diff --git "a/checkpoint-852/trainer_state.json" "b/checkpoint-852/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-852/trainer_state.json"
@@ -0,0 +1,6049 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9694835680751175,
+  "eval_steps": 107,
+  "global_step": 852,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.8741932511329651,
+      "learning_rate": 1e-05,
+      "loss": 1.2567,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "eval_loss": 1.3469510078430176,
+      "eval_runtime": 5.3003,
+      "eval_samples_per_second": 18.867,
+      "eval_steps_per_second": 18.867,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "grad_norm": 0.7848138809204102,
+      "learning_rate": 2e-05,
+      "loss": 1.3328,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.0692198276519775,
+      "learning_rate": 3e-05,
+      "loss": 1.6569,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.4229260683059692,
+      "learning_rate": 4e-05,
+      "loss": 1.5493,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 0.6837282180786133,
+      "learning_rate": 5e-05,
+      "loss": 1.4342,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 1.1194649934768677,
+      "learning_rate": 6e-05,
+      "loss": 1.2668,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.954695463180542,
+      "learning_rate": 7e-05,
+      "loss": 1.4721,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8204368352890015,
+      "learning_rate": 8e-05,
+      "loss": 1.4054,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.8432589769363403,
+      "learning_rate": 9e-05,
+      "loss": 1.2849,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.7272113561630249,
+      "learning_rate": 0.0001,
+      "loss": 1.1438,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8135901093482971,
+      "learning_rate": 0.00011000000000000002,
+      "loss": 1.4665,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 0.8113470673561096,
+      "learning_rate": 0.00012,
+      "loss": 0.9749,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.3105953931808472,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 1.5133,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.0281662940979004,
+      "learning_rate": 0.00014,
+      "loss": 1.1509,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.1005992889404297,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 1.2748,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.453407645225525,
+      "learning_rate": 0.00016,
+      "loss": 1.3313,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0521585941314697,
+      "learning_rate": 0.00017,
+      "loss": 1.2977,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 1.0728052854537964,
+      "learning_rate": 0.00018,
+      "loss": 1.322,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 0.8539354205131531,
+      "learning_rate": 0.00019,
+      "loss": 1.2053,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.6902265548706055,
+      "learning_rate": 0.0002,
+      "loss": 1.1924,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.3770757913589478,
+      "learning_rate": 0.00019999928710990412,
+      "loss": 1.2286,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.358694314956665,
+      "learning_rate": 0.00019999714844978078,
+      "loss": 1.3196,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.4592664241790771,
+      "learning_rate": 0.0001999935840501225,
+      "loss": 1.5923,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.122499942779541,
+      "learning_rate": 0.0001999885939617498,
+      "loss": 1.3107,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0040143728256226,
+      "learning_rate": 0.0001999821782558104,
+      "loss": 1.316,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.0741703510284424,
+      "learning_rate": 0.00019997433702377817,
+      "loss": 1.2527,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 0.9978523850440979,
+      "learning_rate": 0.00019996507037745183,
+      "loss": 1.1989,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.002176284790039,
+      "learning_rate": 0.00019995437844895334,
+      "loss": 1.0503,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.0028811693191528,
+      "learning_rate": 0.0001999422613907262,
+      "loss": 1.356,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 1.3226864337921143,
+      "learning_rate": 0.0001999287193755329,
+      "loss": 1.2571,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.905536949634552,
+      "learning_rate": 0.00019991375259645293,
+      "loss": 1.1576,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.2652262449264526,
+      "learning_rate": 0.00019989736126687963,
+      "loss": 1.1792,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0161558389663696,
+      "learning_rate": 0.00019987954562051725,
+      "loss": 1.0074,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.9903879761695862,
+      "learning_rate": 0.00019986030591137783,
+      "loss": 1.196,
+      "step": 34
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.3282291889190674,
+      "learning_rate": 0.0001998396424137773,
+      "loss": 1.5831,
+      "step": 35
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 1.0442458391189575,
+      "learning_rate": 0.00019981755542233177,
+      "loss": 0.9708,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9892172813415527,
+      "learning_rate": 0.0001997940452519531,
+      "loss": 1.1873,
+      "step": 37
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.6204959154129028,
+      "learning_rate": 0.0001997691122378447,
+      "loss": 1.725,
+      "step": 38
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.9563732147216797,
+      "learning_rate": 0.00019974275673549654,
+      "loss": 1.513,
+      "step": 39
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 1.225378155708313,
+      "learning_rate": 0.00019971497912068013,
+      "loss": 1.1298,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8867512941360474,
+      "learning_rate": 0.00019968577978944323,
+      "loss": 1.2413,
+      "step": 41
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9609911441802979,
+      "learning_rate": 0.0001996551591581041,
+      "loss": 1.2217,
+      "step": 42
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.8636968731880188,
+      "learning_rate": 0.0001996231176632456,
+      "loss": 1.5268,
+      "step": 43
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 0.9208036065101624,
+      "learning_rate": 0.00019958965576170908,
+      "loss": 1.1154,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8316241502761841,
+      "learning_rate": 0.00019955477393058773,
+      "loss": 1.0891,
+      "step": 45
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.0851703882217407,
+      "learning_rate": 0.0001995184726672197,
+      "loss": 1.237,
+      "step": 46
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 1.2032215595245361,
+      "learning_rate": 0.00019948075248918124,
+      "loss": 1.3444,
+      "step": 47
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.8556870818138123,
+      "learning_rate": 0.00019944161393427922,
+      "loss": 0.9632,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.038156270980835,
+      "learning_rate": 0.00019940105756054337,
+      "loss": 1.3897,
+      "step": 49
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1046299934387207,
+      "learning_rate": 0.00019935908394621844,
+      "loss": 1.0211,
+      "step": 50
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1632884740829468,
+      "learning_rate": 0.00019931569368975588,
+      "loss": 1.3098,
+      "step": 51
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.9915144443511963,
+      "learning_rate": 0.0001992708874098054,
+      "loss": 1.1708,
+      "step": 52
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 1.1974948644638062,
+      "learning_rate": 0.00019922466574520608,
+      "loss": 1.248,
+      "step": 53
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.123138189315796,
+      "learning_rate": 0.00019917702935497725,
+      "loss": 0.8789,
+      "step": 54
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 0.7720706462860107,
+      "learning_rate": 0.00019912797891830908,
+      "loss": 1.2278,
+      "step": 55
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.171324610710144,
+      "learning_rate": 0.00019907751513455302,
+      "loss": 1.419,
+      "step": 56
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 1.1821049451828003,
+      "learning_rate": 0.00019902563872321172,
+      "loss": 1.3281,
+      "step": 57
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8871707320213318,
+      "learning_rate": 0.00019897235042392873,
+      "loss": 1.1431,
+      "step": 58
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9106453657150269,
+      "learning_rate": 0.0001989176509964781,
+      "loss": 1.0687,
+      "step": 59
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.8794414401054382,
+      "learning_rate": 0.00019886154122075343,
+      "loss": 1.3627,
+      "step": 60
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.9249592423439026,
+      "learning_rate": 0.00019880402189675678,
+      "loss": 0.5962,
+      "step": 61
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.8355573415756226,
+      "learning_rate": 0.00019874509384458725,
+      "loss": 1.2576,
+      "step": 62
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.3829615116119385,
+      "learning_rate": 0.0001986847579044294,
+      "loss": 1.3844,
+      "step": 63
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.7890944480895996,
+      "learning_rate": 0.00019862301493654108,
+      "loss": 0.9504,
+      "step": 64
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 0.9988391995429993,
+      "learning_rate": 0.00019855986582124126,
+      "loss": 1.1127,
+      "step": 65
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 1.024064302444458,
+      "learning_rate": 0.00019849531145889758,
+      "loss": 1.0418,
+      "step": 66
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 1.1652387380599976,
+      "learning_rate": 0.0001984293527699133,
+      "loss": 1.0524,
+      "step": 67
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.8396044969558716,
+      "learning_rate": 0.00019836199069471437,
+      "loss": 1.1829,
+      "step": 68
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.9767330884933472,
+      "learning_rate": 0.00019829322619373588,
+      "loss": 1.1652,
+      "step": 69
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.7999225854873657,
+      "learning_rate": 0.00019822306024740852,
+      "loss": 1.1075,
+      "step": 70
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9757010340690613,
+      "learning_rate": 0.00019815149385614444,
+      "loss": 1.3074,
+      "step": 71
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.9027532339096069,
+      "learning_rate": 0.00019807852804032305,
+      "loss": 1.308,
+      "step": 72
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 1.5158909559249878,
+      "learning_rate": 0.0001980041638402765,
+      "loss": 1.1594,
+      "step": 73
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.8618407249450684,
+      "learning_rate": 0.00019792840231627482,
+      "loss": 0.7077,
+      "step": 74
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0307233333587646,
+      "learning_rate": 0.00019785124454851084,
+      "loss": 1.0812,
+      "step": 75
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.8295727968215942,
+      "learning_rate": 0.00019777269163708468,
+      "loss": 1.0412,
+      "step": 76
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0772101879119873,
+      "learning_rate": 0.00019769274470198827,
+      "loss": 1.1484,
+      "step": 77
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 1.0130198001861572,
+      "learning_rate": 0.0001976114048830891,
+      "loss": 1.0704,
+      "step": 78
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.6028400659561157,
+      "learning_rate": 0.00019752867334011423,
+      "loss": 0.6347,
+      "step": 79
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8858622312545776,
+      "learning_rate": 0.0001974445512526336,
+      "loss": 1.4117,
+      "step": 80
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.9046239852905273,
+      "learning_rate": 0.00019735903982004324,
+      "loss": 1.0355,
+      "step": 81
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.987149715423584,
+      "learning_rate": 0.00019727214026154827,
+      "loss": 1.117,
+      "step": 82
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 1.066720962524414,
+      "learning_rate": 0.0001971838538161454,
+      "loss": 1.4394,
+      "step": 83
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.7880568504333496,
+      "learning_rate": 0.0001970941817426052,
+      "loss": 1.0591,
+      "step": 84
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8709908723831177,
+      "learning_rate": 0.00019700312531945442,
+      "loss": 0.8653,
+      "step": 85
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.8909516930580139,
+      "learning_rate": 0.00019691068584495742,
+      "loss": 1.0539,
+      "step": 86
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.726899266242981,
+      "learning_rate": 0.000196816864637098,
+      "loss": 1.0426,
+      "step": 87
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.9136834144592285,
+      "learning_rate": 0.00019672166303356028,
+      "loss": 1.0121,
+      "step": 88
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8233316540718079,
+      "learning_rate": 0.0001966250823917099,
+      "loss": 1.1974,
+      "step": 89
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.8916105628013611,
+      "learning_rate": 0.0001965271240885745,
+      "loss": 1.2969,
+      "step": 90
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.7402650713920593,
+      "learning_rate": 0.00019642778952082426,
+      "loss": 1.0409,
+      "step": 91
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.036402702331543,
+      "learning_rate": 0.00019632708010475165,
+      "loss": 1.1482,
+      "step": 92
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 1.50504469871521,
+      "learning_rate": 0.00019622499727625162,
+      "loss": 1.3788,
+      "step": 93
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.7590145468711853,
+      "learning_rate": 0.0001961215424908009,
+      "loss": 1.0139,
+      "step": 94
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.8763208389282227,
+      "learning_rate": 0.00019601671722343738,
+      "loss": 1.066,
+      "step": 95
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.0625308752059937,
+      "learning_rate": 0.00019591052296873888,
+      "loss": 1.1032,
+      "step": 96
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.7704355716705322,
+      "learning_rate": 0.00019580296124080212,
+      "loss": 0.8728,
+      "step": 97
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8164972066879272,
+      "learning_rate": 0.0001956940335732209,
+      "loss": 0.8879,
+      "step": 98
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.114343285560608,
+      "learning_rate": 0.0001955837415190643,
+      "loss": 1.3132,
+      "step": 99
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 0.8190547227859497,
+      "learning_rate": 0.00019547208665085457,
+      "loss": 1.324,
+      "step": 100
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.0214807987213135,
+      "learning_rate": 0.00019535907056054475,
+      "loss": 1.0961,
+      "step": 101
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.0679683685302734,
+      "learning_rate": 0.00019524469485949583,
+      "loss": 1.5574,
+      "step": 102
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 0.7177500128746033,
+      "learning_rate": 0.00019512896117845392,
+      "loss": 1.1553,
+      "step": 103
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 1.2570964097976685,
+      "learning_rate": 0.00019501187116752693,
+      "loss": 1.2929,
+      "step": 104
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8858779668807983,
+      "learning_rate": 0.000194893426496161,
+      "loss": 1.2218,
+      "step": 105
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.9762778282165527,
+      "learning_rate": 0.00019477362885311682,
+      "loss": 0.9456,
+      "step": 106
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8851660490036011,
+      "learning_rate": 0.00019465247994644545,
+      "loss": 1.0032,
+      "step": 107
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 1.1362618207931519,
+      "eval_runtime": 5.3254,
+      "eval_samples_per_second": 18.778,
+      "eval_steps_per_second": 18.778,
+      "step": 107
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.8778185844421387,
+      "learning_rate": 0.00019452998150346401,
+      "loss": 1.1764,
+      "step": 108
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8865760564804077,
+      "learning_rate": 0.00019440613527073105,
+      "loss": 1.0644,
+      "step": 109
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.8576034903526306,
+      "learning_rate": 0.00019428094301402162,
+      "loss": 1.0943,
+      "step": 110
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.810203492641449,
+      "learning_rate": 0.00019415440651830208,
+      "loss": 1.2353,
+      "step": 111
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 1.1911653280258179,
+      "learning_rate": 0.00019402652758770475,
+      "loss": 1.215,
+      "step": 112
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.5166463851928711,
+      "learning_rate": 0.00019389730804550211,
+      "loss": 0.4519,
+      "step": 113
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.983464241027832,
+      "learning_rate": 0.00019376674973408075,
+      "loss": 1.3189,
+      "step": 114
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.7697935700416565,
+      "learning_rate": 0.00019363485451491524,
+      "loss": 1.0372,
+      "step": 115
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.8599738478660583,
+      "learning_rate": 0.0001935016242685415,
+      "loss": 1.1381,
+      "step": 116
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 1.0430530309677124,
+      "learning_rate": 0.00019336706089452996,
+      "loss": 1.2332,
+      "step": 117
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.163251280784607,
+      "learning_rate": 0.0001932311663114586,
+      "loss": 1.1588,
+      "step": 118
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8196917176246643,
+      "learning_rate": 0.0001930939424568854,
+      "loss": 1.1114,
+      "step": 119
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.8848841786384583,
+      "learning_rate": 0.00019295539128732093,
+      "loss": 1.0458,
+      "step": 120
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.9288797974586487,
+      "learning_rate": 0.00019281551477820036,
+      "loss": 0.7388,
+      "step": 121
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8119339942932129,
+      "learning_rate": 0.00019267431492385521,
+      "loss": 1.1095,
+      "step": 122
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9211128354072571,
+      "learning_rate": 0.00019253179373748504,
+      "loss": 1.3183,
+      "step": 123
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9682132601737976,
+      "learning_rate": 0.0001923879532511287,
+      "loss": 1.3786,
+      "step": 124
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.9330196976661682,
+      "learning_rate": 0.00019224279551563532,
+      "loss": 1.4051,
+      "step": 125
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.7206214070320129,
+      "learning_rate": 0.0001920963226006352,
+      "loss": 0.8213,
+      "step": 126
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.6096452474594116,
+      "learning_rate": 0.0001919485365945101,
+      "loss": 0.5982,
+      "step": 127
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 1.0045711994171143,
+      "learning_rate": 0.00019179943960436358,
+      "loss": 1.1399,
+      "step": 128
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 0.8744232058525085,
+      "learning_rate": 0.00019164903375599112,
+      "loss": 1.0176,
+      "step": 129
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8323287963867188,
+      "learning_rate": 0.00019149732119384943,
+      "loss": 1.2937,
+      "step": 130
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.9033424854278564,
+      "learning_rate": 0.00019134430408102615,
+      "loss": 1.3108,
+      "step": 131
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.7392603158950806,
+      "learning_rate": 0.00019118998459920902,
+      "loss": 1.0297,
+      "step": 132
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 1.136338710784912,
+      "learning_rate": 0.0001910343649486546,
+      "loss": 1.3037,
+      "step": 133
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.8202427625656128,
+      "learning_rate": 0.00019087744734815708,
+      "loss": 0.9239,
+      "step": 134
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.039093255996704,
+      "learning_rate": 0.0001907192340350165,
+      "loss": 1.2387,
+      "step": 135
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.1009947061538696,
+      "learning_rate": 0.00019055972726500695,
+      "loss": 1.0234,
+      "step": 136
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 1.057837724685669,
+      "learning_rate": 0.00019039892931234435,
+      "loss": 1.4308,
+      "step": 137
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.8715041279792786,
+      "learning_rate": 0.00019023684246965406,
+      "loss": 1.0642,
+      "step": 138
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.3314340114593506,
+      "learning_rate": 0.00019007346904793818,
+      "loss": 1.1097,
+      "step": 139
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.6808525919914246,
+      "learning_rate": 0.00018990881137654258,
+      "loss": 0.9254,
+      "step": 140
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.8727401494979858,
+      "learning_rate": 0.00018974287180312377,
+      "loss": 1.1062,
+      "step": 141
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 1.0826424360275269,
+      "learning_rate": 0.00018957565269361531,
+      "loss": 1.1528,
+      "step": 142
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8825279474258423,
+      "learning_rate": 0.00018940715643219407,
+      "loss": 1.2208,
+      "step": 143
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 0.8955380320549011,
+      "learning_rate": 0.00018923738542124644,
+      "loss": 1.0918,
+      "step": 144
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.0863996744155884,
+      "learning_rate": 0.00018906634208133385,
+      "loss": 1.2153,
+      "step": 145
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 1.0089964866638184,
+      "learning_rate": 0.00018889402885115833,
+      "loss": 1.0796,
+      "step": 146
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9210363626480103,
+      "learning_rate": 0.0001887204481875278,
+      "loss": 0.8502,
+      "step": 147
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.9592724442481995,
+      "learning_rate": 0.000188545602565321,
+      "loss": 1.4313,
+      "step": 148
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.2299224138259888,
+      "learning_rate": 0.00018836949447745215,
+      "loss": 1.1074,
+      "step": 149
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1486583948135376,
+      "learning_rate": 0.0001881921264348355,
+      "loss": 1.3576,
+      "step": 150
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 1.1912083625793457,
+      "learning_rate": 0.00018801350096634946,
+      "loss": 1.343,
+      "step": 151
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8830162882804871,
+      "learning_rate": 0.00018783362061880062,
+      "loss": 1.6139,
+      "step": 152
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.0919363498687744,
+      "learning_rate": 0.00018765248795688726,
+      "loss": 1.4051,
+      "step": 153
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.8009780049324036,
+      "learning_rate": 0.00018747010556316305,
+      "loss": 1.4095,
+      "step": 154
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.964438796043396,
+      "learning_rate": 0.00018728647603800003,
+      "loss": 1.1634,
+      "step": 155
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.9883137941360474,
+      "learning_rate": 0.00018710160199955156,
+      "loss": 1.1904,
+      "step": 156
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.8936368227005005,
+      "learning_rate": 0.0001869154860837151,
+      "loss": 1.2264,
+      "step": 157
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.6435540914535522,
+      "learning_rate": 0.0001867281309440945,
+      "loss": 0.8426,
+      "step": 158
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.7036202549934387,
+      "learning_rate": 0.00018653953925196225,
+      "loss": 0.8162,
+      "step": 159
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.7669593095779419,
+      "learning_rate": 0.0001863497136962213,
+      "loss": 0.9143,
+      "step": 160
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8047689199447632,
+      "learning_rate": 0.00018615865698336684,
+      "loss": 1.1911,
+      "step": 161
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 1.1324609518051147,
+      "learning_rate": 0.00018596637183744763,
+      "loss": 1.064,
+      "step": 162
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8058610558509827,
+      "learning_rate": 0.00018577286100002723,
+      "loss": 0.7428,
+      "step": 163
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.8588782548904419,
+      "learning_rate": 0.00018557812723014476,
+      "loss": 0.8801,
+      "step": 164
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.0751081705093384,
+      "learning_rate": 0.00018538217330427582,
+      "loss": 1.3674,
+      "step": 165
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.791789174079895,
+      "learning_rate": 0.00018518500201629258,
+      "loss": 1.0103,
+      "step": 166
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.8154894709587097,
+      "learning_rate": 0.00018498661617742426,
+      "loss": 1.1219,
+      "step": 167
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 1.3946795463562012,
+      "learning_rate": 0.00018478701861621686,
+      "loss": 1.0725,
+      "step": 168
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9187031388282776,
+      "learning_rate": 0.00018458621217849286,
+      "loss": 1.2674,
+      "step": 169
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.9884739518165588,
+      "learning_rate": 0.00018438419972731067,
+      "loss": 1.3507,
+      "step": 170
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.4417808055877686,
+      "learning_rate": 0.0001841809841429238,
+      "loss": 1.129,
+      "step": 171
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.0408543348312378,
+      "learning_rate": 0.0001839765683227398,
+      "loss": 1.2038,
+      "step": 172
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.2746024131774902,
+      "learning_rate": 0.00018377095518127897,
+      "loss": 1.2916,
+      "step": 173
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.996474027633667,
+      "learning_rate": 0.00018356414765013267,
+      "loss": 1.3041,
+      "step": 174
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.9435645341873169,
+      "learning_rate": 0.00018335614867792183,
+      "loss": 1.3457,
+      "step": 175
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 1.0813288688659668,
+      "learning_rate": 0.00018314696123025454,
+      "loss": 1.2323,
+      "step": 176
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.1004307270050049,
+      "learning_rate": 0.00018293658828968397,
+      "loss": 1.3084,
+      "step": 177
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.7818973660469055,
+      "learning_rate": 0.00018272503285566587,
+      "loss": 0.88,
+      "step": 178
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8897396326065063,
+      "learning_rate": 0.00018251229794451567,
+      "loss": 1.0124,
+      "step": 179
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 1.0572400093078613,
+      "learning_rate": 0.00018229838658936564,
+      "loss": 1.4603,
+      "step": 180
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.8974701166152954,
+      "learning_rate": 0.0001820833018401215,
+      "loss": 1.1961,
+      "step": 181
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8860751390457153,
+      "learning_rate": 0.00018186704676341898,
+      "loss": 0.9779,
+      "step": 182
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.7719995975494385,
+      "learning_rate": 0.00018164962444258014,
+      "loss": 1.1156,
+      "step": 183
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 1.0823787450790405,
+      "learning_rate": 0.0001814310379775694,
+      "loss": 0.959,
+      "step": 184
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.9932149052619934,
+      "learning_rate": 0.00018121129048494922,
+      "loss": 1.358,
+      "step": 185
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.7363601326942444,
+      "learning_rate": 0.00018099038509783582,
+      "loss": 1.2639,
+      "step": 186
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 1.0355418920516968,
+      "learning_rate": 0.0001807683249658545,
+      "loss": 1.2566,
+      "step": 187
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.8746327757835388,
+      "learning_rate": 0.0001805451132550946,
+      "loss": 1.2159,
+      "step": 188
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.9230121374130249,
+      "learning_rate": 0.00018032075314806448,
+      "loss": 1.1717,
+      "step": 189
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8293251991271973,
+      "learning_rate": 0.00018009524784364615,
+      "loss": 1.1053,
+      "step": 190
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.8819063901901245,
+      "learning_rate": 0.00017986860055704953,
+      "loss": 1.4713,
+      "step": 191
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.843971312046051,
+      "learning_rate": 0.00017964081451976672,
+      "loss": 1.175,
+      "step": 192
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 1.1449857950210571,
+      "learning_rate": 0.00017941189297952597,
+      "loss": 1.0097,
+      "step": 193
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.8204749226570129,
+      "learning_rate": 0.0001791818392002452,
+      "loss": 1.0383,
+      "step": 194
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 1.0437519550323486,
+      "learning_rate": 0.00017895065646198567,
+      "loss": 1.4455,
+      "step": 195
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.803774356842041,
+      "learning_rate": 0.00017871834806090501,
+      "loss": 0.9266,
+      "step": 196
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.991162121295929,
+      "learning_rate": 0.00017848491730921046,
+      "loss": 0.9433,
+      "step": 197
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.9293928146362305,
+      "learning_rate": 0.00017825036753511144,
+      "loss": 1.2432,
+      "step": 198
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6806849241256714,
+      "learning_rate": 0.0001780147020827721,
+      "loss": 0.696,
+      "step": 199
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.2757548093795776,
+      "learning_rate": 0.00017777792431226383,
+      "loss": 1.3426,
+      "step": 200
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.8645375370979309,
+      "learning_rate": 0.00017754003759951715,
+      "loss": 1.0345,
+      "step": 201
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 1.0262008905410767,
+      "learning_rate": 0.0001773010453362737,
+      "loss": 0.935,
+      "step": 202
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.225273609161377,
+      "learning_rate": 0.00017706095093003785,
+      "loss": 1.1271,
+      "step": 203
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9024125337600708,
+      "learning_rate": 0.00017681975780402807,
+      "loss": 1.1307,
+      "step": 204
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.9035881161689758,
+      "learning_rate": 0.00017657746939712815,
+      "loss": 1.3217,
+      "step": 205
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 2.1178483963012695,
+      "learning_rate": 0.00017633408916383826,
+      "loss": 1.4955,
+      "step": 206
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.826454222202301,
+      "learning_rate": 0.00017608962057422549,
+      "loss": 1.187,
+      "step": 207
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.8255906701087952,
+      "learning_rate": 0.00017584406711387463,
+      "loss": 1.0733,
+      "step": 208
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.9498797059059143,
+      "learning_rate": 0.0001755974322838382,
+      "loss": 1.3054,
+      "step": 209
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.7860575914382935,
+      "learning_rate": 0.00017534971960058685,
+      "loss": 1.0231,
+      "step": 210
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.905441403388977,
+      "learning_rate": 0.00017510093259595885,
+      "loss": 1.2928,
+      "step": 211
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.0108212232589722,
+      "learning_rate": 0.00017485107481711012,
+      "loss": 0.8964,
+      "step": 212
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 1.036566972732544,
+      "learning_rate": 0.00017460014982646334,
+      "loss": 1.4823,
+      "step": 213
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.8670862913131714,
+      "learning_rate": 0.00017434816120165728,
+      "loss": 1.0994,
+      "step": 214
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 1.1271697282791138,
+      "eval_runtime": 4.9762,
+      "eval_samples_per_second": 20.096,
+      "eval_steps_per_second": 20.096,
+      "step": 214
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.9359840750694275,
+      "learning_rate": 0.00017409511253549593,
+      "loss": 1.1552,
+      "step": 215
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.0469551086425781,
+      "learning_rate": 0.00017384100743589697,
+      "loss": 1.1175,
+      "step": 216
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.8635923266410828,
+      "learning_rate": 0.0001735858495258406,
+      "loss": 0.9823,
+      "step": 217
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 1.1676996946334839,
+      "learning_rate": 0.00017332964244331776,
+      "loss": 1.2903,
+      "step": 218
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.9774354696273804,
+      "learning_rate": 0.00017307238984127832,
+      "loss": 1.1928,
+      "step": 219
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0486751794815063,
+      "learning_rate": 0.00017281409538757883,
+      "loss": 1.039,
+      "step": 220
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0949512720108032,
+      "learning_rate": 0.00017255476276493056,
+      "loss": 1.148,
+      "step": 221
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.8077650666236877,
+      "learning_rate": 0.0001722943956708466,
+      "loss": 1.1376,
+      "step": 222
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.0595574378967285,
+      "learning_rate": 0.00017203299781758943,
+      "loss": 1.0757,
+      "step": 223
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.813774049282074,
+      "learning_rate": 0.00017177057293211784,
+      "loss": 1.2645,
+      "step": 224
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8764140009880066,
+      "learning_rate": 0.0001715071247560339,
+      "loss": 0.9552,
+      "step": 225
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.9136344194412231,
+      "learning_rate": 0.0001712426570455295,
+      "loss": 1.1841,
+      "step": 226
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 0.8742187023162842,
+      "learning_rate": 0.00017097717357133284,
+      "loss": 1.0314,
+      "step": 227
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8309169411659241,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.9842,
+      "step": 228
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8644968867301941,
+      "learning_rate": 0.00017044317448713461,
+      "loss": 1.3819,
+      "step": 229
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8510339260101318,
+      "learning_rate": 0.0001701746664907862,
+      "loss": 1.2385,
+      "step": 230
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.8174643516540527,
+      "learning_rate": 0.00016990515795794334,
+      "loss": 0.9789,
+      "step": 231
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.9340826272964478,
+      "learning_rate": 0.0001696346527312053,
+      "loss": 1.2472,
+      "step": 232
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9614835977554321,
+      "learning_rate": 0.00016936315466738205,
+      "loss": 1.1588,
+      "step": 233
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 1.1464625597000122,
+      "learning_rate": 0.00016909066763743912,
+      "loss": 0.8365,
+      "step": 234
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.8775334358215332,
+      "learning_rate": 0.00016881719552644273,
+      "loss": 1.1143,
+      "step": 235
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 0.9431893825531006,
+      "learning_rate": 0.00016854274223350397,
+      "loss": 1.362,
+      "step": 236
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7953469157218933,
+      "learning_rate": 0.0001682673116717236,
+      "loss": 1.1568,
+      "step": 237
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7517049908638,
+      "learning_rate": 0.00016799090776813597,
+      "loss": 0.9274,
+      "step": 238
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.796934187412262,
+      "learning_rate": 0.00016771353446365318,
+      "loss": 0.8641,
+      "step": 239
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 0.7946231961250305,
+      "learning_rate": 0.00016743519571300888,
+      "loss": 1.0518,
+      "step": 240
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 1.0859878063201904,
+      "learning_rate": 0.00016715589548470185,
+      "loss": 1.1815,
+      "step": 241
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7418296933174133,
+      "learning_rate": 0.00016687563776093941,
+      "loss": 1.0321,
+      "step": 242
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.8161245584487915,
+      "learning_rate": 0.00016659442653758064,
+      "loss": 1.0931,
+      "step": 243
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.7787612080574036,
+      "learning_rate": 0.00016631226582407952,
+      "loss": 1.2239,
+      "step": 244
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.7161651849746704,
+      "learning_rate": 0.00016602915964342757,
+      "loss": 1.1104,
+      "step": 245
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.786612331867218,
+      "learning_rate": 0.00016574511203209667,
+      "loss": 1.2486,
+      "step": 246
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8251045942306519,
+      "learning_rate": 0.00016546012703998138,
+      "loss": 1.2358,
+      "step": 247
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8045319318771362,
+      "learning_rate": 0.00016517420873034123,
+      "loss": 0.8145,
+      "step": 248
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.8730091452598572,
+      "learning_rate": 0.0001648873611797429,
+      "loss": 0.8832,
+      "step": 249
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.9003087878227234,
+      "learning_rate": 0.00016459958847800187,
+      "loss": 1.1149,
+      "step": 250
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.0912461280822754,
+      "learning_rate": 0.00016431089472812444,
+      "loss": 1.0439,
+      "step": 251
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.7999249696731567,
+      "learning_rate": 0.00016402128404624882,
+      "loss": 0.9821,
+      "step": 252
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 1.0122705698013306,
+      "learning_rate": 0.00016373076056158675,
+      "loss": 1.2302,
+      "step": 253
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.6447519659996033,
+      "learning_rate": 0.00016343932841636456,
+      "loss": 0.6079,
+      "step": 254
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7757092118263245,
+      "learning_rate": 0.00016314699176576402,
+      "loss": 1.0092,
+      "step": 255
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.7445678114891052,
+      "learning_rate": 0.00016285375477786322,
+      "loss": 0.684,
+      "step": 256
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.241065263748169,
+      "learning_rate": 0.000162559621633577,
+      "loss": 1.0321,
+      "step": 257
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.1429563760757446,
+      "learning_rate": 0.00016226459652659753,
+      "loss": 1.1635,
+      "step": 258
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.7441573739051819,
+      "learning_rate": 0.0001619686836633343,
+      "loss": 0.9685,
+      "step": 259
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 1.3199349641799927,
+      "learning_rate": 0.00016167188726285434,
+      "loss": 1.3159,
+      "step": 260
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.8082245588302612,
+      "learning_rate": 0.00016137421155682183,
+      "loss": 1.317,
+      "step": 261
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.9036356210708618,
+      "learning_rate": 0.0001610756607894382,
+      "loss": 0.8672,
+      "step": 262
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.9773459434509277,
+      "learning_rate": 0.00016077623921738102,
+      "loss": 1.1405,
+      "step": 263
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.9840144515037537,
+      "learning_rate": 0.00016047595110974376,
+      "loss": 1.4167,
+      "step": 264
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 1.0358545780181885,
+      "learning_rate": 0.0001601748007479748,
+      "loss": 1.196,
+      "step": 265
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 0.7097404599189758,
+      "learning_rate": 0.0001598727924258164,
+      "loss": 0.791,
+      "step": 266
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 1.7330995798110962,
+      "learning_rate": 0.00015956993044924334,
+      "loss": 1.4283,
+      "step": 267
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.7444025278091431,
+      "learning_rate": 0.0001592662191364017,
+      "loss": 0.7525,
+      "step": 268
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9818642139434814,
+      "learning_rate": 0.0001589616628175472,
+      "loss": 1.2417,
+      "step": 269
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.9218468070030212,
+      "learning_rate": 0.00015865626583498355,
+      "loss": 1.1316,
+      "step": 270
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.9644055366516113,
+      "learning_rate": 0.00015835003254300039,
+      "loss": 1.2594,
+      "step": 271
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.8228254914283752,
+      "learning_rate": 0.00015804296730781135,
+      "loss": 1.2481,
+      "step": 272
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.0708152055740356,
+      "learning_rate": 0.00015773507450749172,
+      "loss": 1.107,
+      "step": 273
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 1.1122934818267822,
+      "learning_rate": 0.00015742635853191608,
+      "loss": 0.8714,
+      "step": 274
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8141905665397644,
+      "learning_rate": 0.00015711682378269565,
+      "loss": 0.9943,
+      "step": 275
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8955541253089905,
+      "learning_rate": 0.00015680647467311557,
+      "loss": 1.3176,
+      "step": 276
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 1.1133729219436646,
+      "learning_rate": 0.000156495315628072,
+      "loss": 1.2602,
+      "step": 277
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8733439445495605,
+      "learning_rate": 0.00015618335108400893,
+      "loss": 1.3639,
+      "step": 278
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.8614795804023743,
+      "learning_rate": 0.00015587058548885505,
+      "loss": 1.1905,
+      "step": 279
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8306368589401245,
+      "learning_rate": 0.00015555702330196023,
+      "loss": 1.1978,
+      "step": 280
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8460854887962341,
+      "learning_rate": 0.00015524266899403206,
+      "loss": 0.9872,
+      "step": 281
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8452059626579285,
+      "learning_rate": 0.000154927527047072,
+      "loss": 0.979,
+      "step": 282
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.8805731534957886,
+      "learning_rate": 0.00015461160195431148,
+      "loss": 1.2885,
+      "step": 283
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9095639586448669,
+      "learning_rate": 0.0001542948982201479,
+      "loss": 1.1156,
+      "step": 284
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9862900376319885,
+      "learning_rate": 0.00015397742036008034,
+      "loss": 1.1571,
+      "step": 285
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.9344744086265564,
+      "learning_rate": 0.0001536591729006453,
+      "loss": 1.2204,
+      "step": 286
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 1.0605379343032837,
+      "learning_rate": 0.00015334016037935196,
+      "loss": 1.3048,
+      "step": 287
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9844763278961182,
+      "learning_rate": 0.0001530203873446177,
+      "loss": 1.0035,
+      "step": 288
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.767954409122467,
+      "learning_rate": 0.0001526998583557031,
+      "loss": 0.9023,
+      "step": 289
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.0622146129608154,
+      "learning_rate": 0.000152378577982647,
+      "loss": 1.4837,
+      "step": 290
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.9536969065666199,
+      "learning_rate": 0.0001520565508062013,
+      "loss": 1.0948,
+      "step": 291
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.9654991030693054,
+      "learning_rate": 0.00015173378141776568,
+      "loss": 1.1913,
+      "step": 292
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.8208710551261902,
+      "learning_rate": 0.00015141027441932216,
+      "loss": 1.1435,
+      "step": 293
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.9273961186408997,
+      "learning_rate": 0.0001510860344233695,
+      "loss": 1.0845,
+      "step": 294
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0316227674484253,
+      "learning_rate": 0.00015076106605285724,
+      "loss": 1.4532,
+      "step": 295
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 1.0121437311172485,
+      "learning_rate": 0.00015043537394112007,
+      "loss": 0.8687,
+      "step": 296
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.0713882446289062,
+      "learning_rate": 0.00015010896273181165,
+      "loss": 1.1097,
+      "step": 297
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.8149722814559937,
+      "learning_rate": 0.00014978183707883827,
+      "loss": 0.8682,
+      "step": 298
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.7118079662322998,
+      "learning_rate": 0.00014945400164629278,
+      "loss": 0.9225,
+      "step": 299
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 1.1042624711990356,
+      "learning_rate": 0.00014912546110838775,
+      "loss": 1.4279,
+      "step": 300
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.947619616985321,
+      "learning_rate": 0.00014879622014938915,
+      "loss": 1.0544,
+      "step": 301
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.9065904021263123,
+      "learning_rate": 0.00014846628346354933,
+      "loss": 1.1642,
+      "step": 302
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.9430320262908936,
+      "learning_rate": 0.00014813565575504022,
+      "loss": 1.2182,
+      "step": 303
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 0.8739117980003357,
+      "learning_rate": 0.00014780434173788617,
+      "loss": 1.0176,
+      "step": 304
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.853125274181366,
+      "learning_rate": 0.00014747234613589685,
+      "loss": 1.1827,
+      "step": 305
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 1.6718727350234985,
+      "learning_rate": 0.0001471396736825998,
+      "loss": 1.2665,
+      "step": 306
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.8566248416900635,
+      "learning_rate": 0.00014680632912117286,
+      "loss": 1.2231,
+      "step": 307
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.6841180324554443,
+      "learning_rate": 0.00014647231720437686,
+      "loss": 0.9366,
+      "step": 308
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9140876531600952,
+      "learning_rate": 0.00014613764269448751,
+      "loss": 1.0711,
+      "step": 309
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9394497275352478,
+      "learning_rate": 0.00014580231036322768,
+      "loss": 1.1159,
+      "step": 310
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 1.1066112518310547,
+      "learning_rate": 0.00014546632499169937,
+      "loss": 1.3487,
+      "step": 311
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9925751090049744,
+      "learning_rate": 0.00014512969137031538,
+      "loss": 1.1207,
+      "step": 312
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 0.9642359018325806,
+      "learning_rate": 0.0001447924142987312,
+      "loss": 1.3772,
+      "step": 313
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.6977396607398987,
+      "learning_rate": 0.0001444544985857766,
+      "loss": 0.8517,
+      "step": 314
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.033882737159729,
+      "learning_rate": 0.00014411594904938682,
+      "loss": 1.0644,
+      "step": 315
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.020871877670288,
+      "learning_rate": 0.00014377677051653404,
+      "loss": 1.2026,
+      "step": 316
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 1.059812068939209,
+      "learning_rate": 0.0001434369678231587,
+      "loss": 1.4181,
+      "step": 317
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8130291104316711,
+      "learning_rate": 0.00014309654581410024,
+      "loss": 1.0691,
+      "step": 318
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.8362820148468018,
+      "learning_rate": 0.00014275550934302823,
+      "loss": 1.0053,
+      "step": 319
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.9266586899757385,
+      "learning_rate": 0.0001424138632723731,
+      "loss": 1.1313,
+      "step": 320
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 1.0162605047225952,
+      "learning_rate": 0.00014207161247325691,
+      "loss": 1.3518,
+      "step": 321
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 1.114696741104126,
+      "eval_runtime": 5.1062,
+      "eval_samples_per_second": 19.584,
+      "eval_steps_per_second": 19.584,
+      "step": 321
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0078846216201782,
+      "learning_rate": 0.00014172876182542372,
+      "loss": 1.0446,
+      "step": 322
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.2844680547714233,
+      "learning_rate": 0.00014138531621717018,
+      "loss": 1.4105,
+      "step": 323
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.0380208492279053,
+      "learning_rate": 0.0001410412805452757,
+      "loss": 1.4212,
+      "step": 324
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.8037036061286926,
+      "learning_rate": 0.00014069665971493274,
+      "loss": 0.8392,
+      "step": 325
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9248948693275452,
+      "learning_rate": 0.00014035145863967692,
+      "loss": 1.3121,
+      "step": 326
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8579298853874207,
+      "learning_rate": 0.0001400056822413167,
+      "loss": 1.1128,
+      "step": 327
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 1.0605120658874512,
+      "learning_rate": 0.0001396593354498635,
+      "loss": 1.543,
+      "step": 328
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.9975443482398987,
+      "learning_rate": 0.0001393124232034613,
+      "loss": 1.1178,
+      "step": 329
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 0.8115065693855286,
+      "learning_rate": 0.0001389649504483162,
+      "loss": 1.1937,
+      "step": 330
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.7796252369880676,
+      "learning_rate": 0.00013861692213862584,
+      "loss": 1.1886,
+      "step": 331
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 1.0133821964263916,
+      "learning_rate": 0.000138268343236509,
+      "loss": 1.4973,
+      "step": 332
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.9557147026062012,
+      "learning_rate": 0.00013791921871193457,
+      "loss": 1.4592,
+      "step": 333
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.9763726592063904,
+      "learning_rate": 0.00013756955354265085,
+      "loss": 0.8502,
+      "step": 334
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.8208116888999939,
+      "learning_rate": 0.00013721935271411464,
+      "loss": 1.1601,
+      "step": 335
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.3176727294921875,
+      "learning_rate": 0.0001368686212194199,
+      "loss": 1.1715,
+      "step": 336
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 1.2329626083374023,
+      "learning_rate": 0.00013651736405922686,
+      "loss": 1.3426,
+      "step": 337
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.9947068691253662,
+      "learning_rate": 0.0001361655862416905,
+      "loss": 1.0623,
+      "step": 338
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 1.176267147064209,
+      "learning_rate": 0.00013581329278238927,
+      "loss": 1.1281,
+      "step": 339
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.909443736076355,
+      "learning_rate": 0.00013546048870425356,
+      "loss": 1.2623,
+      "step": 340
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.8919989466667175,
+      "learning_rate": 0.000135107179037494,
+      "loss": 1.2652,
+      "step": 341
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.7781542539596558,
+      "learning_rate": 0.00013475336881952986,
+      "loss": 0.9857,
+      "step": 342
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.9232913851737976,
+      "learning_rate": 0.00013439906309491712,
+      "loss": 1.0923,
+      "step": 343
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.1160950660705566,
+      "learning_rate": 0.0001340442669152766,
+      "loss": 1.3445,
+      "step": 344
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.084597110748291,
+      "learning_rate": 0.000133688985339222,
+      "loss": 1.7647,
+      "step": 345
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.8420549631118774,
+      "learning_rate": 0.0001333332234322876,
+      "loss": 1.1342,
+      "step": 346
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 1.0362187623977661,
+      "learning_rate": 0.0001329769862668563,
+      "loss": 1.0779,
+      "step": 347
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.902492344379425,
+      "learning_rate": 0.00013262027892208694,
+      "loss": 1.1121,
+      "step": 348
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 1.4322317838668823,
+      "learning_rate": 0.0001322631064838422,
+      "loss": 1.5474,
+      "step": 349
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.8751888275146484,
+      "learning_rate": 0.00013190547404461598,
+      "loss": 1.2055,
+      "step": 350
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.9157432913780212,
+      "learning_rate": 0.0001315473867034608,
+      "loss": 1.3176,
+      "step": 351
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7300966382026672,
+      "learning_rate": 0.0001311888495659149,
+      "loss": 0.9548,
+      "step": 352
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.0954256057739258,
+      "learning_rate": 0.0001308298677439299,
+      "loss": 1.1649,
+      "step": 353
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 1.0646469593048096,
+      "learning_rate": 0.00013047044635579747,
+      "loss": 1.3597,
+      "step": 354
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7668378949165344,
+      "learning_rate": 0.00013011059052607656,
+      "loss": 1.1246,
+      "step": 355
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9135538339614868,
+      "learning_rate": 0.00012975030538552032,
+      "loss": 1.0189,
+      "step": 356
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.7841051816940308,
+      "learning_rate": 0.00012938959607100288,
+      "loss": 1.1396,
+      "step": 357
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.9529784321784973,
+      "learning_rate": 0.00012902846772544624,
+      "loss": 1.4681,
+      "step": 358
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.8711650967597961,
+      "learning_rate": 0.00012866692549774682,
+      "loss": 0.9842,
+      "step": 359
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9562662839889526,
+      "learning_rate": 0.00012830497454270205,
+      "loss": 1.3051,
+      "step": 360
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.0756105184555054,
+      "learning_rate": 0.00012794262002093697,
+      "loss": 1.3275,
+      "step": 361
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.7915710806846619,
+      "learning_rate": 0.0001275798670988306,
+      "loss": 1.0035,
+      "step": 362
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 0.9524595737457275,
+      "learning_rate": 0.0001272167209484422,
+      "loss": 1.2083,
+      "step": 363
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 1.4926435947418213,
+      "learning_rate": 0.0001268531867474377,
+      "loss": 1.3218,
+      "step": 364
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.2689683437347412,
+      "learning_rate": 0.00012648926967901567,
+      "loss": 2.7813,
+      "step": 365
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8361314535140991,
+      "learning_rate": 0.00012612497493183364,
+      "loss": 1.124,
+      "step": 366
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 1.2996618747711182,
+      "learning_rate": 0.00012576030769993393,
+      "loss": 1.3745,
+      "step": 367
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 0.8248890042304993,
+      "learning_rate": 0.0001253952731826697,
+      "loss": 1.1971,
+      "step": 368
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.8044300079345703,
+      "learning_rate": 0.00012502987658463075,
+      "loss": 1.1508,
+      "step": 369
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.196742057800293,
+      "learning_rate": 0.00012466412311556952,
+      "loss": 0.9868,
+      "step": 370
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.9415065050125122,
+      "learning_rate": 0.0001242980179903264,
+      "loss": 1.046,
+      "step": 371
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 1.049695611000061,
+      "learning_rate": 0.0001239315664287558,
+      "loss": 0.9927,
+      "step": 372
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.8266507387161255,
+      "learning_rate": 0.00012356477365565148,
+      "loss": 0.8879,
+      "step": 373
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9163070321083069,
+      "learning_rate": 0.0001231976449006721,
+      "loss": 1.1214,
+      "step": 374
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 1.01756751537323,
+      "learning_rate": 0.00012283018539826685,
+      "loss": 1.1644,
+      "step": 375
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.901319682598114,
+      "learning_rate": 0.00012246240038760043,
+      "loss": 1.1985,
+      "step": 376
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.9721381664276123,
+      "learning_rate": 0.00012209429511247864,
+      "loss": 1.1199,
+      "step": 377
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8883329033851624,
+      "learning_rate": 0.0001217258748212737,
+      "loss": 1.3431,
+      "step": 378
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 1.0698317289352417,
+      "learning_rate": 0.00012135714476684903,
+      "loss": 1.3173,
+      "step": 379
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.8664084076881409,
+      "learning_rate": 0.00012098811020648475,
+      "loss": 1.0441,
+      "step": 380
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 0.9194340109825134,
+      "learning_rate": 0.00012061877640180255,
+      "loss": 1.152,
+      "step": 381
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9599464535713196,
+      "learning_rate": 0.00012024914861869063,
+      "loss": 1.1115,
+      "step": 382
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.9990159273147583,
+      "learning_rate": 0.00011987923212722872,
+      "loss": 1.2534,
+      "step": 383
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.8435646891593933,
+      "learning_rate": 0.00011950903220161285,
+      "loss": 1.1752,
+      "step": 384
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 1.0376098155975342,
+      "learning_rate": 0.00011913855412008023,
+      "loss": 1.4716,
+      "step": 385
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.3249186277389526,
+      "learning_rate": 0.00011876780316483401,
+      "loss": 1.211,
+      "step": 386
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 1.378393292427063,
+      "learning_rate": 0.00011839678462196784,
+      "loss": 1.0357,
+      "step": 387
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7574142217636108,
+      "learning_rate": 0.0001180255037813906,
+      "loss": 0.4137,
+      "step": 388
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.7813417911529541,
+      "learning_rate": 0.00011765396593675097,
+      "loss": 1.1776,
+      "step": 389
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.8787057995796204,
+      "learning_rate": 0.00011728217638536197,
+      "loss": 1.1352,
+      "step": 390
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9643175005912781,
+      "learning_rate": 0.00011691014042812536,
+      "loss": 1.3089,
+      "step": 391
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9101107716560364,
+      "learning_rate": 0.00011653786336945614,
+      "loss": 1.0639,
+      "step": 392
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.018091082572937,
+      "learning_rate": 0.00011616535051720685,
+      "loss": 0.9938,
+      "step": 393
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 0.9708930253982544,
+      "learning_rate": 0.00011579260718259197,
+      "loss": 0.8004,
+      "step": 394
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8909386396408081,
+      "learning_rate": 0.00011541963868011212,
+      "loss": 1.2997,
+      "step": 395
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 1.0622750520706177,
+      "learning_rate": 0.00011504645032747832,
+      "loss": 1.0235,
+      "step": 396
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8857365250587463,
+      "learning_rate": 0.00011467304744553618,
+      "loss": 0.8382,
+      "step": 397
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.8980242013931274,
+      "learning_rate": 0.00011429943535819005,
+      "loss": 1.0877,
+      "step": 398
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.1426031589508057,
+      "learning_rate": 0.00011392561939232706,
+      "loss": 1.3496,
+      "step": 399
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.0347543954849243,
+      "learning_rate": 0.0001135516048777412,
+      "loss": 1.6309,
+      "step": 400
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 1.0121687650680542,
+      "learning_rate": 0.00011317739714705731,
+      "loss": 1.2256,
+      "step": 401
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.8863442540168762,
+      "learning_rate": 0.0001128030015356551,
+      "loss": 0.8687,
+      "step": 402
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.7622981667518616,
+      "learning_rate": 0.00011242842338159309,
+      "loss": 0.7564,
+      "step": 403
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.9527961015701294,
+      "learning_rate": 0.0001120536680255323,
+      "loss": 1.0593,
+      "step": 404
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 1.3481955528259277,
+      "learning_rate": 0.00011167874081066045,
+      "loss": 1.2279,
+      "step": 405
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.8665672540664673,
+      "learning_rate": 0.00011130364708261552,
+      "loss": 1.1677,
+      "step": 406
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.217490553855896,
+      "learning_rate": 0.0001109283921894095,
+      "loss": 1.2617,
+      "step": 407
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8935596942901611,
+      "learning_rate": 0.00011055298148135236,
+      "loss": 1.1184,
+      "step": 408
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.8513955473899841,
+      "learning_rate": 0.00011017742031097563,
+      "loss": 1.2705,
+      "step": 409
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.0295133590698242,
+      "learning_rate": 0.0001098017140329561,
+      "loss": 1.1966,
+      "step": 410
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 1.1029167175292969,
+      "learning_rate": 0.0001094258680040394,
+      "loss": 1.4887,
+      "step": 411
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9045723080635071,
+      "learning_rate": 0.0001090498875829638,
+      "loss": 1.1461,
+      "step": 412
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.8317312002182007,
+      "learning_rate": 0.00010867377813038366,
+      "loss": 1.136,
+      "step": 413
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 1.0023647546768188,
+      "learning_rate": 0.00010829754500879308,
+      "loss": 1.1123,
+      "step": 414
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.9197617769241333,
+      "learning_rate": 0.00010792119358244939,
+      "loss": 1.2792,
+      "step": 415
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.9892452955245972,
+      "learning_rate": 0.00010754472921729661,
+      "loss": 1.634,
+      "step": 416
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.8005648255348206,
+      "learning_rate": 0.00010716815728088912,
+      "loss": 0.7168,
+      "step": 417
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 1.1989247798919678,
+      "learning_rate": 0.00010679148314231504,
+      "loss": 1.2882,
+      "step": 418
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.7820172905921936,
+      "learning_rate": 0.00010641471217211958,
+      "loss": 1.1125,
+      "step": 419
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 1.352563500404358,
+      "learning_rate": 0.00010603784974222861,
+      "loss": 0.9641,
+      "step": 420
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8966504335403442,
+      "learning_rate": 0.000105660901225872,
+      "loss": 1.1155,
+      "step": 421
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8722444176673889,
+      "learning_rate": 0.00010528387199750707,
+      "loss": 1.3011,
+      "step": 422
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.8678218722343445,
+      "learning_rate": 0.00010490676743274181,
+      "loss": 1.2912,
+      "step": 423
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8596826791763306,
+      "learning_rate": 0.00010452959290825846,
+      "loss": 1.3792,
+      "step": 424
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.777655303478241,
+      "learning_rate": 0.00010415235380173662,
+      "loss": 0.9992,
+      "step": 425
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7913762331008911,
+      "learning_rate": 0.00010377505549177682,
+      "loss": 0.8813,
+      "step": 426
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8607832789421082,
+      "learning_rate": 0.00010339770335782359,
+      "loss": 0.9927,
+      "step": 427
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.8529496192932129,
+      "learning_rate": 0.0001030203027800889,
+      "loss": 1.1638,
+      "step": 428
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.109578251838684,
+      "eval_runtime": 5.1972,
+      "eval_samples_per_second": 19.241,
+      "eval_steps_per_second": 19.241,
+      "step": 428
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.0168473720550537,
+      "learning_rate": 0.00010264285913947545,
+      "loss": 1.3069,
+      "step": 429
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.2173975706100464,
+      "learning_rate": 0.00010226537781749987,
+      "loss": 0.9985,
+      "step": 430
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.8822383284568787,
+      "learning_rate": 0.00010188786419621612,
+      "loss": 1.1801,
+      "step": 431
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 1.145887851715088,
+      "learning_rate": 0.00010151032365813859,
+      "loss": 1.2034,
+      "step": 432
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.7805179953575134,
+      "learning_rate": 0.00010113276158616553,
+      "loss": 1.2896,
+      "step": 433
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8651528358459473,
+      "learning_rate": 0.00010075518336350218,
+      "loss": 1.2091,
+      "step": 434
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.0866674184799194,
+      "learning_rate": 0.00010037759437358398,
+      "loss": 1.0422,
+      "step": 435
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.0740783214569092,
+      "learning_rate": 0.0001,
+      "loss": 1.3082,
+      "step": 436
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7619555592536926,
+      "learning_rate": 9.962240562641602e-05,
+      "loss": 0.9208,
+      "step": 437
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.1175497770309448,
+      "learning_rate": 9.924481663649785e-05,
+      "loss": 1.1327,
+      "step": 438
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8330385088920593,
+      "learning_rate": 9.886723841383448e-05,
+      "loss": 0.9008,
+      "step": 439
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7999448776245117,
+      "learning_rate": 9.848967634186142e-05,
+      "loss": 0.8613,
+      "step": 440
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.866118311882019,
+      "learning_rate": 9.81121358037839e-05,
+      "loss": 1.2199,
+      "step": 441
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.882002055644989,
+      "learning_rate": 9.773462218250015e-05,
+      "loss": 0.8658,
+      "step": 442
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.8205627799034119,
+      "learning_rate": 9.735714086052458e-05,
+      "loss": 1.189,
+      "step": 443
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.9855633974075317,
+      "learning_rate": 9.697969721991114e-05,
+      "loss": 0.8399,
+      "step": 444
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.891368567943573,
+      "learning_rate": 9.660229664217642e-05,
+      "loss": 1.0189,
+      "step": 445
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.906017541885376,
+      "learning_rate": 9.62249445082232e-05,
+      "loss": 1.0986,
+      "step": 446
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.08309805393219,
+      "learning_rate": 9.584764619826339e-05,
+      "loss": 1.1887,
+      "step": 447
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.8647370934486389,
+      "learning_rate": 9.547040709174159e-05,
+      "loss": 0.8142,
+      "step": 448
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 1.3828812837600708,
+      "learning_rate": 9.509323256725821e-05,
+      "loss": 0.882,
+      "step": 449
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8859632611274719,
+      "learning_rate": 9.471612800249296e-05,
+      "loss": 0.9624,
+      "step": 450
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.7756340503692627,
+      "learning_rate": 9.433909877412802e-05,
+      "loss": 0.8844,
+      "step": 451
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 1.2693322896957397,
+      "learning_rate": 9.396215025777139e-05,
+      "loss": 1.0696,
+      "step": 452
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.8535933494567871,
+      "learning_rate": 9.358528782788045e-05,
+      "loss": 0.8664,
+      "step": 453
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.8840806484222412,
+      "learning_rate": 9.320851685768497e-05,
+      "loss": 0.744,
+      "step": 454
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.1074801683425903,
+      "learning_rate": 9.283184271911089e-05,
+      "loss": 1.0923,
+      "step": 455
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.0910581350326538,
+      "learning_rate": 9.245527078270341e-05,
+      "loss": 0.8355,
+      "step": 456
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.0176016092300415,
+      "learning_rate": 9.207880641755065e-05,
+      "loss": 0.8598,
+      "step": 457
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.9024606347084045,
+      "learning_rate": 9.170245499120693e-05,
+      "loss": 0.8282,
+      "step": 458
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.70665442943573,
+      "learning_rate": 9.132622186961637e-05,
+      "loss": 0.428,
+      "step": 459
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 0.8624319434165955,
+      "learning_rate": 9.095011241703623e-05,
+      "loss": 0.9361,
+      "step": 460
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.5043039321899414,
+      "learning_rate": 9.057413199596065e-05,
+      "loss": 0.806,
+      "step": 461
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 1.1076829433441162,
+      "learning_rate": 9.019828596704394e-05,
+      "loss": 0.8207,
+      "step": 462
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": NaN,
+      "learning_rate": 9.019828596704394e-05,
+      "loss": 0.8544,
+      "step": 463
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.9130039215087891,
+      "learning_rate": 8.982257968902438e-05,
+      "loss": 0.8329,
+      "step": 464
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.1653575897216797,
+      "learning_rate": 8.944701851864767e-05,
+      "loss": 0.8211,
+      "step": 465
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 1.4047077894210815,
+      "learning_rate": 8.907160781059052e-05,
+      "loss": 1.2467,
+      "step": 466
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.0421421527862549,
+      "learning_rate": 8.869635291738452e-05,
+      "loss": 1.0891,
+      "step": 467
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.8077785968780518,
+      "learning_rate": 8.832125918933954e-05,
+      "loss": 0.6497,
+      "step": 468
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 1.1150001287460327,
+      "learning_rate": 8.79463319744677e-05,
+      "loss": 0.8212,
+      "step": 469
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.704976499080658,
+      "learning_rate": 8.757157661840693e-05,
+      "loss": 0.6064,
+      "step": 470
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.9938413500785828,
+      "learning_rate": 8.719699846434492e-05,
+      "loss": 0.8411,
+      "step": 471
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.8569329380989075,
+      "learning_rate": 8.682260285294271e-05,
+      "loss": 0.7593,
+      "step": 472
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.0257930755615234,
+      "learning_rate": 8.644839512225886e-05,
+      "loss": 0.9576,
+      "step": 473
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.0176326036453247,
+      "learning_rate": 8.607438060767296e-05,
+      "loss": 1.0099,
+      "step": 474
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 1.1403366327285767,
+      "learning_rate": 8.570056464180998e-05,
+      "loss": 0.6884,
+      "step": 475
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.9332993626594543,
+      "learning_rate": 8.532695255446383e-05,
+      "loss": 0.8534,
+      "step": 476
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.0836379528045654,
+      "learning_rate": 8.495354967252169e-05,
+      "loss": 0.9814,
+      "step": 477
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 1.038662075996399,
+      "learning_rate": 8.458036131988792e-05,
+      "loss": 0.8299,
+      "step": 478
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.9515346884727478,
+      "learning_rate": 8.420739281740805e-05,
+      "loss": 0.6605,
+      "step": 479
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.1238003969192505,
+      "learning_rate": 8.383464948279319e-05,
+      "loss": 0.9265,
+      "step": 480
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.245451807975769,
+      "learning_rate": 8.346213663054387e-05,
+      "loss": 1.0276,
+      "step": 481
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.0740351676940918,
+      "learning_rate": 8.308985957187466e-05,
+      "loss": 0.894,
+      "step": 482
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 1.1485562324523926,
+      "learning_rate": 8.271782361463805e-05,
+      "loss": 1.029,
+      "step": 483
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.9134268760681152,
+      "learning_rate": 8.234603406324908e-05,
+      "loss": 0.6061,
+      "step": 484
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.0883495807647705,
+      "learning_rate": 8.197449621860943e-05,
+      "loss": 0.3585,
+      "step": 485
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.11465322971344,
+      "learning_rate": 8.16032153780322e-05,
+      "loss": 1.0824,
+      "step": 486
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 1.1381958723068237,
+      "learning_rate": 8.123219683516603e-05,
+      "loss": 1.2363,
+      "step": 487
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.9504216313362122,
+      "learning_rate": 8.08614458799198e-05,
+      "loss": 0.9756,
+      "step": 488
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.6939163208007812,
+      "learning_rate": 8.049096779838719e-05,
+      "loss": 0.3759,
+      "step": 489
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.0341072082519531,
+      "learning_rate": 8.01207678727713e-05,
+      "loss": 0.8908,
+      "step": 490
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.0956295728683472,
+      "learning_rate": 7.975085138130938e-05,
+      "loss": 0.7801,
+      "step": 491
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 1.07564115524292,
+      "learning_rate": 7.938122359819746e-05,
+      "loss": 0.7834,
+      "step": 492
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.0565071105957031,
+      "learning_rate": 7.901188979351526e-05,
+      "loss": 0.8172,
+      "step": 493
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.0685640573501587,
+      "learning_rate": 7.864285523315096e-05,
+      "loss": 0.7835,
+      "step": 494
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 0.9400045275688171,
+      "learning_rate": 7.827412517872634e-05,
+      "loss": 0.6592,
+      "step": 495
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 1.1758517026901245,
+      "learning_rate": 7.790570488752135e-05,
+      "loss": 0.8823,
+      "step": 496
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.1540248394012451,
+      "learning_rate": 7.753759961239964e-05,
+      "loss": 0.849,
+      "step": 497
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.1850864887237549,
+      "learning_rate": 7.716981460173319e-05,
+      "loss": 0.7586,
+      "step": 498
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.033463954925537,
+      "learning_rate": 7.68023550993279e-05,
+      "loss": 0.9891,
+      "step": 499
+    },
+    {
+      "epoch": 1.14,
+      "grad_norm": 1.2690588235855103,
+      "learning_rate": 7.643522634434856e-05,
+      "loss": 1.1184,
+      "step": 500
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.6630959510803223,
+      "learning_rate": 7.606843357124426e-05,
+      "loss": 1.0677,
+      "step": 501
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.0995733737945557,
+      "learning_rate": 7.570198200967362e-05,
+      "loss": 0.9593,
+      "step": 502
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.1102938652038574,
+      "learning_rate": 7.533587688443049e-05,
+      "loss": 0.7075,
+      "step": 503
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.3560442924499512,
+      "learning_rate": 7.497012341536924e-05,
+      "loss": 1.0318,
+      "step": 504
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 1.0489193201065063,
+      "learning_rate": 7.460472681733031e-05,
+      "loss": 0.7033,
+      "step": 505
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.1091972589492798,
+      "learning_rate": 7.423969230006609e-05,
+      "loss": 0.907,
+      "step": 506
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.096968173980713,
+      "learning_rate": 7.387502506816638e-05,
+      "loss": 0.9167,
+      "step": 507
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 1.8477667570114136,
+      "learning_rate": 7.351073032098437e-05,
+      "loss": 0.9794,
+      "step": 508
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.8208603858947754,
+      "learning_rate": 7.314681325256232e-05,
+      "loss": 0.7489,
+      "step": 509
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.3617076873779297,
+      "learning_rate": 7.278327905155783e-05,
+      "loss": 1.0549,
+      "step": 510
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.2204340696334839,
+      "learning_rate": 7.242013290116944e-05,
+      "loss": 1.0512,
+      "step": 511
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.2859915494918823,
+      "learning_rate": 7.205737997906307e-05,
+      "loss": 0.8753,
+      "step": 512
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 1.184820532798767,
+      "learning_rate": 7.169502545729797e-05,
+      "loss": 0.7513,
+      "step": 513
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.4803907871246338,
+      "learning_rate": 7.133307450225322e-05,
+      "loss": 0.9857,
+      "step": 514
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.8424803614616394,
+      "learning_rate": 7.097153227455379e-05,
+      "loss": 0.586,
+      "step": 515
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.2504682540893555,
+      "learning_rate": 7.061040392899712e-05,
+      "loss": 0.7598,
+      "step": 516
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 1.072726845741272,
+      "learning_rate": 7.024969461447972e-05,
+      "loss": 0.725,
+      "step": 517
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.133083701133728,
+      "learning_rate": 6.988940947392344e-05,
+      "loss": 0.943,
+      "step": 518
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.823098063468933,
+      "learning_rate": 6.952955364420255e-05,
+      "loss": 0.8723,
+      "step": 519
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.2997591495513916,
+      "learning_rate": 6.91701322560701e-05,
+      "loss": 1.0831,
+      "step": 520
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.3265862464904785,
+      "learning_rate": 6.881115043408511e-05,
+      "loss": 0.9224,
+      "step": 521
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 1.3687394857406616,
+      "learning_rate": 6.845261329653922e-05,
+      "loss": 1.2177,
+      "step": 522
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.3694826364517212,
+      "learning_rate": 6.809452595538402e-05,
+      "loss": 0.9281,
+      "step": 523
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.8991251587867737,
+      "learning_rate": 6.77368935161578e-05,
+      "loss": 0.7164,
+      "step": 524
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.059921145439148,
+      "learning_rate": 6.73797210779131e-05,
+      "loss": 0.7976,
+      "step": 525
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 1.2316731214523315,
+      "learning_rate": 6.70230137331437e-05,
+      "loss": 0.9325,
+      "step": 526
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.3245116472244263,
+      "learning_rate": 6.666677656771239e-05,
+      "loss": 0.7446,
+      "step": 527
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.056368112564087,
+      "learning_rate": 6.6311014660778e-05,
+      "loss": 0.6862,
+      "step": 528
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.4571599960327148,
+      "learning_rate": 6.595573308472338e-05,
+      "loss": 1.0019,
+      "step": 529
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 1.2216399908065796,
+      "learning_rate": 6.56009369050829e-05,
+      "loss": 0.8784,
+      "step": 530
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.440184473991394,
+      "learning_rate": 6.524663118047016e-05,
+      "loss": 1.7494,
+      "step": 531
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.9794695973396301,
+      "learning_rate": 6.489282096250601e-05,
+      "loss": 0.7664,
+      "step": 532
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.0644843578338623,
+      "learning_rate": 6.453951129574644e-05,
+      "loss": 1.1805,
+      "step": 533
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 1.155930995941162,
+      "learning_rate": 6.418670721761073e-05,
+      "loss": 0.8715,
+      "step": 534
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.2466217279434204,
+      "learning_rate": 6.383441375830951e-05,
+      "loss": 1.1003,
+      "step": 535
+    },
+    {
+      "epoch": 1.23,
+      "eval_loss": 1.1363521814346313,
+      "eval_runtime": 5.0483,
+      "eval_samples_per_second": 19.808,
+      "eval_steps_per_second": 19.808,
+      "step": 535
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.0918858051300049,
+      "learning_rate": 6.34826359407732e-05,
+      "loss": 1.0192,
+      "step": 536
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.0371659994125366,
+      "learning_rate": 6.313137878058013e-05,
+      "loss": 0.8525,
+      "step": 537
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.3210889101028442,
+      "learning_rate": 6.278064728588542e-05,
+      "loss": 0.9921,
+      "step": 538
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 1.1951533555984497,
+      "learning_rate": 6.243044645734917e-05,
+      "loss": 0.8915,
+      "step": 539
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.4079426527023315,
+      "learning_rate": 6.20807812880655e-05,
+      "loss": 1.0806,
+      "step": 540
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.255631685256958,
+      "learning_rate": 6.173165676349103e-05,
+      "loss": 0.7914,
+      "step": 541
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.08389413356781,
+      "learning_rate": 6.138307786137415e-05,
+      "loss": 0.8829,
+      "step": 542
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 1.3811546564102173,
+      "learning_rate": 6.103504955168382e-05,
+      "loss": 1.0245,
+      "step": 543
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.1965594291687012,
+      "learning_rate": 6.068757679653868e-05,
+      "loss": 1.0113,
+      "step": 544
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.3283885717391968,
+      "learning_rate": 6.034066455013649e-05,
+      "loss": 0.8461,
+      "step": 545
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.3063422441482544,
+      "learning_rate": 5.999431775868329e-05,
+      "loss": 0.7606,
+      "step": 546
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 1.1690608263015747,
+      "learning_rate": 5.9648541360323095e-05,
+      "loss": 0.9931,
+      "step": 547
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.7004899978637695,
+      "learning_rate": 5.930334028506725e-05,
+      "loss": 0.2953,
+      "step": 548
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.0900754928588867,
+      "learning_rate": 5.8958719454724346e-05,
+      "loss": 0.8003,
+      "step": 549
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.1421937942504883,
+      "learning_rate": 5.8614683782829835e-05,
+      "loss": 0.8811,
+      "step": 550
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 1.823502540588379,
+      "learning_rate": 5.8271238174576305e-05,
+      "loss": 1.1777,
+      "step": 551
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.483927607536316,
+      "learning_rate": 5.792838752674309e-05,
+      "loss": 0.9416,
+      "step": 552
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.1196835041046143,
+      "learning_rate": 5.75861367276269e-05,
+      "loss": 0.7103,
+      "step": 553
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.209674596786499,
+      "learning_rate": 5.7244490656971815e-05,
+      "loss": 1.1059,
+      "step": 554
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.0897173881530762,
+      "learning_rate": 5.6903454185899774e-05,
+      "loss": 0.8877,
+      "step": 555
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 1.4691230058670044,
+      "learning_rate": 5.6563032176841324e-05,
+      "loss": 0.8161,
+      "step": 556
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.5138698816299438,
+      "learning_rate": 5.622322948346594e-05,
+      "loss": 1.0267,
+      "step": 557
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.6125860214233398,
+      "learning_rate": 5.588405095061322e-05,
+      "loss": 1.1,
+      "step": 558
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.2148243188858032,
+      "learning_rate": 5.55455014142234e-05,
+      "loss": 0.9657,
+      "step": 559
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.327254295349121,
+      "learning_rate": 5.5207585701268805e-05,
+      "loss": 1.0631,
+      "step": 560
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.1820276975631714,
+      "learning_rate": 5.4870308629684677e-05,
+      "loss": 0.6627,
+      "step": 561
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.6765048503875732,
+      "learning_rate": 5.453367500830069e-05,
+      "loss": 1.0243,
+      "step": 562
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 1.2270108461380005,
+      "learning_rate": 5.4197689636772334e-05,
+      "loss": 0.7945,
+      "step": 563
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.985974133014679,
+      "learning_rate": 5.386235730551252e-05,
+      "loss": 0.7504,
+      "step": 564
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.375940203666687,
+      "learning_rate": 5.3527682795623146e-05,
+      "loss": 0.9097,
+      "step": 565
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.0430001020431519,
+      "learning_rate": 5.319367087882716e-05,
+      "loss": 0.9843,
+      "step": 566
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.0604748725891113,
+      "learning_rate": 5.286032631740023e-05,
+      "loss": 0.6792,
+      "step": 567
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 1.05086088180542,
+      "learning_rate": 5.252765386410312e-05,
+      "loss": 0.632,
+      "step": 568
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.214124083518982,
+      "learning_rate": 5.2195658262113814e-05,
+      "loss": 0.7741,
+      "step": 569
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.1835461854934692,
+      "learning_rate": 5.186434424495979e-05,
+      "loss": 0.9085,
+      "step": 570
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.428858757019043,
+      "learning_rate": 5.1533716536450693e-05,
+      "loss": 1.4804,
+      "step": 571
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.3231133222579956,
+      "learning_rate": 5.1203779850610864e-05,
+      "loss": 0.7928,
+      "step": 572
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 1.2435096502304077,
+      "learning_rate": 5.087453889161229e-05,
+      "loss": 0.8173,
+      "step": 573
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.05045485496521,
+      "learning_rate": 5.054599835370724e-05,
+      "loss": 0.9197,
+      "step": 574
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.5020091533660889,
+      "learning_rate": 5.021816292116175e-05,
+      "loss": 1.0374,
+      "step": 575
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.2051324844360352,
+      "learning_rate": 4.989103726818836e-05,
+      "loss": 0.5828,
+      "step": 576
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 1.2995752096176147,
+      "learning_rate": 4.956462605887994e-05,
+      "loss": 0.9762,
+      "step": 577
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.293330430984497,
+      "learning_rate": 4.923893394714279e-05,
+      "loss": 0.851,
+      "step": 578
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.4019485712051392,
+      "learning_rate": 4.891396557663056e-05,
+      "loss": 0.8867,
+      "step": 579
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.1992961168289185,
+      "learning_rate": 4.8589725580677835e-05,
+      "loss": 0.8609,
+      "step": 580
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 1.3640302419662476,
+      "learning_rate": 4.826621858223431e-05,
+      "loss": 0.8686,
+      "step": 581
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.2899017333984375,
+      "learning_rate": 4.794344919379872e-05,
+      "loss": 0.7867,
+      "step": 582
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.468520164489746,
+      "learning_rate": 4.762142201735299e-05,
+      "loss": 0.8161,
+      "step": 583
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.1621512174606323,
+      "learning_rate": 4.730014164429689e-05,
+      "loss": 0.7486,
+      "step": 584
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 1.599510669708252,
+      "learning_rate": 4.697961265538231e-05,
+      "loss": 0.7786,
+      "step": 585
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.5685900449752808,
+      "learning_rate": 4.6659839620648074e-05,
+      "loss": 0.9098,
+      "step": 586
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.2052661180496216,
+      "learning_rate": 4.634082709935473e-05,
+      "loss": 0.845,
+      "step": 587
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.2109522819519043,
+      "learning_rate": 4.6022579639919695e-05,
+      "loss": 0.7274,
+      "step": 588
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.354245662689209,
+      "learning_rate": 4.5705101779852135e-05,
+      "loss": 0.9229,
+      "step": 589
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 1.398025631904602,
+      "learning_rate": 4.5388398045688566e-05,
+      "loss": 0.7834,
+      "step": 590
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.30653977394104,
+      "learning_rate": 4.507247295292801e-05,
+      "loss": 0.9327,
+      "step": 591
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 0.9323519468307495,
+      "learning_rate": 4.475733100596795e-05,
+      "loss": 0.5555,
+      "step": 592
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.2603791952133179,
+      "learning_rate": 4.444297669803981e-05,
+      "loss": 0.8119,
+      "step": 593
+    },
+    {
+      "epoch": 1.36,
+      "grad_norm": 1.2332367897033691,
+      "learning_rate": 4.412941451114498e-05,
+      "loss": 0.8452,
+      "step": 594
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.3796523809432983,
+      "learning_rate": 4.381664891599111e-05,
+      "loss": 1.0362,
+      "step": 595
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.1836915016174316,
+      "learning_rate": 4.3504684371928006e-05,
+      "loss": 0.7377,
+      "step": 596
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.4385707378387451,
+      "learning_rate": 4.3193525326884435e-05,
+      "loss": 0.9733,
+      "step": 597
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 1.4326701164245605,
+      "learning_rate": 4.288317621730434e-05,
+      "loss": 0.8767,
+      "step": 598
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.1379212141036987,
+      "learning_rate": 4.257364146808393e-05,
+      "loss": 0.8179,
+      "step": 599
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.343801498413086,
+      "learning_rate": 4.226492549250829e-05,
+      "loss": 0.7713,
+      "step": 600
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.1334872245788574,
+      "learning_rate": 4.195703269218868e-05,
+      "loss": 0.9144,
+      "step": 601
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.1683942079544067,
+      "learning_rate": 4.164996745699966e-05,
+      "loss": 0.8409,
+      "step": 602
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 1.074876308441162,
+      "learning_rate": 4.1343734165016514e-05,
+      "loss": 0.5677,
+      "step": 603
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.4278324842453003,
+      "learning_rate": 4.1038337182452826e-05,
+      "loss": 0.8643,
+      "step": 604
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.0636829137802124,
+      "learning_rate": 4.0733780863598335e-05,
+      "loss": 0.7189,
+      "step": 605
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.238641381263733,
+      "learning_rate": 4.0430069550756665e-05,
+      "loss": 0.8048,
+      "step": 606
+    },
+    {
+      "epoch": 1.39,
+      "grad_norm": 1.2493727207183838,
+      "learning_rate": 4.012720757418358e-05,
+      "loss": 0.6849,
+      "step": 607
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.04925537109375,
+      "learning_rate": 3.9825199252025184e-05,
+      "loss": 0.6783,
+      "step": 608
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.0140671730041504,
+      "learning_rate": 3.952404889025626e-05,
+      "loss": 0.6275,
+      "step": 609
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.292979121208191,
+      "learning_rate": 3.9223760782619045e-05,
+      "loss": 0.6685,
+      "step": 610
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.355290174484253,
+      "learning_rate": 3.8924339210561836e-05,
+      "loss": 0.7877,
+      "step": 611
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.2331466674804688,
+      "learning_rate": 3.862578844317817e-05,
+      "loss": 0.8935,
+      "step": 612
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.4775677919387817,
+      "learning_rate": 3.832811273714569e-05,
+      "loss": 0.8254,
+      "step": 613
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.1989951133728027,
+      "learning_rate": 3.803131633666572e-05,
+      "loss": 0.7965,
+      "step": 614
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 1.2056313753128052,
+      "learning_rate": 3.773540347340248e-05,
+      "loss": 1.2453,
+      "step": 615
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.9701207876205444,
+      "learning_rate": 3.7440378366423e-05,
+      "loss": 0.5302,
+      "step": 616
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.1523936986923218,
+      "learning_rate": 3.714624522213681e-05,
+      "loss": 0.6319,
+      "step": 617
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.7994964122772217,
+      "learning_rate": 3.685300823423602e-05,
+      "loss": 0.8799,
+      "step": 618
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.305521011352539,
+      "learning_rate": 3.6560671583635467e-05,
+      "loss": 0.9071,
+      "step": 619
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 1.4707729816436768,
+      "learning_rate": 3.626923943841325e-05,
+      "loss": 1.0193,
+      "step": 620
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.5845611095428467,
+      "learning_rate": 3.597871595375121e-05,
+      "loss": 0.9118,
+      "step": 621
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.0622698068618774,
+      "learning_rate": 3.5689105271875564e-05,
+      "loss": 0.8397,
+      "step": 622
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.2688456773757935,
+      "learning_rate": 3.5400411521998126e-05,
+      "loss": 0.9132,
+      "step": 623
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 1.2782020568847656,
+      "learning_rate": 3.5112638820257115e-05,
+      "loss": 0.5659,
+      "step": 624
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.4183655977249146,
+      "learning_rate": 3.482579126965878e-05,
+      "loss": 0.5982,
+      "step": 625
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.279589056968689,
+      "learning_rate": 3.453987296001866e-05,
+      "loss": 0.6599,
+      "step": 626
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.5479413270950317,
+      "learning_rate": 3.425488796790337e-05,
+      "loss": 0.738,
+      "step": 627
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 1.507043719291687,
+      "learning_rate": 3.397084035657243e-05,
+      "loss": 0.7308,
+      "step": 628
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.5540934801101685,
+      "learning_rate": 3.36877341759205e-05,
+      "loss": 0.9896,
+      "step": 629
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.198010802268982,
+      "learning_rate": 3.340557346241936e-05,
+      "loss": 0.9595,
+      "step": 630
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.4387493133544922,
+      "learning_rate": 3.312436223906062e-05,
+      "loss": 0.9932,
+      "step": 631
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 1.4757272005081177,
+      "learning_rate": 3.2844104515298155e-05,
+      "loss": 0.796,
+      "step": 632
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.2666937112808228,
+      "learning_rate": 3.2564804286991135e-05,
+      "loss": 0.6709,
+      "step": 633
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.187326192855835,
+      "learning_rate": 3.2286465536346854e-05,
+      "loss": 0.8126,
+      "step": 634
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.434696912765503,
+      "learning_rate": 3.2009092231864044e-05,
+      "loss": 0.7374,
+      "step": 635
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.5462465286254883,
+      "learning_rate": 3.173268832827643e-05,
+      "loss": 0.8611,
+      "step": 636
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 1.5220853090286255,
+      "learning_rate": 3.1457257766496015e-05,
+      "loss": 1.0394,
+      "step": 637
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.2815146446228027,
+      "learning_rate": 3.118280447355729e-05,
+      "loss": 1.0586,
+      "step": 638
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.4531899690628052,
+      "learning_rate": 3.090933236256087e-05,
+      "loss": 0.8858,
+      "step": 639
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.7821406126022339,
+      "learning_rate": 3.0636845332617994e-05,
+      "loss": 0.4845,
+      "step": 640
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 1.171593189239502,
+      "learning_rate": 3.036534726879473e-05,
+      "loss": 0.6783,
+      "step": 641
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.9787347316741943,
+      "learning_rate": 3.0094842042056704e-05,
+      "loss": 0.7622,
+      "step": 642
+    },
+    {
+      "epoch": 1.48,
+      "eval_loss": 1.1426820755004883,
+      "eval_runtime": 4.8034,
+      "eval_samples_per_second": 20.819,
+      "eval_steps_per_second": 20.819,
+      "step": 642
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 2.1937968730926514,
+      "learning_rate": 2.9825333509213827e-05,
+      "loss": 0.8779,
+      "step": 643
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.9966670274734497,
+      "learning_rate": 2.9556825512865415e-05,
+      "loss": 0.9429,
+      "step": 644
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.0608552694320679,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.4471,
+      "step": 645
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.5099259614944458,
+      "learning_rate": 2.902282642866716e-05,
+      "loss": 0.9748,
+      "step": 646
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.24240243434906,
+      "learning_rate": 2.8757342954470533e-05,
+      "loss": 0.7204,
+      "step": 647
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 1.631426215171814,
+      "learning_rate": 2.849287524396611e-05,
+      "loss": 1.0639,
+      "step": 648
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.9880529642105103,
+      "learning_rate": 2.8229427067882164e-05,
+      "loss": 0.7009,
+      "step": 649
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.366929531097412,
+      "learning_rate": 2.7967002182410596e-05,
+      "loss": 0.7605,
+      "step": 650
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.3599152565002441,
+      "learning_rate": 2.7705604329153434e-05,
+      "loss": 0.9786,
+      "step": 651
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.316638708114624,
+      "learning_rate": 2.7445237235069455e-05,
+      "loss": 1.0621,
+      "step": 652
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 2.131920099258423,
+      "learning_rate": 2.7185904612421176e-05,
+      "loss": 0.4658,
+      "step": 653
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 1.4175996780395508,
+      "learning_rate": 2.6927610158721706e-05,
+      "loss": 1.1467,
+      "step": 654
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.041159987449646,
+      "learning_rate": 2.6670357556682247e-05,
+      "loss": 0.7331,
+      "step": 655
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.5495882034301758,
+      "learning_rate": 2.6414150474159403e-05,
+      "loss": 1.0374,
+      "step": 656
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.3431997299194336,
+      "learning_rate": 2.6158992564103058e-05,
+      "loss": 0.8734,
+      "step": 657
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.2934917211532593,
+      "learning_rate": 2.5904887464504114e-05,
+      "loss": 0.7493,
+      "step": 658
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2499974966049194,
+      "learning_rate": 2.565183879834272e-05,
+      "loss": 0.6417,
+      "step": 659
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.0983362197875977,
+      "learning_rate": 2.53998501735367e-05,
+      "loss": 0.618,
+      "step": 660
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2114965915679932,
+      "learning_rate": 2.514892518288988e-05,
+      "loss": 0.7808,
+      "step": 661
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.2864103317260742,
+      "learning_rate": 2.4899067404041153e-05,
+      "loss": 0.8773,
+      "step": 662
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.3944170475006104,
+      "learning_rate": 2.465028039941316e-05,
+      "loss": 0.9462,
+      "step": 663
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.3994008302688599,
+      "learning_rate": 2.4402567716161805e-05,
+      "loss": 0.9313,
+      "step": 664
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.3160896301269531,
+      "learning_rate": 2.415593288612541e-05,
+      "loss": 0.724,
+      "step": 665
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 1.348981261253357,
+      "learning_rate": 2.391037942577454e-05,
+      "loss": 0.9646,
+      "step": 666
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.1432273387908936,
+      "learning_rate": 2.3665910836161775e-05,
+      "loss": 0.6094,
+      "step": 667
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.5914933681488037,
+      "learning_rate": 2.3422530602871872e-05,
+      "loss": 0.8079,
+      "step": 668
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.3685251474380493,
+      "learning_rate": 2.318024219597196e-05,
+      "loss": 0.6591,
+      "step": 669
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.1842666864395142,
+      "learning_rate": 2.2939049069962183e-05,
+      "loss": 0.8079,
+      "step": 670
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 1.3472460508346558,
+      "learning_rate": 2.26989546637263e-05,
+      "loss": 0.7606,
+      "step": 671
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.358242392539978,
+      "learning_rate": 2.2459962400482847e-05,
+      "loss": 0.7715,
+      "step": 672
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.2322027683258057,
+      "learning_rate": 2.2222075687736187e-05,
+      "loss": 0.8517,
+      "step": 673
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.5333589315414429,
+      "learning_rate": 2.198529791722792e-05,
+      "loss": 0.6613,
+      "step": 674
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 1.2892917394638062,
+      "learning_rate": 2.1749632464888592e-05,
+      "loss": 0.8702,
+      "step": 675
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.505099892616272,
+      "learning_rate": 2.1515082690789535e-05,
+      "loss": 0.8878,
+      "step": 676
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.266660213470459,
+      "learning_rate": 2.1281651939094992e-05,
+      "loss": 1.0645,
+      "step": 677
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 1.201398491859436,
+      "learning_rate": 2.1049343538014355e-05,
+      "loss": 0.9351,
+      "step": 678
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 0.9400334358215332,
+      "learning_rate": 2.0818160799754828e-05,
+      "loss": 0.4052,
+      "step": 679
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.2123039960861206,
+      "learning_rate": 2.0588107020474056e-05,
+      "loss": 0.6165,
+      "step": 680
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.3787363767623901,
+      "learning_rate": 2.03591854802333e-05,
+      "loss": 0.8574,
+      "step": 681
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 1.328092098236084,
+      "learning_rate": 2.0131399442950505e-05,
+      "loss": 0.7751,
+      "step": 682
+    },
+    {
+      "epoch": 1.57,
+      "grad_norm": 2.2625534534454346,
+      "learning_rate": 1.9904752156353878e-05,
+      "loss": 1.1871,
+      "step": 683
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.0272362232208252,
+      "learning_rate": 1.967924685193552e-05,
+      "loss": 0.5135,
+      "step": 684
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.3563427925109863,
+      "learning_rate": 1.94548867449054e-05,
+      "loss": 0.9451,
+      "step": 685
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.4345487356185913,
+      "learning_rate": 1.9231675034145513e-05,
+      "loss": 0.8185,
+      "step": 686
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 1.4716142416000366,
+      "learning_rate": 1.9009614902164174e-05,
+      "loss": 0.8666,
+      "step": 687
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 2.2574665546417236,
+      "learning_rate": 1.8788709515050808e-05,
+      "loss": 1.1538,
+      "step": 688
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.2094467878341675,
+      "learning_rate": 1.8568962022430636e-05,
+      "loss": 0.9113,
+      "step": 689
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.2264611721038818,
+      "learning_rate": 1.8350375557419875e-05,
+      "loss": 0.693,
+      "step": 690
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.440259337425232,
+      "learning_rate": 1.813295323658103e-05,
+      "loss": 0.702,
+      "step": 691
+    },
+    {
+      "epoch": 1.59,
+      "grad_norm": 1.0737701654434204,
+      "learning_rate": 1.791669815987852e-05,
+      "loss": 0.581,
+      "step": 692
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.3297679424285889,
+      "learning_rate": 1.7701613410634365e-05,
+      "loss": 0.7501,
+      "step": 693
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.1620917320251465,
+      "learning_rate": 1.7487702055484345e-05,
+      "loss": 0.8193,
+      "step": 694
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 2.063955068588257,
+      "learning_rate": 1.7274967144334153e-05,
+      "loss": 1.0735,
+      "step": 695
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.1501412391662598,
+      "learning_rate": 1.7063411710316046e-05,
+      "loss": 0.5951,
+      "step": 696
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.2334132194519043,
+      "learning_rate": 1.6853038769745467e-05,
+      "loss": 0.7601,
+      "step": 697
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.8308238983154297,
+      "learning_rate": 1.6643851322078174e-05,
+      "loss": 1.0234,
+      "step": 698
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.6979902982711792,
+      "learning_rate": 1.643585234986733e-05,
+      "loss": 1.0344,
+      "step": 699
+    },
+    {
+      "epoch": 1.61,
+      "grad_norm": 1.1924549341201782,
+      "learning_rate": 1.622904481872106e-05,
+      "loss": 0.7712,
+      "step": 700
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.3908418416976929,
+      "learning_rate": 1.6023431677260214e-05,
+      "loss": 0.7998,
+      "step": 701
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.5069186687469482,
+      "learning_rate": 1.5819015857076213e-05,
+      "loss": 1.0222,
+      "step": 702
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.5676149129867554,
+      "learning_rate": 1.5615800272689352e-05,
+      "loss": 1.0573,
+      "step": 703
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.4918407201766968,
+      "learning_rate": 1.541378782150714e-05,
+      "loss": 0.7947,
+      "step": 704
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 1.4949313402175903,
+      "learning_rate": 1.5212981383783154e-05,
+      "loss": 0.9044,
+      "step": 705
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.4046580791473389,
+      "learning_rate": 1.5013383822575766e-05,
+      "loss": 0.976,
+      "step": 706
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.247767448425293,
+      "learning_rate": 1.4814997983707458e-05,
+      "loss": 0.8356,
+      "step": 707
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.4620435237884521,
+      "learning_rate": 1.4617826695724223e-05,
+      "loss": 0.9709,
+      "step": 708
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 1.5974092483520508,
+      "learning_rate": 1.442187276985526e-05,
+      "loss": 0.7993,
+      "step": 709
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.4362558126449585,
+      "learning_rate": 1.42271389999728e-05,
+      "loss": 0.8003,
+      "step": 710
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.4817911386489868,
+      "learning_rate": 1.4033628162552359e-05,
+      "loss": 0.7285,
+      "step": 711
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.3569756746292114,
+      "learning_rate": 1.3841343016633167e-05,
+      "loss": 0.6743,
+      "step": 712
+    },
+    {
+      "epoch": 1.64,
+      "grad_norm": 1.1255030632019043,
+      "learning_rate": 1.3650286303778714e-05,
+      "loss": 0.9471,
+      "step": 713
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.5939558744430542,
+      "learning_rate": 1.3460460748037774e-05,
+      "loss": 0.8036,
+      "step": 714
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.3757383823394775,
+      "learning_rate": 1.3271869055905495e-05,
+      "loss": 0.8478,
+      "step": 715
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 1.6052110195159912,
+      "learning_rate": 1.3084513916284913e-05,
+      "loss": 0.902,
+      "step": 716
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 2.1484215259552,
+      "learning_rate": 1.2898398000448443e-05,
+      "loss": 1.0986,
+      "step": 717
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.9963454008102417,
+      "learning_rate": 1.2713523961999996e-05,
+      "loss": 0.5391,
+      "step": 718
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.68670654296875,
+      "learning_rate": 1.2529894436836965e-05,
+      "loss": 0.9895,
+      "step": 719
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.4046229124069214,
+      "learning_rate": 1.2347512043112752e-05,
+      "loss": 0.9313,
+      "step": 720
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 2.2879252433776855,
+      "learning_rate": 1.2166379381199423e-05,
+      "loss": 0.84,
+      "step": 721
+    },
+    {
+      "epoch": 1.66,
+      "grad_norm": 1.3971514701843262,
+      "learning_rate": 1.1986499033650556e-05,
+      "loss": 0.8177,
+      "step": 722
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.1414539813995361,
+      "learning_rate": 1.1807873565164506e-05,
+      "loss": 0.6581,
+      "step": 723
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.2839149236679077,
+      "learning_rate": 1.1630505522547853e-05,
+      "loss": 1.0131,
+      "step": 724
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.4500195980072021,
+      "learning_rate": 1.1454397434679021e-05,
+      "loss": 1.0473,
+      "step": 725
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 1.769698977470398,
+      "learning_rate": 1.12795518124722e-05,
+      "loss": 1.1283,
+      "step": 726
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.6734747886657715,
+      "learning_rate": 1.11059711488417e-05,
+      "loss": 1.0084,
+      "step": 727
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.5987813472747803,
+      "learning_rate": 1.0933657918666174e-05,
+      "loss": 1.0561,
+      "step": 728
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.2631326913833618,
+      "learning_rate": 1.0762614578753572e-05,
+      "loss": 0.9926,
+      "step": 729
+    },
+    {
+      "epoch": 1.68,
+      "grad_norm": 1.380119800567627,
+      "learning_rate": 1.0592843567805943e-05,
+      "loss": 0.9681,
+      "step": 730
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2174406051635742,
+      "learning_rate": 1.0424347306384729e-05,
+      "loss": 0.6559,
+      "step": 731
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.7756376266479492,
+      "learning_rate": 1.025712819687623e-05,
+      "loss": 1.1937,
+      "step": 732
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.4800599813461304,
+      "learning_rate": 1.0091188623457415e-05,
+      "loss": 0.7639,
+      "step": 733
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.4891732931137085,
+      "learning_rate": 9.92653095206183e-06,
+      "loss": 0.7647,
+      "step": 734
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 1.2692780494689941,
+      "learning_rate": 9.763157530345957e-06,
+      "loss": 0.8445,
+      "step": 735
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.459630012512207,
+      "learning_rate": 9.601070687655667e-06,
+      "loss": 1.0512,
+      "step": 736
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.3034725189208984,
+      "learning_rate": 9.440272734993072e-06,
+      "loss": 0.8654,
+      "step": 737
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.2299538850784302,
+      "learning_rate": 9.280765964983529e-06,
+      "loss": 0.6091,
+      "step": 738
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 1.5857917070388794,
+      "learning_rate": 9.12255265184293e-06,
+      "loss": 0.8166,
+      "step": 739
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.3110122680664062,
+      "learning_rate": 8.965635051345411e-06,
+      "loss": 0.8791,
+      "step": 740
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.4379709959030151,
+      "learning_rate": 8.810015400790994e-06,
+      "loss": 0.9836,
+      "step": 741
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.272002100944519,
+      "learning_rate": 8.655695918973862e-06,
+      "loss": 1.0156,
+      "step": 742
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 1.7516756057739258,
+      "learning_rate": 8.502678806150588e-06,
+      "loss": 0.9863,
+      "step": 743
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.4102840423583984,
+      "learning_rate": 8.350966244008895e-06,
+      "loss": 0.9151,
+      "step": 744
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.1626827716827393,
+      "learning_rate": 8.200560395636414e-06,
+      "loss": 0.8854,
+      "step": 745
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.5321221351623535,
+      "learning_rate": 8.051463405489957e-06,
+      "loss": 0.7943,
+      "step": 746
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 1.5110939741134644,
+      "learning_rate": 7.90367739936484e-06,
+      "loss": 0.8067,
+      "step": 747
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.273176908493042,
+      "learning_rate": 7.7572044843647e-06,
+      "loss": 1.1937,
+      "step": 748
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.2125933170318604,
+      "learning_rate": 7.612046748871327e-06,
+      "loss": 1.8904,
+      "step": 749
+    },
+    {
+      "epoch": 1.73,
+      "eval_loss": 1.1425327062606812,
+      "eval_runtime": 5.0157,
+      "eval_samples_per_second": 19.937,
+      "eval_steps_per_second": 19.937,
+      "step": 749
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.541787028312683,
+      "learning_rate": 7.4682062625149655e-06,
+      "loss": 0.9905,
+      "step": 750
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.5736668109893799,
+      "learning_rate": 7.325685076144795e-06,
+      "loss": 1.0864,
+      "step": 751
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 1.21610426902771,
+      "learning_rate": 7.1844852217996305e-06,
+      "loss": 0.7444,
+      "step": 752
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 2.0105512142181396,
+      "learning_rate": 7.0446087126790575e-06,
+      "loss": 0.8589,
+      "step": 753
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.427541732788086,
+      "learning_rate": 6.906057543114619e-06,
+      "loss": 1.0143,
+      "step": 754
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.9788390398025513,
+      "learning_rate": 6.768833688541443e-06,
+      "loss": 1.2519,
+      "step": 755
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 1.646355152130127,
+      "learning_rate": 6.632939105470049e-06,
+      "loss": 1.1151,
+      "step": 756
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.3599865436553955,
+      "learning_rate": 6.498375731458528e-06,
+      "loss": 0.9262,
+      "step": 757
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.618510365486145,
+      "learning_rate": 6.365145485084767e-06,
+      "loss": 0.7766,
+      "step": 758
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.6023226976394653,
+      "learning_rate": 6.233250265919266e-06,
+      "loss": 0.9066,
+      "step": 759
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 1.3686883449554443,
+      "learning_rate": 6.102691954497907e-06,
+      "loss": 1.0172,
+      "step": 760
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.1869922876358032,
+      "learning_rate": 5.973472412295255e-06,
+      "loss": 0.6924,
+      "step": 761
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.10429847240448,
+      "learning_rate": 5.8455934816979305e-06,
+      "loss": 0.768,
+      "step": 762
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.2119660377502441,
+      "learning_rate": 5.719056985978388e-06,
+      "loss": 0.6629,
+      "step": 763
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 1.4681169986724854,
+      "learning_rate": 5.593864729268949e-06,
+      "loss": 0.6561,
+      "step": 764
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.4000868797302246,
+      "learning_rate": 5.470018496535967e-06,
+      "loss": 0.8486,
+      "step": 765
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.7486320734024048,
+      "learning_rate": 5.347520053554545e-06,
+      "loss": 1.0676,
+      "step": 766
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.2215529680252075,
+      "learning_rate": 5.22637114688318e-06,
+      "loss": 0.8262,
+      "step": 767
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.3720446825027466,
+      "learning_rate": 5.106573503839018e-06,
+      "loss": 0.8284,
+      "step": 768
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 1.42399001121521,
+      "learning_rate": 4.9881288324731045e-06,
+      "loss": 0.9733,
+      "step": 769
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.806998610496521,
+      "learning_rate": 4.871038821546103e-06,
+      "loss": 0.87,
+      "step": 770
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.7265794277191162,
+      "learning_rate": 4.755305140504185e-06,
+      "loss": 0.3255,
+      "step": 771
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.4420627355575562,
+      "learning_rate": 4.640929439455277e-06,
+      "loss": 1.0153,
+      "step": 772
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 1.6430668830871582,
+      "learning_rate": 4.527913349145441e-06,
+      "loss": 0.9099,
+      "step": 773
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.2274441719055176,
+      "learning_rate": 4.416258480935731e-06,
+      "loss": 0.8594,
+      "step": 774
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.4887783527374268,
+      "learning_rate": 4.305966426779118e-06,
+      "loss": 0.808,
+      "step": 775
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.1768397092819214,
+      "learning_rate": 4.197038759197869e-06,
+      "loss": 0.561,
+      "step": 776
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 1.266448974609375,
+      "learning_rate": 4.089477031261113e-06,
+      "loss": 0.9325,
+      "step": 777
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.5394303798675537,
+      "learning_rate": 3.9832827765626465e-06,
+      "loss": 0.7159,
+      "step": 778
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.482283115386963,
+      "learning_rate": 3.878457509199107e-06,
+      "loss": 1.103,
+      "step": 779
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.1145195960998535,
+      "learning_rate": 3.7750027237484e-06,
+      "loss": 1.0283,
+      "step": 780
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 1.5757018327713013,
+      "learning_rate": 3.6729198952483724e-06,
+      "loss": 1.162,
+      "step": 781
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.555261492729187,
+      "learning_rate": 3.572210479175753e-06,
+      "loss": 0.9346,
+      "step": 782
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.5401426553726196,
+      "learning_rate": 3.472875911425477e-06,
+      "loss": 1.0614,
+      "step": 783
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.4009335041046143,
+      "learning_rate": 3.3749176082901067e-06,
+      "loss": 0.9636,
+      "step": 784
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.5251446962356567,
+      "learning_rate": 3.2783369664397436e-06,
+      "loss": 0.7538,
+      "step": 785
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 1.2820510864257812,
+      "learning_rate": 3.1831353629020344e-06,
+      "loss": 0.8404,
+      "step": 786
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.2701919078826904,
+      "learning_rate": 3.0893141550425884e-06,
+      "loss": 0.7659,
+      "step": 787
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.619142770767212,
+      "learning_rate": 2.996874680545603e-06,
+      "loss": 0.7558,
+      "step": 788
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.825161099433899,
+      "learning_rate": 2.905818257394799e-06,
+      "loss": 1.0628,
+      "step": 789
+    },
+    {
+      "epoch": 1.82,
+      "grad_norm": 1.2984554767608643,
+      "learning_rate": 2.8161461838546176e-06,
+      "loss": 0.8484,
+      "step": 790
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.8327722549438477,
+      "learning_rate": 2.7278597384517214e-06,
+      "loss": 0.8366,
+      "step": 791
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.1649681329727173,
+      "learning_rate": 2.6409601799567642e-06,
+      "loss": 0.7479,
+      "step": 792
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.212082862854004,
+      "learning_rate": 2.55544874736644e-06,
+      "loss": 1.2608,
+      "step": 793
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 1.675178050994873,
+      "learning_rate": 2.4713266598858086e-06,
+      "loss": 0.7571,
+      "step": 794
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.349767804145813,
+      "learning_rate": 2.3885951169109187e-06,
+      "loss": 0.6354,
+      "step": 795
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.0622371435165405,
+      "learning_rate": 2.3072552980117566e-06,
+      "loss": 0.5172,
+      "step": 796
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.2005047798156738,
+      "learning_rate": 2.2273083629153147e-06,
+      "loss": 1.1696,
+      "step": 797
+    },
+    {
+      "epoch": 1.84,
+      "grad_norm": 1.3214634656906128,
+      "learning_rate": 2.1487554514891704e-06,
+      "loss": 0.9274,
+      "step": 798
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.359706163406372,
+      "learning_rate": 2.071597683725179e-06,
+      "loss": 0.7433,
+      "step": 799
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.3095366954803467,
+      "learning_rate": 1.9958361597235076e-06,
+      "loss": 0.9496,
+      "step": 800
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.5138649940490723,
+      "learning_rate": 1.921471959676957e-06,
+      "loss": 1.0328,
+      "step": 801
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 1.6273144483566284,
+      "learning_rate": 1.848506143855555e-06,
+      "loss": 0.7867,
+      "step": 802
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.9000691175460815,
+      "learning_rate": 1.7769397525914667e-06,
+      "loss": 0.7318,
+      "step": 803
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.6215777397155762,
+      "learning_rate": 1.706773806264106e-06,
+      "loss": 1.1416,
+      "step": 804
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.5041816234588623,
+      "learning_rate": 1.6380093052856483e-06,
+      "loss": 0.9327,
+      "step": 805
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.6266945600509644,
+      "learning_rate": 1.570647230086708e-06,
+      "loss": 1.1518,
+      "step": 806
+    },
+    {
+      "epoch": 1.86,
+      "grad_norm": 1.4679832458496094,
+      "learning_rate": 1.5046885411024391e-06,
+      "loss": 0.8719,
+      "step": 807
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.5392640829086304,
+      "learning_rate": 1.4401341787587453e-06,
+      "loss": 1.0913,
+      "step": 808
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.4636483192443848,
+      "learning_rate": 1.3769850634589354e-06,
+      "loss": 0.9312,
+      "step": 809
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.214938998222351,
+      "learning_rate": 1.3152420955706012e-06,
+      "loss": 0.8939,
+      "step": 810
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 1.2058130502700806,
+      "learning_rate": 1.2549061554127494e-06,
+      "loss": 0.8283,
+      "step": 811
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.2419188022613525,
+      "learning_rate": 1.1959781032432337e-06,
+      "loss": 0.9094,
+      "step": 812
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.7467167377471924,
+      "learning_rate": 1.1384587792465872e-06,
+      "loss": 0.7264,
+      "step": 813
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.167214035987854,
+      "learning_rate": 1.0823490035218987e-06,
+      "loss": 0.6232,
+      "step": 814
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.3175835609436035,
+      "learning_rate": 1.0276495760712767e-06,
+      "loss": 1.0396,
+      "step": 815
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 1.6061522960662842,
+      "learning_rate": 9.743612767882936e-07,
+      "loss": 0.6561,
+      "step": 816
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.4116915464401245,
+      "learning_rate": 9.224848654469931e-07,
+      "loss": 1.2357,
+      "step": 817
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.2765032052993774,
+      "learning_rate": 8.720210816909435e-07,
+      "loss": 0.9708,
+      "step": 818
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.392382025718689,
+      "learning_rate": 8.229706450227803e-07,
+      "loss": 0.9548,
+      "step": 819
+    },
+    {
+      "epoch": 1.89,
+      "grad_norm": 1.3821630477905273,
+      "learning_rate": 7.753342547939357e-07,
+      "loss": 0.7787,
+      "step": 820
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.5395421981811523,
+      "learning_rate": 7.291125901946027e-07,
+      "loss": 0.7223,
+      "step": 821
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.5560616254806519,
+      "learning_rate": 6.843063102441316e-07,
+      "loss": 0.9542,
+      "step": 822
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.2338303327560425,
+      "learning_rate": 6.409160537815817e-07,
+      "loss": 0.9685,
+      "step": 823
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 1.3884745836257935,
+      "learning_rate": 5.989424394566401e-07,
+      "loss": 1.1287,
+      "step": 824
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.1085461378097534,
+      "learning_rate": 5.58386065720784e-07,
+      "loss": 0.7562,
+      "step": 825
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 0.8512900471687317,
+      "learning_rate": 5.192475108187544e-07,
+      "loss": 0.4627,
+      "step": 826
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.3366022109985352,
+      "learning_rate": 4.815273327803182e-07,
+      "loss": 0.8489,
+      "step": 827
+    },
+    {
+      "epoch": 1.91,
+      "grad_norm": 1.066333293914795,
+      "learning_rate": 4.452260694122856e-07,
+      "loss": 0.6867,
+      "step": 828
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.484168291091919,
+      "learning_rate": 4.103442382909051e-07,
+      "loss": 1.1556,
+      "step": 829
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.4741454124450684,
+      "learning_rate": 3.7688233675439166e-07,
+      "loss": 0.9345,
+      "step": 830
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.0482972860336304,
+      "learning_rate": 3.4484084189593257e-07,
+      "loss": 0.7689,
+      "step": 831
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.4800513982772827,
+      "learning_rate": 3.1422021055679265e-07,
+      "loss": 0.8486,
+      "step": 832
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 1.3522624969482422,
+      "learning_rate": 2.850208793198861e-07,
+      "loss": 0.9064,
+      "step": 833
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.6680634021759033,
+      "learning_rate": 2.572432645034817e-07,
+      "loss": 1.0127,
+      "step": 834
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.5550647974014282,
+      "learning_rate": 2.3088776215531848e-07,
+      "loss": 0.8741,
+      "step": 835
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.4581865072250366,
+      "learning_rate": 2.0595474804691038e-07,
+      "loss": 0.9799,
+      "step": 836
+    },
+    {
+      "epoch": 1.93,
+      "grad_norm": 1.3442659378051758,
+      "learning_rate": 1.824445776682504e-07,
+      "loss": 0.8629,
+      "step": 837
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.5211567878723145,
+      "learning_rate": 1.6035758622269247e-07,
+      "loss": 0.6652,
+      "step": 838
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.16912841796875,
+      "learning_rate": 1.3969408862217758e-07,
+      "loss": 0.8838,
+      "step": 839
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.5780668258666992,
+      "learning_rate": 1.204543794827595e-07,
+      "loss": 0.9765,
+      "step": 840
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 1.4517415761947632,
+      "learning_rate": 1.0263873312040818e-07,
+      "loss": 0.6895,
+      "step": 841
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.7025426626205444,
+      "learning_rate": 8.624740354707949e-08,
+      "loss": 0.8764,
+      "step": 842
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.3014923334121704,
+      "learning_rate": 7.128062446709604e-08,
+      "loss": 1.0822,
+      "step": 843
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.501705288887024,
+      "learning_rate": 5.773860927383856e-08,
+      "loss": 0.8872,
+      "step": 844
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 1.6005631685256958,
+      "learning_rate": 4.562155104665955e-08,
+      "loss": 1.1913,
+      "step": 845
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.1631520986557007,
+      "learning_rate": 3.492962254819654e-08,
+      "loss": 0.955,
+      "step": 846
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.285071611404419,
+      "learning_rate": 2.5662976221840773e-08,
+      "loss": 0.8429,
+      "step": 847
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.0921001434326172,
+      "learning_rate": 1.7821744189605582e-08,
+      "loss": 0.7829,
+      "step": 848
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 1.4597008228302002,
+      "learning_rate": 1.1406038250205698e-08,
+      "loss": 1.1112,
+      "step": 849
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 2.1776418685913086,
+      "learning_rate": 6.41594987752514e-09,
+      "loss": 1.2897,
+      "step": 850
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.2497656345367432,
+      "learning_rate": 2.851550219240551e-09,
+      "loss": 0.9627,
+      "step": 851
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 1.291884422302246,
+      "learning_rate": 7.128900958774942e-10,
+      "loss": 0.8828,
+      "step": 852
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 852,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 426,
+  "total_flos": 1.547329324744704e+16,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}