433507794205e-07,
      "loss": 1.611,
      "step": 8082
    },
    {
      "epoch": 2.7549420586230404,
      "grad_norm": 14.04047260996693,
      "learning_rate": 2.02484344343975e-07,
      "loss": 1.7143,
      "step": 8083
    },
    {
      "epoch": 2.7552828902522153,
      "grad_norm": 16.53607040341396,
      "learning_rate": 2.019260925768396e-07,
      "loss": 1.1461,
      "step": 8084
    },
    {
      "epoch": 2.7556237218813906,
      "grad_norm": 12.538006422886921,
      "learning_rate": 2.013685955658301e-07,
      "loss": 0.8306,
      "step": 8085
    },
    {
      "epoch": 2.755964553510566,
      "grad_norm": 20.61115446825965,
      "learning_rate": 2.0081185339864307e-07,
      "loss": 1.2447,
      "step": 8086
    },
    {
      "epoch": 2.7563053851397408,
      "grad_norm": 16.016191724387728,
      "learning_rate": 2.0025586616285564e-07,
      "loss": 1.6995,
      "step": 8087
    },
    {
      "epoch": 2.756646216768916,
      "grad_norm": 14.679644649384212,
      "learning_rate": 1.9970063394592731e-07,
      "loss": 1.7446,
      "step": 8088
    },
    {
      "epoch": 2.7569870483980914,
      "grad_norm": 16.1910583453371,
      "learning_rate": 1.9914615683519821e-07,
      "loss": 1.8297,
      "step": 8089
    },
    {
      "epoch": 2.7573278800272663,
      "grad_norm": 12.82728727464394,
      "learning_rate": 1.9859243491788915e-07,
      "loss": 1.2544,
      "step": 8090
    },
    {
      "epoch": 2.7576687116564416,
      "grad_norm": 15.274887421047548,
      "learning_rate": 1.9803946828110376e-07,
      "loss": 1.6071,
      "step": 8091
    },
    {
      "epoch": 2.758009543285617,
      "grad_norm": 19.60198842198723,
      "learning_rate": 1.974872570118258e-07,
      "loss": 1.8844,
      "step": 8092
    },
    {
      "epoch": 2.758350374914792,
      "grad_norm": 11.205531341747417,
      "learning_rate": 1.9693580119691912e-07,
      "loss": 1.5732,
      "step": 8093
    },
    {
      "epoch": 2.7586912065439675,
      "grad_norm": 16.345092787496416,
      "learning_rate": 1.9638510092313212e-07,
      "loss": 1.2704,
      "step": 8094
    },
    {
      "epoch": 2.7590320381731424,
      "grad_norm": 17.53788784878188,
      "learning_rate": 1.9583515627708993e-07,
      "loss": 1.4721,
      "step": 8095
    },
    {
      "epoch": 2.7593728698023177,
      "grad_norm": 13.00074760463257,
      "learning_rate": 1.952859673453017e-07,
      "loss": 1.4418,
      "step": 8096
    },
    {
      "epoch": 2.759713701431493,
      "grad_norm": 20.376683696667325,
      "learning_rate": 1.9473753421415831e-07,
      "loss": 1.4673,
      "step": 8097
    },
    {
      "epoch": 2.760054533060668,
      "grad_norm": 13.146924266413274,
      "learning_rate": 1.9418985696992853e-07,
      "loss": 1.5924,
      "step": 8098
    },
    {
      "epoch": 2.760395364689843,
      "grad_norm": 15.543786153447835,
      "learning_rate": 1.9364293569876514e-07,
      "loss": 1.4013,
      "step": 8099
    },
    {
      "epoch": 2.7607361963190185,
      "grad_norm": 16.976442657760913,
      "learning_rate": 1.930967704867015e-07,
      "loss": 1.2542,
      "step": 8100
    },
    {
      "epoch": 2.7610770279481933,
      "grad_norm": 15.289762489913526,
      "learning_rate": 1.925513614196506e-07,
      "loss": 1.9644,
      "step": 8101
    },
    {
      "epoch": 2.7614178595773686,
      "grad_norm": 18.86526358065276,
      "learning_rate": 1.920067085834082e-07,
      "loss": 2.0832,
      "step": 8102
    },
    {
      "epoch": 2.761758691206544,
      "grad_norm": 14.714201981417519,
      "learning_rate": 1.914628120636497e-07,
      "loss": 1.4764,
      "step": 8103
    },
    {
      "epoch": 2.7620995228357192,
      "grad_norm": 13.483246743486895,
      "learning_rate": 1.9091967194593274e-07,
      "loss": 1.2286,
      "step": 8104
    },
    {
      "epoch": 2.7624403544648946,
      "grad_norm": 12.065392854583529,
      "learning_rate": 1.9037728831569513e-07,
      "loss": 1.6434,
      "step": 8105
    },
    {
      "epoch": 2.7627811860940694,
      "grad_norm": 11.983975826867335,
      "learning_rate": 1.8983566125825581e-07,
      "loss": 1.1018,
      "step": 8106
    },
    {
      "epoch": 2.7631220177232447,
      "grad_norm": 13.524987476487178,
      "learning_rate": 1.8929479085881553e-07,
      "loss": 1.7251,
      "step": 8107
    },
    {
      "epoch": 2.76346284935242,
      "grad_norm": 14.25839180750318,
      "learning_rate": 1.887546772024551e-07,
      "loss": 1.5016,
      "step": 8108
    },
    {
      "epoch": 2.763803680981595,
      "grad_norm": 11.891673990896397,
      "learning_rate": 1.8821532037413603e-07,
      "loss": 1.2055,
      "step": 8109
    },
    {
      "epoch": 2.76414451261077,
      "grad_norm": 15.560069261157187,
      "learning_rate": 1.876767204587021e-07,
      "loss": 1.699,
      "step": 8110
    },
    {
      "epoch": 2.7644853442399455,
      "grad_norm": 16.48575922073995,
      "learning_rate": 1.8713887754087667e-07,
      "loss": 2.1257,
      "step": 8111
    },
    {
      "epoch": 2.7648261758691204,
      "grad_norm": 11.446483806331441,
      "learning_rate": 1.8660179170526372e-07,
      "loss": 1.304,
      "step": 8112
    },
    {
      "epoch": 2.7651670074982957,
      "grad_norm": 19.71555273362329,
      "learning_rate": 1.860654630363512e-07,
      "loss": 1.3135,
      "step": 8113
    },
    {
      "epoch": 2.765507839127471,
      "grad_norm": 17.795042535201315,
      "learning_rate": 1.8552989161850443e-07,
      "loss": 2.092,
      "step": 8114
    },
    {
      "epoch": 2.7658486707566463,
      "grad_norm": 18.08334423586602,
      "learning_rate": 1.8499507753597046e-07,
      "loss": 1.5266,
      "step": 8115
    },
    {
      "epoch": 2.7661895023858216,
      "grad_norm": 12.689453039573868,
      "learning_rate": 1.8446102087287864e-07,
      "loss": 1.371,
      "step": 8116
    },
    {
      "epoch": 2.7665303340149965,
      "grad_norm": 18.004566040562793,
      "learning_rate": 1.8392772171323737e-07,
      "loss": 1.258,
      "step": 8117
    },
    {
      "epoch": 2.766871165644172,
      "grad_norm": 12.87824797703949,
      "learning_rate": 1.8339518014093726e-07,
      "loss": 1.2489,
      "step": 8118
    },
    {
      "epoch": 2.767211997273347,
      "grad_norm": 13.442599068079128,
      "learning_rate": 1.8286339623974912e-07,
      "loss": 1.2635,
      "step": 8119
    },
    {
      "epoch": 2.767552828902522,
      "grad_norm": 15.220720239286619,
      "learning_rate": 1.8233237009332494e-07,
      "loss": 1.4624,
      "step": 8120
    },
    {
      "epoch": 2.7678936605316973,
      "grad_norm": 21.813804977064922,
      "learning_rate": 1.8180210178519676e-07,
      "loss": 0.94,
      "step": 8121
    },
    {
      "epoch": 2.7682344921608726,
      "grad_norm": 16.186803555152185,
      "learning_rate": 1.8127259139877785e-07,
      "loss": 1.5121,
      "step": 8122
    },
    {
      "epoch": 2.7685753237900474,
      "grad_norm": 18.893541979082446,
      "learning_rate": 1.8074383901736324e-07,
      "loss": 1.509,
      "step": 8123
    },
    {
      "epoch": 2.7689161554192228,
      "grad_norm": 18.509877734930946,
      "learning_rate": 1.802158447241259e-07,
      "loss": 1.1326,
      "step": 8124
    },
    {
      "epoch": 2.769256987048398,
      "grad_norm": 13.588683146774786,
      "learning_rate": 1.7968860860212377e-07,
      "loss": 0.8586,
      "step": 8125
    },
    {
      "epoch": 2.7695978186775734,
      "grad_norm": 11.538901970070764,
      "learning_rate": 1.7916213073429222e-07,
      "loss": 1.387,
      "step": 8126
    },
    {
      "epoch": 2.7699386503067487,
      "grad_norm": 14.089555591440478,
      "learning_rate": 1.7863641120344722e-07,
      "loss": 1.2256,
      "step": 8127
    },
    {
      "epoch": 2.7702794819359235,
      "grad_norm": 12.401098489948426,
      "learning_rate": 1.781114500922876e-07,
      "loss": 1.3816,
      "step": 8128
    },
    {
      "epoch": 2.770620313565099,
      "grad_norm": 16.178396481575074,
      "learning_rate": 1.775872474833923e-07,
      "loss": 0.7967,
      "step": 8129
    },
    {
      "epoch": 2.770961145194274,
      "grad_norm": 13.748991089538498,
      "learning_rate": 1.7706380345921926e-07,
      "loss": 1.1356,
      "step": 8130
    },
    {
      "epoch": 2.771301976823449,
      "grad_norm": 14.58879699690518,
      "learning_rate": 1.7654111810210982e-07,
      "loss": 1.057,
      "step": 8131
    },
    {
      "epoch": 2.7716428084526243,
      "grad_norm": 28.300996730820135,
      "learning_rate": 1.760191914942827e-07,
      "loss": 0.9746,
      "step": 8132
    },
    {
      "epoch": 2.7719836400817996,
      "grad_norm": 16.37358211517568,
      "learning_rate": 1.7549802371783997e-07,
      "loss": 1.5109,
      "step": 8133
    },
    {
      "epoch": 2.7723244717109745,
      "grad_norm": 20.848378455917345,
      "learning_rate": 1.7497761485476384e-07,
      "loss": 1.7734,
      "step": 8134
    },
    {
      "epoch": 2.77266530334015,
      "grad_norm": 12.692546134076267,
      "learning_rate": 1.7445796498691603e-07,
      "loss": 1.3555,
      "step": 8135
    },
    {
      "epoch": 2.773006134969325,
      "grad_norm": 17.833074802218132,
      "learning_rate": 1.739390741960395e-07,
      "loss": 1.3709,
      "step": 8136
    },
    {
      "epoch": 2.7733469665985004,
      "grad_norm": 15.437789537323338,
      "learning_rate": 1.734209425637584e-07,
      "loss": 1.4941,
      "step": 8137
    },
    {
      "epoch": 2.7736877982276757,
      "grad_norm": 12.13539086192995,
      "learning_rate": 1.7290357017157645e-07,
      "loss": 1.4273,
      "step": 8138
    },
    {
      "epoch": 2.7740286298568506,
      "grad_norm": 24.009442513637342,
      "learning_rate": 1.723869571008785e-07,
      "loss": 2.1476,
      "step": 8139
    },
    {
      "epoch": 2.774369461486026,
      "grad_norm": 25.82321412563603,
      "learning_rate": 1.7187110343292902e-07,
      "loss": 1.734,
      "step": 8140
    },
    {
      "epoch": 2.7747102931152012,
      "grad_norm": 16.25665191623066,
      "learning_rate": 1.713560092488753e-07,
      "loss": 1.4015,
      "step": 8141
    },
    {
      "epoch": 2.775051124744376,
      "grad_norm": 11.366317794320885,
      "learning_rate": 1.7084167462974366e-07,
      "loss": 0.7363,
      "step": 8142
    },
    {
      "epoch": 2.7753919563735514,
      "grad_norm": 14.916498077134852,
      "learning_rate": 1.703280996564388e-07,
      "loss": 1.0085,
      "step": 8143
    },
    {
      "epoch": 2.7757327880027267,
      "grad_norm": 19.020505154638137,
      "learning_rate": 1.6981528440975114e-07,
      "loss": 0.9038,
      "step": 8144
    },
    {
      "epoch": 2.7760736196319016,
      "grad_norm": 17.413386921467023,
      "learning_rate": 1.6930322897034667e-07,
      "loss": 1.4372,
      "step": 8145
    },
    {
      "epoch": 2.776414451261077,
      "grad_norm": 22.745872983405768,
      "learning_rate": 1.6879193341877377e-07,
      "loss": 1.2646,
      "step": 8146
    },
    {
      "epoch": 2.776755282890252,
      "grad_norm": 10.82796902878109,
      "learning_rate": 1.6828139783546248e-07,
      "loss": 1.4311,
      "step": 8147
    },
    {
      "epoch": 2.7770961145194275,
      "grad_norm": 17.098897054899176,
      "learning_rate": 1.6777162230072031e-07,
      "loss": 1.8226,
      "step": 8148
    },
    {
      "epoch": 2.777436946148603,
      "grad_norm": 19.30539874085868,
      "learning_rate": 1.6726260689473862e-07,
      "loss": 1.939,
      "step": 8149
    },
    {
      "epoch": 2.7777777777777777,
      "grad_norm": 12.05028778009361,
      "learning_rate": 1.667543516975867e-07,
      "loss": 1.5529,
      "step": 8150
    },
    {
      "epoch": 2.778118609406953,
      "grad_norm": 13.455676138414612,
      "learning_rate": 1.6624685678921503e-07,
      "loss": 1.441,
      "step": 8151
    },
    {
      "epoch": 2.7784594410361283,
      "grad_norm": 12.2758491450673,
      "learning_rate": 1.657401222494548e-07,
      "loss": 1.6013,
      "step": 8152
    },
    {
      "epoch": 2.778800272665303,
      "grad_norm": 13.728680305466733,
      "learning_rate": 1.6523414815801775e-07,
      "loss": 1.0432,
      "step": 8153
    },
    {
      "epoch": 2.7791411042944785,
      "grad_norm": 18.089662123876025,
      "learning_rate": 1.6472893459449523e-07,
      "loss": 1.367,
      "step": 8154
    },
    {
      "epoch": 2.779481935923654,
      "grad_norm": 13.199877463691424,
      "learning_rate": 1.6422448163835923e-07,
      "loss": 1.1024,
      "step": 8155
    },
    {
      "epoch": 2.7798227675528286,
      "grad_norm": 9.673611797313347,
      "learning_rate": 1.637207893689624e-07,
      "loss": 0.8688,
      "step": 8156
    },
    {
      "epoch": 2.780163599182004,
      "grad_norm": 21.88520773195195,
      "learning_rate": 1.6321785786553745e-07,
      "loss": 1.8145,
      "step": 8157
    },
    {
      "epoch": 2.7805044308111793,
      "grad_norm": 25.78202568123999,
      "learning_rate": 1.6271568720719777e-07,
      "loss": 0.9547,
      "step": 8158
    },
    {
      "epoch": 2.7808452624403546,
      "grad_norm": 14.709572367990148,
      "learning_rate": 1.6221427747293682e-07,
      "loss": 1.3391,
      "step": 8159
    },
    {
      "epoch": 2.78118609406953,
      "grad_norm": 17.31092689615749,
      "learning_rate": 1.617136287416282e-07,
      "loss": 1.4305,
      "step": 8160
    },
    {
      "epoch": 2.7815269256987047,
      "grad_norm": 16.208075083189453,
      "learning_rate": 1.6121374109202504e-07,
      "loss": 1.6146,
      "step": 8161
    },
    {
      "epoch": 2.78186775732788,
      "grad_norm": 23.874164187034555,
      "learning_rate": 1.6071461460276328e-07,
      "loss": 1.119,
      "step": 8162
    },
    {
      "epoch": 2.7822085889570554,
      "grad_norm": 12.240660214643773,
      "learning_rate": 1.602162493523568e-07,
      "loss": 0.973,
      "step": 8163
    },
    {
      "epoch": 2.7825494205862302,
      "grad_norm": 15.174533655837587,
      "learning_rate": 1.5971864541919958e-07,
      "loss": 1.7261,
      "step": 8164
    },
    {
      "epoch": 2.7828902522154055,
      "grad_norm": 15.009358694431825,
      "learning_rate": 1.592218028815684e-07,
      "loss": 1.6329,
      "step": 8165
    },
    {
      "epoch": 2.783231083844581,
      "grad_norm": 15.353075975576331,
      "learning_rate": 1.5872572181761693e-07,
      "loss": 1.9149,
      "step": 8166
    },
    {
      "epoch": 2.7835719154737557,
      "grad_norm": 24.349274586489422,
      "learning_rate": 1.58230402305381e-07,
      "loss": 1.4818,
      "step": 8167
    },
    {
      "epoch": 2.783912747102931,
      "grad_norm": 33.61540114448144,
      "learning_rate": 1.5773584442277833e-07,
      "loss": 1.6206,
      "step": 8168
    },
    {
      "epoch": 2.7842535787321063,
      "grad_norm": 10.956268534053383,
      "learning_rate": 1.5724204824760225e-07,
      "loss": 1.1908,
      "step": 8169
    },
    {
      "epoch": 2.7845944103612816,
      "grad_norm": 15.920096004875978,
      "learning_rate": 1.5674901385753006e-07,
      "loss": 1.1008,
      "step": 8170
    },
    {
      "epoch": 2.784935241990457,
      "grad_norm": 13.306461693949208,
      "learning_rate": 1.5625674133011747e-07,
      "loss": 1.5338,
      "step": 8171
    },
    {
      "epoch": 2.785276073619632,
      "grad_norm": 16.777547186671928,
      "learning_rate": 1.5576523074280147e-07,
      "loss": 1.888,
      "step": 8172
    },
    {
      "epoch": 2.785616905248807,
      "grad_norm": 18.12291279081509,
      "learning_rate": 1.5527448217289964e-07,
      "loss": 1.4311,
      "step": 8173
    },
    {
      "epoch": 2.7859577368779824,
      "grad_norm": 16.04256474860708,
      "learning_rate": 1.547844956976058e-07,
      "loss": 1.1867,
      "step": 8174
    },
    {
      "epoch": 2.7862985685071573,
      "grad_norm": 16.34093620529231,
      "learning_rate": 1.542952713939988e-07,
      "loss": 1.1314,
      "step": 8175
    },
    {
      "epoch": 2.7866394001363326,
      "grad_norm": 15.309479909604221,
      "learning_rate": 1.5380680933903657e-07,
      "loss": 1.338,
      "step": 8176
    },
    {
      "epoch": 2.786980231765508,
      "grad_norm": 12.969034876541869,
      "learning_rate": 1.533191096095532e-07,
      "loss": 1.707,
      "step": 8177
    },
    {
      "epoch": 2.7873210633946828,
      "grad_norm": 15.680194945967575,
      "learning_rate": 1.5283217228226843e-07,
      "loss": 1.6934,
      "step": 8178
    },
    {
      "epoch": 2.787661895023858,
      "grad_norm": 13.769872811503648,
      "learning_rate": 1.523459974337782e-07,
      "loss": 1.6648,
      "step": 8179
    },
    {
      "epoch": 2.7880027266530334,
      "grad_norm": 20.594639264707407,
      "learning_rate": 1.5186058514055912e-07,
      "loss": 1.832,
      "step": 8180
    },
    {
      "epoch": 2.7883435582822087,
      "grad_norm": 14.66963157369847,
      "learning_rate": 1.5137593547897013e-07,
      "loss": 1.1305,
      "step": 8181
    },
    {
      "epoch": 2.788684389911384,
      "grad_norm": 17.34911404610959,
      "learning_rate": 1.5089204852524798e-07,
      "loss": 1.3787,
      "step": 8182
    },
    {
      "epoch": 2.789025221540559,
      "grad_norm": 14.648814750549954,
      "learning_rate": 1.5040892435550901e-07,
      "loss": 1.4079,
      "step": 8183
    },
    {
      "epoch": 2.789366053169734,
      "grad_norm": 17.741851781690094,
      "learning_rate": 1.4992656304575192e-07,
      "loss": 1.4321,
      "step": 8184
    },
    {
      "epoch": 2.7897068847989095,
      "grad_norm": 20.919610011527535,
      "learning_rate": 1.4944496467185322e-07,
      "loss": 1.2326,
      "step": 8185
    },
    {
      "epoch": 2.7900477164280844,
      "grad_norm": 12.059932865922773,
      "learning_rate": 1.489641293095706e-07,
      "loss": 1.2192,
      "step": 8186
    },
    {
      "epoch": 2.7903885480572597,
      "grad_norm": 20.742222940802492,
      "learning_rate": 1.4848405703454138e-07,
      "loss": 1.464,
      "step": 8187
    },
    {
      "epoch": 2.790729379686435,
      "grad_norm": 16.431535105848507,
      "learning_rate": 1.4800474792228292e-07,
      "loss": 1.3139,
      "step": 8188
    },
    {
      "epoch": 2.79107021131561,
      "grad_norm": 21.62666517846386,
      "learning_rate": 1.4752620204819158e-07,
      "loss": 0.9921,
      "step": 8189
    },
    {
      "epoch": 2.791411042944785,
      "grad_norm": 15.99158995092772,
      "learning_rate": 1.47048419487546e-07,
      "loss": 1.4136,
      "step": 8190
    },
    {
      "epoch": 2.7917518745739605,
      "grad_norm": 13.556282574373911,
      "learning_rate": 1.4657140031550221e-07,
      "loss": 1.7717,
      "step": 8191
    },
    {
      "epoch": 2.7920927062031358,
      "grad_norm": 16.641194270395196,
      "learning_rate": 1.460951446070974e-07,
      "loss": 1.6968,
      "step": 8192
    },
    {
      "epoch": 2.792433537832311,
      "grad_norm": 10.885461758708278,
      "learning_rate": 1.4561965243724885e-07,
      "loss": 1.3212,
      "step": 8193
    },
    {
      "epoch": 2.792774369461486,
      "grad_norm": 18.182332932971832,
      "learning_rate": 1.4514492388075341e-07,
      "loss": 1.3676,
      "step": 8194
    },
    {
      "epoch": 2.7931152010906612,
      "grad_norm": 21.334825579813312,
      "learning_rate": 1.446709590122869e-07,
      "loss": 1.3119,
      "step": 8195
    },
    {
      "epoch": 2.7934560327198366,
      "grad_norm": 14.445806806986903,
      "learning_rate": 1.4419775790640633e-07,
      "loss": 1.1627,
      "step": 8196
    },
    {
      "epoch": 2.7937968643490114,
      "grad_norm": 17.26907347538002,
      "learning_rate": 1.437253206375494e-07,
      "loss": 1.672,
      "step": 8197
    },
    {
      "epoch": 2.7941376959781867,
      "grad_norm": 15.507726821319102,
      "learning_rate": 1.4325364728003e-07,
      "loss": 1.7855,
      "step": 8198
    },
    {
      "epoch": 2.794478527607362,
      "grad_norm": 13.866977204571521,
      "learning_rate": 1.4278273790804654e-07,
      "loss": 1.3328,
      "step": 8199
    },
    {
      "epoch": 2.794819359236537,
      "grad_norm": 6.862250622485755,
      "learning_rate": 1.423125925956731e-07,
      "loss": 0.6129,
      "step": 8200
    },
    {
      "epoch": 2.795160190865712,
      "grad_norm": 25.287235182396113,
      "learning_rate": 1.4184321141686608e-07,
      "loss": 1.834,
      "step": 8201
    },
    {
      "epoch": 2.7955010224948875,
      "grad_norm": 14.43193680465507,
      "learning_rate": 1.4137459444546198e-07,
      "loss": 1.7952,
      "step": 8202
    },
    {
      "epoch": 2.795841854124063,
      "grad_norm": 23.653452405922764,
      "learning_rate": 1.409067417551746e-07,
      "loss": 1.6094,
      "step": 8203
    },
    {
      "epoch": 2.796182685753238,
      "grad_norm": 23.04660792240722,
      "learning_rate": 1.4043965341959952e-07,
      "loss": 2.0725,
      "step": 8204
    },
    {
      "epoch": 2.796523517382413,
      "grad_norm": 12.773680533695984,
      "learning_rate": 1.3997332951221242e-07,
      "loss": 1.275,
      "step": 8205
    },
    {
      "epoch": 2.7968643490115883,
      "grad_norm": 11.665971548464492,
      "learning_rate": 1.395077701063663e-07,
      "loss": 1.4585,
      "step": 8206
    },
    {
      "epoch": 2.7972051806407636,
      "grad_norm": 16.451612390528506,
      "learning_rate": 1.3904297527529754e-07,
      "loss": 1.7009,
      "step": 8207
    },
    {
      "epoch": 2.7975460122699385,
      "grad_norm": 16.905497351962605,
      "learning_rate": 1.3857894509211822e-07,
      "loss": 1.077,
      "step": 8208
    },
    {
      "epoch": 2.797886843899114,
      "grad_norm": 16.804404952482624,
      "learning_rate": 1.381156796298233e-07,
      "loss": 1.4976,
      "step": 8209
    },
    {
      "epoch": 2.798227675528289,
      "grad_norm": 18.23956799680442,
      "learning_rate": 1.376531789612867e-07,
      "loss": 2.0403,
      "step": 8210
    },
    {
      "epoch": 2.798568507157464,
      "grad_norm": 15.005499711052913,
      "learning_rate": 1.3719144315925968e-07,
      "loss": 1.5192,
      "step": 8211
    },
    {
      "epoch": 2.7989093387866393,
      "grad_norm": 22.576359684317264,
      "learning_rate": 1.367304722963775e-07,
      "loss": 1.2735,
      "step": 8212
    },
    {
      "epoch": 2.7992501704158146,
      "grad_norm": 15.344371330910842,
      "learning_rate": 1.3627026644515097e-07,
      "loss": 0.9488,
      "step": 8213
    },
    {
      "epoch": 2.79959100204499,
      "grad_norm": 15.345259889830702,
      "learning_rate": 1.358108256779722e-07,
      "loss": 1.1782,
      "step": 8214
    },
    {
      "epoch": 2.799931833674165,
      "grad_norm": 13.68619779074325,
      "learning_rate": 1.353521500671151e-07,
      "loss": 1.424,
      "step": 8215
    },
    {
      "epoch": 2.80027266530334,
      "grad_norm": 10.424063523800296,
      "learning_rate": 1.348942396847297e-07,
      "loss": 1.4605,
      "step": 8216
    },
    {
      "epoch": 2.8006134969325154,
      "grad_norm": 25.48032531318921,
      "learning_rate": 1.3443709460284616e-07,
      "loss": 1.485,
      "step": 8217
    },
    {
      "epoch": 2.8009543285616907,
      "grad_norm": 15.04601390894561,
      "learning_rate": 1.339807148933775e-07,
      "loss": 1.3406,
      "step": 8218
    },
    {
      "epoch": 2.8012951601908656,
      "grad_norm": 17.788534784044888,
      "learning_rate": 1.335251006281124e-07,
      "loss": 1.5815,
      "step": 8219
    },
    {
      "epoch": 2.801635991820041,
      "grad_norm": 17.15743956511697,
      "learning_rate": 1.3307025187872135e-07,
      "loss": 1.4299,
      "step": 8220
    },
    {
      "epoch": 2.801976823449216,
      "grad_norm": 12.860507769089491,
      "learning_rate": 1.3261616871675375e-07,
      "loss": 1.5437,
      "step": 8221
    },
    {
      "epoch": 2.802317655078391,
      "grad_norm": 12.030930558230326,
      "learning_rate": 1.3216285121363913e-07,
      "loss": 1.5003,
      "step": 8222
    },
    {
      "epoch": 2.8026584867075663,
      "grad_norm": 10.59522362544998,
      "learning_rate": 1.317102994406849e-07,
      "loss": 1.2976,
      "step": 8223
    },
    {
      "epoch": 2.8029993183367417,
      "grad_norm": 32.829852478217894,
      "learning_rate": 1.3125851346908082e-07,
      "loss": 1.7733,
      "step": 8224
    },
    {
      "epoch": 2.803340149965917,
      "grad_norm": 13.559153641883631,
      "learning_rate": 1.3080749336989384e-07,
      "loss": 1.1825,
      "step": 8225
    },
    {
      "epoch": 2.8036809815950923,
      "grad_norm": 11.467271678237381,
      "learning_rate": 1.3035723921407117e-07,
      "loss": 1.4316,
      "step": 8226
    },
    {
      "epoch": 2.804021813224267,
      "grad_norm": 24.992102370026817,
      "learning_rate": 1.2990775107243947e-07,
      "loss": 0.9822,
      "step": 8227
    },
    {
      "epoch": 2.8043626448534424,
      "grad_norm": 21.233528937745827,
      "learning_rate": 1.2945902901570605e-07,
      "loss": 1.5317,
      "step": 8228
    },
    {
      "epoch": 2.8047034764826178,
      "grad_norm": 11.410600305985529,
      "learning_rate": 1.290110731144545e-07,
      "loss": 1.3397,
      "step": 8229
    },
    {
      "epoch": 2.8050443081117926,
      "grad_norm": 13.86308173974429,
      "learning_rate": 1.2856388343915226e-07,
      "loss": 1.4555,
      "step": 8230
    },
    {
      "epoch": 2.805385139740968,
      "grad_norm": 13.477384208997021,
      "learning_rate": 1.281174600601437e-07,
      "loss": 1.7159,
      "step": 8231
    },
    {
      "epoch": 2.8057259713701432,
      "grad_norm": 14.930078589709218,
      "learning_rate": 1.2767180304765149e-07,
      "loss": 1.5707,
      "step": 8232
    },
    {
      "epoch": 2.806066802999318,
      "grad_norm": 20.43375386680405,
      "learning_rate": 1.272269124717812e-07,
      "loss": 1.0831,
      "step": 8233
    },
    {
      "epoch": 2.8064076346284934,
      "grad_norm": 13.27559774431738,
      "learning_rate": 1.2678278840251467e-07,
      "loss": 1.2217,
      "step": 8234
    },
    {
      "epoch": 2.8067484662576687,
      "grad_norm": 10.699368227882474,
      "learning_rate": 1.2633943090971434e-07,
      "loss": 1.207,
      "step": 8235
    },
    {
      "epoch": 2.807089297886844,
      "grad_norm": 16.32259052207818,
      "learning_rate": 1.2589684006312276e-07,
      "loss": 1.6315,
      "step": 8236
    },
    {
      "epoch": 2.8074301295160193,
      "grad_norm": 10.214735566559185,
      "learning_rate": 1.2545501593236086e-07,
      "loss": 1.516,
      "step": 8237
    },
    {
      "epoch": 2.807770961145194,
      "grad_norm": 11.357863383507132,
      "learning_rate": 1.250139585869292e-07,
      "loss": 1.4676,
      "step": 8238
    },
    {
      "epoch": 2.8081117927743695,
      "grad_norm": 29.7893589148878,
      "learning_rate": 1.2457366809620786e-07,
      "loss": 1.9946,
      "step": 8239
    },
    {
      "epoch": 2.808452624403545,
      "grad_norm": 12.020215000489076,
      "learning_rate": 1.2413414452945693e-07,
      "loss": 1.2356,
      "step": 8240
    },
    {
      "epoch": 2.8087934560327197,
      "grad_norm": 17.79007337655226,
      "learning_rate": 1.2369538795581447e-07,
      "loss": 1.3717,
      "step": 8241
    },
    {
      "epoch": 2.809134287661895,
      "grad_norm": 56.095565943483614,
      "learning_rate": 1.2325739844429919e-07,
      "loss": 1.8355,
      "step": 8242
    },
    {
      "epoch": 2.8094751192910703,
      "grad_norm": 19.792954981304156,
      "learning_rate": 1.2282017606380814e-07,
      "loss": 1.7132,
      "step": 8243
    },
    {
      "epoch": 2.809815950920245,
      "grad_norm": 10.84996252715398,
      "learning_rate": 1.2238372088311913e-07,
      "loss": 0.928,
      "step": 8244
    },
    {
      "epoch": 2.8101567825494205,
      "grad_norm": 14.613616729931715,
      "learning_rate": 1.219480329708861e-07,
      "loss": 1.3954,
      "step": 8245
    },
    {
      "epoch": 2.810497614178596,
      "grad_norm": 17.488664023255833,
      "learning_rate": 1.21513112395647e-07,
      "loss": 1.749,
      "step": 8246
    },
    {
      "epoch": 2.810838445807771,
      "grad_norm": 20.721633169953872,
      "learning_rate": 1.2107895922581547e-07,
      "loss": 2.0875,
      "step": 8247
    },
    {
      "epoch": 2.8111792774369464,
      "grad_norm": 9.94286431974173,
      "learning_rate": 1.2064557352968408e-07,
      "loss": 1.2359,
      "step": 8248
    },
    {
      "epoch": 2.8115201090661213,
      "grad_norm": 30.48215864730469,
      "learning_rate": 1.2021295537542942e-07,
      "loss": 1.7001,
      "step": 8249
    },
    {
      "epoch": 2.8118609406952966,
      "grad_norm": 20.030829789932678,
      "learning_rate": 1.1978110483110096e-07,
      "loss": 1.8595,
      "step": 8250
    },
    {
      "epoch": 2.812201772324472,
      "grad_norm": 17.604134090878006,
      "learning_rate": 1.1935002196463208e-07,
      "loss": 1.2887,
      "step": 8251
    },
    {
      "epoch": 2.8125426039536467,
      "grad_norm": 16.10155947563347,
      "learning_rate": 1.1891970684383359e-07,
      "loss": 1.9301,
      "step": 8252
    },
    {
      "epoch": 2.812883435582822,
      "grad_norm": 17.05795608124555,
      "learning_rate": 1.1849015953639576e-07,
      "loss": 0.9244,
      "step": 8253
    },
    {
      "epoch": 2.8132242672119974,
      "grad_norm": 14.491177995074944,
      "learning_rate": 1.1806138010988788e-07,
      "loss": 1.5347,
      "step": 8254
    },
    {
      "epoch": 2.8135650988411722,
      "grad_norm": 17.38111847487713,
      "learning_rate": 1.1763336863175879e-07,
      "loss": 1.8779,
      "step": 8255
    },
    {
      "epoch": 2.8139059304703475,
      "grad_norm": 14.405274727581789,
      "learning_rate": 1.1720612516933627e-07,
      "loss": 0.7092,
      "step": 8256
    },
    {
      "epoch": 2.814246762099523,
      "grad_norm": 30.04210428779782,
      "learning_rate": 1.1677964978982715e-07,
      "loss": 1.1057,
      "step": 8257
    },
    {
      "epoch": 2.814587593728698,
      "grad_norm": 17.045106264908963,
      "learning_rate": 1.1635394256031884e-07,
      "loss": 1.8347,
      "step": 8258
    },
    {
      "epoch": 2.8149284253578735,
      "grad_norm": 9.621178593425691,
      "learning_rate": 1.1592900354777559e-07,
      "loss": 1.2854,
      "step": 8259
    },
    {
      "epoch": 2.8152692569870483,
      "grad_norm": 10.385211192582364,
      "learning_rate": 1.1550483281904278e-07,
      "loss": 1.279,
      "step": 8260
    },
    {
      "epoch": 2.8156100886162236,
      "grad_norm": 15.956150354432507,
      "learning_rate": 1.1508143044084319e-07,
      "loss": 1.3826,
      "step": 8261
    },
    {
      "epoch": 2.815950920245399,
      "grad_norm": 15.979278255782432,
      "learning_rate": 1.1465879647978128e-07,
      "loss": 1.4937,
      "step": 8262
    },
    {
      "epoch": 2.816291751874574,
      "grad_norm": 11.57109807596676,
      "learning_rate": 1.1423693100233724e-07,
      "loss": 1.0153,
      "step": 8263
    },
    {
      "epoch": 2.816632583503749,
      "grad_norm": 14.014465300016395,
      "learning_rate": 1.1381583407487352e-07,
      "loss": 1.6701,
      "step": 8264
    },
    {
      "epoch": 2.8169734151329244,
      "grad_norm": 13.737091953530468,
      "learning_rate": 1.133955057636299e-07,
      "loss": 1.3068,
      "step": 8265
    },
    {
      "epoch": 2.8173142467620993,
      "grad_norm": 17.73479776885619,
      "learning_rate": 1.1297594613472518e-07,
      "loss": 1.7996,
      "step": 8266
    },
    {
      "epoch": 2.8176550783912746,
      "grad_norm": 11.466213184583415,
      "learning_rate": 1.1255715525415933e-07,
      "loss": 1.3864,
      "step": 8267
    },
    {
      "epoch": 2.81799591002045,
      "grad_norm": 13.97634262270077,
      "learning_rate": 1.12139133187808e-07,
      "loss": 1.5242,
      "step": 8268
    },
    {
      "epoch": 2.818336741649625,
      "grad_norm": 16.696560211728773,
      "learning_rate": 1.1172188000142803e-07,
      "loss": 1.7046,
      "step": 8269
    },
    {
      "epoch": 2.8186775732788005,
      "grad_norm": 16.288142354125686,
      "learning_rate": 1.1130539576065636e-07,
      "loss": 1.8449,
      "step": 8270
    },
    {
      "epoch": 2.8190184049079754,
      "grad_norm": 11.868367389931052,
      "learning_rate": 1.1088968053100613e-07,
      "loss": 0.88,
      "step": 8271
    },
    {
      "epoch": 2.8193592365371507,
      "grad_norm": 15.666214078014766,
      "learning_rate": 1.1047473437787171e-07,
      "loss": 1.1995,
      "step": 8272
    },
    {
      "epoch": 2.819700068166326,
      "grad_norm": 12.729830413381647,
      "learning_rate": 1.1006055736652532e-07,
      "loss": 1.5465,
      "step": 8273
    },
    {
      "epoch": 2.820040899795501,
      "grad_norm": 24.906200654543838,
      "learning_rate": 1.0964714956211931e-07,
      "loss": 1.7233,
      "step": 8274
    },
    {
      "epoch": 2.820381731424676,
      "grad_norm": 14.224330686401801,
      "learning_rate": 1.0923451102968386e-07,
      "loss": 1.3604,
      "step": 8275
    },
    {
      "epoch": 2.8207225630538515,
      "grad_norm": 15.730646404817547,
      "learning_rate": 1.0882264183412928e-07,
      "loss": 1.3671,
      "step": 8276
    },
    {
      "epoch": 2.8210633946830264,
      "grad_norm": 14.415213068503464,
      "learning_rate": 1.0841154204024318e-07,
      "loss": 1.145,
      "step": 8277
    },
    {
      "epoch": 2.8214042263122017,
      "grad_norm": 17.291162758536686,
      "learning_rate": 1.080012117126944e-07,
      "loss": 1.7495,
      "step": 8278
    },
    {
      "epoch": 2.821745057941377,
      "grad_norm": 14.502457458706052,
      "learning_rate": 1.0759165091602852e-07,
      "loss": 1.6,
      "step": 8279
    },
    {
      "epoch": 2.8220858895705523,
      "grad_norm": 14.89627779416749,
      "learning_rate": 1.071828597146718e-07,
      "loss": 1.2145,
      "step": 8280
    },
    {
      "epoch": 2.8224267211997276,
      "grad_norm": 14.047376237823281,
      "learning_rate": 1.0677483817292832e-07,
      "loss": 1.4936,
      "step": 8281
    },
    {
      "epoch": 2.8227675528289025,
      "grad_norm": 16.659148087463063,
      "learning_rate": 1.0636758635498123e-07,
      "loss": 1.0427,
      "step": 8282
    },
    {
      "epoch": 2.8231083844580778,
      "grad_norm": 15.018556627528238,
      "learning_rate": 1.059611043248937e-07,
      "loss": 1.4299,
      "step": 8283
    },
    {
      "epoch": 2.823449216087253,
      "grad_norm": 15.143827898203147,
      "learning_rate": 1.0555539214660682e-07,
      "loss": 1.5391,
      "step": 8284
    },
    {
      "epoch": 2.823790047716428,
      "grad_norm": 15.544348062914478,
      "learning_rate": 1.0515044988393897e-07,
      "loss": 0.8926,
      "step": 8285
    },
    {
      "epoch": 2.8241308793456033,
      "grad_norm": 27.050845986625262,
      "learning_rate": 1.0474627760059253e-07,
      "loss": 2.0421,
      "step": 8286
    },
    {
      "epoch": 2.8244717109747786,
      "grad_norm": 14.825691478261193,
      "learning_rate": 1.043428753601422e-07,
      "loss": 1.6738,
      "step": 8287
    },
    {
      "epoch": 2.8248125426039534,
      "grad_norm": 10.415440467379385,
      "learning_rate": 1.0394024322604667e-07,
      "loss": 0.9944,
      "step": 8288
    },
    {
      "epoch": 2.8251533742331287,
      "grad_norm": 20.613455407819032,
      "learning_rate": 1.0353838126164029e-07,
      "loss": 1.6449,
      "step": 8289
    },
    {
      "epoch": 2.825494205862304,
      "grad_norm": 18.689015411118376,
      "learning_rate": 1.0313728953013912e-07,
      "loss": 1.9172,
      "step": 8290
    },
    {
      "epoch": 2.8258350374914794,
      "grad_norm": 19.514347591201773,
      "learning_rate": 1.0273696809463496e-07,
      "loss": 1.4183,
      "step": 8291
    },
    {
      "epoch": 2.8261758691206547,
      "grad_norm": 14.974139679956037,
      "learning_rate": 1.0233741701810074e-07,
      "loss": 1.5062,
      "step": 8292
    },
    {
      "epoch": 2.8265167007498295,
      "grad_norm": 15.18985240138093,
      "learning_rate": 1.019386363633873e-07,
      "loss": 1.1013,
      "step": 8293
    },
    {
      "epoch": 2.826857532379005,
      "grad_norm": 25.074197869570547,
      "learning_rate": 1.0154062619322447e-07,
      "loss": 1.5517,
      "step": 8294
    },
    {
      "epoch": 2.82719836400818,
      "grad_norm": 15.513846613000494,
      "learning_rate": 1.011433865702205e-07,
      "loss": 1.0701,
      "step": 8295
    },
    {
      "epoch": 2.827539195637355,
      "grad_norm": 17.556290672867668,
      "learning_rate": 1.0074691755686372e-07,
      "loss": 1.0388,
      "step": 8296
    },
    {
      "epoch": 2.8278800272665303,
      "grad_norm": 20.1995249312285,
      "learning_rate": 1.0035121921551872e-07,
      "loss": 1.5416,
      "step": 8297
    },
    {
      "epoch": 2.8282208588957056,
      "grad_norm": 17.707479674864402,
      "learning_rate": 9.995629160843178e-08,
      "loss": 0.9837,
      "step": 8298
    },
    {
      "epoch": 2.8285616905248805,
      "grad_norm": 15.090604757449515,
      "learning_rate": 9.956213479772659e-08,
      "loss": 1.5884,
      "step": 8299
    },
    {
      "epoch": 2.828902522154056,
      "grad_norm": 16.30195152143396,
      "learning_rate": 9.916874884540461e-08,
      "loss": 1.576,
      "step": 8300
    },
    {
      "epoch": 2.829243353783231,
      "grad_norm": 19.678421240139624,
      "learning_rate": 9.87761338133475e-08,
      "loss": 1.387,
      "step": 8301
    },
    {
      "epoch": 2.8295841854124064,
      "grad_norm": 16.091595203328364,
      "learning_rate": 9.83842897633147e-08,
      "loss": 1.6368,
      "step": 8302
    },
    {
      "epoch": 2.8299250170415817,
      "grad_norm": 14.357861485478358,
      "learning_rate": 9.799321675694584e-08,
      "loss": 1.9884,
      "step": 8303
    },
    {
      "epoch": 2.8302658486707566,
      "grad_norm": 13.05494881426643,
      "learning_rate": 9.760291485575779e-08,
      "loss": 1.2804,
      "step": 8304
    },
    {
      "epoch": 2.830606680299932,
      "grad_norm": 12.771030112933394,
      "learning_rate": 9.721338412114645e-08,
      "loss": 1.2111,
      "step": 8305
    },
    {
      "epoch": 2.830947511929107,
      "grad_norm": 15.840166839983452,
      "learning_rate": 9.682462461438669e-08,
      "loss": 1.6872,
      "step": 8306
    },
    {
      "epoch": 2.831288343558282,
      "grad_norm": 13.143594235705484,
      "learning_rate": 9.643663639663181e-08,
      "loss": 1.5148,
      "step": 8307
    },
    {
      "epoch": 2.8316291751874574,
      "grad_norm": 11.323021583793322,
      "learning_rate": 9.604941952891411e-08,
      "loss": 0.8044,
      "step": 8308
    },
    {
      "epoch": 2.8319700068166327,
      "grad_norm": 17.243010781828,
      "learning_rate": 9.566297407214431e-08,
      "loss": 1.7713,
      "step": 8309
    },
    {
      "epoch": 2.8323108384458076,
      "grad_norm": 12.678492379208308,
      "learning_rate": 9.527730008711156e-08,
      "loss": 1.3411,
      "step": 8310
    },
    {
      "epoch": 2.832651670074983,
      "grad_norm": 15.656272679024646,
      "learning_rate": 9.489239763448399e-08,
      "loss": 1.5851,
      "step": 8311
    },
    {
      "epoch": 2.832992501704158,
      "grad_norm": 10.680814806126607,
      "learning_rate": 9.450826677480874e-08,
      "loss": 1.0473,
      "step": 8312
    },
    {
      "epoch": 2.8333333333333335,
      "grad_norm": 20.15711308337298,
      "learning_rate": 9.412490756851022e-08,
      "loss": 1.5046,
      "step": 8313
    },
    {
      "epoch": 2.833674164962509,
      "grad_norm": 18.31043386896112,
      "learning_rate": 9.374232007589302e-08,
      "loss": 1.8175,
      "step": 8314
    },
    {
      "epoch": 2.8340149965916837,
      "grad_norm": 20.80942623777544,
      "learning_rate": 9.336050435714006e-08,
      "loss": 1.8046,
      "step": 8315
    },
    {
      "epoch": 2.834355828220859,
      "grad_norm": 12.53499206989504,
      "learning_rate": 9.29794604723111e-08,
      "loss": 0.9842,
      "step": 8316
    },
    {
      "epoch": 2.8346966598500343,
      "grad_norm": 17.921046678034052,
      "learning_rate": 9.259918848134764e-08,
      "loss": 1.6165,
      "step": 8317
    },
    {
      "epoch": 2.835037491479209,
      "grad_norm": 17.174488169167674,
      "learning_rate": 9.221968844406681e-08,
      "loss": 1.9184,
      "step": 8318
    },
    {
      "epoch": 2.8353783231083844,
      "grad_norm": 17.88123888455869,
      "learning_rate": 9.184096042016533e-08,
      "loss": 1.3134,
      "step": 8319
    },
    {
      "epoch": 2.8357191547375598,
      "grad_norm": 11.897018175693635,
      "learning_rate": 9.146300446921941e-08,
      "loss": 1.2825,
      "step": 8320
    },
    {
      "epoch": 2.8360599863667346,
      "grad_norm": 16.597004628731124,
      "learning_rate": 9.10858206506826e-08,
      "loss": 1.5684,
      "step": 8321
    },
    {
      "epoch": 2.83640081799591,
      "grad_norm": 10.624192387809481,
      "learning_rate": 9.070940902388748e-08,
      "loss": 0.9301,
      "step": 8322
    },
    {
      "epoch": 2.8367416496250852,
      "grad_norm": 17.07181807443578,
      "learning_rate": 9.033376964804497e-08,
      "loss": 1.343,
      "step": 8323
    },
    {
      "epoch": 2.8370824812542605,
      "grad_norm": 21.229396559520367,
      "learning_rate": 8.995890258224449e-08,
      "loss": 1.0703,
      "step": 8324
    },
    {
      "epoch": 2.837423312883436,
      "grad_norm": 15.878518165004843,
      "learning_rate": 8.958480788545499e-08,
      "loss": 1.1107,
      "step": 8325
    },
    {
      "epoch": 2.8377641445126107,
      "grad_norm": 17.04258312775287,
      "learning_rate": 8.921148561652216e-08,
      "loss": 1.946,
      "step": 8326
    },
    {
      "epoch": 2.838104976141786,
      "grad_norm": 15.988715966711233,
      "learning_rate": 8.883893583417125e-08,
      "loss": 1.7105,
      "step": 8327
    },
    {
      "epoch": 2.8384458077709613,
      "grad_norm": 18.13675934950191,
      "learning_rate": 8.846715859700595e-08,
      "loss": 2.0403,
      "step": 8328
    },
    {
      "epoch": 2.838786639400136,
      "grad_norm": 12.557680083795747,
      "learning_rate": 8.80961539635078e-08,
      "loss": 0.9557,
      "step": 8329
    },
    {
      "epoch": 2.8391274710293115,
      "grad_norm": 16.315648952971717,
      "learning_rate": 8.7725921992039e-08,
      "loss": 1.1675,
      "step": 8330
    },
    {
      "epoch": 2.839468302658487,
      "grad_norm": 14.673537556478621,
      "learning_rate": 8.735646274083631e-08,
      "loss": 1.3642,
      "step": 8331
    },
    {
      "epoch": 2.8398091342876617,
      "grad_norm": 16.493782441670522,
      "learning_rate": 8.698777626801824e-08,
      "loss": 1.8308,
      "step": 8332
    },
    {
      "epoch": 2.840149965916837,
      "grad_norm": 11.596640413292967,
      "learning_rate": 8.661986263158118e-08,
      "loss": 0.5732,
      "step": 8333
    },
    {
      "epoch": 2.8404907975460123,
      "grad_norm": 13.52512298627983,
      "learning_rate": 8.625272188939826e-08,
      "loss": 1.9182,
      "step": 8334
    },
    {
      "epoch": 2.8408316291751876,
      "grad_norm": 17.886532232614073,
      "learning_rate": 8.588635409922219e-08,
      "loss": 1.8457,
      "step": 8335
    },
    {
      "epoch": 2.841172460804363,
      "grad_norm": 19.81683975047375,
      "learning_rate": 8.552075931868576e-08,
      "loss": 1.3658,
      "step": 8336
    },
    {
      "epoch": 2.841513292433538,
      "grad_norm": 12.011842183621471,
      "learning_rate": 8.515593760529628e-08,
      "loss": 1.4196,
      "step": 8337
    },
    {
      "epoch": 2.841854124062713,
      "grad_norm": 16.02533976649134,
      "learning_rate": 8.479188901644286e-08,
      "loss": 1.0307,
      "step": 8338
    },
    {
      "epoch": 2.8421949556918884,
      "grad_norm": 15.333089956009555,
      "learning_rate": 8.442861360939192e-08,
      "loss": 1.4138,
      "step": 8339
    },
    {
      "epoch": 2.8425357873210633,
      "grad_norm": 24.171611145420794,
      "learning_rate": 8.406611144128774e-08,
      "loss": 1.5787,
      "step": 8340
    },
    {
      "epoch": 2.8428766189502386,
      "grad_norm": 10.750314781082961,
      "learning_rate": 8.370438256915359e-08,
      "loss": 0.8305,
      "step": 8341
    },
    {
      "epoch": 2.843217450579414,
      "grad_norm": 15.65574273137029,
      "learning_rate": 8.334342704989006e-08,
      "loss": 1.1973,
      "step": 8342
    },
    {
      "epoch": 2.8435582822085887,
      "grad_norm": 19.989696539145793,
      "learning_rate": 8.298324494027788e-08,
      "loss": 1.4668,
      "step": 8343
    },
    {
      "epoch": 2.843899113837764,
      "grad_norm": 14.389069255908973,
      "learning_rate": 8.262383629697502e-08,
      "loss": 1.6472,
      "step": 8344
    },
    {
      "epoch": 2.8442399454669394,
      "grad_norm": 13.523646052040549,
      "learning_rate": 8.226520117651737e-08,
      "loss": 1.3953,
      "step": 8345
    },
    {
      "epoch": 2.8445807770961147,
      "grad_norm": 15.999642880604428,
      "learning_rate": 8.190733963532038e-08,
      "loss": 1.4307,
      "step": 8346
    },
    {
      "epoch": 2.8449216087252895,
      "grad_norm": 10.267150892893753,
      "learning_rate": 8.155025172967624e-08,
      "loss": 1.2,
      "step": 8347
    },
    {
      "epoch": 2.845262440354465,
      "grad_norm": 13.898784335695028,
      "learning_rate": 8.119393751575666e-08,
      "loss": 1.311,
      "step": 8348
    },
    {
      "epoch": 2.84560327198364,
      "grad_norm": 14.82572689515187,
      "learning_rate": 8.083839704961183e-08,
      "loss": 1.3334,
      "step": 8349
    },
    {
      "epoch": 2.8459441036128155,
      "grad_norm": 11.613988523296225,
      "learning_rate": 8.048363038716866e-08,
      "loss": 1.5931,
      "step": 8350
    },
    {
      "epoch": 2.8462849352419903,
      "grad_norm": 17.358331479730804,
      "learning_rate": 8.01296375842342e-08,
      "loss": 1.0894,
      "step": 8351
    },
    {
      "epoch": 2.8466257668711656,
      "grad_norm": 15.472521868856697,
      "learning_rate": 7.977641869649278e-08,
      "loss": 1.9457,
      "step": 8352
    },
    {
      "epoch": 2.846966598500341,
      "grad_norm": 13.573665519936068,
      "learning_rate": 7.942397377950606e-08,
      "loss": 1.0465,
      "step": 8353
    },
    {
      "epoch": 2.847307430129516,
      "grad_norm": 16.1660170725371,
      "learning_rate": 7.907230288871748e-08,
      "loss": 1.5724,
      "step": 8354
    },
    {
      "epoch": 2.847648261758691,
      "grad_norm": 16.574719937623822,
      "learning_rate": 7.872140607944334e-08,
      "loss": 1.8747,
      "step": 8355
    },
    {
      "epoch": 2.8479890933878664,
      "grad_norm": 17.750565688127494,
      "learning_rate": 7.837128340688338e-08,
      "loss": 1.8695,
      "step": 8356
    },
    {
      "epoch": 2.8483299250170417,
      "grad_norm": 24.658080806646137,
      "learning_rate": 7.802193492611187e-08,
      "loss": 1.4672,
      "step": 8357
    },
    {
      "epoch": 2.8486707566462166,
      "grad_norm": 14.19087347699965,
      "learning_rate": 7.76733606920832e-08,
      "loss": 1.2929,
      "step": 8358
    },
    {
      "epoch": 2.849011588275392,
      "grad_norm": 15.426612604376055,
      "learning_rate": 7.732556075963016e-08,
      "loss": 1.5222,
      "step": 8359
    },
    {
      "epoch": 2.8493524199045672,
      "grad_norm": 16.669407403149016,
      "learning_rate": 7.697853518346176e-08,
      "loss": 1.4241,
      "step": 8360
    },
    {
      "epoch": 2.8496932515337425,
      "grad_norm": 22.85129948805778,
      "learning_rate": 7.663228401816768e-08,
      "loss": 1.2504,
      "step": 8361
    },
    {
      "epoch": 2.8500340831629174,
      "grad_norm": 12.283543098115507,
      "learning_rate": 7.62868073182138e-08,
      "loss": 1.3587,
      "step": 8362
    },
    {
      "epoch": 2.8503749147920927,
      "grad_norm": 17.9764453170893,
      "learning_rate": 7.594210513794553e-08,
      "loss": 1.0755,
      "step": 8363
    },
    {
      "epoch": 2.850715746421268,
      "grad_norm": 27.26723481640397,
      "learning_rate": 7.55981775315856e-08,
      "loss": 1.0965,
      "step": 8364
    },
    {
      "epoch": 2.851056578050443,
      "grad_norm": 18.236676620028383,
      "learning_rate": 7.525502455323575e-08,
      "loss": 1.1992,
      "step": 8365
    },
    {
      "epoch": 2.851397409679618,
      "grad_norm": 25.42849329023096,
      "learning_rate": 7.491264625687389e-08,
      "loss": 1.9462,
      "step": 8366
    },
    {
      "epoch": 2.8517382413087935,
      "grad_norm": 13.086045567024971,
      "learning_rate": 7.457104269635917e-08,
      "loss": 1.0996,
      "step": 8367
    },
    {
      "epoch": 2.852079072937969,
      "grad_norm": 14.995950298016279,
      "learning_rate": 7.423021392542585e-08,
      "loss": 1.3162,
      "step": 8368
    },
    {
      "epoch": 2.8524199045671437,
      "grad_norm": 18.995435525359163,
      "learning_rate": 7.389015999768823e-08,
      "loss": 1.3321,
      "step": 8369
    },
    {
      "epoch": 2.852760736196319,
      "grad_norm": 23.75078576257384,
      "learning_rate": 7.355088096663854e-08,
      "loss": 1.8234,
      "step": 8370
    },
    {
      "epoch": 2.8531015678254943,
      "grad_norm": 16.920289408859034,
      "learning_rate": 7.32123768856463e-08,
      "loss": 1.1213,
      "step": 8371
    },
    {
      "epoch": 2.8534423994546696,
      "grad_norm": 16.449804736985566,
      "learning_rate": 7.28746478079595e-08,
      "loss": 1.6279,
      "step": 8372
    },
    {
      "epoch": 2.8537832310838445,
      "grad_norm": 13.781941499752453,
      "learning_rate": 7.253769378670395e-08,
      "loss": 1.3934,
      "step": 8373
    },
    {
      "epoch": 2.8541240627130198,
      "grad_norm": 14.062293388327198,
      "learning_rate": 7.220151487488447e-08,
      "loss": 1.7862,
      "step": 8374
    },
    {
      "epoch": 2.854464894342195,
      "grad_norm": 21.443078064176632,
      "learning_rate": 7.186611112538323e-08,
      "loss": 1.5241,
      "step": 8375
    },
    {
      "epoch": 2.85480572597137,
      "grad_norm": 13.730903088028882,
      "learning_rate": 7.153148259096076e-08,
      "loss": 1.3361,
      "step": 8376
    },
    {
      "epoch": 2.8551465576005453,
      "grad_norm": 18.061306798848523,
      "learning_rate": 7.1197629324255e-08,
      "loss": 1.1922,
      "step": 8377
    },
    {
      "epoch": 2.8554873892297206,
      "grad_norm": 17.435140943622972,
      "learning_rate": 7.086455137778226e-08,
      "loss": 1.9982,
      "step": 8378
    },
    {
      "epoch": 2.855828220858896,
      "grad_norm": 17.43812622064147,
      "learning_rate": 7.053224880393783e-08,
      "loss": 1.5015,
      "step": 8379
    },
    {
      "epoch": 2.8561690524880707,
      "grad_norm": 14.484763355626791,
      "learning_rate": 7.020072165499381e-08,
      "loss": 1.1332,
      "step": 8380
    },
    {
      "epoch": 2.856509884117246,
      "grad_norm": 15.129067853163166,
      "learning_rate": 6.986996998310014e-08,
      "loss": 0.8882,
      "step": 8381
    },
    {
      "epoch": 2.8568507157464214,
      "grad_norm": 15.693009268057772,
      "learning_rate": 6.953999384028576e-08,
      "loss": 1.8093,
      "step": 8382
    },
    {
      "epoch": 2.8571915473755967,
      "grad_norm": 18.252988495944656,
      "learning_rate": 6.921079327845803e-08,
      "loss": 1.8729,
      "step": 8383
    },
    {
      "epoch": 2.8575323790047715,
      "grad_norm": 19.200546462843807,
      "learning_rate": 6.888236834940054e-08,
      "loss": 1.3218,
      "step": 8384
    },
    {
      "epoch": 2.857873210633947,
      "grad_norm": 17.120965142077626,
      "learning_rate": 6.855471910477641e-08,
      "loss": 1.5881,
      "step": 8385
    },
    {
      "epoch": 2.858214042263122,
      "grad_norm": 10.921803401348706,
      "learning_rate": 6.82278455961255e-08,
      "loss": 1.1725,
      "step": 8386
    },
    {
      "epoch": 2.858554873892297,
      "grad_norm": 14.040404196909988,
      "learning_rate": 6.790174787486669e-08,
      "loss": 1.6296,
      "step": 8387
    },
    {
      "epoch": 2.8588957055214723,
      "grad_norm": 12.650158729062975,
      "learning_rate": 6.757642599229674e-08,
      "loss": 1.5884,
      "step": 8388
    },
    {
      "epoch": 2.8592365371506476,
      "grad_norm": 13.540435806499223,
      "learning_rate": 6.725187999959026e-08,
      "loss": 1.397,
      "step": 8389
    },
    {
      "epoch": 2.8595773687798225,
      "grad_norm": 18.846869720595244,
      "learning_rate": 6.692810994779863e-08,
      "loss": 1.3682,
      "step": 8390
    },
    {
      "epoch": 2.859918200408998,
      "grad_norm": 14.882241801702468,
      "learning_rate": 6.660511588785279e-08,
      "loss": 0.6665,
      "step": 8391
    },
    {
      "epoch": 2.860259032038173,
      "grad_norm": 13.907625050639338,
      "learning_rate": 6.628289787056096e-08,
      "loss": 1.3159,
      "step": 8392
    },
    {
      "epoch": 2.8605998636673484,
      "grad_norm": 14.607263227885426,
      "learning_rate": 6.59614559466093e-08,
      "loss": 1.2993,
      "step": 8393
    },
    {
      "epoch": 2.8609406952965237,
      "grad_norm": 15.764255240668708,
      "learning_rate": 6.564079016656177e-08,
      "loss": 1.7458,
      "step": 8394
    },
    {
      "epoch": 2.8612815269256986,
      "grad_norm": 14.167157653780208,
      "learning_rate": 6.532090058086027e-08,
      "loss": 1.3899,
      "step": 8395
    },
    {
      "epoch": 2.861622358554874,
      "grad_norm": 20.895547809634046,
      "learning_rate": 6.500178723982564e-08,
      "loss": 1.0913,
      "step": 8396
    },
    {
      "epoch": 2.861963190184049,
      "grad_norm": 17.63280568306113,
      "learning_rate": 6.46834501936544e-08,
      "loss": 1.7421,
      "step": 8397
    },
    {
      "epoch": 2.862304021813224,
      "grad_norm": 12.860577008428628,
      "learning_rate": 6.436588949242262e-08,
      "loss": 1.3548,
      "step": 8398
    },
    {
      "epoch": 2.8626448534423994,
      "grad_norm": 29.00002250632659,
      "learning_rate": 6.404910518608476e-08,
      "loss": 1.2711,
      "step": 8399
    },
    {
      "epoch": 2.8629856850715747,
      "grad_norm": 15.03024202841385,
      "learning_rate": 6.373309732447097e-08,
      "loss": 0.958,
      "step": 8400
    },
    {
      "epoch": 2.8633265167007496,
      "grad_norm": 12.775467534976185,
      "learning_rate": 6.34178659572915e-08,
      "loss": 1.056,
      "step": 8401
    },
    {
      "epoch": 2.863667348329925,
      "grad_norm": 21.989257551230654,
      "learning_rate": 6.310341113413332e-08,
      "loss": 1.4024,
      "step": 8402
    },
    {
      "epoch": 2.8640081799591,
      "grad_norm": 15.791522649363761,
      "learning_rate": 6.278973290446022e-08,
      "loss": 1.5924,
      "step": 8403
    },
    {
      "epoch": 2.8643490115882755,
      "grad_norm": 16.68601879762969,
      "learning_rate": 6.247683131761717e-08,
      "loss": 1.6535,
      "step": 8404
    },
    {
      "epoch": 2.864689843217451,
      "grad_norm": 12.553769938935936,
      "learning_rate": 6.216470642282313e-08,
      "loss": 1.096,
      "step": 8405
    },
    {
      "epoch": 2.8650306748466257,
      "grad_norm": 43.71729096531252,
      "learning_rate": 6.185335826917715e-08,
      "loss": 1.5698,
      "step": 8406
    },
    {
      "epoch": 2.865371506475801,
      "grad_norm": 20.96630215946244,
      "learning_rate": 6.154278690565562e-08,
      "loss": 1.5897,
      "step": 8407
    },
    {
      "epoch": 2.8657123381049763,
      "grad_norm": 31.48575750178133,
      "learning_rate": 6.123299238111224e-08,
      "loss": 1.3451,
      "step": 8408
    },
    {
      "epoch": 2.866053169734151,
      "grad_norm": 13.216202007060252,
      "learning_rate": 6.092397474427969e-08,
      "loss": 0.8031,
      "step": 8409
    },
    {
      "epoch": 2.8663940013633264,
      "grad_norm": 17.601480750886914,
      "learning_rate": 6.061573404376742e-08,
      "loss": 1.6364,
      "step": 8410
    },
    {
      "epoch": 2.8667348329925018,
      "grad_norm": 19.1304943379796,
      "learning_rate": 6.030827032806219e-08,
      "loss": 1.4749,
      "step": 8411
    },
    {
      "epoch": 2.8670756646216766,
      "grad_norm": 18.452779646493852,
      "learning_rate": 6.000158364552977e-08,
      "loss": 1.1529,
      "step": 8412
    },
    {
      "epoch": 2.867416496250852,
      "grad_norm": 14.69799375569531,
      "learning_rate": 5.969567404441379e-08,
      "loss": 1.4858,
      "step": 8413
    },
    {
      "epoch": 2.8677573278800272,
      "grad_norm": 18.829827877976776,
      "learning_rate": 5.939054157283464e-08,
      "loss": 1.5336,
      "step": 8414
    },
    {
      "epoch": 2.8680981595092025,
      "grad_norm": 15.532524599263466,
      "learning_rate": 5.908618627879004e-08,
      "loss": 1.4543,
      "step": 8415
    },
    {
      "epoch": 2.868438991138378,
      "grad_norm": 13.786290107010572,
      "learning_rate": 5.878260821015669e-08,
      "loss": 1.3437,
      "step": 8416
    },
    {
      "epoch": 2.8687798227675527,
      "grad_norm": 15.745544387797732,
      "learning_rate": 5.8479807414690284e-08,
      "loss": 1.1697,
      "step": 8417
    },
    {
      "epoch": 2.869120654396728,
      "grad_norm": 16.718101485207587,
      "learning_rate": 5.817778394001994e-08,
      "loss": 1.5091,
      "step": 8418
    },
    {
      "epoch": 2.8694614860259033,
      "grad_norm": 13.710174860414487,
      "learning_rate": 5.7876537833657097e-08,
      "loss": 0.9766,
      "step": 8419
    },
    {
      "epoch": 2.869802317655078,
      "grad_norm": 46.733975825429276,
      "learning_rate": 5.75760691429883e-08,
      "loss": 1.3394,
      "step": 8420
    },
    {
      "epoch": 2.8701431492842535,
      "grad_norm": 13.798049631051926,
      "learning_rate": 5.7276377915277956e-08,
      "loss": 1.8173,
      "step": 8421
    },
    {
      "epoch": 2.870483980913429,
      "grad_norm": 15.921797845236066,
      "learning_rate": 5.6977464197669475e-08,
      "loss": 1.252,
      "step": 8422
    },
    {
      "epoch": 2.8708248125426037,
      "grad_norm": 22.802474215951676,
      "learning_rate": 5.667932803718301e-08,
      "loss": 1.2649,
      "step": 8423
    },
    {
      "epoch": 2.871165644171779,
      "grad_norm": 22.795079629571035,
      "learning_rate": 5.638196948071606e-08,
      "loss": 1.2721,
      "step": 8424
    },
    {
      "epoch": 2.8715064758009543,
      "grad_norm": 14.27778974974975,
      "learning_rate": 5.608538857504564e-08,
      "loss": 1.0583,
      "step": 8425
    },
    {
      "epoch": 2.8718473074301296,
      "grad_norm": 21.309088069398737,
      "learning_rate": 5.5789585366823885e-08,
      "loss": 1.6431,
      "step": 8426
    },
    {
      "epoch": 2.872188139059305,
      "grad_norm": 15.105806623243124,
      "learning_rate": 5.5494559902581345e-08,
      "loss": 1.4269,
      "step": 8427
    },
    {
      "epoch": 2.87252897068848,
      "grad_norm": 11.70790203576845,
      "learning_rate": 5.520031222872812e-08,
      "loss": 1.1219,
      "step": 8428
    },
    {
      "epoch": 2.872869802317655,
      "grad_norm": 16.490669975041634,
      "learning_rate": 5.490684239154997e-08,
      "loss": 0.9806,
      "step": 8429
    },
    {
      "epoch": 2.8732106339468304,
      "grad_norm": 17.252104311707065,
      "learning_rate": 5.461415043721108e-08,
      "loss": 1.4795,
      "step": 8430
    },
    {
      "epoch": 2.8735514655760053,
      "grad_norm": 16.00800127097942,
      "learning_rate": 5.4322236411751825e-08,
      "loss": 1.955,
      "step": 8431
    },
    {
      "epoch": 2.8738922972051806,
      "grad_norm": 15.180145136007328,
      "learning_rate": 5.4031100361092714e-08,
      "loss": 0.7515,
      "step": 8432
    },
    {
      "epoch": 2.874233128834356,
      "grad_norm": 17.32183409083827,
      "learning_rate": 5.374074233103044e-08,
      "loss": 1.58,
      "step": 8433
    },
    {
      "epoch": 2.8745739604635308,
      "grad_norm": 14.929315777634194,
      "learning_rate": 5.345116236723902e-08,
      "loss": 1.7061,
      "step": 8434
    },
    {
      "epoch": 2.874914792092706,
      "grad_norm": 16.15255451062592,
      "learning_rate": 5.316236051527146e-08,
      "loss": 1.7457,
      "step": 8435
    },
    {
      "epoch": 2.8752556237218814,
      "grad_norm": 18.288367259893022,
      "learning_rate": 5.2874336820556427e-08,
      "loss": 1.1903,
      "step": 8436
    },
    {
      "epoch": 2.8755964553510567,
      "grad_norm": 19.004901507394745,
      "learning_rate": 5.258709132840156e-08,
      "loss": 1.8613,
      "step": 8437
    },
    {
      "epoch": 2.875937286980232,
      "grad_norm": 14.647524423471973,
      "learning_rate": 5.230062408399183e-08,
      "loss": 1.4736,
      "step": 8438
    },
    {
      "epoch": 2.876278118609407,
      "grad_norm": 19.067604844223652,
      "learning_rate": 5.201493513238953e-08,
      "loss": 1.2344,
      "step": 8439
    },
    {
      "epoch": 2.876618950238582,
      "grad_norm": 18.206570411526027,
      "learning_rate": 5.173002451853482e-08,
      "loss": 0.948,
      "step": 8440
    },
    {
      "epoch": 2.8769597818677575,
      "grad_norm": 15.659666689476822,
      "learning_rate": 5.1445892287245166e-08,
      "loss": 1.4352,
      "step": 8441
    },
    {
      "epoch": 2.8773006134969323,
      "grad_norm": 18.68152811873611,
      "learning_rate": 5.1162538483215393e-08,
      "loss": 1.9717,
      "step": 8442
    },
    {
      "epoch": 2.8776414451261076,
      "grad_norm": 14.854479237855594,
      "learning_rate": 5.087996315101873e-08,
      "loss": 1.264,
      "step": 8443
    },
    {
      "epoch": 2.877982276755283,
      "grad_norm": 17.034715399072848,
      "learning_rate": 5.059816633510517e-08,
      "loss": 1.5506,
      "step": 8444
    },
    {
      "epoch": 2.878323108384458,
      "grad_norm": 18.802915383985052,
      "learning_rate": 5.0317148079802594e-08,
      "loss": 1.5393,
      "step": 8445
    },
    {
      "epoch": 2.878663940013633,
      "grad_norm": 19.171390951841314,
      "learning_rate": 5.0036908429315636e-08,
      "loss": 1.7684,
      "step": 8446
    },
    {
      "epoch": 2.8790047716428084,
      "grad_norm": 18.748907409648105,
      "learning_rate": 4.975744742772848e-08,
      "loss": 1.7856,
      "step": 8447
    },
    {
      "epoch": 2.8793456032719837,
      "grad_norm": 16.503738215868566,
      "learning_rate": 4.947876511899985e-08,
      "loss": 1.5649,
      "step": 8448
    },
    {
      "epoch": 2.879686434901159,
      "grad_norm": 17.651059517218684,
      "learning_rate": 4.920086154696857e-08,
      "loss": 1.2485,
      "step": 8449
    },
    {
      "epoch": 2.880027266530334,
      "grad_norm": 30.757221196132544,
      "learning_rate": 4.8923736755350224e-08,
      "loss": 1.0016,
      "step": 8450
    },
    {
      "epoch": 2.8803680981595092,
      "grad_norm": 11.18420916240282,
      "learning_rate": 4.864739078773717e-08,
      "loss": 1.3712,
      "step": 8451
    },
    {
      "epoch": 2.8807089297886845,
      "grad_norm": 13.885192588165802,
      "learning_rate": 4.837182368759907e-08,
      "loss": 1.8337,
      "step": 8452
    },
    {
      "epoch": 2.8810497614178594,
      "grad_norm": 9.947024151879797,
      "learning_rate": 4.809703549828459e-08,
      "loss": 0.9492,
      "step": 8453
    },
    {
      "epoch": 2.8813905930470347,
      "grad_norm": 21.109868762671073,
      "learning_rate": 4.7823026263019155e-08,
      "loss": 1.2386,
      "step": 8454
    },
    {
      "epoch": 2.88173142467621,
      "grad_norm": 14.418995757925682,
      "learning_rate": 4.7549796024904946e-08,
      "loss": 1.1359,
      "step": 8455
    },
    {
      "epoch": 2.882072256305385,
      "grad_norm": 16.8203590906096,
      "learning_rate": 4.727734482692259e-08,
      "loss": 1.1043,
      "step": 8456
    },
    {
      "epoch": 2.88241308793456,
      "grad_norm": 23.599184348212734,
      "learning_rate": 4.7005672711929464e-08,
      "loss": 1.2202,
      "step": 8457
    },
    {
      "epoch": 2.8827539195637355,
      "grad_norm": 14.527696422684436,
      "learning_rate": 4.6734779722660274e-08,
      "loss": 1.6076,
      "step": 8458
    },
    {
      "epoch": 2.883094751192911,
      "grad_norm": 20.46979457770489,
      "learning_rate": 4.646466590172871e-08,
      "loss": 1.6775,
      "step": 8459
    },
    {
      "epoch": 2.883435582822086,
      "grad_norm": 11.993145930162838,
      "learning_rate": 4.619533129162357e-08,
      "loss": 1.3565,
      "step": 8460
    },
    {
      "epoch": 2.883776414451261,
      "grad_norm": 14.577269347944524,
      "learning_rate": 4.5926775934713175e-08,
      "loss": 1.4573,
      "step": 8461
    },
    {
      "epoch": 2.8841172460804363,
      "grad_norm": 11.051753112229575,
      "learning_rate": 4.565899987324152e-08,
      "loss": 1.1118,
      "step": 8462
    },
    {
      "epoch": 2.8844580777096116,
      "grad_norm": 19.25229714432816,
      "learning_rate": 4.539200314933156e-08,
      "loss": 1.9694,
      "step": 8463
    },
    {
      "epoch": 2.8847989093387865,
      "grad_norm": 22.528393283329194,
      "learning_rate": 4.512578580498306e-08,
      "loss": 1.3606,
      "step": 8464
    },
    {
      "epoch": 2.8851397409679618,
      "grad_norm": 9.915647970588964,
      "learning_rate": 4.486034788207194e-08,
      "loss": 1.0066,
      "step": 8465
    },
    {
      "epoch": 2.885480572597137,
      "grad_norm": 13.70683682775047,
      "learning_rate": 4.4595689422353704e-08,
      "loss": 1.0109,
      "step": 8466
    },
    {
      "epoch": 2.885821404226312,
      "grad_norm": 21.538262101207845,
      "learning_rate": 4.433181046746005e-08,
      "loss": 1.6137,
      "step": 8467
    },
    {
      "epoch": 2.8861622358554873,
      "grad_norm": 14.04998604676097,
      "learning_rate": 4.406871105889943e-08,
      "loss": 1.3961,
      "step": 8468
    },
    {
      "epoch": 2.8865030674846626,
      "grad_norm": 12.36013019316959,
      "learning_rate": 4.3806391238059296e-08,
      "loss": 1.3811,
      "step": 8469
    },
    {
      "epoch": 2.886843899113838,
      "grad_norm": 20.258866757178073,
      "learning_rate": 4.354485104620332e-08,
      "loss": 2.1531,
      "step": 8470
    },
    {
      "epoch": 2.887184730743013,
      "grad_norm": 10.456018928775213,
      "learning_rate": 4.3284090524472466e-08,
      "loss": 1.2759,
      "step": 8471
    },
    {
      "epoch": 2.887525562372188,
      "grad_norm": 10.075969701024677,
      "learning_rate": 4.302410971388615e-08,
      "loss": 0.9555,
      "step": 8472
    },
    {
      "epoch": 2.8878663940013634,
      "grad_norm": 14.843440532229327,
      "learning_rate": 4.276490865534e-08,
      "loss": 1.2165,
      "step": 8473
    },
    {
      "epoch": 2.8882072256305387,
      "grad_norm": 15.482083939334546,
      "learning_rate": 4.250648738960694e-08,
      "loss": 1.3199,
      "step": 8474
    },
    {
      "epoch": 2.8885480572597135,
      "grad_norm": 29.802202730175694,
      "learning_rate": 4.224884595733836e-08,
      "loss": 0.9042,
      "step": 8475
    },
    {
      "epoch": 2.888888888888889,
      "grad_norm": 14.114660527238856,
      "learning_rate": 4.199198439906182e-08,
      "loss": 1.6313,
      "step": 8476
    },
    {
      "epoch": 2.889229720518064,
      "grad_norm": 15.047302141868467,
      "learning_rate": 4.17359027551828e-08,
      "loss": 1.8258,
      "step": 8477
    },
    {
      "epoch": 2.889570552147239,
      "grad_norm": 12.733770833388384,
      "learning_rate": 4.148060106598406e-08,
      "loss": 1.0744,
      "step": 8478
    },
    {
      "epoch": 2.8899113837764143,
      "grad_norm": 13.38530820636626,
      "learning_rate": 4.122607937162515e-08,
      "loss": 1.0907,
      "step": 8479
    },
    {
      "epoch": 2.8902522154055896,
      "grad_norm": 13.599509872446655,
      "learning_rate": 4.097233771214404e-08,
      "loss": 1.2826,
      "step": 8480
    },
    {
      "epoch": 2.890593047034765,
      "grad_norm": 16.63642288491717,
      "learning_rate": 4.071937612745491e-08,
      "loss": 1.7476,
      "step": 8481
    },
    {
      "epoch": 2.8909338786639402,
      "grad_norm": 12.737215660833234,
      "learning_rate": 4.046719465734983e-08,
      "loss": 1.3849,
      "step": 8482
    },
    {
      "epoch": 2.891274710293115,
      "grad_norm": 14.571529202143621,
      "learning_rate": 4.021579334149761e-08,
      "loss": 0.9629,
      "step": 8483
    },
    {
      "epoch": 2.8916155419222904,
      "grad_norm": 17.106338506945786,
      "learning_rate": 3.996517221944496e-08,
      "loss": 1.2462,
      "step": 8484
    },
    {
      "epoch": 2.8919563735514657,
      "grad_norm": 18.001141819809217,
      "learning_rate": 3.971533133061589e-08,
      "loss": 1.62,
      "step": 8485
    },
    {
      "epoch": 2.8922972051806406,
      "grad_norm": 12.358786336266135,
      "learning_rate": 3.946627071431008e-08,
      "loss": 1.5734,
      "step": 8486
    },
    {
      "epoch": 2.892638036809816,
      "grad_norm": 15.711485580426448,
      "learning_rate": 3.921799040970675e-08,
      "loss": 1.766,
      "step": 8487
    },
    {
      "epoch": 2.892978868438991,
      "grad_norm": 18.224574186058202,
      "learning_rate": 3.897049045586188e-08,
      "loss": 1.8236,
      "step": 8488
    },
    {
      "epoch": 2.893319700068166,
      "grad_norm": 16.40398090204466,
      "learning_rate": 3.872377089170709e-08,
      "loss": 1.4799,
      "step": 8489
    },
    {
      "epoch": 2.8936605316973414,
      "grad_norm": 20.674135557302606,
      "learning_rate": 3.847783175605302e-08,
      "loss": 1.293,
      "step": 8490
    },
    {
      "epoch": 2.8940013633265167,
      "grad_norm": 16.047946873696606,
      "learning_rate": 3.823267308758649e-08,
      "loss": 1.171,
      "step": 8491
    },
    {
      "epoch": 2.894342194955692,
      "grad_norm": 13.109156077111171,
      "learning_rate": 3.7988294924872215e-08,
      "loss": 1.2528,
      "step": 8492
    },
    {
      "epoch": 2.8946830265848673,
      "grad_norm": 19.906350591462743,
      "learning_rate": 3.774469730635222e-08,
      "loss": 1.1636,
      "step": 8493
    },
    {
      "epoch": 2.895023858214042,
      "grad_norm": 18.012659282768993,
      "learning_rate": 3.7501880270344735e-08,
      "loss": 1.8389,
      "step": 8494
    },
    {
      "epoch": 2.8953646898432175,
      "grad_norm": 16.828929448443237,
      "learning_rate": 3.725984385504588e-08,
      "loss": 1.8891,
      "step": 8495
    },
    {
      "epoch": 2.895705521472393,
      "grad_norm": 12.24292900037543,
      "learning_rate": 3.70185880985291e-08,
      "loss": 1.5868,
      "step": 8496
    },
    {
      "epoch": 2.8960463531015677,
      "grad_norm": 14.774360875630633,
      "learning_rate": 3.677811303874457e-08,
      "loss": 1.465,
      "step": 8497
    },
    {
      "epoch": 2.896387184730743,
      "grad_norm": 13.674827742775463,
      "learning_rate": 3.653841871352093e-08,
      "loss": 1.3467,
      "step": 8498
    },
    {
      "epoch": 2.8967280163599183,
      "grad_norm": 15.724302237650683,
      "learning_rate": 3.629950516056191e-08,
      "loss": 1.7879,
      "step": 8499
    },
    {
      "epoch": 2.897068847989093,
      "grad_norm": 12.59241821210436,
      "learning_rate": 3.606137241745022e-08,
      "loss": 0.9418,
      "step": 8500
    },
    {
      "epoch": 2.8974096796182685,
      "grad_norm": 17.533075296230663,
      "learning_rate": 3.582402052164535e-08,
      "loss": 1.8739,
      "step": 8501
    },
    {
      "epoch": 2.8977505112474438,
      "grad_norm": 17.35371864223056,
      "learning_rate": 3.558744951048243e-08,
      "loss": 1.4909,
      "step": 8502
    },
    {
      "epoch": 2.898091342876619,
      "grad_norm": 9.785790039280375,
      "learning_rate": 3.535165942117669e-08,
      "loss": 1.2014,
      "step": 8503
    },
    {
      "epoch": 2.8984321745057944,
      "grad_norm": 19.321648524781175,
      "learning_rate": 3.51166502908179e-08,
      "loss": 1.5066,
      "step": 8504
    },
    {
      "epoch": 2.8987730061349692,
      "grad_norm": 13.825923888380867,
      "learning_rate": 3.488242215637372e-08,
      "loss": 1.62,
      "step": 8505
    },
    {
      "epoch": 2.8991138377641446,
      "grad_norm": 17.232753049977433,
      "learning_rate": 3.464897505469023e-08,
      "loss": 1.408,
      "step": 8506
    },
    {
      "epoch": 2.89945466939332,
      "grad_norm": 11.389547531028953,
      "learning_rate": 3.441630902248805e-08,
      "loss": 1.2117,
      "step": 8507
    },
    {
      "epoch": 2.8997955010224947,
      "grad_norm": 13.512811725861628,
      "learning_rate": 3.4184424096367904e-08,
      "loss": 0.9783,
      "step": 8508
    },
    {
      "epoch": 2.90013633265167,
      "grad_norm": 20.789827332127615,
      "learning_rate": 3.3953320312805074e-08,
      "loss": 1.7075,
      "step": 8509
    },
    {
      "epoch": 2.9004771642808453,
      "grad_norm": 18.698878012710562,
      "learning_rate": 3.372299770815435e-08,
      "loss": 2.1461,
      "step": 8510
    },
    {
      "epoch": 2.90081799591002,
      "grad_norm": 16.390705359776057,
      "learning_rate": 3.349345631864509e-08,
      "loss": 1.3143,
      "step": 8511
    },
    {
      "epoch": 2.9011588275391955,
      "grad_norm": 14.154075051152994,
      "learning_rate": 3.3264696180385635e-08,
      "loss": 1.3235,
      "step": 8512
    },
    {
      "epoch": 2.901499659168371,
      "grad_norm": 16.300364181886565,
      "learning_rate": 3.303671732936109e-08,
      "loss": 1.3438,
      "step": 8513
    },
    {
      "epoch": 2.901840490797546,
      "grad_norm": 11.488865756926726,
      "learning_rate": 3.280951980143332e-08,
      "loss": 1.1084,
      "step": 8514
    },
    {
      "epoch": 2.9021813224267214,
      "grad_norm": 16.774047400693387,
      "learning_rate": 3.258310363234152e-08,
      "loss": 2.0656,
      "step": 8515
    },
    {
      "epoch": 2.9025221540558963,
      "grad_norm": 16.420893179179167,
      "learning_rate": 3.235746885770163e-08,
      "loss": 1.7141,
      "step": 8516
    },
    {
      "epoch": 2.9028629856850716,
      "grad_norm": 11.226292248323459,
      "learning_rate": 3.213261551300639e-08,
      "loss": 1.0448,
      "step": 8517
    },
    {
      "epoch": 2.903203817314247,
      "grad_norm": 23.70300515981716,
      "learning_rate": 3.190854363362694e-08,
      "loss": 1.2234,
      "step": 8518
    },
    {
      "epoch": 2.903544648943422,
      "grad_norm": 24.827698303546647,
      "learning_rate": 3.1685253254810645e-08,
      "loss": 1.5049,
      "step": 8519
    },
    {
      "epoch": 2.903885480572597,
      "grad_norm": 19.107373633678275,
      "learning_rate": 3.146274441168162e-08,
      "loss": 1.4261,
      "step": 8520
    },
    {
      "epoch": 2.9042263122017724,
      "grad_norm": 16.252311888807718,
      "learning_rate": 3.124101713924133e-08,
      "loss": 1.8897,
      "step": 8521
    },
    {
      "epoch": 2.9045671438309473,
      "grad_norm": 16.486816482042176,
      "learning_rate": 3.1020071472369074e-08,
      "loss": 1.2799,
      "step": 8522
    },
    {
      "epoch": 2.9049079754601226,
      "grad_norm": 14.37740480419204,
      "learning_rate": 3.0799907445819286e-08,
      "loss": 1.0299,
      "step": 8523
    },
    {
      "epoch": 2.905248807089298,
      "grad_norm": 15.077652299601008,
      "learning_rate": 3.058052509422538e-08,
      "loss": 1.9165,
      "step": 8524
    },
    {
      "epoch": 2.905589638718473,
      "grad_norm": 20.366257358359626,
      "learning_rate": 3.0361924452097516e-08,
      "loss": 1.5694,
      "step": 8525
    },
    {
      "epoch": 2.9059304703476485,
      "grad_norm": 13.224864017487441,
      "learning_rate": 3.014410555382097e-08,
      "loss": 1.1462,
      "step": 8526
    },
    {
      "epoch": 2.9062713019768234,
      "grad_norm": 16.478616282026422,
      "learning_rate": 2.992706843366167e-08,
      "loss": 1.5697,
      "step": 8527
    },
    {
      "epoch": 2.9066121336059987,
      "grad_norm": 32.04292326173052,
      "learning_rate": 2.971081312575841e-08,
      "loss": 0.809,
      "step": 8528
    },
    {
      "epoch": 2.906952965235174,
      "grad_norm": 26.044509143206522,
      "learning_rate": 2.9495339664130095e-08,
      "loss": 0.9337,
      "step": 8529
    },
    {
      "epoch": 2.907293796864349,
      "grad_norm": 15.356498881595268,
      "learning_rate": 2.9280648082670726e-08,
      "loss": 1.463,
      "step": 8530
    },
    {
      "epoch": 2.907634628493524,
      "grad_norm": 18.38807079594237,
      "learning_rate": 2.9066738415153285e-08,
      "loss": 1.8922,
      "step": 8531
    },
    {
      "epoch": 2.9079754601226995,
      "grad_norm": 21.210370833842354,
      "learning_rate": 2.8853610695226408e-08,
      "loss": 1.6522,
      "step": 8532
    },
    {
      "epoch": 2.9083162917518743,
      "grad_norm": 19.794760014632708,
      "learning_rate": 2.864126495641495e-08,
      "loss": 1.6261,
      "step": 8533
    },
    {
      "epoch": 2.9086571233810496,
      "grad_norm": 29.075158760591723,
      "learning_rate": 2.8429701232122187e-08,
      "loss": 2.0802,
      "step": 8534
    },
    {
      "epoch": 2.908997955010225,
      "grad_norm": 16.663585844025008,
      "learning_rate": 2.8218919555628722e-08,
      "loss": 1.963,
      "step": 8535
    },
    {
      "epoch": 2.9093387866394003,
      "grad_norm": 18.485579846962164,
      "learning_rate": 2.8008919960090253e-08,
      "loss": 1.2401,
      "step": 8536
    },
    {
      "epoch": 2.9096796182685756,
      "grad_norm": 12.766100254135631,
      "learning_rate": 2.7799702478540914e-08,
      "loss": 0.9483,
      "step": 8537
    },
    {
      "epoch": 2.9100204498977504,
      "grad_norm": 9.667753034189479,
      "learning_rate": 2.759126714389215e-08,
      "loss": 0.9948,
      "step": 8538
    },
    {
      "epoch": 2.9103612815269257,
      "grad_norm": 15.48795130757094,
      "learning_rate": 2.7383613988930525e-08,
      "loss": 0.7564,
      "step": 8539
    },
    {
      "epoch": 2.910702113156101,
      "grad_norm": 18.57032994633064,
      "learning_rate": 2.717674304632156e-08,
      "loss": 1.7795,
      "step": 8540
    },
    {
      "epoch": 2.911042944785276,
      "grad_norm": 12.046281011856411,
      "learning_rate": 2.697065434860646e-08,
      "loss": 1.5831,
      "step": 8541
    },
    {
      "epoch": 2.9113837764144512,
      "grad_norm": 19.514470906902257,
      "learning_rate": 2.6765347928203734e-08,
      "loss": 0.8347,
      "step": 8542
    },
    {
      "epoch": 2.9117246080436265,
      "grad_norm": 9.67037147784539,
      "learning_rate": 2.656082381740921e-08,
      "loss": 1.1755,
      "step": 8543
    },
    {
      "epoch": 2.9120654396728014,
      "grad_norm": 14.666335952648044,
      "learning_rate": 2.6357082048395488e-08,
      "loss": 1.6043,
      "step": 8544
    },
    {
      "epoch": 2.9124062713019767,
      "grad_norm": 11.731992025215652,
      "learning_rate": 2.6154122653211377e-08,
      "loss": 0.8668,
      "step": 8545
    },
    {
      "epoch": 2.912747102931152,
      "grad_norm": 10.727060272273487,
      "learning_rate": 2.5951945663783562e-08,
      "loss": 1.2057,
      "step": 8546
    },
    {
      "epoch": 2.9130879345603273,
      "grad_norm": 13.83171317363536,
      "learning_rate": 2.5750551111915488e-08,
      "loss": 1.4279,
      "step": 8547
    },
    {
      "epoch": 2.9134287661895026,
      "grad_norm": 10.783987063979742,
      "learning_rate": 2.554993902928682e-08,
      "loss": 1.1919,
      "step": 8548
    },
    {
      "epoch": 2.9137695978186775,
      "grad_norm": 19.065677396983393,
      "learning_rate": 2.5350109447455084e-08,
      "loss": 1.5895,
      "step": 8549
    },
    {
      "epoch": 2.914110429447853,
      "grad_norm": 18.180484675140992,
      "learning_rate": 2.515106239785403e-08,
      "loss": 1.7875,
      "step": 8550
    },
    {
      "epoch": 2.914451261077028,
      "grad_norm": 16.132342631289813,
      "learning_rate": 2.4952797911794168e-08,
      "loss": 1.2325,
      "step": 8551
    },
    {
      "epoch": 2.914792092706203,
      "grad_norm": 17.67119757629895,
      "learning_rate": 2.4755316020463883e-08,
      "loss": 1.435,
      "step": 8552
    },
    {
      "epoch": 2.9151329243353783,
      "grad_norm": 14.205208623847932,
      "learning_rate": 2.455861675492832e-08,
      "loss": 1.8742,
      "step": 8553
    },
    {
      "epoch": 2.9154737559645536,
      "grad_norm": 16.40950145246869,
      "learning_rate": 2.4362700146127737e-08,
      "loss": 1.0581,
      "step": 8554
    },
    {
      "epoch": 2.9158145875937285,
      "grad_norm": 18.4654699838022,
      "learning_rate": 2.4167566224881367e-08,
      "loss": 1.3191,
      "step": 8555
    },
    {
      "epoch": 2.9161554192229038,
      "grad_norm": 10.775491348491641,
      "learning_rate": 2.3973215021884656e-08,
      "loss": 1.391,
      "step": 8556
    },
    {
      "epoch": 2.916496250852079,
      "grad_norm": 26.221405131584312,
      "learning_rate": 2.377964656770926e-08,
      "loss": 1.9159,
      "step": 8557
    },
    {
      "epoch": 2.9168370824812544,
      "grad_norm": 11.585239817636948,
      "learning_rate": 2.3586860892805264e-08,
      "loss": 1.6216,
      "step": 8558
    },
    {
      "epoch": 2.9171779141104297,
      "grad_norm": 18.661426894510413,
      "learning_rate": 2.339485802749786e-08,
      "loss": 1.3217,
      "step": 8559
    },
    {
      "epoch": 2.9175187457396046,
      "grad_norm": 13.87107507695412,
      "learning_rate": 2.320363800198955e-08,
      "loss": 1.6715,
      "step": 8560
    },
    {
      "epoch": 2.91785957736878,
      "grad_norm": 14.380670539582674,
      "learning_rate": 2.3013200846360717e-08,
      "loss": 1.5202,
      "step": 8561
    },
    {
      "epoch": 2.918200408997955,
      "grad_norm": 15.197060604546717,
      "learning_rate": 2.2823546590567958e-08,
      "loss": 0.9059,
      "step": 8562
    },
    {
      "epoch": 2.91854124062713,
      "grad_norm": 14.195243838343858,
      "learning_rate": 2.2634675264444073e-08,
      "loss": 1.6501,
      "step": 8563
    },
    {
      "epoch": 2.9188820722563054,
      "grad_norm": 14.793538743063742,
      "learning_rate": 2.2446586897699185e-08,
      "loss": 1.486,
      "step": 8564
    },
    {
      "epoch": 2.9192229038854807,
      "grad_norm": 13.152414093466783,
      "learning_rate": 2.2259281519921295e-08,
      "loss": 1.2486,
      "step": 8565
    },
    {
      "epoch": 2.9195637355146555,
      "grad_norm": 14.0001850368025,
      "learning_rate": 2.2072759160572944e-08,
      "loss": 0.8972,
      "step": 8566
    },
    {
      "epoch": 2.919904567143831,
      "grad_norm": 18.147437335371677,
      "learning_rate": 2.1887019848996216e-08,
      "loss": 1.015,
      "step": 8567
    },
    {
      "epoch": 2.920245398773006,
      "grad_norm": 33.252835309421855,
      "learning_rate": 2.170206361440774e-08,
      "loss": 1.9784,
      "step": 8568
    },
    {
      "epoch": 2.9205862304021815,
      "grad_norm": 19.543140273995856,
      "learning_rate": 2.1517890485902028e-08,
      "loss": 1.199,
      "step": 8569
    },
    {
      "epoch": 2.9209270620313568,
      "grad_norm": 16.042624675030755,
      "learning_rate": 2.133450049245034e-08,
      "loss": 1.7063,
      "step": 8570
    },
    {
      "epoch": 2.9212678936605316,
      "grad_norm": 14.679622655965144,
      "learning_rate": 2.1151893662900712e-08,
      "loss": 1.6599,
      "step": 8571
    },
    {
      "epoch": 2.921608725289707,
      "grad_norm": 13.843459398872325,
      "learning_rate": 2.0970070025977952e-08,
      "loss": 1.2798,
      "step": 8572
    },
    {
      "epoch": 2.9219495569188823,
      "grad_norm": 14.241016694438674,
      "learning_rate": 2.0789029610283064e-08,
      "loss": 1.3597,
      "step": 8573
    },
    {
      "epoch": 2.922290388548057,
      "grad_norm": 14.234500712330973,
      "learning_rate": 2.060877244429549e-08,
      "loss": 1.7335,
      "step": 8574
    },
    {
      "epoch": 2.9226312201772324,
      "grad_norm": 17.26196366258333,
      "learning_rate": 2.042929855636977e-08,
      "loss": 1.6363,
      "step": 8575
    },
    {
      "epoch": 2.9229720518064077,
      "grad_norm": 14.525610302152266,
      "learning_rate": 2.0250607974737214e-08,
      "loss": 1.487,
      "step": 8576
    },
    {
      "epoch": 2.9233128834355826,
      "grad_norm": 23.10031589203348,
      "learning_rate": 2.007270072750811e-08,
      "loss": 1.0803,
      "step": 8577
    },
    {
      "epoch": 2.923653715064758,
      "grad_norm": 17.04693954521189,
      "learning_rate": 1.9895576842666743e-08,
      "loss": 1.9102,
      "step": 8578
    },
    {
      "epoch": 2.923994546693933,
      "grad_norm": 16.856590509496062,
      "learning_rate": 1.971923634807582e-08,
      "loss": 1.8996,
      "step": 8579
    },
    {
      "epoch": 2.9243353783231085,
      "grad_norm": 15.507179967053363,
      "learning_rate": 1.9543679271474824e-08,
      "loss": 1.7114,
      "step": 8580
    },
    {
      "epoch": 2.924676209952284,
      "grad_norm": 14.181383205545426,
      "learning_rate": 1.936890564047833e-08,
      "loss": 1.368,
      "step": 8581
    },
    {
      "epoch": 2.9250170415814587,
      "grad_norm": 10.800216804612969,
      "learning_rate": 1.9194915482580455e-08,
      "loss": 0.9333,
      "step": 8582
    },
    {
      "epoch": 2.925357873210634,
      "grad_norm": 14.914334616181398,
      "learning_rate": 1.902170882514931e-08,
      "loss": 1.9051,
      "step": 8583
    },
    {
      "epoch": 2.9256987048398093,
      "grad_norm": 19.58111025412017,
      "learning_rate": 1.884928569543143e-08,
      "loss": 1.158,
      "step": 8584
    },
    {
      "epoch": 2.926039536468984,
      "grad_norm": 16.26026063371142,
      "learning_rate": 1.867764612055012e-08,
      "loss": 1.273,
      "step": 8585
    },
    {
      "epoch": 2.9263803680981595,
      "grad_norm": 15.978463034887712,
      "learning_rate": 1.8506790127504337e-08,
      "loss": 1.7219,
      "step": 8586
    },
    {
      "epoch": 2.926721199727335,
      "grad_norm": 11.263761488376462,
      "learning_rate": 1.833671774317092e-08,
      "loss": 1.1236,
      "step": 8587
    },
    {
      "epoch": 2.9270620313565097,
      "grad_norm": 15.816236566880669,
      "learning_rate": 1.816742899430235e-08,
      "loss": 1.6076,
      "step": 8588
    },
    {
      "epoch": 2.927402862985685,
      "grad_norm": 17.440587121405194,
      "learning_rate": 1.7998923907528444e-08,
      "loss": 1.9577,
      "step": 8589
    },
    {
      "epoch": 2.9277436946148603,
      "grad_norm": 14.845926241210684,
      "learning_rate": 1.783120250935688e-08,
      "loss": 1.6416,
      "step": 8590
    },
    {
      "epoch": 2.9280845262440356,
      "grad_norm": 13.545197967426947,
      "learning_rate": 1.766426482616934e-08,
      "loss": 1.8114,
      "step": 8591
    },
    {
      "epoch": 2.928425357873211,
      "grad_norm": 12.987617746018705,
      "learning_rate": 1.749811088422704e-08,
      "loss": 1.6007,
      "step": 8592
    },
    {
      "epoch": 2.9287661895023858,
      "grad_norm": 15.159695334511643,
      "learning_rate": 1.7332740709665174e-08,
      "loss": 0.9089,
      "step": 8593
    },
    {
      "epoch": 2.929107021131561,
      "grad_norm": 17.83428146721345,
      "learning_rate": 1.7168154328499055e-08,
      "loss": 1.013,
      "step": 8594
    },
    {
      "epoch": 2.9294478527607364,
      "grad_norm": 15.52656885238594,
      "learning_rate": 1.7004351766616854e-08,
      "loss": 1.965,
      "step": 8595
    },
    {
      "epoch": 2.9297886843899112,
      "grad_norm": 13.500446902156487,
      "learning_rate": 1.6841333049786855e-08,
      "loss": 1.3432,
      "step": 8596
    },
    {
      "epoch": 2.9301295160190866,
      "grad_norm": 16.361778906533495,
      "learning_rate": 1.6679098203651878e-08,
      "loss": 0.8819,
      "step": 8597
    },
    {
      "epoch": 2.930470347648262,
      "grad_norm": 19.331850925432136,
      "learning_rate": 1.651764725373206e-08,
      "loss": 1.1618,
      "step": 8598
    },
    {
      "epoch": 2.9308111792774367,
      "grad_norm": 30.148657224108412,
      "learning_rate": 1.6356980225424868e-08,
      "loss": 1.5087,
      "step": 8599
    },
    {
      "epoch": 2.931152010906612,
      "grad_norm": 20.255590750428993,
      "learning_rate": 1.619709714400286e-08,
      "loss": 1.827,
      "step": 8600
    },
    {
      "epoch": 2.9314928425357873,
      "grad_norm": 16.221503034435525,
      "learning_rate": 1.603799803461703e-08,
      "loss": 1.5123,
      "step": 8601
    },
    {
      "epoch": 2.9318336741649627,
      "grad_norm": 11.535754447865019,
      "learning_rate": 1.5879682922294027e-08,
      "loss": 1.0872,
      "step": 8602
    },
    {
      "epoch": 2.932174505794138,
      "grad_norm": 13.97257949982635,
      "learning_rate": 1.5722151831937816e-08,
      "loss": 1.3945,
      "step": 8603
    },
    {
      "epoch": 2.932515337423313,
      "grad_norm": 24.26248444148853,
      "learning_rate": 1.556540478832802e-08,
      "loss": 1.6107,
      "step": 8604
    },
    {
      "epoch": 2.932856169052488,
      "grad_norm": 16.39051192630273,
      "learning_rate": 1.5409441816122138e-08,
      "loss": 0.8853,
      "step": 8605
    },
    {
      "epoch": 2.9331970006816634,
      "grad_norm": 13.760881221303123,
      "learning_rate": 1.525426293985388e-08,
      "loss": 1.5612,
      "step": 8606
    },
    {
      "epoch": 2.9335378323108383,
      "grad_norm": 16.743581868499096,
      "learning_rate": 1.509986818393261e-08,
      "loss": 1.425,
      "step": 8607
    },
    {
      "epoch": 2.9338786639400136,
      "grad_norm": 18.911542388345026,
      "learning_rate": 1.4946257572646116e-08,
      "loss": 1.4418,
      "step": 8608
    },
    {
      "epoch": 2.934219495569189,
      "grad_norm": 29.472454774355914,
      "learning_rate": 1.4793431130157854e-08,
      "loss": 1.8294,
      "step": 8609
    },
    {
      "epoch": 2.934560327198364,
      "grad_norm": 19.1493044228161,
      "learning_rate": 1.4641388880507479e-08,
      "loss": 1.1564,
      "step": 8610
    },
    {
      "epoch": 2.934901158827539,
      "grad_norm": 17.20699852448855,
      "learning_rate": 1.4490130847612527e-08,
      "loss": 1.1155,
      "step": 8611
    },
    {
      "epoch": 2.9352419904567144,
      "grad_norm": 16.67961299379523,
      "learning_rate": 1.4339657055265632e-08,
      "loss": 1.4131,
      "step": 8612
    },
    {
      "epoch": 2.9355828220858897,
      "grad_norm": 20.093846906275523,
      "learning_rate": 1.418996752713786e-08,
      "loss": 1.2476,
      "step": 8613
    },
    {
      "epoch": 2.935923653715065,
      "grad_norm": 12.418595472841353,
      "learning_rate": 1.4041062286775375e-08,
      "loss": 1.6949,
      "step": 8614
    },
    {
      "epoch": 2.93626448534424,
      "grad_norm": 17.368803637122426,
      "learning_rate": 1.389294135760222e-08,
      "loss": 1.7027,
      "step": 8615
    },
    {
      "epoch": 2.936605316973415,
      "grad_norm": 12.371924537407596,
      "learning_rate": 1.374560476291753e-08,
      "loss": 1.3868,
      "step": 8616
    },
    {
      "epoch": 2.9369461486025905,
      "grad_norm": 13.823815592355194,
      "learning_rate": 1.3599052525898327e-08,
      "loss": 1.2486,
      "step": 8617
    },
    {
      "epoch": 2.9372869802317654,
      "grad_norm": 12.836353327258843,
      "learning_rate": 1.3453284669597833e-08,
      "loss": 1.4046,
      "step": 8618
    },
    {
      "epoch": 2.9376278118609407,
      "grad_norm": 10.778397294689825,
      "learning_rate": 1.330830121694604e-08,
      "loss": 0.8515,
      "step": 8619
    },
    {
      "epoch": 2.937968643490116,
      "grad_norm": 16.54153290488683,
      "learning_rate": 1.316410219074915e-08,
      "loss": 1.4021,
      "step": 8620
    },
    {
      "epoch": 2.938309475119291,
      "grad_norm": 12.201115779137458,
      "learning_rate": 1.3020687613690685e-08,
      "loss": 1.3311,
      "step": 8621
    },
    {
      "epoch": 2.938650306748466,
      "grad_norm": 14.885679169789656,
      "learning_rate": 1.2878057508329823e-08,
      "loss": 1.2737,
      "step": 8622
    },
    {
      "epoch": 2.9389911383776415,
      "grad_norm": 11.894284993349144,
      "learning_rate": 1.2736211897103613e-08,
      "loss": 1.0235,
      "step": 8623
    },
    {
      "epoch": 2.939331970006817,
      "grad_norm": 12.908963708539318,
      "learning_rate": 1.2595150802324207e-08,
      "loss": 1.3134,
      "step": 8624
    },
    {
      "epoch": 2.939672801635992,
      "grad_norm": 12.854405694826342,
      "learning_rate": 1.2454874246181081e-08,
      "loss": 0.9604,
      "step": 8625
    },
    {
      "epoch": 2.940013633265167,
      "grad_norm": 9.001557313884804,
      "learning_rate": 1.231538225074047e-08,
      "loss": 1.0381,
      "step": 8626
    },
    {
      "epoch": 2.9403544648943423,
      "grad_norm": 21.796673732147852,
      "learning_rate": 1.217667483794538e-08,
      "loss": 1.123,
      "step": 8627
    },
    {
      "epoch": 2.9406952965235176,
      "grad_norm": 13.917167948494303,
      "learning_rate": 1.2038752029614464e-08,
      "loss": 1.421,
      "step": 8628
    },
    {
      "epoch": 2.9410361281526924,
      "grad_norm": 15.995164295997728,
      "learning_rate": 1.1901613847443706e-08,
      "loss": 1.4732,
      "step": 8629
    },
    {
      "epoch": 2.9413769597818678,
      "grad_norm": 20.752417998906317,
      "learning_rate": 1.1765260313005844e-08,
      "loss": 1.1574,
      "step": 8630
    },
    {
      "epoch": 2.941717791411043,
      "grad_norm": 14.922126841516329,
      "learning_rate": 1.162969144774928e-08,
      "loss": 1.7117,
      "step": 8631
    },
    {
      "epoch": 2.942058623040218,
      "grad_norm": 18.92300045395376,
      "learning_rate": 1.1494907272999733e-08,
      "loss": 0.8706,
      "step": 8632
    },
    {
      "epoch": 2.9423994546693932,
      "grad_norm": 13.320803452364126,
      "learning_rate": 1.1360907809959687e-08,
      "loss": 1.546,
      "step": 8633
    },
    {
      "epoch": 2.9427402862985685,
      "grad_norm": 23.503239157882177,
      "learning_rate": 1.122769307970728e-08,
      "loss": 0.9436,
      "step": 8634
    },
    {
      "epoch": 2.943081117927744,
      "grad_norm": 11.894010569996643,
      "learning_rate": 1.1095263103197973e-08,
      "loss": 1.1001,
      "step": 8635
    },
    {
      "epoch": 2.943421949556919,
      "grad_norm": 11.660001747452846,
      "learning_rate": 1.0963617901263434e-08,
      "loss": 0.9267,
      "step": 8636
    },
    {
      "epoch": 2.943762781186094,
      "grad_norm": 15.777766658537628,
      "learning_rate": 1.08327574946121e-08,
      "loss": 1.6118,
      "step": 8637
    },
    {
      "epoch": 2.9441036128152693,
      "grad_norm": 12.660326600414482,
      "learning_rate": 1.0702681903828615e-08,
      "loss": 1.8174,
      "step": 8638
    },
    {
      "epoch": 2.9444444444444446,
      "grad_norm": 12.994033857651692,
      "learning_rate": 1.057339114937439e-08,
      "loss": 1.2878,
      "step": 8639
    },
    {
      "epoch": 2.9447852760736195,
      "grad_norm": 15.739674970759928,
      "learning_rate": 1.04448852515876e-08,
      "loss": 1.1095,
      "step": 8640
    },
    {
      "epoch": 2.945126107702795,
      "grad_norm": 11.434976591648086,
      "learning_rate": 1.031716423068263e-08,
      "loss": 1.446,
      "step": 8641
    },
    {
      "epoch": 2.94546693933197,
      "grad_norm": 13.917060989151137,
      "learning_rate": 1.0190228106750633e-08,
      "loss": 1.2087,
      "step": 8642
    },
    {
      "epoch": 2.945807770961145,
      "grad_norm": 18.638080835625615,
      "learning_rate": 1.0064076899758968e-08,
      "loss": 1.2174,
      "step": 8643
    },
    {
      "epoch": 2.9461486025903203,
      "grad_norm": 10.402169140323856,
      "learning_rate": 9.938710629551763e-09,
      "loss": 1.4282,
      "step": 8644
    },
    {
      "epoch": 2.9464894342194956,
      "grad_norm": 11.47390879455159,
      "learning_rate": 9.814129315849908e-09,
      "loss": 1.4103,
      "step": 8645
    },
    {
      "epoch": 2.946830265848671,
      "grad_norm": 15.685189311277485,
      "learning_rate": 9.69033297824995e-09,
      "loss": 1.6045,
      "step": 8646
    },
    {
      "epoch": 2.947171097477846,
      "grad_norm": 16.046300767971534,
      "learning_rate": 9.567321636225757e-09,
      "loss": 0.7149,
      "step": 8647
    },
    {
      "epoch": 2.947511929107021,
      "grad_norm": 14.796919036008703,
      "learning_rate": 9.445095309127961e-09,
      "loss": 1.5703,
      "step": 8648
    },
    {
      "epoch": 2.9478527607361964,
      "grad_norm": 15.054574786216534,
      "learning_rate": 9.323654016182848e-09,
      "loss": 1.8019,
      "step": 8649
    },
    {
      "epoch": 2.9481935923653717,
      "grad_norm": 16.23664189110524,
      "learning_rate": 9.202997776493472e-09,
      "loss": 1.409,
      "step": 8650
    },
    {
      "epoch": 2.9485344239945466,
      "grad_norm": 10.574074741040537,
      "learning_rate": 9.083126609040206e-09,
      "loss": 0.9976,
      "step": 8651
    },
    {
      "epoch": 2.948875255623722,
      "grad_norm": 16.139278570743258,
      "learning_rate": 8.96404053267852e-09,
      "loss": 1.6064,
      "step": 8652
    },
    {
      "epoch": 2.949216087252897,
      "grad_norm": 12.792669250067183,
      "learning_rate": 8.84573956614121e-09,
      "loss": 1.3869,
      "step": 8653
    },
    {
      "epoch": 2.949556918882072,
      "grad_norm": 12.245682733401445,
      "learning_rate": 8.728223728038388e-09,
      "loss": 1.0543,
      "step": 8654
    },
    {
      "epoch": 2.9498977505112474,
      "grad_norm": 14.510615822644022,
      "learning_rate": 8.611493036854157e-09,
      "loss": 1.8609,
      "step": 8655
    },
    {
      "epoch": 2.9502385821404227,
      "grad_norm": 14.36889300575775,
      "learning_rate": 8.495547510952163e-09,
      "loss": 1.5218,
      "step": 8656
    },
    {
      "epoch": 2.950579413769598,
      "grad_norm": 18.139075817700494,
      "learning_rate": 8.380387168570592e-09,
      "loss": 1.2239,
      "step": 8657
    },
    {
      "epoch": 2.950920245398773,
      "grad_norm": 13.114127164223571,
      "learning_rate": 8.266012027824954e-09,
      "loss": 1.4188,
      "step": 8658
    },
    {
      "epoch": 2.951261077027948,
      "grad_norm": 12.252060062976652,
      "learning_rate": 8.15242210670586e-09,
      "loss": 1.0566,
      "step": 8659
    },
    {
      "epoch": 2.9516019086571235,
      "grad_norm": 14.401278042428666,
      "learning_rate": 8.039617423082346e-09,
      "loss": 1.7395,
      "step": 8660
    },
    {
      "epoch": 2.9519427402862988,
      "grad_norm": 13.009425070688584,
      "learning_rate": 7.92759799469911e-09,
      "loss": 1.2238,
      "step": 8661
    },
    {
      "epoch": 2.9522835719154736,
      "grad_norm": 22.018122390743613,
      "learning_rate": 7.816363839177054e-09,
      "loss": 1.2757,
      "step": 8662
    },
    {
      "epoch": 2.952624403544649,
      "grad_norm": 12.953120185737616,
      "learning_rate": 7.705914974013296e-09,
      "loss": 1.5164,
      "step": 8663
    },
    {
      "epoch": 2.9529652351738243,
      "grad_norm": 42.039641954630525,
      "learning_rate": 7.596251416582822e-09,
      "loss": 1.2412,
      "step": 8664
    },
    {
      "epoch": 2.953306066802999,
      "grad_norm": 48.91256893337498,
      "learning_rate": 7.487373184135171e-09,
      "loss": 1.1311,
      "step": 8665
    },
    {
      "epoch": 2.9536468984321744,
      "grad_norm": 19.53535799287134,
      "learning_rate": 7.379280293797753e-09,
      "loss": 1.6843,
      "step": 8666
    },
    {
      "epoch": 2.9539877300613497,
      "grad_norm": 12.814526661312811,
      "learning_rate": 7.271972762574187e-09,
      "loss": 1.2731,
      "step": 8667
    },
    {
      "epoch": 2.954328561690525,
      "grad_norm": 19.17491488549641,
      "learning_rate": 7.165450607344304e-09,
      "loss": 0.9876,
      "step": 8668
    },
    {
      "epoch": 2.9546693933197,
      "grad_norm": 16.52254916454829,
      "learning_rate": 7.059713844864147e-09,
      "loss": 1.9505,
      "step": 8669
    },
    {
      "epoch": 2.955010224948875,
      "grad_norm": 14.508045637576299,
      "learning_rate": 6.954762491767075e-09,
      "loss": 1.4495,
      "step": 8670
    },
    {
      "epoch": 2.9553510565780505,
      "grad_norm": 13.134310111946034,
      "learning_rate": 6.850596564561551e-09,
      "loss": 1.6138,
      "step": 8671
    },
    {
      "epoch": 2.955691888207226,
      "grad_norm": 13.811437175344023,
      "learning_rate": 6.747216079633911e-09,
      "loss": 1.3598,
      "step": 8672
    },
    {
      "epoch": 2.9560327198364007,
      "grad_norm": 17.123777687389698,
      "learning_rate": 6.644621053246702e-09,
      "loss": 1.2517,
      "step": 8673
    },
    {
      "epoch": 2.956373551465576,
      "grad_norm": 18.81309701992254,
      "learning_rate": 6.542811501537571e-09,
      "loss": 1.068,
      "step": 8674
    },
    {
      "epoch": 2.9567143830947513,
      "grad_norm": 13.269690078259375,
      "learning_rate": 6.441787440521485e-09,
      "loss": 1.5072,
      "step": 8675
    },
    {
      "epoch": 2.957055214723926,
      "grad_norm": 22.789193904174475,
      "learning_rate": 6.341548886091286e-09,
      "loss": 1.3938,
      "step": 8676
    },
    {
      "epoch": 2.9573960463531015,
      "grad_norm": 13.71253497308214,
      "learning_rate": 6.242095854013808e-09,
      "loss": 1.2105,
      "step": 8677
    },
    {
      "epoch": 2.957736877982277,
      "grad_norm": 17.594498798793598,
      "learning_rate": 6.143428359933201e-09,
      "loss": 1.3587,
      "step": 8678
    },
    {
      "epoch": 2.9580777096114517,
      "grad_norm": 14.183136634183352,
      "learning_rate": 6.045546419371496e-09,
      "loss": 1.5091,
      "step": 8679
    },
    {
      "epoch": 2.958418541240627,
      "grad_norm": 19.9473363463764,
      "learning_rate": 5.9484500477247075e-09,
      "loss": 1.2338,
      "step": 8680
    },
    {
      "epoch": 2.9587593728698023,
      "grad_norm": 16.28830225110535,
      "learning_rate": 5.8521392602667314e-09,
      "loss": 1.2302,
      "step": 8681
    },
    {
      "epoch": 2.9591002044989776,
      "grad_norm": 10.535966445017367,
      "learning_rate": 5.756614072147671e-09,
      "loss": 1.2325,
      "step": 8682
    },
    {
      "epoch": 2.959441036128153,
      "grad_norm": 23.066292903098244,
      "learning_rate": 5.661874498394393e-09,
      "loss": 1.2344,
      "step": 8683
    },
    {
      "epoch": 2.9597818677573278,
      "grad_norm": 18.289654450956192,
      "learning_rate": 5.567920553909978e-09,
      "loss": 1.0006,
      "step": 8684
    },
    {
      "epoch": 2.960122699386503,
      "grad_norm": 19.29515633394969,
      "learning_rate": 5.474752253472604e-09,
      "loss": 1.247,
      "step": 8685
    },
    {
      "epoch": 2.9604635310156784,
      "grad_norm": 11.197093384867285,
      "learning_rate": 5.382369611739435e-09,
      "loss": 1.1371,
      "step": 8686
    },
    {
      "epoch": 2.9608043626448532,
      "grad_norm": 15.014400457347097,
      "learning_rate": 5.290772643241626e-09,
      "loss": 1.0969,
      "step": 8687
    },
    {
      "epoch": 2.9611451942740286,
      "grad_norm": 17.329205615622232,
      "learning_rate": 5.199961362387651e-09,
      "loss": 1.4793,
      "step": 8688
    },
    {
      "epoch": 2.961486025903204,
      "grad_norm": 17.36743352180151,
      "learning_rate": 5.109935783463305e-09,
      "loss": 1.2795,
      "step": 8689
    },
    {
      "epoch": 2.9618268575323787,
      "grad_norm": 17.78295320893461,
      "learning_rate": 5.0206959206300364e-09,
      "loss": 1.2827,
      "step": 8690
    },
    {
      "epoch": 2.962167689161554,
      "grad_norm": 13.384235586109837,
      "learning_rate": 4.9322417879243965e-09,
      "loss": 1.6259,
      "step": 8691
    },
    {
      "epoch": 2.9625085207907293,
      "grad_norm": 15.914181173616491,
      "learning_rate": 4.84457339926192e-09,
      "loss": 1.6242,
      "step": 8692
    },
    {
      "epoch": 2.9628493524199047,
      "grad_norm": 17.484786054924836,
      "learning_rate": 4.757690768432133e-09,
      "loss": 1.073,
      "step": 8693
    },
    {
      "epoch": 2.96319018404908,
      "grad_norm": 15.989498758071397,
      "learning_rate": 4.6715939091024344e-09,
      "loss": 1.3036,
      "step": 8694
    },
    {
      "epoch": 2.963531015678255,
      "grad_norm": 9.071651192899399,
      "learning_rate": 4.586282834816991e-09,
      "loss": 1.0327,
      "step": 8695
    },
    {
      "epoch": 2.96387184730743,
      "grad_norm": 22.70079678069397,
      "learning_rate": 4.501757558993958e-09,
      "loss": 1.4697,
      "step": 8696
    },
    {
      "epoch": 2.9642126789366054,
      "grad_norm": 16.772055267569474,
      "learning_rate": 4.418018094931032e-09,
      "loss": 1.1554,
      "step": 8697
    },
    {
      "epoch": 2.9645535105657803,
      "grad_norm": 15.876814461257688,
      "learning_rate": 4.335064455799898e-09,
      "loss": 1.2923,
      "step": 8698
    },
    {
      "epoch": 2.9648943421949556,
      "grad_norm": 14.24914999817876,
      "learning_rate": 4.2528966546495634e-09,
      "loss": 1.5015,
      "step": 8699
    },
    {
      "epoch": 2.965235173824131,
      "grad_norm": 15.691152807107361,
      "learning_rate": 4.171514704405799e-09,
      "loss": 1.916,
      "step": 8700
    },
    {
      "epoch": 2.965576005453306,
      "grad_norm": 18.468593879896456,
      "learning_rate": 4.0909186178700325e-09,
      "loss": 1.3798,
      "step": 8701
    },
    {
      "epoch": 2.965916837082481,
      "grad_norm": 13.018833352749837,
      "learning_rate": 4.011108407719899e-09,
      "loss": 1.4807,
      "step": 8702
    },
    {
      "epoch": 2.9662576687116564,
      "grad_norm": 10.167267202927293,
      "learning_rate": 3.932084086510912e-09,
      "loss": 1.089,
      "step": 8703
    },
    {
      "epoch": 2.9665985003408317,
      "grad_norm": 16.955101139888555,
      "learning_rate": 3.8538456666725736e-09,
      "loss": 1.1487,
      "step": 8704
    },
    {
      "epoch": 2.966939331970007,
      "grad_norm": 16.80441577963747,
      "learning_rate": 3.7763931605133694e-09,
      "loss": 1.8121,
      "step": 8705
    },
    {
      "epoch": 2.967280163599182,
      "grad_norm": 14.68001659111009,
      "learning_rate": 3.699726580216334e-09,
      "loss": 1.0802,
      "step": 8706
    },
    {
      "epoch": 2.967620995228357,
      "grad_norm": 9.98982755038774,
      "learning_rate": 3.623845937841264e-09,
      "loss": 1.2851,
      "step": 8707
    },
    {
      "epoch": 2.9679618268575325,
      "grad_norm": 28.4403213855046,
      "learning_rate": 3.548751245324722e-09,
      "loss": 1.3708,
      "step": 8708
    },
    {
      "epoch": 2.9683026584867074,
      "grad_norm": 12.522838137007149,
      "learning_rate": 3.4744425144789262e-09,
      "loss": 1.5277,
      "step": 8709
    },
    {
      "epoch": 2.9686434901158827,
      "grad_norm": 13.036411057043248,
      "learning_rate": 3.40091975699397e-09,
      "loss": 1.1401,
      "step": 8710
    },
    {
      "epoch": 2.968984321745058,
      "grad_norm": 14.002368013613793,
      "learning_rate": 3.3281829844344917e-09,
      "loss": 1.482,
      "step": 8711
    },
    {
      "epoch": 2.969325153374233,
      "grad_norm": 15.40560779480544,
      "learning_rate": 3.256232208241894e-09,
      "loss": 1.9392,
      "step": 8712
    },
    {
      "epoch": 2.969665985003408,
      "grad_norm": 11.402353154309171,
      "learning_rate": 3.1850674397354563e-09,
      "loss": 1.0036,
      "step": 8713
    },
    {
      "epoch": 2.9700068166325835,
      "grad_norm": 15.548131149146515,
      "learning_rate": 3.1146886901090024e-09,
      "loss": 1.5951,
      "step": 8714
    },
    {
      "epoch": 2.970347648261759,
      "grad_norm": 36.819427247516835,
      "learning_rate": 3.0450959704331207e-09,
      "loss": 1.6158,
      "step": 8715
    },
    {
      "epoch": 2.970688479890934,
      "grad_norm": 13.641395289755538,
      "learning_rate": 2.9762892916557205e-09,
      "loss": 1.7867,
      "step": 8716
    },
    {
      "epoch": 2.971029311520109,
      "grad_norm": 20.700161598521817,
      "learning_rate": 2.9082686645998113e-09,
      "loss": 2.0366,
      "step": 8717
    },
    {
      "epoch": 2.9713701431492843,
      "grad_norm": 45.706940006095024,
      "learning_rate": 2.8410340999657226e-09,
      "loss": 1.5135,
      "step": 8718
    },
    {
      "epoch": 2.9717109747784596,
      "grad_norm": 13.551935798355906,
      "learning_rate": 2.7745856083294386e-09,
      "loss": 1.5252,
      "step": 8719
    },
    {
      "epoch": 2.9720518064076344,
      "grad_norm": 21.699153490944383,
      "learning_rate": 2.7089232001431543e-09,
      "loss": 1.4525,
      "step": 8720
    },
    {
      "epoch": 2.9723926380368098,
      "grad_norm": 37.0888857504136,
      "learning_rate": 2.6440468857369396e-09,
      "loss": 1.7783,
      "step": 8721
    },
    {
      "epoch": 2.972733469665985,
      "grad_norm": 19.773664006700134,
      "learning_rate": 2.5799566753154094e-09,
      "loss": 1.2775,
      "step": 8722
    },
    {
      "epoch": 2.97307430129516,
      "grad_norm": 25.06010456748568,
      "learning_rate": 2.516652578959944e-09,
      "loss": 1.4918,
      "step": 8723
    },
    {
      "epoch": 2.9734151329243352,
      "grad_norm": 15.25037868842791,
      "learning_rate": 2.4541346066292437e-09,
      "loss": 1.7249,
      "step": 8724
    },
    {
      "epoch": 2.9737559645535105,
      "grad_norm": 17.26107512486261,
      "learning_rate": 2.3924027681571093e-09,
      "loss": 1.2717,
      "step": 8725
    },
    {
      "epoch": 2.974096796182686,
      "grad_norm": 14.993020431531003,
      "learning_rate": 2.3314570732541063e-09,
      "loss": 1.7361,
      "step": 8726
    },
    {
      "epoch": 2.974437627811861,
      "grad_norm": 15.010723043096037,
      "learning_rate": 2.2712975315075658e-09,
      "loss": 1.0909,
      "step": 8727
    },
    {
      "epoch": 2.974778459441036,
      "grad_norm": 16.247327811717707,
      "learning_rate": 2.2119241523810287e-09,
      "loss": 1.3751,
      "step": 8728
    },
    {
      "epoch": 2.9751192910702113,
      "grad_norm": 12.74991591906319,
      "learning_rate": 2.153336945213691e-09,
      "loss": 1.7517,
      "step": 8729
    },
    {
      "epoch": 2.9754601226993866,
      "grad_norm": 11.99642490312561,
      "learning_rate": 2.0955359192215143e-09,
      "loss": 1.0799,
      "step": 8730
    },
    {
      "epoch": 2.9758009543285615,
      "grad_norm": 21.781742986694297,
      "learning_rate": 2.0385210834972245e-09,
      "loss": 1.8693,
      "step": 8731
    },
    {
      "epoch": 2.976141785957737,
      "grad_norm": 16.701900721563774,
      "learning_rate": 1.982292447009204e-09,
      "loss": 1.605,
      "step": 8732
    },
    {
      "epoch": 2.976482617586912,
      "grad_norm": 11.385649081727555,
      "learning_rate": 1.9268500186025994e-09,
      "loss": 1.4428,
      "step": 8733
    },
    {
      "epoch": 2.976823449216087,
      "grad_norm": 12.537288480308787,
      "learning_rate": 1.8721938069987676e-09,
      "loss": 1.1865,
      "step": 8734
    },
    {
      "epoch": 2.9771642808452623,
      "grad_norm": 16.414246243353716,
      "learning_rate": 1.8183238207952759e-09,
      "loss": 1.6654,
      "step": 8735
    },
    {
      "epoch": 2.9775051124744376,
      "grad_norm": 13.247236961257771,
      "learning_rate": 1.765240068465901e-09,
      "loss": 1.2454,
      "step": 8736
    },
    {
      "epoch": 2.977845944103613,
      "grad_norm": 17.099696628245884,
      "learning_rate": 1.7129425583611857e-09,
      "loss": 1.158,
      "step": 8737
    },
    {
      "epoch": 2.9781867757327882,
      "grad_norm": 35.4342241643858,
      "learning_rate": 1.6614312987078828e-09,
      "loss": 1.3809,
      "step": 8738
    },
    {
      "epoch": 2.978527607361963,
      "grad_norm": 16.366758283759978,
      "learning_rate": 1.6107062976083998e-09,
      "loss": 1.5384,
      "step": 8739
    },
    {
      "epoch": 2.9788684389911384,
      "grad_norm": 31.616451754245432,
      "learning_rate": 1.5607675630419094e-09,
      "loss": 2.0056,
      "step": 8740
    },
    {
      "epoch": 2.9792092706203137,
      "grad_norm": 9.08069967640011,
      "learning_rate": 1.5116151028649052e-09,
      "loss": 1.3891,
      "step": 8741
    },
    {
      "epoch": 2.9795501022494886,
      "grad_norm": 7.8722954420131765,
      "learning_rate": 1.4632489248084247e-09,
      "loss": 1.0138,
      "step": 8742
    },
    {
      "epoch": 2.979890933878664,
      "grad_norm": 17.699351124924032,
      "learning_rate": 1.4156690364808267e-09,
      "loss": 1.5058,
      "step": 8743
    },
    {
      "epoch": 2.980231765507839,
      "grad_norm": 15.508553959367113,
      "learning_rate": 1.3688754453666797e-09,
      "loss": 1.0433,
      "step": 8744
    },
    {
      "epoch": 2.980572597137014,
      "grad_norm": 10.647124325623421,
      "learning_rate": 1.3228681588267623e-09,
      "loss": 1.0818,
      "step": 8745
    },
    {
      "epoch": 2.9809134287661894,
      "grad_norm": 18.815185437967003,
      "learning_rate": 1.2776471840986182e-09,
      "loss": 1.0788,
      "step": 8746
    },
    {
      "epoch": 2.9812542603953647,
      "grad_norm": 18.21682129905519,
      "learning_rate": 1.2332125282948915e-09,
      "loss": 1.0344,
      "step": 8747
    },
    {
      "epoch": 2.98159509202454,
      "grad_norm": 16.0548507210014,
      "learning_rate": 1.1895641984061012e-09,
      "loss": 1.347,
      "step": 8748
    },
    {
      "epoch": 2.9819359236537153,
      "grad_norm": 12.628833390077013,
      "learning_rate": 1.1467022012978667e-09,
      "loss": 1.6541,
      "step": 8749
    },
    {
      "epoch": 2.98227675528289,
      "grad_norm": 14.413029137103704,
      "learning_rate": 1.1046265437131276e-09,
      "loss": 1.3072,
      "step": 8750
    },
    {
      "epoch": 2.9826175869120655,
      "grad_norm": 14.882569236851317,
      "learning_rate": 1.063337232269368e-09,
      "loss": 1.0298,
      "step": 8751
    },
    {
      "epoch": 2.9829584185412408,
      "grad_norm": 15.878146160799863,
      "learning_rate": 1.022834273462503e-09,
      "loss": 1.4061,
      "step": 8752
    },
    {
      "epoch": 2.9832992501704156,
      "grad_norm": 17.480825167331275,
      "learning_rate": 9.831176736635473e-10,
      "loss": 2.0342,
      "step": 8753
    },
    {
      "epoch": 2.983640081799591,
      "grad_norm": 14.6003607295353,
      "learning_rate": 9.44187439120281e-10,
      "loss": 1.4236,
      "step": 8754
    },
    {
      "epoch": 2.9839809134287663,
      "grad_norm": 13.022309975826323,
      "learning_rate": 9.060435759566944e-10,
      "loss": 1.538,
      "step": 8755
    },
    {
      "epoch": 2.984321745057941,
      "grad_norm": 20.52988288870748,
      "learning_rate": 8.686860901718775e-10,
      "loss": 2.3399,
      "step": 8756
    },
    {
      "epoch": 2.9846625766871164,
      "grad_norm": 13.109524388123395,
      "learning_rate": 8.32114987643906e-10,
      "loss": 1.5327,
      "step": 8757
    },
    {
      "epoch": 2.9850034083162917,
      "grad_norm": 14.978285743690895,
      "learning_rate": 7.963302741242906e-10,
      "loss": 1.199,
      "step": 8758
    },
    {
      "epoch": 2.985344239945467,
      "grad_norm": 15.07117059362515,
      "learning_rate": 7.613319552429721e-10,
      "loss": 1.4839,
      "step": 8759
    },
    {
      "epoch": 2.9856850715746424,
      "grad_norm": 207.57297653500828,
      "learning_rate": 7.271200365049913e-10,
      "loss": 1.3884,
      "step": 8760
    },
    {
      "epoch": 2.9860259032038172,
      "grad_norm": 11.792015859776921,
      "learning_rate": 6.936945232915993e-10,
      "loss": 1.3367,
      "step": 8761
    },
    {
      "epoch": 2.9863667348329925,
      "grad_norm": 13.755168853019155,
      "learning_rate": 6.610554208613673e-10,
      "loss": 1.3477,
      "step": 8762
    },
    {
      "epoch": 2.986707566462168,
      "grad_norm": 13.034924800363322,
      "learning_rate": 6.292027343479668e-10,
      "loss": 1.5284,
      "step": 8763
    },
    {
      "epoch": 2.9870483980913427,
      "grad_norm": 19.346684337656228,
      "learning_rate": 5.981364687629443e-10,
      "loss": 1.8468,
      "step": 8764
    },
    {
      "epoch": 2.987389229720518,
      "grad_norm": 18.686691550434507,
      "learning_rate": 5.678566289918363e-10,
      "loss": 1.4882,
      "step": 8765
    },
    {
      "epoch": 2.9877300613496933,
      "grad_norm": 14.46547381950829,
      "learning_rate": 5.383632197991651e-10,
      "loss": 1.2136,
      "step": 8766
    },
    {
      "epoch": 2.988070892978868,
      "grad_norm": 20.568440299912247,
      "learning_rate": 5.096562458234422e-10,
      "loss": 1.2858,
      "step": 8767
    },
    {
      "epoch": 2.9884117246080435,
      "grad_norm": 22.78960819418655,
      "learning_rate": 4.817357115804999e-10,
      "loss": 1.6174,
      "step": 8768
    },
    {
      "epoch": 2.988752556237219,
      "grad_norm": 15.796673862941324,
      "learning_rate": 4.546016214623805e-10,
      "loss": 1.1207,
      "step": 8769
    },
    {
      "epoch": 2.989093387866394,
      "grad_norm": 21.356990194560566,
      "learning_rate": 4.282539797373364e-10,
      "loss": 1.1166,
      "step": 8770
    },
    {
      "epoch": 2.9894342194955694,
      "grad_norm": 13.33039359521358,
      "learning_rate": 4.0269279055038525e-10,
      "loss": 1.1683,
      "step": 8771
    },
    {
      "epoch": 2.9897750511247443,
      "grad_norm": 44.765580566339956,
      "learning_rate": 3.779180579221997e-10,
      "loss": 1.6973,
      "step": 8772
    },
    {
      "epoch": 2.9901158827539196,
      "grad_norm": 15.165777421899378,
      "learning_rate": 3.5392978574966263e-10,
      "loss": 1.3061,
      "step": 8773
    },
    {
      "epoch": 2.990456714383095,
      "grad_norm": 15.143257472111436,
      "learning_rate": 3.3072797780642207e-10,
      "loss": 1.0042,
      "step": 8774
    },
    {
      "epoch": 2.9907975460122698,
      "grad_norm": 13.643589023309422,
      "learning_rate": 3.083126377417811e-10,
      "loss": 0.9342,
      "step": 8775
    },
    {
      "epoch": 2.991138377641445,
      "grad_norm": 15.775919554617062,
      "learning_rate": 2.866837690829183e-10,
      "loss": 1.2619,
      "step": 8776
    },
    {
      "epoch": 2.9914792092706204,
      "grad_norm": 16.52639959580398,
      "learning_rate": 2.6584137523100183e-10,
      "loss": 1.0829,
      "step": 8777
    },
    {
      "epoch": 2.9918200408997953,
      "grad_norm": 15.932087749396446,
      "learning_rate": 2.4578545946507546e-10,
      "loss": 1.7002,
      "step": 8778
    },
    {
      "epoch": 2.9921608725289706,
      "grad_norm": 14.072840619944818,
      "learning_rate": 2.2651602493983792e-10,
      "loss": 1.1174,
      "step": 8779
    },
    {
      "epoch": 2.992501704158146,
      "grad_norm": 13.730307105529477,
      "learning_rate": 2.0803307468675317e-10,
      "loss": 1.0698,
      "step": 8780
    },
    {
      "epoch": 2.992842535787321,
      "grad_norm": 18.684302368506653,
      "learning_rate": 1.9033661161349525e-10,
      "loss": 1.5043,
      "step": 8781
    },
    {
      "epoch": 2.9931833674164965,
      "grad_norm": 15.322318548911477,
      "learning_rate": 1.7342663850283826e-10,
      "loss": 1.1541,
      "step": 8782
    },
    {
      "epoch": 2.9935241990456714,
      "grad_norm": 13.545635350404503,
      "learning_rate": 1.5730315801543161e-10,
      "loss": 1.4986,
      "step": 8783
    },
    {
      "epoch": 2.9938650306748467,
      "grad_norm": 11.664291990762784,
      "learning_rate": 1.4196617268757984e-10,
      "loss": 1.2646,
      "step": 8784
    },
    {
      "epoch": 2.994205862304022,
      "grad_norm": 12.164670326259502,
      "learning_rate": 1.2741568493124247e-10,
      "loss": 1.2813,
      "step": 8785
    },
    {
      "epoch": 2.994546693933197,
      "grad_norm": 9.442762887153371,
      "learning_rate": 1.1365169703625445e-10,
      "loss": 0.9142,
      "step": 8786
    },
    {
      "epoch": 2.994887525562372,
      "grad_norm": 12.710259895980363,
      "learning_rate": 1.0067421116755071e-10,
      "loss": 1.148,
      "step": 8787
    },
    {
      "epoch": 2.9952283571915475,
      "grad_norm": 12.352054924200603,
      "learning_rate": 8.848322936572118e-11,
      "loss": 1.1761,
      "step": 8788
    },
    {
      "epoch": 2.9955691888207223,
      "grad_norm": 24.70548176979912,
      "learning_rate": 7.707875354867612e-11,
      "loss": 1.2595,
      "step": 8789
    },
    {
      "epoch": 2.9959100204498976,
      "grad_norm": 13.900626520482637,
      "learning_rate": 6.646078551109103e-11,
      "loss": 1.3354,
      "step": 8790
    },
    {
      "epoch": 2.996250852079073,
      "grad_norm": 26.336106030519122,
      "learning_rate": 5.662932692329648e-11,
      "loss": 0.9432,
      "step": 8791
    },
    {
      "epoch": 2.9965916837082482,
      "grad_norm": 12.595606589872276,
      "learning_rate": 4.7584379330722906e-11,
      "loss": 1.5888,
      "step": 8792
    },
    {
      "epoch": 2.9969325153374236,
      "grad_norm": 15.768712374316397,
      "learning_rate": 3.9325944156676234e-11,
      "loss": 1.7218,
      "step": 8793
    },
    {
      "epoch": 2.9972733469665984,
      "grad_norm": 15.457723230951412,
      "learning_rate": 3.18540227006725e-11,
      "loss": 1.1009,
      "step": 8794
    },
    {
      "epoch": 2.9976141785957737,
      "grad_norm": 16.30660002086329,
      "learning_rate": 2.5168616137882796e-11,
      "loss": 1.7677,
      "step": 8795
    },
    {
      "epoch": 2.997955010224949,
      "grad_norm": 17.234665224552728,
      "learning_rate": 1.926972551968831e-11,
      "loss": 1.2316,
      "step": 8796
    },
    {
      "epoch": 2.998295841854124,
      "grad_norm": 15.877628755801904,
      "learning_rate": 1.4157351773680383e-11,
      "loss": 1.4115,
      "step": 8797
    },
    {
      "epoch": 2.998636673483299,
      "grad_norm": 17.182786530807526,
      "learning_rate": 9.83149570477071e-12,
      "loss": 1.6813,
      "step": 8798
    },
    {
      "epoch": 2.9989775051124745,
      "grad_norm": 17.82533384489834,
      "learning_rate": 6.292157993526005e-12,
      "loss": 1.7146,
      "step": 8799
    },
    {
      "epoch": 2.9993183367416494,
      "grad_norm": 18.57568378832831,
      "learning_rate": 3.539339196168001e-12,
      "loss": 2.1321,
      "step": 8800
    },
    {
      "epoch": 2.9996591683708247,
      "grad_norm": 16.289174318712707,
      "learning_rate": 1.5730397456836799e-12,
      "loss": 1.4318,
      "step": 8801
    },
    {
      "epoch": 3.0,
      "grad_norm": 22.72764068673402,
      "learning_rate": 3.9325995182526445e-13,
      "loss": 1.3212,
      "step": 8802
    },
    {
      "epoch": 3.0,
      "step": 8802,
      "total_flos": 7592842027008.0,
      "train_loss": 2.352717310000593,
      "train_runtime": 5422.7235,
      "train_samples_per_second": 12.981,
      "train_steps_per_second": 1.623
    }
  ],
  "logging_steps": 1,
  "max_steps": 8802,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7592842027008.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}