| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 1350, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0022222222222222222, | |
| "grad_norm": 0.06687259674072266, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.6478, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0044444444444444444, | |
| "grad_norm": 0.08385849744081497, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.6389, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.006666666666666667, | |
| "grad_norm": 0.06483206152915955, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.6552, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.008888888888888889, | |
| "grad_norm": 0.06315144151449203, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.8175, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.011111111111111112, | |
| "grad_norm": 0.06796158105134964, | |
| "learning_rate": 1.111111111111111e-05, | |
| "loss": 0.6561, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.013333333333333334, | |
| "grad_norm": 0.0593683160841465, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.6677, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.015555555555555555, | |
| "grad_norm": 0.0792003720998764, | |
| "learning_rate": 1.5555555555555555e-05, | |
| "loss": 0.782, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.017777777777777778, | |
| "grad_norm": 0.0665215253829956, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 0.752, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 0.07640073448419571, | |
| "learning_rate": 1.9999999999999998e-05, | |
| "loss": 0.7552, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.022222222222222223, | |
| "grad_norm": 0.11807028204202652, | |
| "learning_rate": 2.222222222222222e-05, | |
| "loss": 0.8112, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.024444444444444446, | |
| "grad_norm": 0.08529076725244522, | |
| "learning_rate": 2.4444444444444445e-05, | |
| "loss": 0.8115, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02666666666666667, | |
| "grad_norm": 0.09069886803627014, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 0.7827, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.028888888888888888, | |
| "grad_norm": 0.08439790457487106, | |
| "learning_rate": 2.8888888888888885e-05, | |
| "loss": 0.6591, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.03111111111111111, | |
| "grad_norm": 0.049057431519031525, | |
| "learning_rate": 3.111111111111111e-05, | |
| "loss": 0.4825, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.03333333333333333, | |
| "grad_norm": 0.06010900437831879, | |
| "learning_rate": 3.333333333333333e-05, | |
| "loss": 0.6372, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.035555555555555556, | |
| "grad_norm": 0.09451697021722794, | |
| "learning_rate": 3.555555555555555e-05, | |
| "loss": 0.7133, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.03777777777777778, | |
| "grad_norm": 0.10815142095088959, | |
| "learning_rate": 3.777777777777777e-05, | |
| "loss": 0.9101, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.08878195285797119, | |
| "learning_rate": 3.9999999999999996e-05, | |
| "loss": 0.695, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.042222222222222223, | |
| "grad_norm": 0.13616155087947845, | |
| "learning_rate": 4.222222222222222e-05, | |
| "loss": 0.8215, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.044444444444444446, | |
| "grad_norm": 0.07087361067533493, | |
| "learning_rate": 4.444444444444444e-05, | |
| "loss": 0.7639, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.04666666666666667, | |
| "grad_norm": 0.07879175245761871, | |
| "learning_rate": 4.6666666666666665e-05, | |
| "loss": 0.6547, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.04888888888888889, | |
| "grad_norm": 0.08119899034500122, | |
| "learning_rate": 4.888888888888889e-05, | |
| "loss": 0.69, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.051111111111111114, | |
| "grad_norm": 0.05957357957959175, | |
| "learning_rate": 5.111111111111111e-05, | |
| "loss": 0.5669, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.05333333333333334, | |
| "grad_norm": 0.08532346785068512, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 0.6756, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.05555555555555555, | |
| "grad_norm": 0.10510731488466263, | |
| "learning_rate": 5.5555555555555545e-05, | |
| "loss": 0.7982, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.057777777777777775, | |
| "grad_norm": 0.09259792417287827, | |
| "learning_rate": 5.777777777777777e-05, | |
| "loss": 0.7915, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.2204543501138687, | |
| "learning_rate": 5.9999999999999995e-05, | |
| "loss": 0.7374, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.06222222222222222, | |
| "grad_norm": 0.0847523957490921, | |
| "learning_rate": 6.222222222222222e-05, | |
| "loss": 0.7315, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.06444444444444444, | |
| "grad_norm": 0.06072792410850525, | |
| "learning_rate": 6.444444444444444e-05, | |
| "loss": 0.6173, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.06666666666666667, | |
| "grad_norm": 0.08779124915599823, | |
| "learning_rate": 6.666666666666666e-05, | |
| "loss": 0.7824, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.06888888888888889, | |
| "grad_norm": 0.07153160125017166, | |
| "learning_rate": 6.888888888888888e-05, | |
| "loss": 0.7171, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.07111111111111111, | |
| "grad_norm": 0.06395824253559113, | |
| "learning_rate": 7.11111111111111e-05, | |
| "loss": 0.7462, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.07333333333333333, | |
| "grad_norm": 0.08911176025867462, | |
| "learning_rate": 7.333333333333332e-05, | |
| "loss": 0.6821, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.07555555555555556, | |
| "grad_norm": 0.0638173520565033, | |
| "learning_rate": 7.555555555555554e-05, | |
| "loss": 0.6373, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.07777777777777778, | |
| "grad_norm": 0.07565649598836899, | |
| "learning_rate": 7.777777777777777e-05, | |
| "loss": 0.7063, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.05962090566754341, | |
| "learning_rate": 7.999999999999999e-05, | |
| "loss": 0.5242, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.08222222222222222, | |
| "grad_norm": 0.09105438739061356, | |
| "learning_rate": 8.222222222222222e-05, | |
| "loss": 0.7797, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.08444444444444445, | |
| "grad_norm": 0.06883728504180908, | |
| "learning_rate": 8.444444444444444e-05, | |
| "loss": 0.6209, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.08666666666666667, | |
| "grad_norm": 0.09342409670352936, | |
| "learning_rate": 8.666666666666665e-05, | |
| "loss": 0.7694, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.08888888888888889, | |
| "grad_norm": 0.045902229845523834, | |
| "learning_rate": 8.888888888888888e-05, | |
| "loss": 0.6035, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.09111111111111111, | |
| "grad_norm": 0.0726340040564537, | |
| "learning_rate": 9.11111111111111e-05, | |
| "loss": 0.7127, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.09333333333333334, | |
| "grad_norm": 0.08810717612504959, | |
| "learning_rate": 9.333333333333333e-05, | |
| "loss": 0.838, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.09555555555555556, | |
| "grad_norm": 0.08479885011911392, | |
| "learning_rate": 9.555555555555555e-05, | |
| "loss": 0.8645, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.09777777777777778, | |
| "grad_norm": 0.06431703269481659, | |
| "learning_rate": 9.777777777777778e-05, | |
| "loss": 0.8542, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.07358094304800034, | |
| "learning_rate": 9.999999999999999e-05, | |
| "loss": 0.7753, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.10222222222222223, | |
| "grad_norm": 0.09305302053689957, | |
| "learning_rate": 0.00010222222222222222, | |
| "loss": 0.6795, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.10444444444444445, | |
| "grad_norm": 0.08080010116100311, | |
| "learning_rate": 0.00010444444444444444, | |
| "loss": 0.5749, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.10666666666666667, | |
| "grad_norm": 0.06824497878551483, | |
| "learning_rate": 0.00010666666666666667, | |
| "loss": 0.6968, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.10888888888888888, | |
| "grad_norm": 0.09318830817937851, | |
| "learning_rate": 0.00010888888888888889, | |
| "loss": 0.7178, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.1111111111111111, | |
| "grad_norm": 0.07186290621757507, | |
| "learning_rate": 0.00011111111111111109, | |
| "loss": 0.742, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.11333333333333333, | |
| "grad_norm": 0.07324495166540146, | |
| "learning_rate": 0.00011333333333333331, | |
| "loss": 0.6345, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.11555555555555555, | |
| "grad_norm": 0.06657472252845764, | |
| "learning_rate": 0.00011555555555555554, | |
| "loss": 0.62, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.11777777777777777, | |
| "grad_norm": 0.06139868125319481, | |
| "learning_rate": 0.00011777777777777776, | |
| "loss": 0.7513, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.06240592524409294, | |
| "learning_rate": 0.00011999999999999999, | |
| "loss": 0.6174, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.12222222222222222, | |
| "grad_norm": 0.07715508341789246, | |
| "learning_rate": 0.00012222222222222221, | |
| "loss": 0.8125, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.12444444444444444, | |
| "grad_norm": 0.0686756893992424, | |
| "learning_rate": 0.00012444444444444444, | |
| "loss": 0.5399, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.12666666666666668, | |
| "grad_norm": 0.07324881106615067, | |
| "learning_rate": 0.00012666666666666666, | |
| "loss": 0.6211, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.1288888888888889, | |
| "grad_norm": 0.08367079496383667, | |
| "learning_rate": 0.0001288888888888889, | |
| "loss": 0.6632, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.13111111111111112, | |
| "grad_norm": 0.06640663743019104, | |
| "learning_rate": 0.00013111111111111111, | |
| "loss": 0.6054, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.13333333333333333, | |
| "grad_norm": 0.06046535447239876, | |
| "learning_rate": 0.0001333333333333333, | |
| "loss": 0.7288, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.13555555555555557, | |
| "grad_norm": 0.06970898061990738, | |
| "learning_rate": 0.00013555555555555554, | |
| "loss": 0.7491, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.13777777777777778, | |
| "grad_norm": 0.06068103015422821, | |
| "learning_rate": 0.00013777777777777776, | |
| "loss": 0.6525, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.08266304433345795, | |
| "learning_rate": 0.00014, | |
| "loss": 0.7896, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.14222222222222222, | |
| "grad_norm": 0.05625743046402931, | |
| "learning_rate": 0.0001422222222222222, | |
| "loss": 0.6877, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.14444444444444443, | |
| "grad_norm": 0.06797035783529282, | |
| "learning_rate": 0.0001444444444444444, | |
| "loss": 0.6664, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.14666666666666667, | |
| "grad_norm": 0.06103198602795601, | |
| "learning_rate": 0.00014666666666666664, | |
| "loss": 0.7097, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.14888888888888888, | |
| "grad_norm": 0.0863097533583641, | |
| "learning_rate": 0.00014888888888888886, | |
| "loss": 0.8366, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.1511111111111111, | |
| "grad_norm": 0.04689197614789009, | |
| "learning_rate": 0.00015111111111111109, | |
| "loss": 0.6539, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.15333333333333332, | |
| "grad_norm": 0.08514045178890228, | |
| "learning_rate": 0.0001533333333333333, | |
| "loss": 0.8184, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.15555555555555556, | |
| "grad_norm": 0.06352313607931137, | |
| "learning_rate": 0.00015555555555555554, | |
| "loss": 0.735, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.15777777777777777, | |
| "grad_norm": 0.05917006731033325, | |
| "learning_rate": 0.00015777777777777776, | |
| "loss": 0.6767, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.07596120238304138, | |
| "learning_rate": 0.00015999999999999999, | |
| "loss": 0.6356, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.1622222222222222, | |
| "grad_norm": 0.07308676093816757, | |
| "learning_rate": 0.0001622222222222222, | |
| "loss": 0.7718, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.16444444444444445, | |
| "grad_norm": 0.06971703469753265, | |
| "learning_rate": 0.00016444444444444444, | |
| "loss": 0.7646, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.06212179735302925, | |
| "learning_rate": 0.00016666666666666666, | |
| "loss": 0.607, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1688888888888889, | |
| "grad_norm": 0.08927149325609207, | |
| "learning_rate": 0.00016888888888888889, | |
| "loss": 0.838, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.1711111111111111, | |
| "grad_norm": 0.12827588617801666, | |
| "learning_rate": 0.0001711111111111111, | |
| "loss": 0.7376, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.17333333333333334, | |
| "grad_norm": 0.06566973775625229, | |
| "learning_rate": 0.0001733333333333333, | |
| "loss": 0.6345, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.17555555555555555, | |
| "grad_norm": 0.07192882150411606, | |
| "learning_rate": 0.00017555555555555553, | |
| "loss": 0.8029, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.17777777777777778, | |
| "grad_norm": 0.04837195575237274, | |
| "learning_rate": 0.00017777777777777776, | |
| "loss": 0.6689, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.1018751710653305, | |
| "learning_rate": 0.00017999999999999998, | |
| "loss": 0.6286, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.18222222222222223, | |
| "grad_norm": 0.05638294294476509, | |
| "learning_rate": 0.0001822222222222222, | |
| "loss": 0.5445, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.18444444444444444, | |
| "grad_norm": 0.05647343024611473, | |
| "learning_rate": 0.00018444444444444443, | |
| "loss": 0.6552, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.18666666666666668, | |
| "grad_norm": 0.07809357345104218, | |
| "learning_rate": 0.00018666666666666666, | |
| "loss": 0.722, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.18888888888888888, | |
| "grad_norm": 0.06055169925093651, | |
| "learning_rate": 0.00018888888888888888, | |
| "loss": 0.8052, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.19111111111111112, | |
| "grad_norm": 0.05945683270692825, | |
| "learning_rate": 0.0001911111111111111, | |
| "loss": 0.7048, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.19333333333333333, | |
| "grad_norm": 0.07493992894887924, | |
| "learning_rate": 0.00019333333333333333, | |
| "loss": 0.8066, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.19555555555555557, | |
| "grad_norm": 0.07349304109811783, | |
| "learning_rate": 0.00019555555555555556, | |
| "loss": 0.6287, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.19777777777777777, | |
| "grad_norm": 0.06694339215755463, | |
| "learning_rate": 0.00019777777777777776, | |
| "loss": 0.6973, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.06596689671278, | |
| "learning_rate": 0.00019999999999999998, | |
| "loss": 0.6139, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.20222222222222222, | |
| "grad_norm": 0.059765566140413284, | |
| "learning_rate": 0.0002022222222222222, | |
| "loss": 0.7106, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.20444444444444446, | |
| "grad_norm": 0.06259062141180038, | |
| "learning_rate": 0.00020444444444444443, | |
| "loss": 0.7255, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.20666666666666667, | |
| "grad_norm": 0.060403451323509216, | |
| "learning_rate": 0.00020666666666666666, | |
| "loss": 0.6794, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.2088888888888889, | |
| "grad_norm": 0.08180979639291763, | |
| "learning_rate": 0.00020888888888888888, | |
| "loss": 0.7074, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.2111111111111111, | |
| "grad_norm": 0.0711866170167923, | |
| "learning_rate": 0.0002111111111111111, | |
| "loss": 0.7665, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.21333333333333335, | |
| "grad_norm": 0.06080659106373787, | |
| "learning_rate": 0.00021333333333333333, | |
| "loss": 0.7143, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.21555555555555556, | |
| "grad_norm": 0.06005050614476204, | |
| "learning_rate": 0.00021555555555555556, | |
| "loss": 0.6225, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.21777777777777776, | |
| "grad_norm": 0.05671238526701927, | |
| "learning_rate": 0.00021777777777777778, | |
| "loss": 0.6115, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.04517147317528725, | |
| "learning_rate": 0.00021999999999999995, | |
| "loss": 0.6039, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.2222222222222222, | |
| "grad_norm": 0.06307649612426758, | |
| "learning_rate": 0.00022222222222222218, | |
| "loss": 0.6816, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.22444444444444445, | |
| "grad_norm": 0.08459043502807617, | |
| "learning_rate": 0.0002244444444444444, | |
| "loss": 0.742, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.22666666666666666, | |
| "grad_norm": 0.07087739557027817, | |
| "learning_rate": 0.00022666666666666663, | |
| "loss": 0.5635, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.2288888888888889, | |
| "grad_norm": 0.0924144759774208, | |
| "learning_rate": 0.00022888888888888885, | |
| "loss": 0.6658, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.2311111111111111, | |
| "grad_norm": 0.07960222661495209, | |
| "learning_rate": 0.00023111111111111108, | |
| "loss": 0.7541, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.23333333333333334, | |
| "grad_norm": 0.05279611423611641, | |
| "learning_rate": 0.0002333333333333333, | |
| "loss": 0.6112, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.23555555555555555, | |
| "grad_norm": 0.06858152151107788, | |
| "learning_rate": 0.00023555555555555553, | |
| "loss": 0.5877, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.23777777777777778, | |
| "grad_norm": 0.07217252254486084, | |
| "learning_rate": 0.00023777777777777775, | |
| "loss": 0.7188, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.06040087342262268, | |
| "learning_rate": 0.00023999999999999998, | |
| "loss": 0.7232, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.24222222222222223, | |
| "grad_norm": 0.07673678547143936, | |
| "learning_rate": 0.0002422222222222222, | |
| "loss": 0.6434, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.24444444444444444, | |
| "grad_norm": 0.04461099952459335, | |
| "learning_rate": 0.00024444444444444443, | |
| "loss": 0.539, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.24666666666666667, | |
| "grad_norm": 0.07912620157003403, | |
| "learning_rate": 0.0002466666666666666, | |
| "loss": 0.6223, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.24888888888888888, | |
| "grad_norm": 0.061267413198947906, | |
| "learning_rate": 0.0002488888888888889, | |
| "loss": 0.645, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.2511111111111111, | |
| "grad_norm": 0.05509350448846817, | |
| "learning_rate": 0.0002511111111111111, | |
| "loss": 0.5941, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.25333333333333335, | |
| "grad_norm": 0.07612048834562302, | |
| "learning_rate": 0.00025333333333333333, | |
| "loss": 0.7886, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.25555555555555554, | |
| "grad_norm": 0.045813750475645065, | |
| "learning_rate": 0.00025555555555555553, | |
| "loss": 0.5798, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.2577777777777778, | |
| "grad_norm": 0.07017998397350311, | |
| "learning_rate": 0.0002577777777777778, | |
| "loss": 0.664, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.059138622134923935, | |
| "learning_rate": 0.00026, | |
| "loss": 0.6002, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.26222222222222225, | |
| "grad_norm": 0.06197448447346687, | |
| "learning_rate": 0.00026222222222222223, | |
| "loss": 0.7018, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.2644444444444444, | |
| "grad_norm": 0.055604707449674606, | |
| "learning_rate": 0.00026444444444444443, | |
| "loss": 0.7665, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.26666666666666666, | |
| "grad_norm": 0.06360765546560287, | |
| "learning_rate": 0.0002666666666666666, | |
| "loss": 0.701, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.2688888888888889, | |
| "grad_norm": 0.06214721128344536, | |
| "learning_rate": 0.0002688888888888889, | |
| "loss": 0.567, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.27111111111111114, | |
| "grad_norm": 0.05708202347159386, | |
| "learning_rate": 0.0002711111111111111, | |
| "loss": 0.7378, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.2733333333333333, | |
| "grad_norm": 0.05343154817819595, | |
| "learning_rate": 0.00027333333333333333, | |
| "loss": 0.6234, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.27555555555555555, | |
| "grad_norm": 0.12749263644218445, | |
| "learning_rate": 0.0002755555555555555, | |
| "loss": 0.7841, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "grad_norm": 0.13472327589988708, | |
| "learning_rate": 0.0002777777777777778, | |
| "loss": 0.807, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.08000756055116653, | |
| "learning_rate": 0.00028, | |
| "loss": 0.8359, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.2822222222222222, | |
| "grad_norm": 0.057078856974840164, | |
| "learning_rate": 0.00028222222222222223, | |
| "loss": 0.6232, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.28444444444444444, | |
| "grad_norm": 0.07158028334379196, | |
| "learning_rate": 0.0002844444444444444, | |
| "loss": 0.517, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.2866666666666667, | |
| "grad_norm": 0.05929658189415932, | |
| "learning_rate": 0.0002866666666666667, | |
| "loss": 0.6644, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.28888888888888886, | |
| "grad_norm": 0.05705413967370987, | |
| "learning_rate": 0.0002888888888888888, | |
| "loss": 0.5647, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.2911111111111111, | |
| "grad_norm": 0.03960437327623367, | |
| "learning_rate": 0.0002911111111111111, | |
| "loss": 0.5516, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.29333333333333333, | |
| "grad_norm": 0.06845048069953918, | |
| "learning_rate": 0.00029333333333333327, | |
| "loss": 0.8321, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.29555555555555557, | |
| "grad_norm": 0.06761569529771805, | |
| "learning_rate": 0.0002955555555555555, | |
| "loss": 0.6994, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.29777777777777775, | |
| "grad_norm": 0.055330973118543625, | |
| "learning_rate": 0.0002977777777777777, | |
| "loss": 0.5582, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.06866293400526047, | |
| "learning_rate": 0.0003, | |
| "loss": 0.5497, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3022222222222222, | |
| "grad_norm": 0.06994841247797012, | |
| "learning_rate": 0.00029975308641975307, | |
| "loss": 0.71, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.30444444444444446, | |
| "grad_norm": 0.061404068022966385, | |
| "learning_rate": 0.00029950617283950616, | |
| "loss": 0.7004, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.30666666666666664, | |
| "grad_norm": 0.08630826324224472, | |
| "learning_rate": 0.00029925925925925926, | |
| "loss": 0.685, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.3088888888888889, | |
| "grad_norm": 0.05255487933754921, | |
| "learning_rate": 0.0002990123456790123, | |
| "loss": 0.5405, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.3111111111111111, | |
| "grad_norm": 0.050141241401433945, | |
| "learning_rate": 0.0002987654320987654, | |
| "loss": 0.6132, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.31333333333333335, | |
| "grad_norm": 0.053194086998701096, | |
| "learning_rate": 0.0002985185185185185, | |
| "loss": 0.6057, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.31555555555555553, | |
| "grad_norm": 0.0686035305261612, | |
| "learning_rate": 0.0002982716049382716, | |
| "loss": 0.7368, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.31777777777777777, | |
| "grad_norm": 0.08060168474912643, | |
| "learning_rate": 0.0002980246913580247, | |
| "loss": 0.7203, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.06388822197914124, | |
| "learning_rate": 0.0002977777777777777, | |
| "loss": 0.6401, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.32222222222222224, | |
| "grad_norm": 0.08037619292736053, | |
| "learning_rate": 0.00029753086419753087, | |
| "loss": 0.5519, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.3244444444444444, | |
| "grad_norm": 0.05287360027432442, | |
| "learning_rate": 0.0002972839506172839, | |
| "loss": 0.634, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.32666666666666666, | |
| "grad_norm": 0.11447149515151978, | |
| "learning_rate": 0.000297037037037037, | |
| "loss": 0.6378, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.3288888888888889, | |
| "grad_norm": 0.052581414580345154, | |
| "learning_rate": 0.0002967901234567901, | |
| "loss": 0.6137, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.33111111111111113, | |
| "grad_norm": 0.05176948010921478, | |
| "learning_rate": 0.0002965432098765432, | |
| "loss": 0.6561, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.06692127138376236, | |
| "learning_rate": 0.00029629629629629624, | |
| "loss": 0.7146, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.33555555555555555, | |
| "grad_norm": 0.060980744659900665, | |
| "learning_rate": 0.00029604938271604933, | |
| "loss": 0.6315, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.3377777777777778, | |
| "grad_norm": 0.057845454663038254, | |
| "learning_rate": 0.0002958024691358025, | |
| "loss": 0.586, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.05785061791539192, | |
| "learning_rate": 0.0002955555555555555, | |
| "loss": 0.6386, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.3422222222222222, | |
| "grad_norm": 0.07744596898555756, | |
| "learning_rate": 0.0002953086419753086, | |
| "loss": 0.7732, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.34444444444444444, | |
| "grad_norm": 0.051941994577646255, | |
| "learning_rate": 0.0002950617283950617, | |
| "loss": 0.5876, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.3466666666666667, | |
| "grad_norm": 0.05806094780564308, | |
| "learning_rate": 0.0002948148148148148, | |
| "loss": 0.7254, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.3488888888888889, | |
| "grad_norm": 0.06293324381113052, | |
| "learning_rate": 0.00029456790123456785, | |
| "loss": 0.7451, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.3511111111111111, | |
| "grad_norm": 0.0668523982167244, | |
| "learning_rate": 0.00029432098765432095, | |
| "loss": 0.5934, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.35333333333333333, | |
| "grad_norm": 0.07634396851062775, | |
| "learning_rate": 0.00029407407407407404, | |
| "loss": 0.7957, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.35555555555555557, | |
| "grad_norm": 0.07272932678461075, | |
| "learning_rate": 0.00029382716049382714, | |
| "loss": 0.6655, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.35777777777777775, | |
| "grad_norm": 0.059514183551073074, | |
| "learning_rate": 0.00029358024691358023, | |
| "loss": 0.6436, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.058966901153326035, | |
| "learning_rate": 0.00029333333333333327, | |
| "loss": 0.5661, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.3622222222222222, | |
| "grad_norm": 0.08566144853830338, | |
| "learning_rate": 0.0002930864197530864, | |
| "loss": 0.6154, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.36444444444444446, | |
| "grad_norm": 0.091421939432621, | |
| "learning_rate": 0.00029283950617283946, | |
| "loss": 0.7903, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.36666666666666664, | |
| "grad_norm": 0.06168731674551964, | |
| "learning_rate": 0.00029259259259259256, | |
| "loss": 0.6711, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3688888888888889, | |
| "grad_norm": 0.0587494820356369, | |
| "learning_rate": 0.00029234567901234565, | |
| "loss": 0.6277, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.3711111111111111, | |
| "grad_norm": 0.05911793187260628, | |
| "learning_rate": 0.00029209876543209875, | |
| "loss": 0.6871, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.37333333333333335, | |
| "grad_norm": 0.06315206736326218, | |
| "learning_rate": 0.00029185185185185184, | |
| "loss": 0.6521, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.37555555555555553, | |
| "grad_norm": 0.07214728742837906, | |
| "learning_rate": 0.0002916049382716049, | |
| "loss": 0.6457, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.37777777777777777, | |
| "grad_norm": 0.058888014405965805, | |
| "learning_rate": 0.000291358024691358, | |
| "loss": 0.6962, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.05860765278339386, | |
| "learning_rate": 0.0002911111111111111, | |
| "loss": 0.6141, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.38222222222222224, | |
| "grad_norm": 0.04580352082848549, | |
| "learning_rate": 0.00029086419753086417, | |
| "loss": 0.6696, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.3844444444444444, | |
| "grad_norm": 0.0666603222489357, | |
| "learning_rate": 0.00029061728395061726, | |
| "loss": 0.6104, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.38666666666666666, | |
| "grad_norm": 0.05495892092585564, | |
| "learning_rate": 0.00029037037037037036, | |
| "loss": 0.6219, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.3888888888888889, | |
| "grad_norm": 0.05430325120687485, | |
| "learning_rate": 0.00029012345679012345, | |
| "loss": 0.6611, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.39111111111111113, | |
| "grad_norm": 0.06881535798311234, | |
| "learning_rate": 0.0002898765432098765, | |
| "loss": 0.686, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.3933333333333333, | |
| "grad_norm": 0.06852874904870987, | |
| "learning_rate": 0.0002896296296296296, | |
| "loss": 0.7312, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.39555555555555555, | |
| "grad_norm": 0.07395198196172714, | |
| "learning_rate": 0.0002893827160493827, | |
| "loss": 0.6799, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.3977777777777778, | |
| "grad_norm": 0.07263553887605667, | |
| "learning_rate": 0.0002891358024691358, | |
| "loss": 0.7232, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.05263182893395424, | |
| "learning_rate": 0.0002888888888888888, | |
| "loss": 0.5936, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4022222222222222, | |
| "grad_norm": 0.047391194850206375, | |
| "learning_rate": 0.00028864197530864197, | |
| "loss": 0.5515, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.40444444444444444, | |
| "grad_norm": 0.056658029556274414, | |
| "learning_rate": 0.00028839506172839507, | |
| "loss": 0.6422, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.4066666666666667, | |
| "grad_norm": 0.053307462483644485, | |
| "learning_rate": 0.0002881481481481481, | |
| "loss": 0.6596, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.4088888888888889, | |
| "grad_norm": 0.05152831971645355, | |
| "learning_rate": 0.0002879012345679012, | |
| "loss": 0.6162, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.4111111111111111, | |
| "grad_norm": 0.05862154811620712, | |
| "learning_rate": 0.0002876543209876543, | |
| "loss": 0.6215, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.41333333333333333, | |
| "grad_norm": 0.05871991813182831, | |
| "learning_rate": 0.0002874074074074074, | |
| "loss": 0.6423, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.41555555555555557, | |
| "grad_norm": 0.0672457218170166, | |
| "learning_rate": 0.00028716049382716043, | |
| "loss": 0.6997, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.4177777777777778, | |
| "grad_norm": 0.059497442096471786, | |
| "learning_rate": 0.00028691358024691353, | |
| "loss": 0.7856, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.08486013859510422, | |
| "learning_rate": 0.0002866666666666667, | |
| "loss": 0.6585, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.4222222222222222, | |
| "grad_norm": 0.057826027274131775, | |
| "learning_rate": 0.0002864197530864197, | |
| "loss": 0.7507, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.42444444444444446, | |
| "grad_norm": 0.06144357845187187, | |
| "learning_rate": 0.0002861728395061728, | |
| "loss": 0.6617, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.4266666666666667, | |
| "grad_norm": 0.05544937029480934, | |
| "learning_rate": 0.0002859259259259259, | |
| "loss": 0.7459, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.4288888888888889, | |
| "grad_norm": 0.07672414183616638, | |
| "learning_rate": 0.000285679012345679, | |
| "loss": 0.7217, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.4311111111111111, | |
| "grad_norm": 0.07053379714488983, | |
| "learning_rate": 0.00028543209876543204, | |
| "loss": 0.7123, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.43333333333333335, | |
| "grad_norm": 0.05894757807254791, | |
| "learning_rate": 0.00028518518518518514, | |
| "loss": 0.6724, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.43555555555555553, | |
| "grad_norm": 0.06975255906581879, | |
| "learning_rate": 0.00028493827160493824, | |
| "loss": 0.6617, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.43777777777777777, | |
| "grad_norm": 0.06969016790390015, | |
| "learning_rate": 0.00028469135802469133, | |
| "loss": 0.6081, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.06051938980817795, | |
| "learning_rate": 0.0002844444444444444, | |
| "loss": 0.5882, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.44222222222222224, | |
| "grad_norm": 0.054278962314128876, | |
| "learning_rate": 0.0002841975308641975, | |
| "loss": 0.5356, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 0.0599193349480629, | |
| "learning_rate": 0.0002839506172839506, | |
| "loss": 0.6907, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.44666666666666666, | |
| "grad_norm": 0.1046207845211029, | |
| "learning_rate": 0.00028370370370370366, | |
| "loss": 0.5205, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.4488888888888889, | |
| "grad_norm": 0.05720556154847145, | |
| "learning_rate": 0.00028345679012345675, | |
| "loss": 0.6213, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.45111111111111113, | |
| "grad_norm": 0.07286486029624939, | |
| "learning_rate": 0.00028320987654320985, | |
| "loss": 0.7178, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.4533333333333333, | |
| "grad_norm": 0.057378239929676056, | |
| "learning_rate": 0.00028296296296296294, | |
| "loss": 0.5544, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.45555555555555555, | |
| "grad_norm": 0.0836499035358429, | |
| "learning_rate": 0.00028271604938271604, | |
| "loss": 0.6861, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.4577777777777778, | |
| "grad_norm": 0.060671668499708176, | |
| "learning_rate": 0.0002824691358024691, | |
| "loss": 0.5515, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.0478038564324379, | |
| "learning_rate": 0.00028222222222222223, | |
| "loss": 0.4805, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.4622222222222222, | |
| "grad_norm": 0.05711014196276665, | |
| "learning_rate": 0.00028197530864197527, | |
| "loss": 0.6406, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.46444444444444444, | |
| "grad_norm": 0.06043294817209244, | |
| "learning_rate": 0.00028172839506172836, | |
| "loss": 0.7391, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.4666666666666667, | |
| "grad_norm": 0.046299222856760025, | |
| "learning_rate": 0.00028148148148148146, | |
| "loss": 0.5542, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.4688888888888889, | |
| "grad_norm": 0.08454018831253052, | |
| "learning_rate": 0.00028123456790123455, | |
| "loss": 0.6525, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.4711111111111111, | |
| "grad_norm": 0.050603095442056656, | |
| "learning_rate": 0.00028098765432098765, | |
| "loss": 0.5568, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.47333333333333333, | |
| "grad_norm": 0.052967581897974014, | |
| "learning_rate": 0.0002807407407407407, | |
| "loss": 0.6809, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.47555555555555556, | |
| "grad_norm": 0.05539657548069954, | |
| "learning_rate": 0.0002804938271604938, | |
| "loss": 0.5629, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.4777777777777778, | |
| "grad_norm": 0.05519342049956322, | |
| "learning_rate": 0.0002802469135802469, | |
| "loss": 0.5795, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.0779273733496666, | |
| "learning_rate": 0.00028, | |
| "loss": 0.7076, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.4822222222222222, | |
| "grad_norm": 0.047700658440589905, | |
| "learning_rate": 0.00027975308641975307, | |
| "loss": 0.5524, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.48444444444444446, | |
| "grad_norm": 0.05553275719285011, | |
| "learning_rate": 0.00027950617283950617, | |
| "loss": 0.5827, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.4866666666666667, | |
| "grad_norm": 0.04649887606501579, | |
| "learning_rate": 0.00027925925925925926, | |
| "loss": 0.6617, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.4888888888888889, | |
| "grad_norm": 0.0619090273976326, | |
| "learning_rate": 0.0002790123456790123, | |
| "loss": 0.8373, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.4911111111111111, | |
| "grad_norm": 0.06357704102993011, | |
| "learning_rate": 0.0002787654320987654, | |
| "loss": 0.5782, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.49333333333333335, | |
| "grad_norm": 0.06721944361925125, | |
| "learning_rate": 0.0002785185185185185, | |
| "loss": 0.5343, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.4955555555555556, | |
| "grad_norm": 0.0589229092001915, | |
| "learning_rate": 0.0002782716049382716, | |
| "loss": 0.6475, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.49777777777777776, | |
| "grad_norm": 0.04312542825937271, | |
| "learning_rate": 0.00027802469135802463, | |
| "loss": 0.5189, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.0846293643116951, | |
| "learning_rate": 0.0002777777777777778, | |
| "loss": 0.6499, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5022222222222222, | |
| "grad_norm": 0.059431884437799454, | |
| "learning_rate": 0.00027753086419753087, | |
| "loss": 0.5793, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.5044444444444445, | |
| "grad_norm": 0.064137764275074, | |
| "learning_rate": 0.0002772839506172839, | |
| "loss": 0.7592, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.5066666666666667, | |
| "grad_norm": 0.06620758026838303, | |
| "learning_rate": 0.000277037037037037, | |
| "loss": 0.6093, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.5088888888888888, | |
| "grad_norm": 0.061502937227487564, | |
| "learning_rate": 0.0002767901234567901, | |
| "loss": 0.7111, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.5111111111111111, | |
| "grad_norm": 0.05012490227818489, | |
| "learning_rate": 0.0002765432098765432, | |
| "loss": 0.8083, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.5133333333333333, | |
| "grad_norm": 0.07317467033863068, | |
| "learning_rate": 0.00027629629629629624, | |
| "loss": 0.6893, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.5155555555555555, | |
| "grad_norm": 0.05900180712342262, | |
| "learning_rate": 0.00027604938271604934, | |
| "loss": 0.7233, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.5177777777777778, | |
| "grad_norm": 0.08215516805648804, | |
| "learning_rate": 0.00027580246913580243, | |
| "loss": 0.7605, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.06336875259876251, | |
| "learning_rate": 0.0002755555555555555, | |
| "loss": 0.6947, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.5222222222222223, | |
| "grad_norm": 0.05653493478894234, | |
| "learning_rate": 0.0002753086419753086, | |
| "loss": 0.5903, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.5244444444444445, | |
| "grad_norm": 0.05590517818927765, | |
| "learning_rate": 0.0002750617283950617, | |
| "loss": 0.5845, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.5266666666666666, | |
| "grad_norm": 0.06814274936914444, | |
| "learning_rate": 0.0002748148148148148, | |
| "loss": 0.8476, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.5288888888888889, | |
| "grad_norm": 0.05073162913322449, | |
| "learning_rate": 0.00027456790123456785, | |
| "loss": 0.5527, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.5311111111111111, | |
| "grad_norm": 0.07849614322185516, | |
| "learning_rate": 0.00027432098765432095, | |
| "loss": 0.6355, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.5333333333333333, | |
| "grad_norm": 0.04955737665295601, | |
| "learning_rate": 0.00027407407407407404, | |
| "loss": 0.6487, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.5355555555555556, | |
| "grad_norm": 0.061425868421792984, | |
| "learning_rate": 0.00027382716049382714, | |
| "loss": 0.4973, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.5377777777777778, | |
| "grad_norm": 0.07901917397975922, | |
| "learning_rate": 0.00027358024691358023, | |
| "loss": 0.7306, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.06443935632705688, | |
| "learning_rate": 0.00027333333333333333, | |
| "loss": 0.687, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.5422222222222223, | |
| "grad_norm": 0.070670947432518, | |
| "learning_rate": 0.0002730864197530864, | |
| "loss": 0.9579, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.5444444444444444, | |
| "grad_norm": 0.05495050176978111, | |
| "learning_rate": 0.00027283950617283946, | |
| "loss": 0.4887, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.5466666666666666, | |
| "grad_norm": 0.07778953015804291, | |
| "learning_rate": 0.00027259259259259256, | |
| "loss": 0.6945, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.5488888888888889, | |
| "grad_norm": 0.04747968167066574, | |
| "learning_rate": 0.00027234567901234565, | |
| "loss": 0.642, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.5511111111111111, | |
| "grad_norm": 0.060923799872398376, | |
| "learning_rate": 0.00027209876543209875, | |
| "loss": 0.6945, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.5533333333333333, | |
| "grad_norm": 0.047542307525873184, | |
| "learning_rate": 0.00027185185185185184, | |
| "loss": 0.4812, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 0.0694783627986908, | |
| "learning_rate": 0.0002716049382716049, | |
| "loss": 0.5538, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.5577777777777778, | |
| "grad_norm": 0.04970422014594078, | |
| "learning_rate": 0.00027135802469135803, | |
| "loss": 0.6646, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.07231730967760086, | |
| "learning_rate": 0.0002711111111111111, | |
| "loss": 0.7271, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.5622222222222222, | |
| "grad_norm": 0.055860117077827454, | |
| "learning_rate": 0.00027086419753086417, | |
| "loss": 0.5625, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.5644444444444444, | |
| "grad_norm": 0.06579416990280151, | |
| "learning_rate": 0.00027061728395061727, | |
| "loss": 0.6468, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.5666666666666667, | |
| "grad_norm": 0.06695139408111572, | |
| "learning_rate": 0.00027037037037037036, | |
| "loss": 0.5721, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.5688888888888889, | |
| "grad_norm": 0.07215917110443115, | |
| "learning_rate": 0.00027012345679012346, | |
| "loss": 0.6384, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.5711111111111111, | |
| "grad_norm": 0.11614429950714111, | |
| "learning_rate": 0.0002698765432098765, | |
| "loss": 0.7965, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.5733333333333334, | |
| "grad_norm": 0.06437898427248001, | |
| "learning_rate": 0.0002696296296296296, | |
| "loss": 0.575, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.5755555555555556, | |
| "grad_norm": 0.051053401082754135, | |
| "learning_rate": 0.0002693827160493827, | |
| "loss": 0.5971, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.5777777777777777, | |
| "grad_norm": 0.05523311719298363, | |
| "learning_rate": 0.0002691358024691358, | |
| "loss": 0.5934, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 0.05467135086655617, | |
| "learning_rate": 0.0002688888888888889, | |
| "loss": 0.569, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.5822222222222222, | |
| "grad_norm": 0.06978314369916916, | |
| "learning_rate": 0.00026864197530864197, | |
| "loss": 0.6239, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.5844444444444444, | |
| "grad_norm": 0.07661367952823639, | |
| "learning_rate": 0.000268395061728395, | |
| "loss": 0.8036, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.5866666666666667, | |
| "grad_norm": 0.06496507674455643, | |
| "learning_rate": 0.0002681481481481481, | |
| "loss": 0.7783, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.5888888888888889, | |
| "grad_norm": 0.05868912860751152, | |
| "learning_rate": 0.0002679012345679012, | |
| "loss": 0.6631, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.5911111111111111, | |
| "grad_norm": 0.09481733292341232, | |
| "learning_rate": 0.0002676543209876543, | |
| "loss": 0.6108, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.5933333333333334, | |
| "grad_norm": 0.06981168687343597, | |
| "learning_rate": 0.0002674074074074074, | |
| "loss": 0.5884, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.5955555555555555, | |
| "grad_norm": 0.05572844669222832, | |
| "learning_rate": 0.00026716049382716043, | |
| "loss": 0.7082, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.5977777777777777, | |
| "grad_norm": 0.08810676634311676, | |
| "learning_rate": 0.0002669135802469136, | |
| "loss": 0.7245, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.06569419801235199, | |
| "learning_rate": 0.0002666666666666666, | |
| "loss": 0.5489, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.6022222222222222, | |
| "grad_norm": 0.08496364206075668, | |
| "learning_rate": 0.0002664197530864197, | |
| "loss": 0.6375, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.6044444444444445, | |
| "grad_norm": 0.05873265862464905, | |
| "learning_rate": 0.0002661728395061728, | |
| "loss": 0.7489, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.6066666666666667, | |
| "grad_norm": 0.07921940833330154, | |
| "learning_rate": 0.0002659259259259259, | |
| "loss": 0.5877, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.6088888888888889, | |
| "grad_norm": 0.05226515978574753, | |
| "learning_rate": 0.000265679012345679, | |
| "loss": 0.5937, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.6111111111111112, | |
| "grad_norm": 0.06759991496801376, | |
| "learning_rate": 0.00026543209876543205, | |
| "loss": 0.6726, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.6133333333333333, | |
| "grad_norm": 0.06234094873070717, | |
| "learning_rate": 0.00026518518518518514, | |
| "loss": 0.6509, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.6155555555555555, | |
| "grad_norm": 0.07358703017234802, | |
| "learning_rate": 0.00026493827160493824, | |
| "loss": 0.5465, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.6177777777777778, | |
| "grad_norm": 0.07200343906879425, | |
| "learning_rate": 0.00026469135802469133, | |
| "loss": 0.5876, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.0803864523768425, | |
| "learning_rate": 0.00026444444444444443, | |
| "loss": 0.6187, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.6222222222222222, | |
| "grad_norm": 0.07174916565418243, | |
| "learning_rate": 0.0002641975308641975, | |
| "loss": 0.5527, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.6244444444444445, | |
| "grad_norm": 0.058686818927526474, | |
| "learning_rate": 0.0002639506172839506, | |
| "loss": 0.64, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.6266666666666667, | |
| "grad_norm": 0.050265416502952576, | |
| "learning_rate": 0.00026370370370370366, | |
| "loss": 0.5849, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.6288888888888889, | |
| "grad_norm": 0.0491122230887413, | |
| "learning_rate": 0.00026345679012345675, | |
| "loss": 0.5673, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.6311111111111111, | |
| "grad_norm": 0.0580422468483448, | |
| "learning_rate": 0.00026320987654320985, | |
| "loss": 0.6451, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.6333333333333333, | |
| "grad_norm": 0.04680923372507095, | |
| "learning_rate": 0.00026296296296296294, | |
| "loss": 0.6308, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.6355555555555555, | |
| "grad_norm": 0.06512854248285294, | |
| "learning_rate": 0.00026271604938271604, | |
| "loss": 0.6047, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.6377777777777778, | |
| "grad_norm": 0.04747498780488968, | |
| "learning_rate": 0.00026246913580246913, | |
| "loss": 0.5444, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.05045843496918678, | |
| "learning_rate": 0.00026222222222222223, | |
| "loss": 0.5943, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.6422222222222222, | |
| "grad_norm": 0.06047791615128517, | |
| "learning_rate": 0.00026197530864197527, | |
| "loss": 0.6508, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.6444444444444445, | |
| "grad_norm": 0.05897914618253708, | |
| "learning_rate": 0.00026172839506172837, | |
| "loss": 0.6656, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.6466666666666666, | |
| "grad_norm": 0.06382246315479279, | |
| "learning_rate": 0.00026148148148148146, | |
| "loss": 0.6152, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.6488888888888888, | |
| "grad_norm": 0.05804605409502983, | |
| "learning_rate": 0.00026123456790123456, | |
| "loss": 0.7588, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.6511111111111111, | |
| "grad_norm": 0.06460540741682053, | |
| "learning_rate": 0.0002609876543209876, | |
| "loss": 0.7051, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.6533333333333333, | |
| "grad_norm": 0.05697052925825119, | |
| "learning_rate": 0.0002607407407407407, | |
| "loss": 0.6507, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.6555555555555556, | |
| "grad_norm": 0.08563496172428131, | |
| "learning_rate": 0.00026049382716049384, | |
| "loss": 0.5777, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.6577777777777778, | |
| "grad_norm": 0.053184833377599716, | |
| "learning_rate": 0.0002602469135802469, | |
| "loss": 0.5745, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.07504092901945114, | |
| "learning_rate": 0.00026, | |
| "loss": 0.5631, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.6622222222222223, | |
| "grad_norm": 0.060315705835819244, | |
| "learning_rate": 0.00025975308641975307, | |
| "loss": 0.5452, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.6644444444444444, | |
| "grad_norm": 0.0928531289100647, | |
| "learning_rate": 0.00025950617283950617, | |
| "loss": 0.7384, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.05180056020617485, | |
| "learning_rate": 0.0002592592592592592, | |
| "loss": 0.4523, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.6688888888888889, | |
| "grad_norm": 0.0784502848982811, | |
| "learning_rate": 0.0002590123456790123, | |
| "loss": 0.8033, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.6711111111111111, | |
| "grad_norm": 0.06314003467559814, | |
| "learning_rate": 0.0002587654320987654, | |
| "loss": 0.6854, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.6733333333333333, | |
| "grad_norm": 0.0726189836859703, | |
| "learning_rate": 0.0002585185185185185, | |
| "loss": 0.6991, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.6755555555555556, | |
| "grad_norm": 0.08663801848888397, | |
| "learning_rate": 0.0002582716049382716, | |
| "loss": 0.6959, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.6777777777777778, | |
| "grad_norm": 0.04849329963326454, | |
| "learning_rate": 0.0002580246913580247, | |
| "loss": 0.5358, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.05882357433438301, | |
| "learning_rate": 0.0002577777777777778, | |
| "loss": 0.5991, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.6822222222222222, | |
| "grad_norm": 0.04300856962800026, | |
| "learning_rate": 0.0002575308641975308, | |
| "loss": 0.5293, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.6844444444444444, | |
| "grad_norm": 0.058613792061805725, | |
| "learning_rate": 0.0002572839506172839, | |
| "loss": 0.6661, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.6866666666666666, | |
| "grad_norm": 0.04765408858656883, | |
| "learning_rate": 0.000257037037037037, | |
| "loss": 0.5902, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.6888888888888889, | |
| "grad_norm": 0.06682875007390976, | |
| "learning_rate": 0.0002567901234567901, | |
| "loss": 0.6213, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.6911111111111111, | |
| "grad_norm": 0.05194618180394173, | |
| "learning_rate": 0.0002565432098765432, | |
| "loss": 0.5554, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.6933333333333334, | |
| "grad_norm": 0.06991297751665115, | |
| "learning_rate": 0.00025629629629629624, | |
| "loss": 0.5905, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.6955555555555556, | |
| "grad_norm": 0.07291572540998459, | |
| "learning_rate": 0.0002560493827160494, | |
| "loss": 0.6945, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.6977777777777778, | |
| "grad_norm": 0.056158553808927536, | |
| "learning_rate": 0.00025580246913580243, | |
| "loss": 0.6289, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.056562066078186035, | |
| "learning_rate": 0.00025555555555555553, | |
| "loss": 0.5093, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.7022222222222222, | |
| "grad_norm": 0.0728093609213829, | |
| "learning_rate": 0.0002553086419753086, | |
| "loss": 0.5912, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.7044444444444444, | |
| "grad_norm": 0.0464496985077858, | |
| "learning_rate": 0.0002550617283950617, | |
| "loss": 0.6498, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.7066666666666667, | |
| "grad_norm": 0.05788656324148178, | |
| "learning_rate": 0.0002548148148148148, | |
| "loss": 0.6619, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.7088888888888889, | |
| "grad_norm": 0.07232210785150528, | |
| "learning_rate": 0.00025456790123456785, | |
| "loss": 0.6081, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.7111111111111111, | |
| "grad_norm": 0.04878592863678932, | |
| "learning_rate": 0.00025432098765432095, | |
| "loss": 0.5144, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.7133333333333334, | |
| "grad_norm": 0.062463223934173584, | |
| "learning_rate": 0.00025407407407407404, | |
| "loss": 0.6477, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.7155555555555555, | |
| "grad_norm": 0.0489543080329895, | |
| "learning_rate": 0.00025382716049382714, | |
| "loss": 0.5673, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.7177777777777777, | |
| "grad_norm": 0.05130515620112419, | |
| "learning_rate": 0.00025358024691358023, | |
| "loss": 0.5205, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.09179674088954926, | |
| "learning_rate": 0.00025333333333333333, | |
| "loss": 0.7485, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.7222222222222222, | |
| "grad_norm": 0.05300341546535492, | |
| "learning_rate": 0.0002530864197530864, | |
| "loss": 0.6421, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.7244444444444444, | |
| "grad_norm": 0.05454908311367035, | |
| "learning_rate": 0.00025283950617283947, | |
| "loss": 0.5707, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.7266666666666667, | |
| "grad_norm": 0.06438113749027252, | |
| "learning_rate": 0.00025259259259259256, | |
| "loss": 0.5568, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.7288888888888889, | |
| "grad_norm": 0.052538372576236725, | |
| "learning_rate": 0.00025234567901234566, | |
| "loss": 0.589, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.7311111111111112, | |
| "grad_norm": 0.07397031784057617, | |
| "learning_rate": 0.00025209876543209875, | |
| "loss": 0.5477, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.7333333333333333, | |
| "grad_norm": 0.06466836482286453, | |
| "learning_rate": 0.0002518518518518518, | |
| "loss": 0.6557, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.7355555555555555, | |
| "grad_norm": 0.10031090676784515, | |
| "learning_rate": 0.00025160493827160494, | |
| "loss": 0.6862, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.7377777777777778, | |
| "grad_norm": 0.061177365481853485, | |
| "learning_rate": 0.00025135802469135804, | |
| "loss": 0.6014, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.0572127141058445, | |
| "learning_rate": 0.0002511111111111111, | |
| "loss": 0.5435, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.7422222222222222, | |
| "grad_norm": 0.06117907911539078, | |
| "learning_rate": 0.00025086419753086417, | |
| "loss": 0.5665, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.7444444444444445, | |
| "grad_norm": 0.053214021027088165, | |
| "learning_rate": 0.00025061728395061727, | |
| "loss": 0.587, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.7466666666666667, | |
| "grad_norm": 0.06741708517074585, | |
| "learning_rate": 0.00025037037037037036, | |
| "loss": 0.6778, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.7488888888888889, | |
| "grad_norm": 0.06096634268760681, | |
| "learning_rate": 0.0002501234567901234, | |
| "loss": 0.7982, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.7511111111111111, | |
| "grad_norm": 0.05581033229827881, | |
| "learning_rate": 0.0002498765432098765, | |
| "loss": 0.5297, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.7533333333333333, | |
| "grad_norm": 0.07700909674167633, | |
| "learning_rate": 0.00024962962962962965, | |
| "loss": 0.7023, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.7555555555555555, | |
| "grad_norm": 0.04708324000239372, | |
| "learning_rate": 0.0002493827160493827, | |
| "loss": 0.6108, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.7577777777777778, | |
| "grad_norm": 0.05553693324327469, | |
| "learning_rate": 0.0002491358024691358, | |
| "loss": 0.5974, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.0789913535118103, | |
| "learning_rate": 0.0002488888888888889, | |
| "loss": 0.6418, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.7622222222222222, | |
| "grad_norm": 0.057143647223711014, | |
| "learning_rate": 0.000248641975308642, | |
| "loss": 0.6978, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.7644444444444445, | |
| "grad_norm": 0.05981820449233055, | |
| "learning_rate": 0.000248395061728395, | |
| "loss": 0.7412, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.7666666666666667, | |
| "grad_norm": 0.07758378982543945, | |
| "learning_rate": 0.0002481481481481481, | |
| "loss": 0.5357, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.7688888888888888, | |
| "grad_norm": 0.0765530914068222, | |
| "learning_rate": 0.0002479012345679012, | |
| "loss": 0.6916, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.7711111111111111, | |
| "grad_norm": 0.07973485440015793, | |
| "learning_rate": 0.0002476543209876543, | |
| "loss": 0.675, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.7733333333333333, | |
| "grad_norm": 0.07663662731647491, | |
| "learning_rate": 0.0002474074074074074, | |
| "loss": 0.5767, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.7755555555555556, | |
| "grad_norm": 0.044524773955345154, | |
| "learning_rate": 0.0002471604938271605, | |
| "loss": 0.6424, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.7777777777777778, | |
| "grad_norm": 0.06698901206254959, | |
| "learning_rate": 0.0002469135802469136, | |
| "loss": 0.5833, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.04637622833251953, | |
| "learning_rate": 0.0002466666666666666, | |
| "loss": 0.4859, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.7822222222222223, | |
| "grad_norm": 0.06135889142751694, | |
| "learning_rate": 0.0002464197530864197, | |
| "loss": 0.6461, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.7844444444444445, | |
| "grad_norm": 0.05274899676442146, | |
| "learning_rate": 0.0002461728395061728, | |
| "loss": 0.5364, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.7866666666666666, | |
| "grad_norm": 0.04430905729532242, | |
| "learning_rate": 0.0002459259259259259, | |
| "loss": 0.5564, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.7888888888888889, | |
| "grad_norm": 0.08678912371397018, | |
| "learning_rate": 0.000245679012345679, | |
| "loss": 0.6058, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.7911111111111111, | |
| "grad_norm": 0.07081649452447891, | |
| "learning_rate": 0.00024543209876543205, | |
| "loss": 0.6037, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.7933333333333333, | |
| "grad_norm": 0.054697491228580475, | |
| "learning_rate": 0.0002451851851851852, | |
| "loss": 0.6191, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.7955555555555556, | |
| "grad_norm": 0.05636013299226761, | |
| "learning_rate": 0.00024493827160493824, | |
| "loss": 0.5038, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.7977777777777778, | |
| "grad_norm": 0.05813833326101303, | |
| "learning_rate": 0.00024469135802469133, | |
| "loss": 0.5779, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.08280492573976517, | |
| "learning_rate": 0.00024444444444444443, | |
| "loss": 0.8622, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.8022222222222222, | |
| "grad_norm": 0.07269857823848724, | |
| "learning_rate": 0.0002441975308641975, | |
| "loss": 0.5954, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.8044444444444444, | |
| "grad_norm": 0.05752701312303543, | |
| "learning_rate": 0.00024395061728395062, | |
| "loss": 0.6726, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.8066666666666666, | |
| "grad_norm": 0.05924740061163902, | |
| "learning_rate": 0.0002437037037037037, | |
| "loss": 0.5368, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.8088888888888889, | |
| "grad_norm": 0.09675303101539612, | |
| "learning_rate": 0.00024345679012345678, | |
| "loss": 0.7003, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.8111111111111111, | |
| "grad_norm": 0.0829503983259201, | |
| "learning_rate": 0.00024320987654320985, | |
| "loss": 0.6148, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.8133333333333334, | |
| "grad_norm": 0.04949120804667473, | |
| "learning_rate": 0.00024296296296296295, | |
| "loss": 0.5993, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.8155555555555556, | |
| "grad_norm": 0.0907658115029335, | |
| "learning_rate": 0.00024271604938271601, | |
| "loss": 0.7222, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.8177777777777778, | |
| "grad_norm": 0.0548752024769783, | |
| "learning_rate": 0.0002424691358024691, | |
| "loss": 0.5262, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.05041273310780525, | |
| "learning_rate": 0.0002422222222222222, | |
| "loss": 0.5163, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.8222222222222222, | |
| "grad_norm": 0.06130453571677208, | |
| "learning_rate": 0.00024197530864197527, | |
| "loss": 0.5394, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.8244444444444444, | |
| "grad_norm": 0.0681861937046051, | |
| "learning_rate": 0.0002417283950617284, | |
| "loss": 0.6466, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.8266666666666667, | |
| "grad_norm": 0.0512981116771698, | |
| "learning_rate": 0.00024148148148148146, | |
| "loss": 0.5321, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.8288888888888889, | |
| "grad_norm": 0.06226026266813278, | |
| "learning_rate": 0.00024123456790123456, | |
| "loss": 0.5056, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.8311111111111111, | |
| "grad_norm": 0.05910397320985794, | |
| "learning_rate": 0.00024098765432098763, | |
| "loss": 0.6231, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 0.060540203005075455, | |
| "learning_rate": 0.00024074074074074072, | |
| "loss": 0.5323, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.8355555555555556, | |
| "grad_norm": 0.061202920973300934, | |
| "learning_rate": 0.0002404938271604938, | |
| "loss": 0.5317, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.8377777777777777, | |
| "grad_norm": 0.05394279956817627, | |
| "learning_rate": 0.00024024691358024688, | |
| "loss": 0.5134, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.04831009730696678, | |
| "learning_rate": 0.00023999999999999998, | |
| "loss": 0.5485, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.8422222222222222, | |
| "grad_norm": 0.05570105463266373, | |
| "learning_rate": 0.00023975308641975305, | |
| "loss": 0.6335, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.8444444444444444, | |
| "grad_norm": 0.057373929768800735, | |
| "learning_rate": 0.00023950617283950617, | |
| "loss": 0.6775, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.8466666666666667, | |
| "grad_norm": 0.0687582939863205, | |
| "learning_rate": 0.00023925925925925924, | |
| "loss": 0.5954, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.8488888888888889, | |
| "grad_norm": 0.08149206638336182, | |
| "learning_rate": 0.00023901234567901233, | |
| "loss": 0.6245, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.8511111111111112, | |
| "grad_norm": 0.07159999012947083, | |
| "learning_rate": 0.0002387654320987654, | |
| "loss": 0.7488, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.8533333333333334, | |
| "grad_norm": 0.06775209307670593, | |
| "learning_rate": 0.0002385185185185185, | |
| "loss": 0.6536, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.8555555555555555, | |
| "grad_norm": 0.05568854138255119, | |
| "learning_rate": 0.0002382716049382716, | |
| "loss": 0.4267, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.8577777777777778, | |
| "grad_norm": 0.05042643845081329, | |
| "learning_rate": 0.00023802469135802466, | |
| "loss": 0.5114, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.05210336297750473, | |
| "learning_rate": 0.00023777777777777775, | |
| "loss": 0.7252, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.8622222222222222, | |
| "grad_norm": 0.07702841609716415, | |
| "learning_rate": 0.00023753086419753082, | |
| "loss": 0.6754, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.8644444444444445, | |
| "grad_norm": 0.04862853139638901, | |
| "learning_rate": 0.00023728395061728394, | |
| "loss": 0.6162, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.8666666666666667, | |
| "grad_norm": 0.06977728009223938, | |
| "learning_rate": 0.000237037037037037, | |
| "loss": 0.6521, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.8688888888888889, | |
| "grad_norm": 0.06286180019378662, | |
| "learning_rate": 0.0002367901234567901, | |
| "loss": 0.6411, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.8711111111111111, | |
| "grad_norm": 0.06280764937400818, | |
| "learning_rate": 0.0002365432098765432, | |
| "loss": 0.5349, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.8733333333333333, | |
| "grad_norm": 0.05416274443268776, | |
| "learning_rate": 0.00023629629629629627, | |
| "loss": 0.5094, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.8755555555555555, | |
| "grad_norm": 0.06306446343660355, | |
| "learning_rate": 0.00023604938271604937, | |
| "loss": 0.6075, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.8777777777777778, | |
| "grad_norm": 0.06415388733148575, | |
| "learning_rate": 0.00023580246913580243, | |
| "loss": 0.5358, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.11125559359788895, | |
| "learning_rate": 0.00023555555555555553, | |
| "loss": 0.6107, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.8822222222222222, | |
| "grad_norm": 0.07780521363019943, | |
| "learning_rate": 0.0002353086419753086, | |
| "loss": 0.6397, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.8844444444444445, | |
| "grad_norm": 0.06937316805124283, | |
| "learning_rate": 0.00023506172839506172, | |
| "loss": 0.6575, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.8866666666666667, | |
| "grad_norm": 0.1283538043498993, | |
| "learning_rate": 0.0002348148148148148, | |
| "loss": 0.5819, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.8888888888888888, | |
| "grad_norm": 0.0543513298034668, | |
| "learning_rate": 0.00023456790123456788, | |
| "loss": 0.5547, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.8911111111111111, | |
| "grad_norm": 0.054155413061380386, | |
| "learning_rate": 0.00023432098765432098, | |
| "loss": 0.4027, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.8933333333333333, | |
| "grad_norm": 0.06638886779546738, | |
| "learning_rate": 0.00023407407407407405, | |
| "loss": 0.6455, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.8955555555555555, | |
| "grad_norm": 0.12468775361776352, | |
| "learning_rate": 0.00023382716049382714, | |
| "loss": 0.6439, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.8977777777777778, | |
| "grad_norm": 0.04853810369968414, | |
| "learning_rate": 0.0002335802469135802, | |
| "loss": 0.5014, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.17347636818885803, | |
| "learning_rate": 0.0002333333333333333, | |
| "loss": 0.7411, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.9022222222222223, | |
| "grad_norm": 0.11848603188991547, | |
| "learning_rate": 0.00023308641975308637, | |
| "loss": 0.6649, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.9044444444444445, | |
| "grad_norm": 0.059788476675748825, | |
| "learning_rate": 0.0002328395061728395, | |
| "loss": 0.4632, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.9066666666666666, | |
| "grad_norm": 0.06635454297065735, | |
| "learning_rate": 0.0002325925925925926, | |
| "loss": 0.6958, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.9088888888888889, | |
| "grad_norm": 0.057197559624910355, | |
| "learning_rate": 0.00023234567901234566, | |
| "loss": 0.6016, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.9111111111111111, | |
| "grad_norm": 0.049528613686561584, | |
| "learning_rate": 0.00023209876543209875, | |
| "loss": 0.6648, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.9133333333333333, | |
| "grad_norm": 0.06839912384748459, | |
| "learning_rate": 0.00023185185185185182, | |
| "loss": 0.5886, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.9155555555555556, | |
| "grad_norm": 0.06785445660352707, | |
| "learning_rate": 0.00023160493827160492, | |
| "loss": 0.615, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.9177777777777778, | |
| "grad_norm": 0.043476469814777374, | |
| "learning_rate": 0.00023135802469135798, | |
| "loss": 0.4486, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.06550905853509903, | |
| "learning_rate": 0.00023111111111111108, | |
| "loss": 0.6641, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.9222222222222223, | |
| "grad_norm": 0.0671965703368187, | |
| "learning_rate": 0.0002308641975308642, | |
| "loss": 0.6494, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.9244444444444444, | |
| "grad_norm": 0.07037396728992462, | |
| "learning_rate": 0.00023061728395061727, | |
| "loss": 0.6273, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.9266666666666666, | |
| "grad_norm": 0.04188195616006851, | |
| "learning_rate": 0.00023037037037037036, | |
| "loss": 0.5256, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.9288888888888889, | |
| "grad_norm": 0.05789446830749512, | |
| "learning_rate": 0.00023012345679012343, | |
| "loss": 0.5553, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.9311111111111111, | |
| "grad_norm": 0.047859832644462585, | |
| "learning_rate": 0.00022987654320987653, | |
| "loss": 0.5206, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.9333333333333333, | |
| "grad_norm": 0.08361068367958069, | |
| "learning_rate": 0.0002296296296296296, | |
| "loss": 0.6268, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.9355555555555556, | |
| "grad_norm": 0.08475058525800705, | |
| "learning_rate": 0.0002293827160493827, | |
| "loss": 0.6478, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.9377777777777778, | |
| "grad_norm": 0.042017385363578796, | |
| "learning_rate": 0.00022913580246913579, | |
| "loss": 0.5547, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.08007270842790604, | |
| "learning_rate": 0.00022888888888888885, | |
| "loss": 0.767, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.9422222222222222, | |
| "grad_norm": 0.052083924412727356, | |
| "learning_rate": 0.00022864197530864198, | |
| "loss": 0.4849, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.9444444444444444, | |
| "grad_norm": 0.053211141377687454, | |
| "learning_rate": 0.00022839506172839504, | |
| "loss": 0.5743, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.9466666666666667, | |
| "grad_norm": 0.07235322892665863, | |
| "learning_rate": 0.00022814814814814814, | |
| "loss": 0.6326, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.9488888888888889, | |
| "grad_norm": 0.08066584169864655, | |
| "learning_rate": 0.0002279012345679012, | |
| "loss": 0.4816, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.9511111111111111, | |
| "grad_norm": 0.0743255764245987, | |
| "learning_rate": 0.0002276543209876543, | |
| "loss": 0.6265, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.9533333333333334, | |
| "grad_norm": 0.048771534115076065, | |
| "learning_rate": 0.00022740740740740737, | |
| "loss": 0.5854, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.9555555555555556, | |
| "grad_norm": 0.0896950364112854, | |
| "learning_rate": 0.00022716049382716047, | |
| "loss": 0.5222, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.9577777777777777, | |
| "grad_norm": 0.05134379863739014, | |
| "learning_rate": 0.00022691358024691356, | |
| "loss": 0.5088, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.07395590096712112, | |
| "learning_rate": 0.00022666666666666663, | |
| "loss": 0.673, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.9622222222222222, | |
| "grad_norm": 0.11836867779493332, | |
| "learning_rate": 0.00022641975308641975, | |
| "loss": 0.7372, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.9644444444444444, | |
| "grad_norm": 0.07460740953683853, | |
| "learning_rate": 0.00022617283950617282, | |
| "loss": 0.6902, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.9666666666666667, | |
| "grad_norm": 0.07469192892313004, | |
| "learning_rate": 0.00022592592592592591, | |
| "loss": 0.8642, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.9688888888888889, | |
| "grad_norm": 0.06623450666666031, | |
| "learning_rate": 0.00022567901234567898, | |
| "loss": 0.5888, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.9711111111111111, | |
| "grad_norm": 0.09601905941963196, | |
| "learning_rate": 0.00022543209876543208, | |
| "loss": 0.5912, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.9733333333333334, | |
| "grad_norm": 0.06865989416837692, | |
| "learning_rate": 0.00022518518518518517, | |
| "loss": 0.6455, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.9755555555555555, | |
| "grad_norm": 0.05189220979809761, | |
| "learning_rate": 0.00022493827160493824, | |
| "loss": 0.5397, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.9777777777777777, | |
| "grad_norm": 0.05525451898574829, | |
| "learning_rate": 0.00022469135802469134, | |
| "loss": 0.6784, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 0.057762518525123596, | |
| "learning_rate": 0.0002244444444444444, | |
| "loss": 0.4479, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.9822222222222222, | |
| "grad_norm": 0.060063641518354416, | |
| "learning_rate": 0.00022419753086419753, | |
| "loss": 0.64, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.9844444444444445, | |
| "grad_norm": 0.06219639629125595, | |
| "learning_rate": 0.0002239506172839506, | |
| "loss": 0.5714, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.9866666666666667, | |
| "grad_norm": 0.06326323747634888, | |
| "learning_rate": 0.0002237037037037037, | |
| "loss": 0.6054, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.9888888888888889, | |
| "grad_norm": 0.04752006009221077, | |
| "learning_rate": 0.00022345679012345678, | |
| "loss": 0.5744, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.9911111111111112, | |
| "grad_norm": 0.056707534939050674, | |
| "learning_rate": 0.00022320987654320985, | |
| "loss": 0.6122, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.9933333333333333, | |
| "grad_norm": 0.06692970544099808, | |
| "learning_rate": 0.00022296296296296295, | |
| "loss": 0.6779, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.9955555555555555, | |
| "grad_norm": 0.050331197679042816, | |
| "learning_rate": 0.00022271604938271602, | |
| "loss": 0.4721, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.9977777777777778, | |
| "grad_norm": 0.06355144828557968, | |
| "learning_rate": 0.0002224691358024691, | |
| "loss": 0.564, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.04859111085534096, | |
| "learning_rate": 0.00022222222222222218, | |
| "loss": 0.5267, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.0022222222222221, | |
| "grad_norm": 0.07078340649604797, | |
| "learning_rate": 0.0002219753086419753, | |
| "loss": 0.5857, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 1.0044444444444445, | |
| "grad_norm": 0.061178382486104965, | |
| "learning_rate": 0.0002217283950617284, | |
| "loss": 0.5948, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 1.0066666666666666, | |
| "grad_norm": 0.07282081246376038, | |
| "learning_rate": 0.00022148148148148146, | |
| "loss": 0.5487, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 1.008888888888889, | |
| "grad_norm": 0.08245845139026642, | |
| "learning_rate": 0.00022123456790123456, | |
| "loss": 0.6801, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 1.011111111111111, | |
| "grad_norm": 0.06969266384840012, | |
| "learning_rate": 0.00022098765432098763, | |
| "loss": 0.7193, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 1.0133333333333334, | |
| "grad_norm": 0.05709267035126686, | |
| "learning_rate": 0.00022074074074074072, | |
| "loss": 0.5754, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 1.0155555555555555, | |
| "grad_norm": 0.06781706213951111, | |
| "learning_rate": 0.0002204938271604938, | |
| "loss": 0.5097, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 1.0177777777777777, | |
| "grad_norm": 0.05709918588399887, | |
| "learning_rate": 0.00022024691358024689, | |
| "loss": 0.5642, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "grad_norm": 0.07334735989570618, | |
| "learning_rate": 0.00021999999999999995, | |
| "loss": 0.6115, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 1.0222222222222221, | |
| "grad_norm": 0.07981806248426437, | |
| "learning_rate": 0.00021975308641975308, | |
| "loss": 0.7185, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 1.0244444444444445, | |
| "grad_norm": 0.05394122749567032, | |
| "learning_rate": 0.00021950617283950617, | |
| "loss": 0.5799, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 1.0266666666666666, | |
| "grad_norm": 0.058062098920345306, | |
| "learning_rate": 0.00021925925925925924, | |
| "loss": 0.5288, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 1.028888888888889, | |
| "grad_norm": 0.1541467308998108, | |
| "learning_rate": 0.00021901234567901233, | |
| "loss": 0.5756, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 1.031111111111111, | |
| "grad_norm": 0.06938309967517853, | |
| "learning_rate": 0.0002187654320987654, | |
| "loss": 0.6833, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 1.0333333333333334, | |
| "grad_norm": 0.08871179074048996, | |
| "learning_rate": 0.0002185185185185185, | |
| "loss": 0.7262, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 1.0355555555555556, | |
| "grad_norm": 0.06083621084690094, | |
| "learning_rate": 0.00021827160493827157, | |
| "loss": 0.5472, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 1.0377777777777777, | |
| "grad_norm": 0.06976111978292465, | |
| "learning_rate": 0.00021802469135802466, | |
| "loss": 0.691, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "grad_norm": 0.12717045843601227, | |
| "learning_rate": 0.00021777777777777778, | |
| "loss": 0.6161, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 1.0422222222222222, | |
| "grad_norm": 0.06936180591583252, | |
| "learning_rate": 0.00021753086419753085, | |
| "loss": 0.6569, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 1.0444444444444445, | |
| "grad_norm": 0.06263519823551178, | |
| "learning_rate": 0.00021728395061728395, | |
| "loss": 0.5334, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 1.0466666666666666, | |
| "grad_norm": 0.08511345088481903, | |
| "learning_rate": 0.00021703703703703701, | |
| "loss": 0.5885, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 1.048888888888889, | |
| "grad_norm": 0.06053664907813072, | |
| "learning_rate": 0.0002167901234567901, | |
| "loss": 0.6264, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 1.051111111111111, | |
| "grad_norm": 0.05959416553378105, | |
| "learning_rate": 0.00021654320987654318, | |
| "loss": 0.5635, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 1.0533333333333332, | |
| "grad_norm": 0.06442887336015701, | |
| "learning_rate": 0.00021629629629629627, | |
| "loss": 0.5492, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 1.0555555555555556, | |
| "grad_norm": 0.06876346468925476, | |
| "learning_rate": 0.00021604938271604937, | |
| "loss": 0.6686, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 1.0577777777777777, | |
| "grad_norm": 0.049920275807380676, | |
| "learning_rate": 0.00021580246913580244, | |
| "loss": 0.605, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "grad_norm": 0.06928721815347672, | |
| "learning_rate": 0.00021555555555555556, | |
| "loss": 0.5062, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 1.0622222222222222, | |
| "grad_norm": 0.05916568636894226, | |
| "learning_rate": 0.00021530864197530863, | |
| "loss": 0.4742, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 1.0644444444444445, | |
| "grad_norm": 0.053693048655986786, | |
| "learning_rate": 0.00021506172839506172, | |
| "loss": 0.6089, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 1.0666666666666667, | |
| "grad_norm": 0.10113474726676941, | |
| "learning_rate": 0.0002148148148148148, | |
| "loss": 0.5749, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 1.068888888888889, | |
| "grad_norm": 0.05808864161372185, | |
| "learning_rate": 0.00021456790123456788, | |
| "loss": 0.6241, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 1.0711111111111111, | |
| "grad_norm": 0.059021301567554474, | |
| "learning_rate": 0.00021432098765432095, | |
| "loss": 0.5746, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 1.0733333333333333, | |
| "grad_norm": 0.06783812493085861, | |
| "learning_rate": 0.00021407407407407405, | |
| "loss": 0.593, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 1.0755555555555556, | |
| "grad_norm": 0.05687486007809639, | |
| "learning_rate": 0.00021382716049382714, | |
| "loss": 0.6513, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 1.0777777777777777, | |
| "grad_norm": 0.06547362357378006, | |
| "learning_rate": 0.0002135802469135802, | |
| "loss": 0.5432, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "grad_norm": 0.06113448739051819, | |
| "learning_rate": 0.00021333333333333333, | |
| "loss": 0.6199, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 1.0822222222222222, | |
| "grad_norm": 0.055431269109249115, | |
| "learning_rate": 0.0002130864197530864, | |
| "loss": 0.5542, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 1.0844444444444445, | |
| "grad_norm": 0.06054743006825447, | |
| "learning_rate": 0.0002128395061728395, | |
| "loss": 0.5033, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 1.0866666666666667, | |
| "grad_norm": 0.0743783488869667, | |
| "learning_rate": 0.00021259259259259256, | |
| "loss": 0.5311, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 1.0888888888888888, | |
| "grad_norm": 0.05679158866405487, | |
| "learning_rate": 0.00021234567901234566, | |
| "loss": 0.4519, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 1.0911111111111111, | |
| "grad_norm": 0.05509325861930847, | |
| "learning_rate": 0.00021209876543209875, | |
| "loss": 0.6373, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 1.0933333333333333, | |
| "grad_norm": 0.07969211786985397, | |
| "learning_rate": 0.00021185185185185182, | |
| "loss": 0.6296, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 1.0955555555555556, | |
| "grad_norm": 0.06158541887998581, | |
| "learning_rate": 0.00021160493827160492, | |
| "loss": 0.7126, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 1.0977777777777777, | |
| "grad_norm": 0.06029195338487625, | |
| "learning_rate": 0.00021135802469135799, | |
| "loss": 0.5298, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "grad_norm": 0.05442732945084572, | |
| "learning_rate": 0.0002111111111111111, | |
| "loss": 0.5408, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 1.1022222222222222, | |
| "grad_norm": 0.057536616921424866, | |
| "learning_rate": 0.00021086419753086418, | |
| "loss": 0.6603, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 1.1044444444444443, | |
| "grad_norm": 0.062051836401224136, | |
| "learning_rate": 0.00021061728395061727, | |
| "loss": 0.5411, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 1.1066666666666667, | |
| "grad_norm": 0.06361360847949982, | |
| "learning_rate": 0.00021037037037037037, | |
| "loss": 0.5855, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 1.1088888888888888, | |
| "grad_norm": 0.05750907585024834, | |
| "learning_rate": 0.00021012345679012343, | |
| "loss": 0.5023, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 0.05303119495511055, | |
| "learning_rate": 0.00020987654320987653, | |
| "loss": 0.7192, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.1133333333333333, | |
| "grad_norm": 0.05355321615934372, | |
| "learning_rate": 0.0002096296296296296, | |
| "loss": 0.5047, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 1.1155555555555556, | |
| "grad_norm": 0.04647836089134216, | |
| "learning_rate": 0.0002093827160493827, | |
| "loss": 0.5495, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 1.1177777777777778, | |
| "grad_norm": 0.06943508982658386, | |
| "learning_rate": 0.00020913580246913576, | |
| "loss": 0.5198, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "grad_norm": 0.07622205466032028, | |
| "learning_rate": 0.00020888888888888888, | |
| "loss": 0.6394, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 1.1222222222222222, | |
| "grad_norm": 0.0701436847448349, | |
| "learning_rate": 0.00020864197530864198, | |
| "loss": 0.6032, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 1.1244444444444444, | |
| "grad_norm": 0.06063472479581833, | |
| "learning_rate": 0.00020839506172839505, | |
| "loss": 0.5935, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 1.1266666666666667, | |
| "grad_norm": 0.05640608072280884, | |
| "learning_rate": 0.00020814814814814814, | |
| "loss": 0.5701, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 1.1288888888888888, | |
| "grad_norm": 0.06712471693754196, | |
| "learning_rate": 0.0002079012345679012, | |
| "loss": 0.6875, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 1.1311111111111112, | |
| "grad_norm": 0.05165605619549751, | |
| "learning_rate": 0.0002076543209876543, | |
| "loss": 0.6435, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 1.1333333333333333, | |
| "grad_norm": 0.10354224592447281, | |
| "learning_rate": 0.00020740740740740737, | |
| "loss": 0.762, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 1.1355555555555557, | |
| "grad_norm": 0.06619168072938919, | |
| "learning_rate": 0.00020716049382716047, | |
| "loss": 0.6213, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 1.1377777777777778, | |
| "grad_norm": 0.057507146149873734, | |
| "learning_rate": 0.00020691358024691354, | |
| "loss": 0.5507, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 1.1400000000000001, | |
| "grad_norm": 0.07064547389745712, | |
| "learning_rate": 0.00020666666666666666, | |
| "loss": 0.5473, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 1.1422222222222222, | |
| "grad_norm": 0.05132355913519859, | |
| "learning_rate": 0.00020641975308641975, | |
| "loss": 0.5543, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 1.1444444444444444, | |
| "grad_norm": 0.07275696098804474, | |
| "learning_rate": 0.00020617283950617282, | |
| "loss": 0.5865, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 1.1466666666666667, | |
| "grad_norm": 0.08267077058553696, | |
| "learning_rate": 0.00020592592592592592, | |
| "loss": 0.616, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 1.1488888888888888, | |
| "grad_norm": 0.0553496852517128, | |
| "learning_rate": 0.00020567901234567898, | |
| "loss": 0.505, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 1.1511111111111112, | |
| "grad_norm": 0.08034535497426987, | |
| "learning_rate": 0.00020543209876543208, | |
| "loss": 0.6024, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 1.1533333333333333, | |
| "grad_norm": 0.04105300456285477, | |
| "learning_rate": 0.00020518518518518515, | |
| "loss": 0.4681, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 1.1555555555555554, | |
| "grad_norm": 0.061053916811943054, | |
| "learning_rate": 0.00020493827160493824, | |
| "loss": 0.5733, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 1.1577777777777778, | |
| "grad_norm": 0.0702579915523529, | |
| "learning_rate": 0.00020469135802469136, | |
| "loss": 0.6121, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "grad_norm": 0.055256616324186325, | |
| "learning_rate": 0.00020444444444444443, | |
| "loss": 0.5665, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 1.1622222222222223, | |
| "grad_norm": 0.05480768159031868, | |
| "learning_rate": 0.00020419753086419753, | |
| "loss": 0.6217, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 1.1644444444444444, | |
| "grad_norm": 0.053458116948604584, | |
| "learning_rate": 0.0002039506172839506, | |
| "loss": 0.5462, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 1.1666666666666667, | |
| "grad_norm": 0.057166215032339096, | |
| "learning_rate": 0.0002037037037037037, | |
| "loss": 0.6562, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 1.1688888888888889, | |
| "grad_norm": 0.0650937482714653, | |
| "learning_rate": 0.00020345679012345676, | |
| "loss": 0.5649, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 1.1711111111111112, | |
| "grad_norm": 0.06261788308620453, | |
| "learning_rate": 0.00020320987654320985, | |
| "loss": 0.6047, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 1.1733333333333333, | |
| "grad_norm": 0.05728009343147278, | |
| "learning_rate": 0.00020296296296296295, | |
| "loss": 0.626, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 1.1755555555555555, | |
| "grad_norm": 0.057623572647571564, | |
| "learning_rate": 0.00020271604938271602, | |
| "loss": 0.5539, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 1.1777777777777778, | |
| "grad_norm": 0.06340766698122025, | |
| "learning_rate": 0.00020246913580246914, | |
| "loss": 0.6803, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "grad_norm": 0.06497206538915634, | |
| "learning_rate": 0.0002022222222222222, | |
| "loss": 0.5192, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 1.1822222222222223, | |
| "grad_norm": 0.06043071672320366, | |
| "learning_rate": 0.0002019753086419753, | |
| "loss": 0.6507, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 1.1844444444444444, | |
| "grad_norm": 0.061710089445114136, | |
| "learning_rate": 0.00020172839506172837, | |
| "loss": 0.477, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 1.1866666666666668, | |
| "grad_norm": 0.060189537703990936, | |
| "learning_rate": 0.00020148148148148147, | |
| "loss": 0.5794, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 1.1888888888888889, | |
| "grad_norm": 0.06140688434243202, | |
| "learning_rate": 0.00020123456790123456, | |
| "loss": 0.7886, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 1.1911111111111112, | |
| "grad_norm": 0.0800878182053566, | |
| "learning_rate": 0.00020098765432098763, | |
| "loss": 0.6257, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 1.1933333333333334, | |
| "grad_norm": 0.09335595369338989, | |
| "learning_rate": 0.00020074074074074072, | |
| "loss": 0.5105, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 1.1955555555555555, | |
| "grad_norm": 0.07600431144237518, | |
| "learning_rate": 0.0002004938271604938, | |
| "loss": 0.5055, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 1.1977777777777778, | |
| "grad_norm": 0.06713638454675674, | |
| "learning_rate": 0.00020024691358024691, | |
| "loss": 0.5703, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "grad_norm": 0.06339897960424423, | |
| "learning_rate": 0.00019999999999999998, | |
| "loss": 0.6407, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 1.2022222222222223, | |
| "grad_norm": 0.07118365913629532, | |
| "learning_rate": 0.00019975308641975308, | |
| "loss": 0.5867, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 1.2044444444444444, | |
| "grad_norm": 0.060945894569158554, | |
| "learning_rate": 0.00019950617283950615, | |
| "loss": 0.6236, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 1.2066666666666666, | |
| "grad_norm": 0.0749896839261055, | |
| "learning_rate": 0.00019925925925925924, | |
| "loss": 0.6901, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 1.208888888888889, | |
| "grad_norm": 0.06734532117843628, | |
| "learning_rate": 0.00019901234567901234, | |
| "loss": 0.5685, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 1.211111111111111, | |
| "grad_norm": 0.051820721477270126, | |
| "learning_rate": 0.0001987654320987654, | |
| "loss": 0.5625, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 1.2133333333333334, | |
| "grad_norm": 0.07281868159770966, | |
| "learning_rate": 0.0001985185185185185, | |
| "loss": 0.7227, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 1.2155555555555555, | |
| "grad_norm": 0.05526493117213249, | |
| "learning_rate": 0.00019827160493827157, | |
| "loss": 0.5238, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 1.2177777777777778, | |
| "grad_norm": 0.05460463836789131, | |
| "learning_rate": 0.0001980246913580247, | |
| "loss": 0.6307, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "grad_norm": 0.07580731064081192, | |
| "learning_rate": 0.00019777777777777776, | |
| "loss": 0.5489, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 1.2222222222222223, | |
| "grad_norm": 0.05781114846467972, | |
| "learning_rate": 0.00019753086419753085, | |
| "loss": 0.5373, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 1.2244444444444444, | |
| "grad_norm": 0.05181167647242546, | |
| "learning_rate": 0.00019728395061728395, | |
| "loss": 0.5223, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 1.2266666666666666, | |
| "grad_norm": 0.09700439125299454, | |
| "learning_rate": 0.00019703703703703702, | |
| "loss": 0.5627, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 1.228888888888889, | |
| "grad_norm": 0.055470336228609085, | |
| "learning_rate": 0.0001967901234567901, | |
| "loss": 0.4962, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 1.231111111111111, | |
| "grad_norm": 0.04977557808160782, | |
| "learning_rate": 0.00019654320987654318, | |
| "loss": 0.5354, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 1.2333333333333334, | |
| "grad_norm": 0.0558207705616951, | |
| "learning_rate": 0.00019629629629629627, | |
| "loss": 0.6646, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 1.2355555555555555, | |
| "grad_norm": 0.06280802190303802, | |
| "learning_rate": 0.00019604938271604934, | |
| "loss": 0.5616, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 1.2377777777777779, | |
| "grad_norm": 0.06775356829166412, | |
| "learning_rate": 0.00019580246913580246, | |
| "loss": 0.5433, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "grad_norm": 0.05544356256723404, | |
| "learning_rate": 0.00019555555555555556, | |
| "loss": 0.6169, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 1.2422222222222223, | |
| "grad_norm": 0.07850141078233719, | |
| "learning_rate": 0.00019530864197530863, | |
| "loss": 0.5894, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 1.2444444444444445, | |
| "grad_norm": 0.048309750854969025, | |
| "learning_rate": 0.00019506172839506172, | |
| "loss": 0.5247, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 1.2466666666666666, | |
| "grad_norm": 0.1482103317975998, | |
| "learning_rate": 0.0001948148148148148, | |
| "loss": 0.6559, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 1.248888888888889, | |
| "grad_norm": 0.04897948354482651, | |
| "learning_rate": 0.00019456790123456789, | |
| "loss": 0.528, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 1.251111111111111, | |
| "grad_norm": 0.05171951279044151, | |
| "learning_rate": 0.00019432098765432095, | |
| "loss": 0.5279, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 1.2533333333333334, | |
| "grad_norm": 0.06654342263936996, | |
| "learning_rate": 0.00019407407407407405, | |
| "loss": 0.5316, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 1.2555555555555555, | |
| "grad_norm": 0.07495734095573425, | |
| "learning_rate": 0.00019382716049382712, | |
| "loss": 0.6979, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 1.2577777777777777, | |
| "grad_norm": 0.0822639912366867, | |
| "learning_rate": 0.00019358024691358024, | |
| "loss": 0.5646, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "grad_norm": 0.04671519622206688, | |
| "learning_rate": 0.00019333333333333333, | |
| "loss": 0.4708, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 1.2622222222222224, | |
| "grad_norm": 0.05543937534093857, | |
| "learning_rate": 0.0001930864197530864, | |
| "loss": 0.6776, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 1.2644444444444445, | |
| "grad_norm": 0.06297334283590317, | |
| "learning_rate": 0.0001928395061728395, | |
| "loss": 0.6146, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 1.2666666666666666, | |
| "grad_norm": 0.06672172993421555, | |
| "learning_rate": 0.00019259259259259257, | |
| "loss": 0.5809, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 1.268888888888889, | |
| "grad_norm": 0.06314659118652344, | |
| "learning_rate": 0.00019234567901234566, | |
| "loss": 0.6398, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 1.271111111111111, | |
| "grad_norm": 0.07397986948490143, | |
| "learning_rate": 0.00019209876543209873, | |
| "loss": 0.5926, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 1.2733333333333334, | |
| "grad_norm": 0.05730567127466202, | |
| "learning_rate": 0.00019185185185185182, | |
| "loss": 0.502, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 1.2755555555555556, | |
| "grad_norm": 0.05761762708425522, | |
| "learning_rate": 0.00019160493827160495, | |
| "loss": 0.6246, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 1.2777777777777777, | |
| "grad_norm": 0.05256739631295204, | |
| "learning_rate": 0.00019135802469135801, | |
| "loss": 0.4969, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "grad_norm": 0.07907039672136307, | |
| "learning_rate": 0.0001911111111111111, | |
| "loss": 0.5639, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 1.2822222222222222, | |
| "grad_norm": 0.06401896476745605, | |
| "learning_rate": 0.00019086419753086418, | |
| "loss": 0.455, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 1.2844444444444445, | |
| "grad_norm": 0.06785224378108978, | |
| "learning_rate": 0.00019061728395061727, | |
| "loss": 0.6206, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 1.2866666666666666, | |
| "grad_norm": 0.05500558391213417, | |
| "learning_rate": 0.00019037037037037034, | |
| "loss": 0.5703, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 1.2888888888888888, | |
| "grad_norm": 0.13497020304203033, | |
| "learning_rate": 0.00019012345679012344, | |
| "loss": 0.6413, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 1.291111111111111, | |
| "grad_norm": 0.05647628381848335, | |
| "learning_rate": 0.00018987654320987653, | |
| "loss": 0.4631, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 1.2933333333333334, | |
| "grad_norm": 0.06448817253112793, | |
| "learning_rate": 0.0001896296296296296, | |
| "loss": 0.7013, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 1.2955555555555556, | |
| "grad_norm": 0.05188887193799019, | |
| "learning_rate": 0.00018938271604938272, | |
| "loss": 0.5185, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 1.2977777777777777, | |
| "grad_norm": 0.07357285916805267, | |
| "learning_rate": 0.0001891358024691358, | |
| "loss": 0.6057, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "grad_norm": 0.057298485189676285, | |
| "learning_rate": 0.00018888888888888888, | |
| "loss": 0.6387, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 1.3022222222222222, | |
| "grad_norm": 0.05352728068828583, | |
| "learning_rate": 0.00018864197530864195, | |
| "loss": 0.5425, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 1.3044444444444445, | |
| "grad_norm": 0.05617798492312431, | |
| "learning_rate": 0.00018839506172839505, | |
| "loss": 0.6855, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 1.3066666666666666, | |
| "grad_norm": 0.05917446315288544, | |
| "learning_rate": 0.00018814814814814814, | |
| "loss": 0.7752, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 1.3088888888888888, | |
| "grad_norm": 0.07065349817276001, | |
| "learning_rate": 0.0001879012345679012, | |
| "loss": 0.6745, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 1.3111111111111111, | |
| "grad_norm": 0.05729060247540474, | |
| "learning_rate": 0.0001876543209876543, | |
| "loss": 0.5004, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 1.3133333333333335, | |
| "grad_norm": 0.050993286073207855, | |
| "learning_rate": 0.00018740740740740737, | |
| "loss": 0.5734, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 1.3155555555555556, | |
| "grad_norm": 0.050808437168598175, | |
| "learning_rate": 0.0001871604938271605, | |
| "loss": 0.5089, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 1.3177777777777777, | |
| "grad_norm": 0.06159992888569832, | |
| "learning_rate": 0.00018691358024691356, | |
| "loss": 0.5593, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "grad_norm": 0.057330355048179626, | |
| "learning_rate": 0.00018666666666666666, | |
| "loss": 0.5363, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 1.3222222222222222, | |
| "grad_norm": 0.07226645946502686, | |
| "learning_rate": 0.00018641975308641973, | |
| "loss": 0.62, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 1.3244444444444445, | |
| "grad_norm": 0.0795191153883934, | |
| "learning_rate": 0.00018617283950617282, | |
| "loss": 0.6801, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 1.3266666666666667, | |
| "grad_norm": 0.06860774755477905, | |
| "learning_rate": 0.00018592592592592592, | |
| "loss": 0.5357, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 1.3288888888888888, | |
| "grad_norm": 0.07353827357292175, | |
| "learning_rate": 0.00018567901234567899, | |
| "loss": 0.6449, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 1.3311111111111111, | |
| "grad_norm": 0.08865563571453094, | |
| "learning_rate": 0.00018543209876543208, | |
| "loss": 0.6238, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.0657055452466011, | |
| "learning_rate": 0.00018518518518518515, | |
| "loss": 0.6608, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 1.3355555555555556, | |
| "grad_norm": 0.052916329354047775, | |
| "learning_rate": 0.00018493827160493827, | |
| "loss": 0.5771, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 1.3377777777777777, | |
| "grad_norm": 0.06563440710306168, | |
| "learning_rate": 0.00018469135802469134, | |
| "loss": 0.5596, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "grad_norm": 0.05654449015855789, | |
| "learning_rate": 0.00018444444444444443, | |
| "loss": 0.6079, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 1.3422222222222222, | |
| "grad_norm": 0.06601669639348984, | |
| "learning_rate": 0.00018419753086419753, | |
| "loss": 0.5052, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 1.3444444444444446, | |
| "grad_norm": 0.05436199903488159, | |
| "learning_rate": 0.0001839506172839506, | |
| "loss": 0.5443, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 1.3466666666666667, | |
| "grad_norm": 0.05751197785139084, | |
| "learning_rate": 0.0001837037037037037, | |
| "loss": 0.5621, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 1.3488888888888888, | |
| "grad_norm": 0.056823842227458954, | |
| "learning_rate": 0.00018345679012345676, | |
| "loss": 0.5466, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 1.3511111111111112, | |
| "grad_norm": 0.08196792006492615, | |
| "learning_rate": 0.00018320987654320986, | |
| "loss": 0.6071, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 1.3533333333333333, | |
| "grad_norm": 0.07378371059894562, | |
| "learning_rate": 0.00018296296296296292, | |
| "loss": 0.6775, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 1.3555555555555556, | |
| "grad_norm": 0.0697040930390358, | |
| "learning_rate": 0.00018271604938271605, | |
| "loss": 0.6275, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 1.3577777777777778, | |
| "grad_norm": 0.05658598616719246, | |
| "learning_rate": 0.00018246913580246914, | |
| "loss": 0.4926, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 1.3599999999999999, | |
| "grad_norm": 0.05035831779241562, | |
| "learning_rate": 0.0001822222222222222, | |
| "loss": 0.5523, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 1.3622222222222222, | |
| "grad_norm": 0.08606016635894775, | |
| "learning_rate": 0.0001819753086419753, | |
| "loss": 0.5475, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 1.3644444444444446, | |
| "grad_norm": 0.07070509344339371, | |
| "learning_rate": 0.00018172839506172837, | |
| "loss": 0.7618, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 1.3666666666666667, | |
| "grad_norm": 0.054081596434116364, | |
| "learning_rate": 0.00018148148148148147, | |
| "loss": 0.5402, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 1.3688888888888888, | |
| "grad_norm": 0.06346143782138824, | |
| "learning_rate": 0.00018123456790123454, | |
| "loss": 0.5518, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 1.3711111111111112, | |
| "grad_norm": 0.07378754019737244, | |
| "learning_rate": 0.00018098765432098763, | |
| "loss": 0.617, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 1.3733333333333333, | |
| "grad_norm": 0.08497416228055954, | |
| "learning_rate": 0.00018074074074074075, | |
| "loss": 0.7788, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 1.3755555555555556, | |
| "grad_norm": 0.07911262661218643, | |
| "learning_rate": 0.00018049382716049382, | |
| "loss": 0.5803, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 1.3777777777777778, | |
| "grad_norm": 0.04942178353667259, | |
| "learning_rate": 0.00018024691358024692, | |
| "loss": 0.4716, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "grad_norm": 0.047186482697725296, | |
| "learning_rate": 0.00017999999999999998, | |
| "loss": 0.5332, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 1.3822222222222222, | |
| "grad_norm": 0.06078922376036644, | |
| "learning_rate": 0.00017975308641975308, | |
| "loss": 0.5501, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 1.3844444444444444, | |
| "grad_norm": 0.052392344921827316, | |
| "learning_rate": 0.00017950617283950615, | |
| "loss": 0.5307, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 1.3866666666666667, | |
| "grad_norm": 0.06864654272794724, | |
| "learning_rate": 0.00017925925925925924, | |
| "loss": 0.5512, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "grad_norm": 0.06133788079023361, | |
| "learning_rate": 0.0001790123456790123, | |
| "loss": 0.4992, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 1.3911111111111112, | |
| "grad_norm": 0.06265698373317719, | |
| "learning_rate": 0.0001787654320987654, | |
| "loss": 0.5094, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 1.3933333333333333, | |
| "grad_norm": 0.05738376826047897, | |
| "learning_rate": 0.00017851851851851853, | |
| "loss": 0.6811, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 1.3955555555555557, | |
| "grad_norm": 0.07373687624931335, | |
| "learning_rate": 0.0001782716049382716, | |
| "loss": 0.6168, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 1.3977777777777778, | |
| "grad_norm": 0.052827995270490646, | |
| "learning_rate": 0.0001780246913580247, | |
| "loss": 0.5009, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "grad_norm": 0.12193891406059265, | |
| "learning_rate": 0.00017777777777777776, | |
| "loss": 0.5653, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 1.4022222222222223, | |
| "grad_norm": 0.0757482722401619, | |
| "learning_rate": 0.00017753086419753085, | |
| "loss": 0.5405, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 1.4044444444444444, | |
| "grad_norm": 0.06812596321105957, | |
| "learning_rate": 0.00017728395061728392, | |
| "loss": 0.5486, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 1.4066666666666667, | |
| "grad_norm": 0.04514603316783905, | |
| "learning_rate": 0.00017703703703703702, | |
| "loss": 0.5271, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 1.4088888888888889, | |
| "grad_norm": 0.05779915302991867, | |
| "learning_rate": 0.0001767901234567901, | |
| "loss": 0.7327, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 1.411111111111111, | |
| "grad_norm": 0.07517620921134949, | |
| "learning_rate": 0.00017654320987654318, | |
| "loss": 0.4626, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 1.4133333333333333, | |
| "grad_norm": 0.0513499453663826, | |
| "learning_rate": 0.0001762962962962963, | |
| "loss": 0.5631, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 1.4155555555555557, | |
| "grad_norm": 0.06641045212745667, | |
| "learning_rate": 0.00017604938271604937, | |
| "loss": 0.5825, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 1.4177777777777778, | |
| "grad_norm": 0.07333184778690338, | |
| "learning_rate": 0.00017580246913580247, | |
| "loss": 0.6719, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "grad_norm": 0.06464029848575592, | |
| "learning_rate": 0.00017555555555555553, | |
| "loss": 0.6932, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 1.4222222222222223, | |
| "grad_norm": 0.06149693578481674, | |
| "learning_rate": 0.00017530864197530863, | |
| "loss": 0.5776, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 1.4244444444444444, | |
| "grad_norm": 0.05443153157830238, | |
| "learning_rate": 0.00017506172839506172, | |
| "loss": 0.5091, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 1.4266666666666667, | |
| "grad_norm": 0.07074269652366638, | |
| "learning_rate": 0.0001748148148148148, | |
| "loss": 0.6371, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 1.4288888888888889, | |
| "grad_norm": 0.10091938078403473, | |
| "learning_rate": 0.0001745679012345679, | |
| "loss": 0.423, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 1.431111111111111, | |
| "grad_norm": 0.05455939099192619, | |
| "learning_rate": 0.00017432098765432096, | |
| "loss": 0.5415, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 1.4333333333333333, | |
| "grad_norm": 0.05939517170190811, | |
| "learning_rate": 0.00017407407407407408, | |
| "loss": 0.6316, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 1.4355555555555555, | |
| "grad_norm": 0.07641264796257019, | |
| "learning_rate": 0.00017382716049382715, | |
| "loss": 0.5573, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 1.4377777777777778, | |
| "grad_norm": 0.08066412806510925, | |
| "learning_rate": 0.00017358024691358024, | |
| "loss": 0.6718, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "grad_norm": 0.07123071700334549, | |
| "learning_rate": 0.0001733333333333333, | |
| "loss": 0.538, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 1.4422222222222223, | |
| "grad_norm": 0.0550350546836853, | |
| "learning_rate": 0.0001730864197530864, | |
| "loss": 0.5855, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 1.4444444444444444, | |
| "grad_norm": 0.054279692471027374, | |
| "learning_rate": 0.0001728395061728395, | |
| "loss": 0.6959, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 1.4466666666666668, | |
| "grad_norm": 0.051811859011650085, | |
| "learning_rate": 0.00017259259259259257, | |
| "loss": 0.4472, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 1.448888888888889, | |
| "grad_norm": 0.059803206473588943, | |
| "learning_rate": 0.00017234567901234566, | |
| "loss": 0.5405, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 1.451111111111111, | |
| "grad_norm": 0.07218248397111893, | |
| "learning_rate": 0.00017209876543209873, | |
| "loss": 0.6648, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 1.4533333333333334, | |
| "grad_norm": 0.05009540542960167, | |
| "learning_rate": 0.00017185185185185185, | |
| "loss": 0.5352, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 1.4555555555555555, | |
| "grad_norm": 0.06544394046068192, | |
| "learning_rate": 0.00017160493827160492, | |
| "loss": 0.5009, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 1.4577777777777778, | |
| "grad_norm": 0.05920146033167839, | |
| "learning_rate": 0.00017135802469135802, | |
| "loss": 0.5075, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "grad_norm": 0.04878571629524231, | |
| "learning_rate": 0.0001711111111111111, | |
| "loss": 0.4529, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 1.462222222222222, | |
| "grad_norm": 0.04539019986987114, | |
| "learning_rate": 0.00017086419753086418, | |
| "loss": 0.451, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 1.4644444444444444, | |
| "grad_norm": 0.05576840788125992, | |
| "learning_rate": 0.00017061728395061727, | |
| "loss": 0.5135, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 1.4666666666666668, | |
| "grad_norm": 0.05507802218198776, | |
| "learning_rate": 0.00017037037037037034, | |
| "loss": 0.6366, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 1.468888888888889, | |
| "grad_norm": 0.08249601721763611, | |
| "learning_rate": 0.00017012345679012344, | |
| "loss": 0.6801, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 1.471111111111111, | |
| "grad_norm": 0.07689239829778671, | |
| "learning_rate": 0.0001698765432098765, | |
| "loss": 0.6374, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 1.4733333333333334, | |
| "grad_norm": 0.10602860897779465, | |
| "learning_rate": 0.00016962962962962963, | |
| "loss": 0.5077, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 1.4755555555555555, | |
| "grad_norm": 0.06070888042449951, | |
| "learning_rate": 0.00016938271604938272, | |
| "loss": 0.7006, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 1.4777777777777779, | |
| "grad_norm": 0.05798992142081261, | |
| "learning_rate": 0.0001691358024691358, | |
| "loss": 0.493, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "grad_norm": 0.08377783000469208, | |
| "learning_rate": 0.00016888888888888889, | |
| "loss": 0.4825, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 1.482222222222222, | |
| "grad_norm": 0.06520260870456696, | |
| "learning_rate": 0.00016864197530864195, | |
| "loss": 0.5424, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 1.4844444444444445, | |
| "grad_norm": 0.07959870994091034, | |
| "learning_rate": 0.00016839506172839505, | |
| "loss": 0.6104, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 1.4866666666666668, | |
| "grad_norm": 0.06556657701730728, | |
| "learning_rate": 0.00016814814814814812, | |
| "loss": 0.5864, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 1.488888888888889, | |
| "grad_norm": 0.06736795604228973, | |
| "learning_rate": 0.0001679012345679012, | |
| "loss": 0.5733, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 1.491111111111111, | |
| "grad_norm": 0.057398103177547455, | |
| "learning_rate": 0.00016765432098765433, | |
| "loss": 0.4716, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 1.4933333333333334, | |
| "grad_norm": 0.09340738505125046, | |
| "learning_rate": 0.0001674074074074074, | |
| "loss": 0.5637, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 1.4955555555555555, | |
| "grad_norm": 0.06669988483190536, | |
| "learning_rate": 0.0001671604938271605, | |
| "loss": 0.651, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 1.4977777777777779, | |
| "grad_norm": 0.08472840487957001, | |
| "learning_rate": 0.00016691358024691357, | |
| "loss": 0.58, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 0.057394471019506454, | |
| "learning_rate": 0.00016666666666666666, | |
| "loss": 0.5119, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 1.5022222222222221, | |
| "grad_norm": 0.06451477110385895, | |
| "learning_rate": 0.00016641975308641973, | |
| "loss": 0.5004, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 1.5044444444444445, | |
| "grad_norm": 0.07595914602279663, | |
| "learning_rate": 0.00016617283950617282, | |
| "loss": 0.6697, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 1.5066666666666668, | |
| "grad_norm": 0.06590458005666733, | |
| "learning_rate": 0.0001659259259259259, | |
| "loss": 0.5613, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 1.508888888888889, | |
| "grad_norm": 0.07247265428304672, | |
| "learning_rate": 0.000165679012345679, | |
| "loss": 0.6653, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 1.511111111111111, | |
| "grad_norm": 0.07311250269412994, | |
| "learning_rate": 0.0001654320987654321, | |
| "loss": 0.5959, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 1.5133333333333332, | |
| "grad_norm": 0.08032218366861343, | |
| "learning_rate": 0.00016518518518518518, | |
| "loss": 0.6595, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 1.5155555555555555, | |
| "grad_norm": 0.04984492063522339, | |
| "learning_rate": 0.00016493827160493827, | |
| "loss": 0.535, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 1.517777777777778, | |
| "grad_norm": 0.06561774015426636, | |
| "learning_rate": 0.00016469135802469134, | |
| "loss": 0.5198, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "grad_norm": 0.08288616687059402, | |
| "learning_rate": 0.00016444444444444444, | |
| "loss": 0.674, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 1.5222222222222221, | |
| "grad_norm": 0.0601021908223629, | |
| "learning_rate": 0.0001641975308641975, | |
| "loss": 0.5975, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 1.5244444444444445, | |
| "grad_norm": 0.12323024868965149, | |
| "learning_rate": 0.0001639506172839506, | |
| "loss": 0.5378, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 1.5266666666666666, | |
| "grad_norm": 0.06298123300075531, | |
| "learning_rate": 0.0001637037037037037, | |
| "loss": 0.554, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 1.528888888888889, | |
| "grad_norm": 0.06339532881975174, | |
| "learning_rate": 0.00016345679012345676, | |
| "loss": 0.5133, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 1.531111111111111, | |
| "grad_norm": 0.0558561347424984, | |
| "learning_rate": 0.00016320987654320988, | |
| "loss": 0.5032, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 1.5333333333333332, | |
| "grad_norm": 0.055791474878787994, | |
| "learning_rate": 0.00016296296296296295, | |
| "loss": 0.5456, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 1.5355555555555556, | |
| "grad_norm": 0.04931165277957916, | |
| "learning_rate": 0.00016271604938271605, | |
| "loss": 0.4454, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 1.537777777777778, | |
| "grad_norm": 0.07284574955701828, | |
| "learning_rate": 0.00016246913580246912, | |
| "loss": 0.5862, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "grad_norm": 0.1155928522348404, | |
| "learning_rate": 0.0001622222222222222, | |
| "loss": 0.527, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 1.5422222222222222, | |
| "grad_norm": 0.054718051105737686, | |
| "learning_rate": 0.0001619753086419753, | |
| "loss": 0.5375, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 1.5444444444444443, | |
| "grad_norm": 0.05141949653625488, | |
| "learning_rate": 0.00016172839506172837, | |
| "loss": 0.5186, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 1.5466666666666666, | |
| "grad_norm": 0.07433553040027618, | |
| "learning_rate": 0.00016148148148148147, | |
| "loss": 0.5112, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 1.548888888888889, | |
| "grad_norm": 0.05488400533795357, | |
| "learning_rate": 0.00016123456790123454, | |
| "loss": 0.5465, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 1.551111111111111, | |
| "grad_norm": 0.07929231971502304, | |
| "learning_rate": 0.00016098765432098766, | |
| "loss": 0.6853, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 1.5533333333333332, | |
| "grad_norm": 0.05515801161527634, | |
| "learning_rate": 0.00016074074074074073, | |
| "loss": 0.681, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 1.5555555555555556, | |
| "grad_norm": 0.04796384274959564, | |
| "learning_rate": 0.00016049382716049382, | |
| "loss": 0.4594, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 1.557777777777778, | |
| "grad_norm": 0.06954047083854675, | |
| "learning_rate": 0.00016024691358024692, | |
| "loss": 0.5117, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "grad_norm": 0.04393395036458969, | |
| "learning_rate": 0.00015999999999999999, | |
| "loss": 0.4467, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 1.5622222222222222, | |
| "grad_norm": 0.07525661587715149, | |
| "learning_rate": 0.00015975308641975308, | |
| "loss": 0.5913, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 1.5644444444444443, | |
| "grad_norm": 0.07192367315292358, | |
| "learning_rate": 0.00015950617283950615, | |
| "loss": 0.4955, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 1.5666666666666667, | |
| "grad_norm": 0.07459937036037445, | |
| "learning_rate": 0.00015925925925925924, | |
| "loss": 0.6316, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 1.568888888888889, | |
| "grad_norm": 0.06207813322544098, | |
| "learning_rate": 0.0001590123456790123, | |
| "loss": 0.6553, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 1.5711111111111111, | |
| "grad_norm": 0.07175853848457336, | |
| "learning_rate": 0.00015876543209876543, | |
| "loss": 0.5826, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 1.5733333333333333, | |
| "grad_norm": 0.05727192759513855, | |
| "learning_rate": 0.0001585185185185185, | |
| "loss": 0.6622, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 1.5755555555555556, | |
| "grad_norm": 0.06073378771543503, | |
| "learning_rate": 0.0001582716049382716, | |
| "loss": 0.4843, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 1.5777777777777777, | |
| "grad_norm": 0.1481916457414627, | |
| "learning_rate": 0.0001580246913580247, | |
| "loss": 0.6315, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "grad_norm": 0.07267245650291443, | |
| "learning_rate": 0.00015777777777777776, | |
| "loss": 0.577, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 1.5822222222222222, | |
| "grad_norm": 0.1542692631483078, | |
| "learning_rate": 0.00015753086419753086, | |
| "loss": 0.5453, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 1.5844444444444443, | |
| "grad_norm": 0.08447085320949554, | |
| "learning_rate": 0.00015728395061728392, | |
| "loss": 0.6688, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 1.5866666666666667, | |
| "grad_norm": 0.05692017823457718, | |
| "learning_rate": 0.00015703703703703702, | |
| "loss": 0.4807, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 1.588888888888889, | |
| "grad_norm": 0.07300254702568054, | |
| "learning_rate": 0.0001567901234567901, | |
| "loss": 0.5531, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 1.5911111111111111, | |
| "grad_norm": 0.07361926138401031, | |
| "learning_rate": 0.0001565432098765432, | |
| "loss": 0.5973, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 1.5933333333333333, | |
| "grad_norm": 0.09326552599668503, | |
| "learning_rate": 0.0001562962962962963, | |
| "loss": 0.68, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 1.5955555555555554, | |
| "grad_norm": 0.06923039257526398, | |
| "learning_rate": 0.00015604938271604937, | |
| "loss": 0.5468, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 1.5977777777777777, | |
| "grad_norm": 0.05969049409031868, | |
| "learning_rate": 0.00015580246913580247, | |
| "loss": 0.5393, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "grad_norm": 0.06347218155860901, | |
| "learning_rate": 0.00015555555555555554, | |
| "loss": 0.5553, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 1.6022222222222222, | |
| "grad_norm": 0.04843074828386307, | |
| "learning_rate": 0.00015530864197530863, | |
| "loss": 0.4702, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 1.6044444444444443, | |
| "grad_norm": 0.05043656378984451, | |
| "learning_rate": 0.0001550617283950617, | |
| "loss": 0.6072, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 1.6066666666666667, | |
| "grad_norm": 0.06668213754892349, | |
| "learning_rate": 0.0001548148148148148, | |
| "loss": 0.5995, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 1.608888888888889, | |
| "grad_norm": 0.05539967119693756, | |
| "learning_rate": 0.00015456790123456792, | |
| "loss": 0.5543, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 1.6111111111111112, | |
| "grad_norm": 0.05010242387652397, | |
| "learning_rate": 0.00015432098765432098, | |
| "loss": 0.549, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 1.6133333333333333, | |
| "grad_norm": 0.08259212225675583, | |
| "learning_rate": 0.00015407407407407408, | |
| "loss": 0.6413, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 1.6155555555555554, | |
| "grad_norm": 0.06118133291602135, | |
| "learning_rate": 0.00015382716049382715, | |
| "loss": 0.5549, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 1.6177777777777778, | |
| "grad_norm": 0.053836848586797714, | |
| "learning_rate": 0.00015358024691358024, | |
| "loss": 0.5361, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "grad_norm": 0.1261414736509323, | |
| "learning_rate": 0.0001533333333333333, | |
| "loss": 0.5703, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 1.6222222222222222, | |
| "grad_norm": 0.05367890000343323, | |
| "learning_rate": 0.0001530864197530864, | |
| "loss": 0.555, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 1.6244444444444444, | |
| "grad_norm": 0.09554382413625717, | |
| "learning_rate": 0.00015283950617283947, | |
| "loss": 0.3854, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 1.6266666666666667, | |
| "grad_norm": 0.0832231193780899, | |
| "learning_rate": 0.00015259259259259257, | |
| "loss": 0.6498, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 1.628888888888889, | |
| "grad_norm": 0.09230218082666397, | |
| "learning_rate": 0.0001523456790123457, | |
| "loss": 0.5641, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 1.6311111111111112, | |
| "grad_norm": 0.057532064616680145, | |
| "learning_rate": 0.00015209876543209873, | |
| "loss": 0.5838, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 1.6333333333333333, | |
| "grad_norm": 0.13021503388881683, | |
| "learning_rate": 0.00015185185185185185, | |
| "loss": 0.5665, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 1.6355555555555554, | |
| "grad_norm": 0.06547033041715622, | |
| "learning_rate": 0.00015160493827160492, | |
| "loss": 0.5572, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 1.6377777777777778, | |
| "grad_norm": 0.06860460340976715, | |
| "learning_rate": 0.00015135802469135802, | |
| "loss": 0.5439, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 1.6400000000000001, | |
| "grad_norm": 0.054408688098192215, | |
| "learning_rate": 0.00015111111111111109, | |
| "loss": 0.5768, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 1.6422222222222222, | |
| "grad_norm": 0.051527902483940125, | |
| "learning_rate": 0.00015086419753086418, | |
| "loss": 0.6987, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 1.6444444444444444, | |
| "grad_norm": 0.07253484427928925, | |
| "learning_rate": 0.00015061728395061728, | |
| "loss": 0.6066, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 1.6466666666666665, | |
| "grad_norm": 0.07275417447090149, | |
| "learning_rate": 0.00015037037037037034, | |
| "loss": 0.6965, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 1.6488888888888888, | |
| "grad_norm": 0.06611516326665878, | |
| "learning_rate": 0.00015012345679012347, | |
| "loss": 0.5983, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 1.6511111111111112, | |
| "grad_norm": 0.06346430629491806, | |
| "learning_rate": 0.00014987654320987653, | |
| "loss": 0.506, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 1.6533333333333333, | |
| "grad_norm": 0.0587560310959816, | |
| "learning_rate": 0.00014962962962962963, | |
| "loss": 0.3811, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 1.6555555555555554, | |
| "grad_norm": 0.061339814215898514, | |
| "learning_rate": 0.0001493827160493827, | |
| "loss": 0.617, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 1.6577777777777778, | |
| "grad_norm": 0.06670242547988892, | |
| "learning_rate": 0.0001491358024691358, | |
| "loss": 0.4602, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 1.6600000000000001, | |
| "grad_norm": 0.06967601925134659, | |
| "learning_rate": 0.00014888888888888886, | |
| "loss": 0.4374, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.6622222222222223, | |
| "grad_norm": 0.06698095798492432, | |
| "learning_rate": 0.00014864197530864196, | |
| "loss": 0.4996, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 1.6644444444444444, | |
| "grad_norm": 0.073794424533844, | |
| "learning_rate": 0.00014839506172839505, | |
| "loss": 0.6215, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 0.0930507704615593, | |
| "learning_rate": 0.00014814814814814812, | |
| "loss": 0.6421, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.6688888888888889, | |
| "grad_norm": 0.05774183198809624, | |
| "learning_rate": 0.00014790123456790124, | |
| "loss": 0.4443, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.6711111111111112, | |
| "grad_norm": 0.04958115518093109, | |
| "learning_rate": 0.0001476543209876543, | |
| "loss": 0.4481, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.6733333333333333, | |
| "grad_norm": 0.06320313364267349, | |
| "learning_rate": 0.0001474074074074074, | |
| "loss": 0.4927, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.6755555555555555, | |
| "grad_norm": 0.05753500014543533, | |
| "learning_rate": 0.00014716049382716047, | |
| "loss": 0.6196, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 1.6777777777777778, | |
| "grad_norm": 0.05950382724404335, | |
| "learning_rate": 0.00014691358024691357, | |
| "loss": 0.656, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 1.6800000000000002, | |
| "grad_norm": 0.059872355312108994, | |
| "learning_rate": 0.00014666666666666664, | |
| "loss": 0.5287, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 1.6822222222222223, | |
| "grad_norm": 0.07446783781051636, | |
| "learning_rate": 0.00014641975308641973, | |
| "loss": 0.4874, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 1.6844444444444444, | |
| "grad_norm": 0.09308301657438278, | |
| "learning_rate": 0.00014617283950617283, | |
| "loss": 0.6015, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 1.6866666666666665, | |
| "grad_norm": 0.07845073938369751, | |
| "learning_rate": 0.00014592592592592592, | |
| "loss": 0.6242, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 1.6888888888888889, | |
| "grad_norm": 0.05412207916378975, | |
| "learning_rate": 0.000145679012345679, | |
| "loss": 0.5224, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 1.6911111111111112, | |
| "grad_norm": 0.06393057107925415, | |
| "learning_rate": 0.00014543209876543208, | |
| "loss": 0.6104, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 1.6933333333333334, | |
| "grad_norm": 0.0477752685546875, | |
| "learning_rate": 0.00014518518518518518, | |
| "loss": 0.6407, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 1.6955555555555555, | |
| "grad_norm": 0.05183986574411392, | |
| "learning_rate": 0.00014493827160493825, | |
| "loss": 0.4581, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 1.6977777777777778, | |
| "grad_norm": 0.07449138909578323, | |
| "learning_rate": 0.00014469135802469134, | |
| "loss": 0.5091, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "grad_norm": 0.06958594918251038, | |
| "learning_rate": 0.0001444444444444444, | |
| "loss": 0.7315, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 1.7022222222222223, | |
| "grad_norm": 0.05860864371061325, | |
| "learning_rate": 0.00014419753086419753, | |
| "loss": 0.4559, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 1.7044444444444444, | |
| "grad_norm": 0.06844347715377808, | |
| "learning_rate": 0.0001439506172839506, | |
| "loss": 0.5453, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 1.7066666666666666, | |
| "grad_norm": 0.06979213654994965, | |
| "learning_rate": 0.0001437037037037037, | |
| "loss": 0.5099, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 1.708888888888889, | |
| "grad_norm": 0.05852322652935982, | |
| "learning_rate": 0.00014345679012345676, | |
| "loss": 0.4338, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 1.7111111111111112, | |
| "grad_norm": 0.05683210492134094, | |
| "learning_rate": 0.00014320987654320986, | |
| "loss": 0.5247, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 1.7133333333333334, | |
| "grad_norm": 0.05876768380403519, | |
| "learning_rate": 0.00014296296296296295, | |
| "loss": 0.4864, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 1.7155555555555555, | |
| "grad_norm": 0.08508090674877167, | |
| "learning_rate": 0.00014271604938271602, | |
| "loss": 0.6039, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 1.7177777777777776, | |
| "grad_norm": 0.052276160567998886, | |
| "learning_rate": 0.00014246913580246912, | |
| "loss": 0.6211, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "grad_norm": 0.05639871582388878, | |
| "learning_rate": 0.0001422222222222222, | |
| "loss": 0.5547, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 1.7222222222222223, | |
| "grad_norm": 0.10673470050096512, | |
| "learning_rate": 0.0001419753086419753, | |
| "loss": 0.5379, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 1.7244444444444444, | |
| "grad_norm": 0.05808853358030319, | |
| "learning_rate": 0.00014172839506172838, | |
| "loss": 0.5664, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 1.7266666666666666, | |
| "grad_norm": 0.08077769726514816, | |
| "learning_rate": 0.00014148148148148147, | |
| "loss": 0.5641, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 1.728888888888889, | |
| "grad_norm": 0.09657977521419525, | |
| "learning_rate": 0.00014123456790123454, | |
| "loss": 0.6948, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 1.7311111111111113, | |
| "grad_norm": 0.06040675938129425, | |
| "learning_rate": 0.00014098765432098763, | |
| "loss": 0.5752, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 1.7333333333333334, | |
| "grad_norm": 0.09027255326509476, | |
| "learning_rate": 0.00014074074074074073, | |
| "loss": 0.571, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 1.7355555555555555, | |
| "grad_norm": 0.08459677547216415, | |
| "learning_rate": 0.00014049382716049382, | |
| "loss": 0.4594, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 1.7377777777777776, | |
| "grad_norm": 0.06038082763552666, | |
| "learning_rate": 0.0001402469135802469, | |
| "loss": 0.5246, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "grad_norm": 0.07737093418836594, | |
| "learning_rate": 0.00014, | |
| "loss": 0.4698, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 1.7422222222222223, | |
| "grad_norm": 0.05954982340335846, | |
| "learning_rate": 0.00013975308641975308, | |
| "loss": 0.661, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 1.7444444444444445, | |
| "grad_norm": 0.05921616032719612, | |
| "learning_rate": 0.00013950617283950615, | |
| "loss": 0.5617, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 1.7466666666666666, | |
| "grad_norm": 0.07192415744066238, | |
| "learning_rate": 0.00013925925925925925, | |
| "loss": 0.5018, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.748888888888889, | |
| "grad_norm": 0.06786806136369705, | |
| "learning_rate": 0.00013901234567901231, | |
| "loss": 0.6252, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 1.751111111111111, | |
| "grad_norm": 0.059251997619867325, | |
| "learning_rate": 0.00013876543209876544, | |
| "loss": 0.5964, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 1.7533333333333334, | |
| "grad_norm": 0.08339954167604446, | |
| "learning_rate": 0.0001385185185185185, | |
| "loss": 0.5799, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 1.7555555555555555, | |
| "grad_norm": 0.06480483710765839, | |
| "learning_rate": 0.0001382716049382716, | |
| "loss": 0.6731, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 1.7577777777777777, | |
| "grad_norm": 0.08144319802522659, | |
| "learning_rate": 0.00013802469135802467, | |
| "loss": 0.5436, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "grad_norm": 0.06506912410259247, | |
| "learning_rate": 0.00013777777777777776, | |
| "loss": 0.4995, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 1.7622222222222224, | |
| "grad_norm": 0.06140586733818054, | |
| "learning_rate": 0.00013753086419753086, | |
| "loss": 0.5528, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 1.7644444444444445, | |
| "grad_norm": 0.07202223688364029, | |
| "learning_rate": 0.00013728395061728393, | |
| "loss": 0.6095, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 1.7666666666666666, | |
| "grad_norm": 0.06542429327964783, | |
| "learning_rate": 0.00013703703703703702, | |
| "loss": 0.6198, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 1.7688888888888887, | |
| "grad_norm": 0.05932031571865082, | |
| "learning_rate": 0.00013679012345679012, | |
| "loss": 0.433, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 1.771111111111111, | |
| "grad_norm": 0.062494318932294846, | |
| "learning_rate": 0.0001365432098765432, | |
| "loss": 0.5766, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 1.7733333333333334, | |
| "grad_norm": 0.0756322592496872, | |
| "learning_rate": 0.00013629629629629628, | |
| "loss": 0.6299, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 1.7755555555555556, | |
| "grad_norm": 0.04970455914735794, | |
| "learning_rate": 0.00013604938271604937, | |
| "loss": 0.5057, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 1.7777777777777777, | |
| "grad_norm": 0.061017874628305435, | |
| "learning_rate": 0.00013580246913580244, | |
| "loss": 0.5729, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "grad_norm": 0.056633949279785156, | |
| "learning_rate": 0.00013555555555555554, | |
| "loss": 0.4932, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 1.7822222222222224, | |
| "grad_norm": 0.05541403591632843, | |
| "learning_rate": 0.00013530864197530863, | |
| "loss": 0.4638, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 1.7844444444444445, | |
| "grad_norm": 0.06685563176870346, | |
| "learning_rate": 0.00013506172839506173, | |
| "loss": 0.4718, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 1.7866666666666666, | |
| "grad_norm": 0.06282782554626465, | |
| "learning_rate": 0.0001348148148148148, | |
| "loss": 0.4503, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 1.7888888888888888, | |
| "grad_norm": 0.08332068473100662, | |
| "learning_rate": 0.0001345679012345679, | |
| "loss": 0.5893, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 1.791111111111111, | |
| "grad_norm": 0.05621097981929779, | |
| "learning_rate": 0.00013432098765432099, | |
| "loss": 0.4665, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 1.7933333333333334, | |
| "grad_norm": 0.07574357837438583, | |
| "learning_rate": 0.00013407407407407405, | |
| "loss": 0.504, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 1.7955555555555556, | |
| "grad_norm": 0.09337465465068817, | |
| "learning_rate": 0.00013382716049382715, | |
| "loss": 0.6057, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 1.7977777777777777, | |
| "grad_norm": 0.052697040140628815, | |
| "learning_rate": 0.00013358024691358022, | |
| "loss": 0.5173, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "grad_norm": 0.05568508058786392, | |
| "learning_rate": 0.0001333333333333333, | |
| "loss": 0.5579, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 1.8022222222222222, | |
| "grad_norm": 0.05045437812805176, | |
| "learning_rate": 0.0001330864197530864, | |
| "loss": 0.6121, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 1.8044444444444445, | |
| "grad_norm": 0.069134920835495, | |
| "learning_rate": 0.0001328395061728395, | |
| "loss": 0.4159, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 1.8066666666666666, | |
| "grad_norm": 0.1204785630106926, | |
| "learning_rate": 0.00013259259259259257, | |
| "loss": 0.6306, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 1.8088888888888888, | |
| "grad_norm": 0.06287333369255066, | |
| "learning_rate": 0.00013234567901234567, | |
| "loss": 0.4427, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 1.8111111111111111, | |
| "grad_norm": 0.05693672224879265, | |
| "learning_rate": 0.00013209876543209876, | |
| "loss": 0.5253, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 1.8133333333333335, | |
| "grad_norm": 0.06401466578245163, | |
| "learning_rate": 0.00013185185185185183, | |
| "loss": 0.4845, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 1.8155555555555556, | |
| "grad_norm": 0.10396970808506012, | |
| "learning_rate": 0.00013160493827160492, | |
| "loss": 0.5815, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 1.8177777777777777, | |
| "grad_norm": 0.04332258924841881, | |
| "learning_rate": 0.00013135802469135802, | |
| "loss": 0.5852, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 1.8199999999999998, | |
| "grad_norm": 0.06762854009866714, | |
| "learning_rate": 0.00013111111111111111, | |
| "loss": 0.5667, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 1.8222222222222222, | |
| "grad_norm": 0.05695168673992157, | |
| "learning_rate": 0.00013086419753086418, | |
| "loss": 0.4791, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 1.8244444444444445, | |
| "grad_norm": 0.0671275183558464, | |
| "learning_rate": 0.00013061728395061728, | |
| "loss": 0.4719, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 1.8266666666666667, | |
| "grad_norm": 0.07296454906463623, | |
| "learning_rate": 0.00013037037037037035, | |
| "loss": 0.6347, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 1.8288888888888888, | |
| "grad_norm": 0.06824342161417007, | |
| "learning_rate": 0.00013012345679012344, | |
| "loss": 0.4157, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 1.8311111111111111, | |
| "grad_norm": 0.06561208516359329, | |
| "learning_rate": 0.00012987654320987654, | |
| "loss": 0.5368, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 1.8333333333333335, | |
| "grad_norm": 0.07323651015758514, | |
| "learning_rate": 0.0001296296296296296, | |
| "loss": 0.4789, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 1.8355555555555556, | |
| "grad_norm": 0.05986982211470604, | |
| "learning_rate": 0.0001293827160493827, | |
| "loss": 0.5182, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 1.8377777777777777, | |
| "grad_norm": 0.06120222806930542, | |
| "learning_rate": 0.0001291358024691358, | |
| "loss": 0.6269, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 1.8399999999999999, | |
| "grad_norm": 0.05974249541759491, | |
| "learning_rate": 0.0001288888888888889, | |
| "loss": 0.4981, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 1.8422222222222222, | |
| "grad_norm": 0.06369238346815109, | |
| "learning_rate": 0.00012864197530864196, | |
| "loss": 0.6065, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 1.8444444444444446, | |
| "grad_norm": 0.06819500029087067, | |
| "learning_rate": 0.00012839506172839505, | |
| "loss": 0.5027, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 1.8466666666666667, | |
| "grad_norm": 0.08742193132638931, | |
| "learning_rate": 0.00012814814814814812, | |
| "loss": 0.4614, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 1.8488888888888888, | |
| "grad_norm": 0.09370777010917664, | |
| "learning_rate": 0.00012790123456790122, | |
| "loss": 0.4883, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 1.8511111111111112, | |
| "grad_norm": 0.04948408156633377, | |
| "learning_rate": 0.0001276543209876543, | |
| "loss": 0.4401, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 1.8533333333333335, | |
| "grad_norm": 0.057232894003391266, | |
| "learning_rate": 0.0001274074074074074, | |
| "loss": 0.4827, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 1.8555555555555556, | |
| "grad_norm": 0.06303838640451431, | |
| "learning_rate": 0.00012716049382716047, | |
| "loss": 0.539, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 1.8577777777777778, | |
| "grad_norm": 0.058734916150569916, | |
| "learning_rate": 0.00012691358024691357, | |
| "loss": 0.5328, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 1.8599999999999999, | |
| "grad_norm": 0.06439004093408585, | |
| "learning_rate": 0.00012666666666666666, | |
| "loss": 0.5914, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 1.8622222222222222, | |
| "grad_norm": 0.06162082031369209, | |
| "learning_rate": 0.00012641975308641973, | |
| "loss": 0.5006, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 1.8644444444444446, | |
| "grad_norm": 0.08706016838550568, | |
| "learning_rate": 0.00012617283950617283, | |
| "loss": 0.5539, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 1.8666666666666667, | |
| "grad_norm": 0.06765034794807434, | |
| "learning_rate": 0.0001259259259259259, | |
| "loss": 0.4778, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 1.8688888888888888, | |
| "grad_norm": 0.09286016970872879, | |
| "learning_rate": 0.00012567901234567902, | |
| "loss": 0.7237, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 1.871111111111111, | |
| "grad_norm": 0.06110518053174019, | |
| "learning_rate": 0.00012543209876543209, | |
| "loss": 0.7471, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 1.8733333333333333, | |
| "grad_norm": 0.07018718868494034, | |
| "learning_rate": 0.00012518518518518518, | |
| "loss": 0.5854, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 1.8755555555555556, | |
| "grad_norm": 0.05846709385514259, | |
| "learning_rate": 0.00012493827160493825, | |
| "loss": 0.516, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 1.8777777777777778, | |
| "grad_norm": 0.07162352651357651, | |
| "learning_rate": 0.00012469135802469134, | |
| "loss": 0.778, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "grad_norm": 0.07620000839233398, | |
| "learning_rate": 0.00012444444444444444, | |
| "loss": 0.5572, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 1.8822222222222222, | |
| "grad_norm": 0.05490870773792267, | |
| "learning_rate": 0.0001241975308641975, | |
| "loss": 0.5349, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 1.8844444444444446, | |
| "grad_norm": 0.06622836738824844, | |
| "learning_rate": 0.0001239506172839506, | |
| "loss": 0.5669, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 1.8866666666666667, | |
| "grad_norm": 0.07705485075712204, | |
| "learning_rate": 0.0001237037037037037, | |
| "loss": 0.5705, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 1.8888888888888888, | |
| "grad_norm": 0.06699000298976898, | |
| "learning_rate": 0.0001234567901234568, | |
| "loss": 0.5654, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 1.891111111111111, | |
| "grad_norm": 0.06047092005610466, | |
| "learning_rate": 0.00012320987654320986, | |
| "loss": 0.4673, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 1.8933333333333333, | |
| "grad_norm": 0.07268223911523819, | |
| "learning_rate": 0.00012296296296296296, | |
| "loss": 0.6171, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 1.8955555555555557, | |
| "grad_norm": 0.10248938947916031, | |
| "learning_rate": 0.00012271604938271602, | |
| "loss": 0.5941, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 1.8977777777777778, | |
| "grad_norm": 0.05336601287126541, | |
| "learning_rate": 0.00012246913580246912, | |
| "loss": 0.4696, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "grad_norm": 0.058589834719896317, | |
| "learning_rate": 0.00012222222222222221, | |
| "loss": 0.5427, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 1.9022222222222223, | |
| "grad_norm": 0.06546634435653687, | |
| "learning_rate": 0.00012197530864197531, | |
| "loss": 0.5526, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 1.9044444444444446, | |
| "grad_norm": 0.06999081373214722, | |
| "learning_rate": 0.00012172839506172839, | |
| "loss": 0.5001, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 1.9066666666666667, | |
| "grad_norm": 0.0729992538690567, | |
| "learning_rate": 0.00012148148148148147, | |
| "loss": 0.5502, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 1.9088888888888889, | |
| "grad_norm": 0.05125622823834419, | |
| "learning_rate": 0.00012123456790123455, | |
| "loss": 0.5191, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 1.911111111111111, | |
| "grad_norm": 0.0560213103890419, | |
| "learning_rate": 0.00012098765432098764, | |
| "loss": 0.4794, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 1.9133333333333333, | |
| "grad_norm": 0.06891674548387527, | |
| "learning_rate": 0.00012074074074074073, | |
| "loss": 0.5763, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 1.9155555555555557, | |
| "grad_norm": 0.08869532495737076, | |
| "learning_rate": 0.00012049382716049381, | |
| "loss": 0.5773, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 1.9177777777777778, | |
| "grad_norm": 0.08977396786212921, | |
| "learning_rate": 0.0001202469135802469, | |
| "loss": 0.6123, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "grad_norm": 0.061748404055833817, | |
| "learning_rate": 0.00011999999999999999, | |
| "loss": 0.5121, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 1.9222222222222223, | |
| "grad_norm": 0.06991101056337357, | |
| "learning_rate": 0.00011975308641975308, | |
| "loss": 0.5506, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 1.9244444444444444, | |
| "grad_norm": 0.04891958832740784, | |
| "learning_rate": 0.00011950617283950617, | |
| "loss": 0.5519, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 1.9266666666666667, | |
| "grad_norm": 0.057783063501119614, | |
| "learning_rate": 0.00011925925925925925, | |
| "loss": 0.4751, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 1.9288888888888889, | |
| "grad_norm": 0.05776379629969597, | |
| "learning_rate": 0.00011901234567901233, | |
| "loss": 0.5066, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 1.931111111111111, | |
| "grad_norm": 0.06384903937578201, | |
| "learning_rate": 0.00011876543209876541, | |
| "loss": 0.5378, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 1.9333333333333333, | |
| "grad_norm": 0.05312483012676239, | |
| "learning_rate": 0.0001185185185185185, | |
| "loss": 0.4739, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 1.9355555555555557, | |
| "grad_norm": 0.06334760785102844, | |
| "learning_rate": 0.0001182716049382716, | |
| "loss": 0.4546, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 1.9377777777777778, | |
| "grad_norm": 0.047741394490003586, | |
| "learning_rate": 0.00011802469135802468, | |
| "loss": 0.5654, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "grad_norm": 0.06999547779560089, | |
| "learning_rate": 0.00011777777777777776, | |
| "loss": 0.5412, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 1.942222222222222, | |
| "grad_norm": 0.06686510145664215, | |
| "learning_rate": 0.00011753086419753086, | |
| "loss": 0.5527, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 1.9444444444444444, | |
| "grad_norm": 0.059512909501791, | |
| "learning_rate": 0.00011728395061728394, | |
| "loss": 0.542, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.9466666666666668, | |
| "grad_norm": 0.047394365072250366, | |
| "learning_rate": 0.00011703703703703702, | |
| "loss": 0.4956, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 1.948888888888889, | |
| "grad_norm": 0.08979500085115433, | |
| "learning_rate": 0.0001167901234567901, | |
| "loss": 0.5184, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 1.951111111111111, | |
| "grad_norm": 0.05302036926150322, | |
| "learning_rate": 0.00011654320987654319, | |
| "loss": 0.4902, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 1.9533333333333334, | |
| "grad_norm": 0.07505984604358673, | |
| "learning_rate": 0.0001162962962962963, | |
| "loss": 0.6523, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 1.9555555555555557, | |
| "grad_norm": 0.05769059807062149, | |
| "learning_rate": 0.00011604938271604938, | |
| "loss": 0.5161, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 1.9577777777777778, | |
| "grad_norm": 0.054265160113573074, | |
| "learning_rate": 0.00011580246913580246, | |
| "loss": 0.5704, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "grad_norm": 0.06840366125106812, | |
| "learning_rate": 0.00011555555555555554, | |
| "loss": 0.5586, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 1.962222222222222, | |
| "grad_norm": 0.05442436784505844, | |
| "learning_rate": 0.00011530864197530863, | |
| "loss": 0.5448, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 1.9644444444444444, | |
| "grad_norm": 0.07485603541135788, | |
| "learning_rate": 0.00011506172839506172, | |
| "loss": 0.5699, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 1.9666666666666668, | |
| "grad_norm": 0.05651551112532616, | |
| "learning_rate": 0.0001148148148148148, | |
| "loss": 0.5211, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 1.968888888888889, | |
| "grad_norm": 0.09889449924230576, | |
| "learning_rate": 0.00011456790123456789, | |
| "loss": 0.5635, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 1.971111111111111, | |
| "grad_norm": 0.06663019210100174, | |
| "learning_rate": 0.00011432098765432099, | |
| "loss": 0.6432, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 1.9733333333333334, | |
| "grad_norm": 0.06748384237289429, | |
| "learning_rate": 0.00011407407407407407, | |
| "loss": 0.6222, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 1.9755555555555555, | |
| "grad_norm": 0.06977667659521103, | |
| "learning_rate": 0.00011382716049382715, | |
| "loss": 0.6341, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 1.9777777777777779, | |
| "grad_norm": 0.07906489074230194, | |
| "learning_rate": 0.00011358024691358023, | |
| "loss": 0.5485, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "grad_norm": 0.05564168095588684, | |
| "learning_rate": 0.00011333333333333331, | |
| "loss": 0.5855, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 1.982222222222222, | |
| "grad_norm": 0.06901608407497406, | |
| "learning_rate": 0.00011308641975308641, | |
| "loss": 0.4678, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 1.9844444444444445, | |
| "grad_norm": 0.056001629680395126, | |
| "learning_rate": 0.00011283950617283949, | |
| "loss": 0.5875, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 1.9866666666666668, | |
| "grad_norm": 0.06345464289188385, | |
| "learning_rate": 0.00011259259259259259, | |
| "loss": 0.6737, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 1.988888888888889, | |
| "grad_norm": 0.06477094441652298, | |
| "learning_rate": 0.00011234567901234567, | |
| "loss": 0.467, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 1.991111111111111, | |
| "grad_norm": 0.06361080706119537, | |
| "learning_rate": 0.00011209876543209876, | |
| "loss": 0.5965, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 1.9933333333333332, | |
| "grad_norm": 0.08266133815050125, | |
| "learning_rate": 0.00011185185185185184, | |
| "loss": 0.6367, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 1.9955555555555555, | |
| "grad_norm": 0.07403647899627686, | |
| "learning_rate": 0.00011160493827160493, | |
| "loss": 0.4981, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 1.9977777777777779, | |
| "grad_norm": 0.06103847920894623, | |
| "learning_rate": 0.00011135802469135801, | |
| "loss": 0.6276, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 0.044981278479099274, | |
| "learning_rate": 0.00011111111111111109, | |
| "loss": 0.4664, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 2.002222222222222, | |
| "grad_norm": 0.05929357558488846, | |
| "learning_rate": 0.0001108641975308642, | |
| "loss": 0.4904, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 2.0044444444444443, | |
| "grad_norm": 0.05011327192187309, | |
| "learning_rate": 0.00011061728395061728, | |
| "loss": 0.5205, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 2.006666666666667, | |
| "grad_norm": 0.09381725639104843, | |
| "learning_rate": 0.00011037037037037036, | |
| "loss": 0.6632, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 2.008888888888889, | |
| "grad_norm": 0.05319998413324356, | |
| "learning_rate": 0.00011012345679012344, | |
| "loss": 0.5231, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 2.011111111111111, | |
| "grad_norm": 0.07952640950679779, | |
| "learning_rate": 0.00010987654320987654, | |
| "loss": 0.5011, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 2.013333333333333, | |
| "grad_norm": 0.07917460799217224, | |
| "learning_rate": 0.00010962962962962962, | |
| "loss": 0.5721, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 2.0155555555555558, | |
| "grad_norm": 0.06674852222204208, | |
| "learning_rate": 0.0001093827160493827, | |
| "loss": 0.588, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 2.017777777777778, | |
| "grad_norm": 0.05087682232260704, | |
| "learning_rate": 0.00010913580246913578, | |
| "loss": 0.5812, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "grad_norm": 0.06378140300512314, | |
| "learning_rate": 0.00010888888888888889, | |
| "loss": 0.5039, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 2.022222222222222, | |
| "grad_norm": 0.06806355714797974, | |
| "learning_rate": 0.00010864197530864197, | |
| "loss": 0.7617, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 2.0244444444444443, | |
| "grad_norm": 0.05110703036189079, | |
| "learning_rate": 0.00010839506172839505, | |
| "loss": 0.4773, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 2.026666666666667, | |
| "grad_norm": 0.08746683597564697, | |
| "learning_rate": 0.00010814814814814814, | |
| "loss": 0.511, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 2.028888888888889, | |
| "grad_norm": 0.06579595804214478, | |
| "learning_rate": 0.00010790123456790122, | |
| "loss": 0.6538, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 2.031111111111111, | |
| "grad_norm": 0.05616295337677002, | |
| "learning_rate": 0.00010765432098765431, | |
| "loss": 0.5027, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 2.033333333333333, | |
| "grad_norm": 0.08300033956766129, | |
| "learning_rate": 0.0001074074074074074, | |
| "loss": 0.6057, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 2.0355555555555553, | |
| "grad_norm": 0.05325921252369881, | |
| "learning_rate": 0.00010716049382716048, | |
| "loss": 0.5811, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 2.037777777777778, | |
| "grad_norm": 0.06213977560400963, | |
| "learning_rate": 0.00010691358024691357, | |
| "loss": 0.5295, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "grad_norm": 0.050137948244810104, | |
| "learning_rate": 0.00010666666666666667, | |
| "loss": 0.5571, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 2.042222222222222, | |
| "grad_norm": 0.15031930804252625, | |
| "learning_rate": 0.00010641975308641975, | |
| "loss": 0.6984, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 2.0444444444444443, | |
| "grad_norm": 0.06720391660928726, | |
| "learning_rate": 0.00010617283950617283, | |
| "loss": 0.4943, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 2.046666666666667, | |
| "grad_norm": 0.06666924804449081, | |
| "learning_rate": 0.00010592592592592591, | |
| "loss": 0.4965, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 2.048888888888889, | |
| "grad_norm": 0.06180984526872635, | |
| "learning_rate": 0.00010567901234567899, | |
| "loss": 0.4507, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 2.051111111111111, | |
| "grad_norm": 0.07479197531938553, | |
| "learning_rate": 0.00010543209876543209, | |
| "loss": 0.7286, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 2.0533333333333332, | |
| "grad_norm": 0.06343316286802292, | |
| "learning_rate": 0.00010518518518518518, | |
| "loss": 0.5466, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 2.0555555555555554, | |
| "grad_norm": 0.05978839099407196, | |
| "learning_rate": 0.00010493827160493826, | |
| "loss": 0.4675, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 2.057777777777778, | |
| "grad_norm": 0.06739088892936707, | |
| "learning_rate": 0.00010469135802469135, | |
| "loss": 0.5894, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "grad_norm": 0.08694419264793396, | |
| "learning_rate": 0.00010444444444444444, | |
| "loss": 0.5999, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 2.062222222222222, | |
| "grad_norm": 0.06224658340215683, | |
| "learning_rate": 0.00010419753086419752, | |
| "loss": 0.4674, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 2.0644444444444443, | |
| "grad_norm": 0.05369469150900841, | |
| "learning_rate": 0.0001039506172839506, | |
| "loss": 0.6258, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 2.066666666666667, | |
| "grad_norm": 0.08104464411735535, | |
| "learning_rate": 0.00010370370370370369, | |
| "loss": 0.6806, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 2.068888888888889, | |
| "grad_norm": 0.07722575217485428, | |
| "learning_rate": 0.00010345679012345677, | |
| "loss": 0.7507, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 2.071111111111111, | |
| "grad_norm": 0.06280424445867538, | |
| "learning_rate": 0.00010320987654320988, | |
| "loss": 0.6387, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 2.0733333333333333, | |
| "grad_norm": 0.0635446161031723, | |
| "learning_rate": 0.00010296296296296296, | |
| "loss": 0.519, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 2.0755555555555554, | |
| "grad_norm": 0.0843411311507225, | |
| "learning_rate": 0.00010271604938271604, | |
| "loss": 0.5354, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 2.077777777777778, | |
| "grad_norm": 0.058972544968128204, | |
| "learning_rate": 0.00010246913580246912, | |
| "loss": 0.6468, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "grad_norm": 0.08231063932180405, | |
| "learning_rate": 0.00010222222222222222, | |
| "loss": 0.4815, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 2.082222222222222, | |
| "grad_norm": 0.04656016454100609, | |
| "learning_rate": 0.0001019753086419753, | |
| "loss": 0.5225, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 2.0844444444444443, | |
| "grad_norm": 0.06876270473003387, | |
| "learning_rate": 0.00010172839506172838, | |
| "loss": 0.676, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 2.086666666666667, | |
| "grad_norm": 0.0567607656121254, | |
| "learning_rate": 0.00010148148148148147, | |
| "loss": 0.5187, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 2.088888888888889, | |
| "grad_norm": 0.07258874177932739, | |
| "learning_rate": 0.00010123456790123457, | |
| "loss": 0.5526, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 2.091111111111111, | |
| "grad_norm": 0.0688469335436821, | |
| "learning_rate": 0.00010098765432098765, | |
| "loss": 0.4792, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 2.0933333333333333, | |
| "grad_norm": 0.1316925436258316, | |
| "learning_rate": 0.00010074074074074073, | |
| "loss": 0.5259, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 2.0955555555555554, | |
| "grad_norm": 0.06025873124599457, | |
| "learning_rate": 0.00010049382716049381, | |
| "loss": 0.4789, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 2.097777777777778, | |
| "grad_norm": 0.07988710701465607, | |
| "learning_rate": 0.0001002469135802469, | |
| "loss": 0.6522, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "grad_norm": 0.06128866225481033, | |
| "learning_rate": 9.999999999999999e-05, | |
| "loss": 0.4895, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 2.102222222222222, | |
| "grad_norm": 0.05999578535556793, | |
| "learning_rate": 9.975308641975307e-05, | |
| "loss": 0.6414, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 2.1044444444444443, | |
| "grad_norm": 0.059802647680044174, | |
| "learning_rate": 9.950617283950617e-05, | |
| "loss": 0.4996, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 2.1066666666666665, | |
| "grad_norm": 0.058329831808805466, | |
| "learning_rate": 9.925925925925925e-05, | |
| "loss": 0.7335, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 2.108888888888889, | |
| "grad_norm": 0.054771069437265396, | |
| "learning_rate": 9.901234567901234e-05, | |
| "loss": 0.525, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 2.111111111111111, | |
| "grad_norm": 0.05562947317957878, | |
| "learning_rate": 9.876543209876543e-05, | |
| "loss": 0.4125, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 2.1133333333333333, | |
| "grad_norm": 0.05977302044630051, | |
| "learning_rate": 9.851851851851851e-05, | |
| "loss": 0.4594, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 2.1155555555555554, | |
| "grad_norm": 0.0554046556353569, | |
| "learning_rate": 9.827160493827159e-05, | |
| "loss": 0.5191, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 2.117777777777778, | |
| "grad_norm": 0.058400023728609085, | |
| "learning_rate": 9.802469135802467e-05, | |
| "loss": 0.6893, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "grad_norm": 0.07758837938308716, | |
| "learning_rate": 9.777777777777778e-05, | |
| "loss": 0.5919, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 2.1222222222222222, | |
| "grad_norm": 0.0872005820274353, | |
| "learning_rate": 9.753086419753086e-05, | |
| "loss": 0.6871, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 2.1244444444444444, | |
| "grad_norm": 0.08081839233636856, | |
| "learning_rate": 9.728395061728394e-05, | |
| "loss": 0.6952, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 2.1266666666666665, | |
| "grad_norm": 0.06615129858255386, | |
| "learning_rate": 9.703703703703702e-05, | |
| "loss": 0.5317, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 2.128888888888889, | |
| "grad_norm": 0.09356512874364853, | |
| "learning_rate": 9.679012345679012e-05, | |
| "loss": 0.6803, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 2.131111111111111, | |
| "grad_norm": 0.05351172760128975, | |
| "learning_rate": 9.65432098765432e-05, | |
| "loss": 0.4233, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 2.1333333333333333, | |
| "grad_norm": 0.0722617581486702, | |
| "learning_rate": 9.629629629629628e-05, | |
| "loss": 0.6503, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 2.1355555555555554, | |
| "grad_norm": 0.060257527977228165, | |
| "learning_rate": 9.604938271604936e-05, | |
| "loss": 0.4915, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 2.137777777777778, | |
| "grad_norm": 0.10558751225471497, | |
| "learning_rate": 9.580246913580247e-05, | |
| "loss": 0.5252, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "grad_norm": 0.05734710022807121, | |
| "learning_rate": 9.555555555555555e-05, | |
| "loss": 0.6058, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 2.1422222222222222, | |
| "grad_norm": 0.06799523532390594, | |
| "learning_rate": 9.530864197530864e-05, | |
| "loss": 0.5999, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 2.1444444444444444, | |
| "grad_norm": 0.07418831437826157, | |
| "learning_rate": 9.506172839506172e-05, | |
| "loss": 0.6482, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 2.1466666666666665, | |
| "grad_norm": 0.050821445882320404, | |
| "learning_rate": 9.48148148148148e-05, | |
| "loss": 0.441, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 2.148888888888889, | |
| "grad_norm": 0.05051890015602112, | |
| "learning_rate": 9.45679012345679e-05, | |
| "loss": 0.4902, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 2.151111111111111, | |
| "grad_norm": 0.10794457793235779, | |
| "learning_rate": 9.432098765432098e-05, | |
| "loss": 0.5588, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 2.1533333333333333, | |
| "grad_norm": 0.07309269160032272, | |
| "learning_rate": 9.407407407407407e-05, | |
| "loss": 0.5948, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 2.1555555555555554, | |
| "grad_norm": 0.049595270305871964, | |
| "learning_rate": 9.382716049382715e-05, | |
| "loss": 0.5163, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 2.1577777777777776, | |
| "grad_norm": 0.09771185368299484, | |
| "learning_rate": 9.358024691358025e-05, | |
| "loss": 0.6503, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "grad_norm": 0.05198473855853081, | |
| "learning_rate": 9.333333333333333e-05, | |
| "loss": 0.5181, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 2.1622222222222223, | |
| "grad_norm": 0.07002071291208267, | |
| "learning_rate": 9.308641975308641e-05, | |
| "loss": 0.6796, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 2.1644444444444444, | |
| "grad_norm": 0.11276919394731522, | |
| "learning_rate": 9.283950617283949e-05, | |
| "loss": 0.5194, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 2.1666666666666665, | |
| "grad_norm": 0.05655023828148842, | |
| "learning_rate": 9.259259259259257e-05, | |
| "loss": 0.5285, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 2.168888888888889, | |
| "grad_norm": 0.08103014528751373, | |
| "learning_rate": 9.234567901234567e-05, | |
| "loss": 0.5859, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 2.171111111111111, | |
| "grad_norm": 0.044183384627103806, | |
| "learning_rate": 9.209876543209876e-05, | |
| "loss": 0.4693, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 2.1733333333333333, | |
| "grad_norm": 0.0533757321536541, | |
| "learning_rate": 9.185185185185185e-05, | |
| "loss": 0.5282, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 2.1755555555555555, | |
| "grad_norm": 0.048276450484991074, | |
| "learning_rate": 9.160493827160493e-05, | |
| "loss": 0.4395, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 2.1777777777777776, | |
| "grad_norm": 0.10359742492437363, | |
| "learning_rate": 9.135802469135802e-05, | |
| "loss": 0.5895, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "grad_norm": 0.05386464670300484, | |
| "learning_rate": 9.11111111111111e-05, | |
| "loss": 0.4812, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 2.1822222222222223, | |
| "grad_norm": 0.072238028049469, | |
| "learning_rate": 9.086419753086419e-05, | |
| "loss": 0.7056, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 2.1844444444444444, | |
| "grad_norm": 0.06497480720281601, | |
| "learning_rate": 9.061728395061727e-05, | |
| "loss": 0.5694, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 2.1866666666666665, | |
| "grad_norm": 0.0650085061788559, | |
| "learning_rate": 9.037037037037038e-05, | |
| "loss": 0.535, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 2.188888888888889, | |
| "grad_norm": 0.08351404964923859, | |
| "learning_rate": 9.012345679012346e-05, | |
| "loss": 0.6464, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 2.1911111111111112, | |
| "grad_norm": 0.060226235538721085, | |
| "learning_rate": 8.987654320987654e-05, | |
| "loss": 0.4206, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 2.1933333333333334, | |
| "grad_norm": 0.07895997911691666, | |
| "learning_rate": 8.962962962962962e-05, | |
| "loss": 0.532, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 2.1955555555555555, | |
| "grad_norm": 0.06073553487658501, | |
| "learning_rate": 8.93827160493827e-05, | |
| "loss": 0.3753, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 2.1977777777777776, | |
| "grad_norm": 0.06916957348585129, | |
| "learning_rate": 8.91358024691358e-05, | |
| "loss": 0.5444, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "grad_norm": 0.05763095244765282, | |
| "learning_rate": 8.888888888888888e-05, | |
| "loss": 0.5598, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 2.2022222222222223, | |
| "grad_norm": 0.07719336450099945, | |
| "learning_rate": 8.864197530864196e-05, | |
| "loss": 0.6593, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 2.2044444444444444, | |
| "grad_norm": 0.04980287328362465, | |
| "learning_rate": 8.839506172839506e-05, | |
| "loss": 0.4634, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 2.2066666666666666, | |
| "grad_norm": 0.07499777525663376, | |
| "learning_rate": 8.814814814814815e-05, | |
| "loss": 0.6515, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 2.2088888888888887, | |
| "grad_norm": 0.06283349543809891, | |
| "learning_rate": 8.790123456790123e-05, | |
| "loss": 0.554, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 2.2111111111111112, | |
| "grad_norm": 0.07238056510686874, | |
| "learning_rate": 8.765432098765431e-05, | |
| "loss": 0.5551, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 2.2133333333333334, | |
| "grad_norm": 0.06402061879634857, | |
| "learning_rate": 8.74074074074074e-05, | |
| "loss": 0.4701, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 2.2155555555555555, | |
| "grad_norm": 0.06097758188843727, | |
| "learning_rate": 8.716049382716048e-05, | |
| "loss": 0.5056, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 2.2177777777777776, | |
| "grad_norm": 0.066048763692379, | |
| "learning_rate": 8.691358024691357e-05, | |
| "loss": 0.5434, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "grad_norm": 0.04893951117992401, | |
| "learning_rate": 8.666666666666665e-05, | |
| "loss": 0.5412, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 0.061332616955041885, | |
| "learning_rate": 8.641975308641975e-05, | |
| "loss": 0.5636, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.2244444444444444, | |
| "grad_norm": 0.08975567668676376, | |
| "learning_rate": 8.617283950617283e-05, | |
| "loss": 0.4669, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 2.2266666666666666, | |
| "grad_norm": 0.04541618376970291, | |
| "learning_rate": 8.592592592592593e-05, | |
| "loss": 0.4714, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 2.2288888888888887, | |
| "grad_norm": 0.07543133199214935, | |
| "learning_rate": 8.567901234567901e-05, | |
| "loss": 0.5177, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 2.2311111111111113, | |
| "grad_norm": 0.056555863469839096, | |
| "learning_rate": 8.543209876543209e-05, | |
| "loss": 0.5839, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 2.2333333333333334, | |
| "grad_norm": 0.0559513233602047, | |
| "learning_rate": 8.518518518518517e-05, | |
| "loss": 0.4586, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 2.2355555555555555, | |
| "grad_norm": 0.05552906543016434, | |
| "learning_rate": 8.493827160493825e-05, | |
| "loss": 0.5174, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 2.2377777777777776, | |
| "grad_norm": 0.0614011324942112, | |
| "learning_rate": 8.469135802469136e-05, | |
| "loss": 0.6988, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "grad_norm": 0.06229771301150322, | |
| "learning_rate": 8.444444444444444e-05, | |
| "loss": 0.4308, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 2.2422222222222223, | |
| "grad_norm": 0.0725821703672409, | |
| "learning_rate": 8.419753086419752e-05, | |
| "loss": 0.5323, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 2.2444444444444445, | |
| "grad_norm": 0.08226242661476135, | |
| "learning_rate": 8.39506172839506e-05, | |
| "loss": 0.5034, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 2.2466666666666666, | |
| "grad_norm": 0.06674276292324066, | |
| "learning_rate": 8.37037037037037e-05, | |
| "loss": 0.4633, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 2.2488888888888887, | |
| "grad_norm": 0.060701869428157806, | |
| "learning_rate": 8.345679012345678e-05, | |
| "loss": 0.4909, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 2.2511111111111113, | |
| "grad_norm": 0.063337542116642, | |
| "learning_rate": 8.320987654320986e-05, | |
| "loss": 0.5566, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 2.2533333333333334, | |
| "grad_norm": 0.05131515860557556, | |
| "learning_rate": 8.296296296296295e-05, | |
| "loss": 0.492, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 2.2555555555555555, | |
| "grad_norm": 0.056889623403549194, | |
| "learning_rate": 8.271604938271605e-05, | |
| "loss": 0.4611, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 2.2577777777777777, | |
| "grad_norm": 0.06834527850151062, | |
| "learning_rate": 8.246913580246914e-05, | |
| "loss": 0.7076, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "grad_norm": 0.06255058944225311, | |
| "learning_rate": 8.222222222222222e-05, | |
| "loss": 0.5511, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 2.2622222222222224, | |
| "grad_norm": 0.062491778284311295, | |
| "learning_rate": 8.19753086419753e-05, | |
| "loss": 0.6179, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 2.2644444444444445, | |
| "grad_norm": 0.05383693054318428, | |
| "learning_rate": 8.172839506172838e-05, | |
| "loss": 0.4916, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 2.2666666666666666, | |
| "grad_norm": 0.0685378834605217, | |
| "learning_rate": 8.148148148148148e-05, | |
| "loss": 0.6815, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 2.2688888888888887, | |
| "grad_norm": 0.05592293292284012, | |
| "learning_rate": 8.123456790123456e-05, | |
| "loss": 0.6527, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 2.2711111111111113, | |
| "grad_norm": 0.06338860839605331, | |
| "learning_rate": 8.098765432098765e-05, | |
| "loss": 0.6013, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 2.2733333333333334, | |
| "grad_norm": 0.06278155744075775, | |
| "learning_rate": 8.074074074074073e-05, | |
| "loss": 0.591, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 2.2755555555555556, | |
| "grad_norm": 0.07255859673023224, | |
| "learning_rate": 8.049382716049383e-05, | |
| "loss": 0.5748, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 2.2777777777777777, | |
| "grad_norm": 0.06974730640649796, | |
| "learning_rate": 8.024691358024691e-05, | |
| "loss": 0.489, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 2.2800000000000002, | |
| "grad_norm": 0.06803488731384277, | |
| "learning_rate": 7.999999999999999e-05, | |
| "loss": 0.4965, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 2.2822222222222224, | |
| "grad_norm": 0.06226198375225067, | |
| "learning_rate": 7.975308641975307e-05, | |
| "loss": 0.5525, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 2.2844444444444445, | |
| "grad_norm": 0.05195463076233864, | |
| "learning_rate": 7.950617283950616e-05, | |
| "loss": 0.5943, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 2.2866666666666666, | |
| "grad_norm": 0.06786032766103745, | |
| "learning_rate": 7.925925925925925e-05, | |
| "loss": 0.5846, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 2.2888888888888888, | |
| "grad_norm": 0.055401552468538284, | |
| "learning_rate": 7.901234567901235e-05, | |
| "loss": 0.4741, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 2.2911111111111113, | |
| "grad_norm": 0.06985199451446533, | |
| "learning_rate": 7.876543209876543e-05, | |
| "loss": 0.5871, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 2.2933333333333334, | |
| "grad_norm": 0.060605403035879135, | |
| "learning_rate": 7.851851851851851e-05, | |
| "loss": 0.587, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 2.2955555555555556, | |
| "grad_norm": 0.058030206710100174, | |
| "learning_rate": 7.82716049382716e-05, | |
| "loss": 0.5585, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 2.2977777777777777, | |
| "grad_norm": 0.0632559284567833, | |
| "learning_rate": 7.802469135802469e-05, | |
| "loss": 0.5816, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "grad_norm": 0.045592762529850006, | |
| "learning_rate": 7.777777777777777e-05, | |
| "loss": 0.5184, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 2.3022222222222224, | |
| "grad_norm": 0.07075291126966476, | |
| "learning_rate": 7.753086419753085e-05, | |
| "loss": 0.5876, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 2.3044444444444445, | |
| "grad_norm": 0.07266212999820709, | |
| "learning_rate": 7.728395061728396e-05, | |
| "loss": 0.5915, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 2.3066666666666666, | |
| "grad_norm": 0.0763152539730072, | |
| "learning_rate": 7.703703703703704e-05, | |
| "loss": 0.4925, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 2.3088888888888888, | |
| "grad_norm": 0.05008356273174286, | |
| "learning_rate": 7.679012345679012e-05, | |
| "loss": 0.5566, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 2.311111111111111, | |
| "grad_norm": 0.058586686849594116, | |
| "learning_rate": 7.65432098765432e-05, | |
| "loss": 0.5645, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 2.3133333333333335, | |
| "grad_norm": 0.06809350848197937, | |
| "learning_rate": 7.629629629629628e-05, | |
| "loss": 0.683, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 2.3155555555555556, | |
| "grad_norm": 0.06814820319414139, | |
| "learning_rate": 7.604938271604937e-05, | |
| "loss": 0.4951, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 2.3177777777777777, | |
| "grad_norm": 0.06426460295915604, | |
| "learning_rate": 7.580246913580246e-05, | |
| "loss": 0.5607, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "grad_norm": 0.06859628111124039, | |
| "learning_rate": 7.555555555555554e-05, | |
| "loss": 0.5394, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 2.3222222222222224, | |
| "grad_norm": 0.08079138398170471, | |
| "learning_rate": 7.530864197530864e-05, | |
| "loss": 0.591, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 2.3244444444444445, | |
| "grad_norm": 0.06668093800544739, | |
| "learning_rate": 7.506172839506173e-05, | |
| "loss": 0.5569, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 2.3266666666666667, | |
| "grad_norm": 0.05966291204094887, | |
| "learning_rate": 7.481481481481481e-05, | |
| "loss": 0.4233, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 2.328888888888889, | |
| "grad_norm": 0.058995746076107025, | |
| "learning_rate": 7.45679012345679e-05, | |
| "loss": 0.5384, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 2.3311111111111114, | |
| "grad_norm": 0.050750378519296646, | |
| "learning_rate": 7.432098765432098e-05, | |
| "loss": 0.5719, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 2.3333333333333335, | |
| "grad_norm": 0.054379820823669434, | |
| "learning_rate": 7.407407407407406e-05, | |
| "loss": 0.5022, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 2.3355555555555556, | |
| "grad_norm": 0.06929490715265274, | |
| "learning_rate": 7.382716049382715e-05, | |
| "loss": 0.6144, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 2.3377777777777777, | |
| "grad_norm": 0.0617499053478241, | |
| "learning_rate": 7.358024691358024e-05, | |
| "loss": 0.4842, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "grad_norm": 0.1699656844139099, | |
| "learning_rate": 7.333333333333332e-05, | |
| "loss": 0.4776, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 2.3422222222222224, | |
| "grad_norm": 0.070797398686409, | |
| "learning_rate": 7.308641975308641e-05, | |
| "loss": 0.4883, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 2.3444444444444446, | |
| "grad_norm": 0.061980705708265305, | |
| "learning_rate": 7.28395061728395e-05, | |
| "loss": 0.5367, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 2.3466666666666667, | |
| "grad_norm": 0.0650290697813034, | |
| "learning_rate": 7.259259259259259e-05, | |
| "loss": 0.4806, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 2.348888888888889, | |
| "grad_norm": 0.05629928782582283, | |
| "learning_rate": 7.234567901234567e-05, | |
| "loss": 0.4375, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 2.351111111111111, | |
| "grad_norm": 0.0574677512049675, | |
| "learning_rate": 7.209876543209877e-05, | |
| "loss": 0.4422, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 2.3533333333333335, | |
| "grad_norm": 0.045402076095342636, | |
| "learning_rate": 7.185185185185185e-05, | |
| "loss": 0.527, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 2.3555555555555556, | |
| "grad_norm": 0.047060877084732056, | |
| "learning_rate": 7.160493827160493e-05, | |
| "loss": 0.4543, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 2.3577777777777778, | |
| "grad_norm": 0.05593902990221977, | |
| "learning_rate": 7.135802469135801e-05, | |
| "loss": 0.486, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "grad_norm": 0.042419929057359695, | |
| "learning_rate": 7.11111111111111e-05, | |
| "loss": 0.5189, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 2.362222222222222, | |
| "grad_norm": 0.05823193117976189, | |
| "learning_rate": 7.086419753086419e-05, | |
| "loss": 0.5521, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 2.3644444444444446, | |
| "grad_norm": 0.07853607833385468, | |
| "learning_rate": 7.061728395061727e-05, | |
| "loss": 0.5679, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 2.3666666666666667, | |
| "grad_norm": 0.056772686541080475, | |
| "learning_rate": 7.037037037037036e-05, | |
| "loss": 0.5089, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 2.368888888888889, | |
| "grad_norm": 0.059823885560035706, | |
| "learning_rate": 7.012345679012345e-05, | |
| "loss": 0.601, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 2.371111111111111, | |
| "grad_norm": 0.10615638643503189, | |
| "learning_rate": 6.987654320987654e-05, | |
| "loss": 0.5864, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 2.3733333333333335, | |
| "grad_norm": 0.06523310393095016, | |
| "learning_rate": 6.962962962962962e-05, | |
| "loss": 0.5912, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 2.3755555555555556, | |
| "grad_norm": 0.05638803914189339, | |
| "learning_rate": 6.938271604938272e-05, | |
| "loss": 0.4248, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 2.3777777777777778, | |
| "grad_norm": 0.05251540243625641, | |
| "learning_rate": 6.91358024691358e-05, | |
| "loss": 0.4581, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "grad_norm": 0.05011064559221268, | |
| "learning_rate": 6.888888888888888e-05, | |
| "loss": 0.5094, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 2.3822222222222225, | |
| "grad_norm": 0.06625556200742722, | |
| "learning_rate": 6.864197530864196e-05, | |
| "loss": 0.5462, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 2.3844444444444446, | |
| "grad_norm": 0.06464938819408417, | |
| "learning_rate": 6.839506172839506e-05, | |
| "loss": 0.5221, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 2.3866666666666667, | |
| "grad_norm": 0.06818525493144989, | |
| "learning_rate": 6.814814814814814e-05, | |
| "loss": 0.5493, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 2.388888888888889, | |
| "grad_norm": 0.08517734706401825, | |
| "learning_rate": 6.790123456790122e-05, | |
| "loss": 0.5241, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 2.391111111111111, | |
| "grad_norm": 0.049852389842271805, | |
| "learning_rate": 6.765432098765432e-05, | |
| "loss": 0.5092, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 2.3933333333333335, | |
| "grad_norm": 0.06493844836950302, | |
| "learning_rate": 6.74074074074074e-05, | |
| "loss": 0.7021, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 2.3955555555555557, | |
| "grad_norm": 0.04971351847052574, | |
| "learning_rate": 6.716049382716049e-05, | |
| "loss": 0.3808, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 2.397777777777778, | |
| "grad_norm": 0.07863829284906387, | |
| "learning_rate": 6.691358024691357e-05, | |
| "loss": 0.6705, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "grad_norm": 0.05218340456485748, | |
| "learning_rate": 6.666666666666666e-05, | |
| "loss": 0.504, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 2.402222222222222, | |
| "grad_norm": 0.05992546305060387, | |
| "learning_rate": 6.641975308641975e-05, | |
| "loss": 0.5528, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 2.4044444444444446, | |
| "grad_norm": 0.05170424282550812, | |
| "learning_rate": 6.617283950617283e-05, | |
| "loss": 0.5887, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 2.4066666666666667, | |
| "grad_norm": 0.06698056310415268, | |
| "learning_rate": 6.592592592592591e-05, | |
| "loss": 0.5743, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 2.408888888888889, | |
| "grad_norm": 0.0655369758605957, | |
| "learning_rate": 6.567901234567901e-05, | |
| "loss": 0.5041, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 2.411111111111111, | |
| "grad_norm": 0.05607493594288826, | |
| "learning_rate": 6.543209876543209e-05, | |
| "loss": 0.5527, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 2.413333333333333, | |
| "grad_norm": 0.05267152935266495, | |
| "learning_rate": 6.518518518518517e-05, | |
| "loss": 0.405, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 2.4155555555555557, | |
| "grad_norm": 0.051925208419561386, | |
| "learning_rate": 6.493827160493827e-05, | |
| "loss": 0.4746, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 2.417777777777778, | |
| "grad_norm": 0.0693216323852539, | |
| "learning_rate": 6.469135802469135e-05, | |
| "loss": 0.5639, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "grad_norm": 0.04773171246051788, | |
| "learning_rate": 6.444444444444444e-05, | |
| "loss": 0.5445, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 2.422222222222222, | |
| "grad_norm": 0.059276122599840164, | |
| "learning_rate": 6.419753086419753e-05, | |
| "loss": 0.5707, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 2.4244444444444446, | |
| "grad_norm": 0.05703291296958923, | |
| "learning_rate": 6.395061728395061e-05, | |
| "loss": 0.6186, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 2.4266666666666667, | |
| "grad_norm": 0.059713978320360184, | |
| "learning_rate": 6.37037037037037e-05, | |
| "loss": 0.5472, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 2.428888888888889, | |
| "grad_norm": 0.07642186433076859, | |
| "learning_rate": 6.345679012345678e-05, | |
| "loss": 0.6143, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 2.431111111111111, | |
| "grad_norm": 0.06594318151473999, | |
| "learning_rate": 6.320987654320987e-05, | |
| "loss": 0.4702, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 2.4333333333333336, | |
| "grad_norm": 0.07548554241657257, | |
| "learning_rate": 6.296296296296295e-05, | |
| "loss": 0.5768, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 2.4355555555555557, | |
| "grad_norm": 0.06707516312599182, | |
| "learning_rate": 6.271604938271604e-05, | |
| "loss": 0.6927, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 2.437777777777778, | |
| "grad_norm": 0.0729343593120575, | |
| "learning_rate": 6.246913580246912e-05, | |
| "loss": 0.7841, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "grad_norm": 0.05313663184642792, | |
| "learning_rate": 6.222222222222222e-05, | |
| "loss": 0.5131, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 2.442222222222222, | |
| "grad_norm": 0.07384145259857178, | |
| "learning_rate": 6.19753086419753e-05, | |
| "loss": 0.6907, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 2.4444444444444446, | |
| "grad_norm": 0.051853831857442856, | |
| "learning_rate": 6.17283950617284e-05, | |
| "loss": 0.6614, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 2.4466666666666668, | |
| "grad_norm": 0.04776390641927719, | |
| "learning_rate": 6.148148148148148e-05, | |
| "loss": 0.4446, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 2.448888888888889, | |
| "grad_norm": 0.06168276071548462, | |
| "learning_rate": 6.123456790123456e-05, | |
| "loss": 0.5067, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 2.451111111111111, | |
| "grad_norm": 0.05856132134795189, | |
| "learning_rate": 6.0987654320987655e-05, | |
| "loss": 0.5644, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 2.453333333333333, | |
| "grad_norm": 0.06515628844499588, | |
| "learning_rate": 6.0740740740740736e-05, | |
| "loss": 0.6779, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 2.4555555555555557, | |
| "grad_norm": 0.06323693692684174, | |
| "learning_rate": 6.049382716049382e-05, | |
| "loss": 0.577, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 2.457777777777778, | |
| "grad_norm": 0.05601757392287254, | |
| "learning_rate": 6.0246913580246906e-05, | |
| "loss": 0.6251, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "grad_norm": 0.06924091279506683, | |
| "learning_rate": 5.9999999999999995e-05, | |
| "loss": 0.4927, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 2.462222222222222, | |
| "grad_norm": 0.1901812106370926, | |
| "learning_rate": 5.975308641975308e-05, | |
| "loss": 0.6125, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 2.464444444444444, | |
| "grad_norm": 0.06211036071181297, | |
| "learning_rate": 5.9506172839506165e-05, | |
| "loss": 0.741, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 2.466666666666667, | |
| "grad_norm": 0.05459253862500191, | |
| "learning_rate": 5.925925925925925e-05, | |
| "loss": 0.5158, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 2.468888888888889, | |
| "grad_norm": 0.06459296494722366, | |
| "learning_rate": 5.901234567901234e-05, | |
| "loss": 0.6143, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 2.471111111111111, | |
| "grad_norm": 0.0627160295844078, | |
| "learning_rate": 5.876543209876543e-05, | |
| "loss": 0.4867, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 2.473333333333333, | |
| "grad_norm": 0.05840280279517174, | |
| "learning_rate": 5.851851851851851e-05, | |
| "loss": 0.6229, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 2.4755555555555557, | |
| "grad_norm": 0.07121528685092926, | |
| "learning_rate": 5.827160493827159e-05, | |
| "loss": 0.4508, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 2.477777777777778, | |
| "grad_norm": 0.08437851071357727, | |
| "learning_rate": 5.802469135802469e-05, | |
| "loss": 0.6107, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "grad_norm": 0.09962430596351624, | |
| "learning_rate": 5.777777777777777e-05, | |
| "loss": 0.6031, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 2.482222222222222, | |
| "grad_norm": 0.06262155622243881, | |
| "learning_rate": 5.753086419753086e-05, | |
| "loss": 0.4977, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 2.4844444444444447, | |
| "grad_norm": 0.05871571972966194, | |
| "learning_rate": 5.7283950617283946e-05, | |
| "loss": 0.5218, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 2.486666666666667, | |
| "grad_norm": 0.05485491827130318, | |
| "learning_rate": 5.7037037037037035e-05, | |
| "loss": 0.5382, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 2.488888888888889, | |
| "grad_norm": 0.07214649766683578, | |
| "learning_rate": 5.6790123456790116e-05, | |
| "loss": 0.5695, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 2.491111111111111, | |
| "grad_norm": 0.06216902658343315, | |
| "learning_rate": 5.6543209876543205e-05, | |
| "loss": 0.5668, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 2.493333333333333, | |
| "grad_norm": 0.0855700746178627, | |
| "learning_rate": 5.629629629629629e-05, | |
| "loss": 0.5041, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 2.4955555555555557, | |
| "grad_norm": 0.05714331939816475, | |
| "learning_rate": 5.604938271604938e-05, | |
| "loss": 0.3766, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 2.497777777777778, | |
| "grad_norm": 0.06301258504390717, | |
| "learning_rate": 5.580246913580246e-05, | |
| "loss": 0.5554, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 0.07956915348768234, | |
| "learning_rate": 5.5555555555555545e-05, | |
| "loss": 0.566, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 2.502222222222222, | |
| "grad_norm": 0.06741563230752945, | |
| "learning_rate": 5.530864197530864e-05, | |
| "loss": 0.6758, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 2.5044444444444443, | |
| "grad_norm": 0.07749062776565552, | |
| "learning_rate": 5.506172839506172e-05, | |
| "loss": 0.5189, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 2.506666666666667, | |
| "grad_norm": 0.06751803308725357, | |
| "learning_rate": 5.481481481481481e-05, | |
| "loss": 0.5469, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 2.508888888888889, | |
| "grad_norm": 0.07618889212608337, | |
| "learning_rate": 5.456790123456789e-05, | |
| "loss": 0.6026, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 2.511111111111111, | |
| "grad_norm": 0.07808917760848999, | |
| "learning_rate": 5.4320987654320986e-05, | |
| "loss": 0.5355, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 2.513333333333333, | |
| "grad_norm": 0.07308845967054367, | |
| "learning_rate": 5.407407407407407e-05, | |
| "loss": 0.5843, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 2.5155555555555553, | |
| "grad_norm": 0.0436067134141922, | |
| "learning_rate": 5.3827160493827156e-05, | |
| "loss": 0.4646, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 2.517777777777778, | |
| "grad_norm": 0.054867230355739594, | |
| "learning_rate": 5.358024691358024e-05, | |
| "loss": 0.5209, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "grad_norm": 0.05941241979598999, | |
| "learning_rate": 5.333333333333333e-05, | |
| "loss": 0.4341, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 2.522222222222222, | |
| "grad_norm": 0.08015015721321106, | |
| "learning_rate": 5.3086419753086415e-05, | |
| "loss": 0.454, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 2.5244444444444447, | |
| "grad_norm": 0.07355854660272598, | |
| "learning_rate": 5.2839506172839496e-05, | |
| "loss": 0.5666, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 2.5266666666666664, | |
| "grad_norm": 0.051169779151678085, | |
| "learning_rate": 5.259259259259259e-05, | |
| "loss": 0.4666, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 2.528888888888889, | |
| "grad_norm": 0.06035088375210762, | |
| "learning_rate": 5.234567901234567e-05, | |
| "loss": 0.6008, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 2.531111111111111, | |
| "grad_norm": 0.06222086399793625, | |
| "learning_rate": 5.209876543209876e-05, | |
| "loss": 0.4605, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 2.533333333333333, | |
| "grad_norm": 0.06256081163883209, | |
| "learning_rate": 5.185185185185184e-05, | |
| "loss": 0.5614, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 2.535555555555556, | |
| "grad_norm": 0.0504666231572628, | |
| "learning_rate": 5.160493827160494e-05, | |
| "loss": 0.4592, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 2.537777777777778, | |
| "grad_norm": 0.057697102427482605, | |
| "learning_rate": 5.135802469135802e-05, | |
| "loss": 0.4808, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "grad_norm": 0.05934306979179382, | |
| "learning_rate": 5.111111111111111e-05, | |
| "loss": 0.6204, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 2.542222222222222, | |
| "grad_norm": 0.05678641051054001, | |
| "learning_rate": 5.086419753086419e-05, | |
| "loss": 0.4813, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 2.5444444444444443, | |
| "grad_norm": 0.049973636865615845, | |
| "learning_rate": 5.0617283950617285e-05, | |
| "loss": 0.4882, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 2.546666666666667, | |
| "grad_norm": 0.05903719365596771, | |
| "learning_rate": 5.0370370370370366e-05, | |
| "loss": 0.5108, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 2.548888888888889, | |
| "grad_norm": 0.06578648090362549, | |
| "learning_rate": 5.012345679012345e-05, | |
| "loss": 0.5471, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 2.551111111111111, | |
| "grad_norm": 0.07062222063541412, | |
| "learning_rate": 4.9876543209876536e-05, | |
| "loss": 0.5773, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 2.5533333333333332, | |
| "grad_norm": 0.06813468784093857, | |
| "learning_rate": 4.9629629629629625e-05, | |
| "loss": 0.5425, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 2.5555555555555554, | |
| "grad_norm": 0.05379043146967888, | |
| "learning_rate": 4.938271604938271e-05, | |
| "loss": 0.5869, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 2.557777777777778, | |
| "grad_norm": 0.06605621427297592, | |
| "learning_rate": 4.9135802469135795e-05, | |
| "loss": 0.5702, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "grad_norm": 0.05059510096907616, | |
| "learning_rate": 4.888888888888889e-05, | |
| "loss": 0.5515, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 2.562222222222222, | |
| "grad_norm": 0.08809015899896622, | |
| "learning_rate": 4.864197530864197e-05, | |
| "loss": 0.5429, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 2.5644444444444443, | |
| "grad_norm": 0.1451186090707779, | |
| "learning_rate": 4.839506172839506e-05, | |
| "loss": 0.5834, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 2.5666666666666664, | |
| "grad_norm": 0.0585605725646019, | |
| "learning_rate": 4.814814814814814e-05, | |
| "loss": 0.4758, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 2.568888888888889, | |
| "grad_norm": 0.07379104197025299, | |
| "learning_rate": 4.7901234567901237e-05, | |
| "loss": 0.5267, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 2.571111111111111, | |
| "grad_norm": 0.12918312847614288, | |
| "learning_rate": 4.765432098765432e-05, | |
| "loss": 0.7243, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 2.5733333333333333, | |
| "grad_norm": 0.0846041664481163, | |
| "learning_rate": 4.74074074074074e-05, | |
| "loss": 0.495, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 2.575555555555556, | |
| "grad_norm": 0.06449992209672928, | |
| "learning_rate": 4.716049382716049e-05, | |
| "loss": 0.6013, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 2.5777777777777775, | |
| "grad_norm": 0.07164155691862106, | |
| "learning_rate": 4.6913580246913576e-05, | |
| "loss": 0.5062, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "grad_norm": 0.057980503886938095, | |
| "learning_rate": 4.6666666666666665e-05, | |
| "loss": 0.5751, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 2.582222222222222, | |
| "grad_norm": 0.05791774019598961, | |
| "learning_rate": 4.6419753086419746e-05, | |
| "loss": 0.4968, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 2.5844444444444443, | |
| "grad_norm": 0.08030201494693756, | |
| "learning_rate": 4.6172839506172835e-05, | |
| "loss": 0.6533, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 2.586666666666667, | |
| "grad_norm": 0.07436682283878326, | |
| "learning_rate": 4.592592592592592e-05, | |
| "loss": 0.6033, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 2.588888888888889, | |
| "grad_norm": 0.08432338386774063, | |
| "learning_rate": 4.567901234567901e-05, | |
| "loss": 0.664, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 2.591111111111111, | |
| "grad_norm": 0.06954263150691986, | |
| "learning_rate": 4.543209876543209e-05, | |
| "loss": 0.6309, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 2.5933333333333333, | |
| "grad_norm": 0.05851338058710098, | |
| "learning_rate": 4.518518518518519e-05, | |
| "loss": 0.7941, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 2.5955555555555554, | |
| "grad_norm": 0.06911693513393402, | |
| "learning_rate": 4.493827160493827e-05, | |
| "loss": 0.5295, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 2.597777777777778, | |
| "grad_norm": 0.055619195103645325, | |
| "learning_rate": 4.469135802469135e-05, | |
| "loss": 0.5289, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "grad_norm": 0.07144150882959366, | |
| "learning_rate": 4.444444444444444e-05, | |
| "loss": 0.7668, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 2.602222222222222, | |
| "grad_norm": 0.05022842809557915, | |
| "learning_rate": 4.419753086419753e-05, | |
| "loss": 0.5514, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 2.6044444444444443, | |
| "grad_norm": 0.09054499119520187, | |
| "learning_rate": 4.3950617283950617e-05, | |
| "loss": 0.5177, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 2.6066666666666665, | |
| "grad_norm": 0.07231871783733368, | |
| "learning_rate": 4.37037037037037e-05, | |
| "loss": 0.5098, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 2.608888888888889, | |
| "grad_norm": 0.06764987111091614, | |
| "learning_rate": 4.3456790123456786e-05, | |
| "loss": 0.5113, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 2.611111111111111, | |
| "grad_norm": 0.06053946167230606, | |
| "learning_rate": 4.3209876543209875e-05, | |
| "loss": 0.5271, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 2.6133333333333333, | |
| "grad_norm": 0.07456693798303604, | |
| "learning_rate": 4.296296296296296e-05, | |
| "loss": 0.5984, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 2.6155555555555554, | |
| "grad_norm": 0.0758151188492775, | |
| "learning_rate": 4.2716049382716045e-05, | |
| "loss": 0.5604, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 2.6177777777777775, | |
| "grad_norm": 0.0626075491309166, | |
| "learning_rate": 4.2469135802469126e-05, | |
| "loss": 0.4546, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "grad_norm": 0.054678939282894135, | |
| "learning_rate": 4.222222222222222e-05, | |
| "loss": 0.5228, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 2.6222222222222222, | |
| "grad_norm": 0.07075554132461548, | |
| "learning_rate": 4.19753086419753e-05, | |
| "loss": 0.5948, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 2.6244444444444444, | |
| "grad_norm": 0.05450223758816719, | |
| "learning_rate": 4.172839506172839e-05, | |
| "loss": 0.5478, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 2.626666666666667, | |
| "grad_norm": 0.09620941430330276, | |
| "learning_rate": 4.148148148148147e-05, | |
| "loss": 0.5981, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 2.628888888888889, | |
| "grad_norm": 0.07030943036079407, | |
| "learning_rate": 4.123456790123457e-05, | |
| "loss": 0.5803, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 2.631111111111111, | |
| "grad_norm": 0.0563332624733448, | |
| "learning_rate": 4.098765432098765e-05, | |
| "loss": 0.5838, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 2.6333333333333333, | |
| "grad_norm": 0.05982625484466553, | |
| "learning_rate": 4.074074074074074e-05, | |
| "loss": 0.4185, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 2.6355555555555554, | |
| "grad_norm": 0.06112497299909592, | |
| "learning_rate": 4.0493827160493827e-05, | |
| "loss": 0.4933, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 2.637777777777778, | |
| "grad_norm": 0.06475501507520676, | |
| "learning_rate": 4.0246913580246915e-05, | |
| "loss": 0.5246, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "grad_norm": 0.05402704328298569, | |
| "learning_rate": 3.9999999999999996e-05, | |
| "loss": 0.5475, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 2.6422222222222222, | |
| "grad_norm": 0.04999251663684845, | |
| "learning_rate": 3.975308641975308e-05, | |
| "loss": 0.3999, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 2.6444444444444444, | |
| "grad_norm": 0.06521523743867874, | |
| "learning_rate": 3.950617283950617e-05, | |
| "loss": 0.6131, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 2.6466666666666665, | |
| "grad_norm": 0.053711630403995514, | |
| "learning_rate": 3.9259259259259255e-05, | |
| "loss": 0.5736, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 2.648888888888889, | |
| "grad_norm": 0.07394678145647049, | |
| "learning_rate": 3.901234567901234e-05, | |
| "loss": 0.4872, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 2.651111111111111, | |
| "grad_norm": 0.11503905057907104, | |
| "learning_rate": 3.8765432098765425e-05, | |
| "loss": 0.5291, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 2.6533333333333333, | |
| "grad_norm": 0.0591617114841938, | |
| "learning_rate": 3.851851851851852e-05, | |
| "loss": 0.5433, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 2.6555555555555554, | |
| "grad_norm": 0.06372568011283875, | |
| "learning_rate": 3.82716049382716e-05, | |
| "loss": 0.5799, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 2.6577777777777776, | |
| "grad_norm": 0.08090534806251526, | |
| "learning_rate": 3.802469135802468e-05, | |
| "loss": 0.5181, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "grad_norm": 0.06028290092945099, | |
| "learning_rate": 3.777777777777777e-05, | |
| "loss": 0.4973, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 2.6622222222222223, | |
| "grad_norm": 0.05448451265692711, | |
| "learning_rate": 3.7530864197530867e-05, | |
| "loss": 0.4895, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 2.6644444444444444, | |
| "grad_norm": 0.05945173278450966, | |
| "learning_rate": 3.728395061728395e-05, | |
| "loss": 0.6438, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.06829828768968582, | |
| "learning_rate": 3.703703703703703e-05, | |
| "loss": 0.3988, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 2.6688888888888886, | |
| "grad_norm": 0.05857190117239952, | |
| "learning_rate": 3.679012345679012e-05, | |
| "loss": 0.8121, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 2.671111111111111, | |
| "grad_norm": 0.0487736314535141, | |
| "learning_rate": 3.6543209876543207e-05, | |
| "loss": 0.4318, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 2.6733333333333333, | |
| "grad_norm": 0.062193140387535095, | |
| "learning_rate": 3.6296296296296295e-05, | |
| "loss": 0.5721, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 2.6755555555555555, | |
| "grad_norm": 0.06735247373580933, | |
| "learning_rate": 3.604938271604938e-05, | |
| "loss": 0.6563, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 2.677777777777778, | |
| "grad_norm": 0.09367570281028748, | |
| "learning_rate": 3.5802469135802465e-05, | |
| "loss": 0.5724, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "grad_norm": 0.11274941265583038, | |
| "learning_rate": 3.555555555555555e-05, | |
| "loss": 0.6182, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 2.6822222222222223, | |
| "grad_norm": 0.04841291159391403, | |
| "learning_rate": 3.5308641975308635e-05, | |
| "loss": 0.5414, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 2.6844444444444444, | |
| "grad_norm": 0.07584474980831146, | |
| "learning_rate": 3.506172839506172e-05, | |
| "loss": 0.3981, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 2.6866666666666665, | |
| "grad_norm": 0.07028737664222717, | |
| "learning_rate": 3.481481481481481e-05, | |
| "loss": 0.4701, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 2.688888888888889, | |
| "grad_norm": 0.06219390779733658, | |
| "learning_rate": 3.45679012345679e-05, | |
| "loss": 0.4585, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 2.6911111111111112, | |
| "grad_norm": 0.07443705201148987, | |
| "learning_rate": 3.432098765432098e-05, | |
| "loss": 0.5342, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 2.6933333333333334, | |
| "grad_norm": 0.07293137907981873, | |
| "learning_rate": 3.407407407407407e-05, | |
| "loss": 0.6419, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 2.6955555555555555, | |
| "grad_norm": 0.05949751287698746, | |
| "learning_rate": 3.382716049382716e-05, | |
| "loss": 0.6161, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 2.6977777777777776, | |
| "grad_norm": 0.055641401559114456, | |
| "learning_rate": 3.3580246913580247e-05, | |
| "loss": 0.5305, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "grad_norm": 0.05257444828748703, | |
| "learning_rate": 3.333333333333333e-05, | |
| "loss": 0.5328, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 2.7022222222222223, | |
| "grad_norm": 0.060353755950927734, | |
| "learning_rate": 3.3086419753086417e-05, | |
| "loss": 0.5574, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 2.7044444444444444, | |
| "grad_norm": 0.062428735196590424, | |
| "learning_rate": 3.2839506172839505e-05, | |
| "loss": 0.57, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 2.7066666666666666, | |
| "grad_norm": 0.1273394227027893, | |
| "learning_rate": 3.2592592592592586e-05, | |
| "loss": 0.6074, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 2.7088888888888887, | |
| "grad_norm": 0.08416179567575455, | |
| "learning_rate": 3.2345679012345675e-05, | |
| "loss": 0.5412, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 2.7111111111111112, | |
| "grad_norm": 0.06812825053930283, | |
| "learning_rate": 3.209876543209876e-05, | |
| "loss": 0.4663, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 2.7133333333333334, | |
| "grad_norm": 0.07209253311157227, | |
| "learning_rate": 3.185185185185185e-05, | |
| "loss": 0.7026, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 2.7155555555555555, | |
| "grad_norm": 0.07327984273433685, | |
| "learning_rate": 3.160493827160493e-05, | |
| "loss": 0.6341, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 2.7177777777777776, | |
| "grad_norm": 0.06341154873371124, | |
| "learning_rate": 3.135802469135802e-05, | |
| "loss": 0.6397, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 2.7199999999999998, | |
| "grad_norm": 0.06885281205177307, | |
| "learning_rate": 3.111111111111111e-05, | |
| "loss": 0.4368, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 2.7222222222222223, | |
| "grad_norm": 0.059726446866989136, | |
| "learning_rate": 3.08641975308642e-05, | |
| "loss": 0.6954, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 2.7244444444444444, | |
| "grad_norm": 0.06625626981258392, | |
| "learning_rate": 3.061728395061728e-05, | |
| "loss": 0.54, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 2.7266666666666666, | |
| "grad_norm": 0.042261507362127304, | |
| "learning_rate": 3.0370370370370368e-05, | |
| "loss": 0.5263, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 2.728888888888889, | |
| "grad_norm": 0.048613209277391434, | |
| "learning_rate": 3.0123456790123453e-05, | |
| "loss": 0.4816, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 2.7311111111111113, | |
| "grad_norm": 0.09477412700653076, | |
| "learning_rate": 2.987654320987654e-05, | |
| "loss": 0.5138, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 2.7333333333333334, | |
| "grad_norm": 0.06048741191625595, | |
| "learning_rate": 2.9629629629629627e-05, | |
| "loss": 0.5746, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 2.7355555555555555, | |
| "grad_norm": 0.05966867506504059, | |
| "learning_rate": 2.9382716049382715e-05, | |
| "loss": 0.5898, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 2.7377777777777776, | |
| "grad_norm": 0.08776303380727768, | |
| "learning_rate": 2.9135802469135796e-05, | |
| "loss": 0.7237, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "grad_norm": 0.07035078853368759, | |
| "learning_rate": 2.8888888888888885e-05, | |
| "loss": 0.6296, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 2.7422222222222223, | |
| "grad_norm": 0.06418696790933609, | |
| "learning_rate": 2.8641975308641973e-05, | |
| "loss": 0.4852, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 2.7444444444444445, | |
| "grad_norm": 0.04925747588276863, | |
| "learning_rate": 2.8395061728395058e-05, | |
| "loss": 0.4677, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 2.7466666666666666, | |
| "grad_norm": 0.06488417088985443, | |
| "learning_rate": 2.8148148148148147e-05, | |
| "loss": 0.5834, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 2.7488888888888887, | |
| "grad_norm": 0.04511519521474838, | |
| "learning_rate": 2.790123456790123e-05, | |
| "loss": 0.4887, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 2.7511111111111113, | |
| "grad_norm": 0.06843352317810059, | |
| "learning_rate": 2.765432098765432e-05, | |
| "loss": 0.4983, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 2.7533333333333334, | |
| "grad_norm": 0.06756050139665604, | |
| "learning_rate": 2.7407407407407405e-05, | |
| "loss": 0.4994, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 2.7555555555555555, | |
| "grad_norm": 0.06538543850183487, | |
| "learning_rate": 2.7160493827160493e-05, | |
| "loss": 0.3782, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 2.7577777777777777, | |
| "grad_norm": 0.057626090943813324, | |
| "learning_rate": 2.6913580246913578e-05, | |
| "loss": 0.563, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "grad_norm": 0.1639692187309265, | |
| "learning_rate": 2.6666666666666667e-05, | |
| "loss": 0.434, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 2.7622222222222224, | |
| "grad_norm": 0.048517338931560516, | |
| "learning_rate": 2.6419753086419748e-05, | |
| "loss": 0.5348, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 2.7644444444444445, | |
| "grad_norm": 0.05313562974333763, | |
| "learning_rate": 2.6172839506172837e-05, | |
| "loss": 0.4687, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 2.7666666666666666, | |
| "grad_norm": 0.05369576811790466, | |
| "learning_rate": 2.592592592592592e-05, | |
| "loss": 0.4959, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 2.7688888888888887, | |
| "grad_norm": 0.06537558138370514, | |
| "learning_rate": 2.567901234567901e-05, | |
| "loss": 0.4862, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 2.771111111111111, | |
| "grad_norm": 0.062362298369407654, | |
| "learning_rate": 2.5432098765432095e-05, | |
| "loss": 0.5051, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 2.7733333333333334, | |
| "grad_norm": 0.07045628130435944, | |
| "learning_rate": 2.5185185185185183e-05, | |
| "loss": 0.5431, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 2.7755555555555556, | |
| "grad_norm": 0.058078449219465256, | |
| "learning_rate": 2.4938271604938268e-05, | |
| "loss": 0.4098, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 0.04137657955288887, | |
| "learning_rate": 2.4691358024691357e-05, | |
| "loss": 0.5553, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 2.7800000000000002, | |
| "grad_norm": 0.08747418969869614, | |
| "learning_rate": 2.4444444444444445e-05, | |
| "loss": 0.5497, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 2.7822222222222224, | |
| "grad_norm": 0.05402607470750809, | |
| "learning_rate": 2.419753086419753e-05, | |
| "loss": 0.7094, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 2.7844444444444445, | |
| "grad_norm": 0.0792861357331276, | |
| "learning_rate": 2.3950617283950618e-05, | |
| "loss": 0.4718, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 2.7866666666666666, | |
| "grad_norm": 0.13500571250915527, | |
| "learning_rate": 2.37037037037037e-05, | |
| "loss": 0.5404, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 2.7888888888888888, | |
| "grad_norm": 0.12560763955116272, | |
| "learning_rate": 2.3456790123456788e-05, | |
| "loss": 0.701, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 2.7911111111111113, | |
| "grad_norm": 0.04671892896294594, | |
| "learning_rate": 2.3209876543209873e-05, | |
| "loss": 0.5641, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 2.7933333333333334, | |
| "grad_norm": 0.043379250913858414, | |
| "learning_rate": 2.296296296296296e-05, | |
| "loss": 0.5214, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 2.7955555555555556, | |
| "grad_norm": 0.06795503944158554, | |
| "learning_rate": 2.2716049382716047e-05, | |
| "loss": 0.7148, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 2.7977777777777777, | |
| "grad_norm": 0.08358605951070786, | |
| "learning_rate": 2.2469135802469135e-05, | |
| "loss": 0.5945, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "grad_norm": 0.05269525200128555, | |
| "learning_rate": 2.222222222222222e-05, | |
| "loss": 0.6251, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 2.8022222222222224, | |
| "grad_norm": 0.059296853840351105, | |
| "learning_rate": 2.1975308641975308e-05, | |
| "loss": 0.5402, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 2.8044444444444445, | |
| "grad_norm": 0.054700687527656555, | |
| "learning_rate": 2.1728395061728393e-05, | |
| "loss": 0.6104, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 2.8066666666666666, | |
| "grad_norm": 0.051813941448926926, | |
| "learning_rate": 2.148148148148148e-05, | |
| "loss": 0.7215, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 2.8088888888888888, | |
| "grad_norm": 0.07439989596605301, | |
| "learning_rate": 2.1234567901234563e-05, | |
| "loss": 0.5278, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 2.811111111111111, | |
| "grad_norm": 0.05079736188054085, | |
| "learning_rate": 2.098765432098765e-05, | |
| "loss": 0.3882, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 2.8133333333333335, | |
| "grad_norm": 0.054003696888685226, | |
| "learning_rate": 2.0740740740740737e-05, | |
| "loss": 0.4078, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 2.8155555555555556, | |
| "grad_norm": 0.05357246473431587, | |
| "learning_rate": 2.0493827160493825e-05, | |
| "loss": 0.504, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 2.8177777777777777, | |
| "grad_norm": 0.05181051790714264, | |
| "learning_rate": 2.0246913580246913e-05, | |
| "loss": 0.4397, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "grad_norm": 0.06633854657411575, | |
| "learning_rate": 1.9999999999999998e-05, | |
| "loss": 0.467, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 2.822222222222222, | |
| "grad_norm": 0.05127658322453499, | |
| "learning_rate": 1.9753086419753087e-05, | |
| "loss": 0.5133, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 2.8244444444444445, | |
| "grad_norm": 0.0671505257487297, | |
| "learning_rate": 1.950617283950617e-05, | |
| "loss": 0.5096, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 2.8266666666666667, | |
| "grad_norm": 0.07473781704902649, | |
| "learning_rate": 1.925925925925926e-05, | |
| "loss": 0.5238, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 2.828888888888889, | |
| "grad_norm": 0.05219605565071106, | |
| "learning_rate": 1.901234567901234e-05, | |
| "loss": 0.5948, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 2.8311111111111114, | |
| "grad_norm": 0.07204823940992355, | |
| "learning_rate": 1.8765432098765433e-05, | |
| "loss": 0.4418, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 2.8333333333333335, | |
| "grad_norm": 0.06105707958340645, | |
| "learning_rate": 1.8518518518518515e-05, | |
| "loss": 0.4878, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 2.8355555555555556, | |
| "grad_norm": 0.06640278548002243, | |
| "learning_rate": 1.8271604938271603e-05, | |
| "loss": 0.6453, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 2.8377777777777777, | |
| "grad_norm": 0.07634621113538742, | |
| "learning_rate": 1.802469135802469e-05, | |
| "loss": 0.4362, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "grad_norm": 0.09639936685562134, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 0.5412, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 2.8422222222222224, | |
| "grad_norm": 0.05341300740838051, | |
| "learning_rate": 1.753086419753086e-05, | |
| "loss": 0.5017, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 2.8444444444444446, | |
| "grad_norm": 0.0719999149441719, | |
| "learning_rate": 1.728395061728395e-05, | |
| "loss": 0.5701, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 2.8466666666666667, | |
| "grad_norm": 0.07877740263938904, | |
| "learning_rate": 1.7037037037037035e-05, | |
| "loss": 0.614, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 2.848888888888889, | |
| "grad_norm": 0.051416411995887756, | |
| "learning_rate": 1.6790123456790123e-05, | |
| "loss": 0.5091, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 2.851111111111111, | |
| "grad_norm": 0.05415206402540207, | |
| "learning_rate": 1.6543209876543208e-05, | |
| "loss": 0.5306, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 2.8533333333333335, | |
| "grad_norm": 0.06409255415201187, | |
| "learning_rate": 1.6296296296296293e-05, | |
| "loss": 0.5477, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 2.8555555555555556, | |
| "grad_norm": 0.07997147738933563, | |
| "learning_rate": 1.604938271604938e-05, | |
| "loss": 0.5713, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 2.8577777777777778, | |
| "grad_norm": 0.06648123264312744, | |
| "learning_rate": 1.5802469135802467e-05, | |
| "loss": 0.4982, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "grad_norm": 0.05242307111620903, | |
| "learning_rate": 1.5555555555555555e-05, | |
| "loss": 0.4074, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 2.862222222222222, | |
| "grad_norm": 0.04754193499684334, | |
| "learning_rate": 1.530864197530864e-05, | |
| "loss": 0.4296, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 2.8644444444444446, | |
| "grad_norm": 0.0683235228061676, | |
| "learning_rate": 1.5061728395061727e-05, | |
| "loss": 0.4713, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 2.8666666666666667, | |
| "grad_norm": 0.05207923427224159, | |
| "learning_rate": 1.4814814814814813e-05, | |
| "loss": 0.4893, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 2.868888888888889, | |
| "grad_norm": 0.056113988161087036, | |
| "learning_rate": 1.4567901234567898e-05, | |
| "loss": 0.4756, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 2.871111111111111, | |
| "grad_norm": 0.06050454080104828, | |
| "learning_rate": 1.4320987654320987e-05, | |
| "loss": 0.3881, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 2.873333333333333, | |
| "grad_norm": 0.06342190504074097, | |
| "learning_rate": 1.4074074074074073e-05, | |
| "loss": 0.4357, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 2.8755555555555556, | |
| "grad_norm": 0.06841931492090225, | |
| "learning_rate": 1.382716049382716e-05, | |
| "loss": 0.5386, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 2.8777777777777778, | |
| "grad_norm": 0.06081142649054527, | |
| "learning_rate": 1.3580246913580247e-05, | |
| "loss": 0.4596, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "grad_norm": 0.07481483370065689, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 0.5804, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 2.8822222222222225, | |
| "grad_norm": 0.050198931246995926, | |
| "learning_rate": 1.3086419753086418e-05, | |
| "loss": 0.5793, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 2.8844444444444446, | |
| "grad_norm": 0.05838954448699951, | |
| "learning_rate": 1.2839506172839505e-05, | |
| "loss": 0.4812, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 2.8866666666666667, | |
| "grad_norm": 0.05184544622898102, | |
| "learning_rate": 1.2592592592592592e-05, | |
| "loss": 0.481, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 2.888888888888889, | |
| "grad_norm": 0.07582832872867584, | |
| "learning_rate": 1.2345679012345678e-05, | |
| "loss": 0.5617, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 2.891111111111111, | |
| "grad_norm": 0.07804284989833832, | |
| "learning_rate": 1.2098765432098765e-05, | |
| "loss": 0.6318, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 2.8933333333333335, | |
| "grad_norm": 0.060935601592063904, | |
| "learning_rate": 1.185185185185185e-05, | |
| "loss": 0.5654, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 2.8955555555555557, | |
| "grad_norm": 0.058726221323013306, | |
| "learning_rate": 1.1604938271604937e-05, | |
| "loss": 0.593, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 2.897777777777778, | |
| "grad_norm": 0.05662497878074646, | |
| "learning_rate": 1.1358024691358023e-05, | |
| "loss": 0.6047, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "grad_norm": 0.05847261846065521, | |
| "learning_rate": 1.111111111111111e-05, | |
| "loss": 0.5682, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 2.902222222222222, | |
| "grad_norm": 0.05384650081396103, | |
| "learning_rate": 1.0864197530864197e-05, | |
| "loss": 0.5402, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 2.9044444444444446, | |
| "grad_norm": 0.05802860110998154, | |
| "learning_rate": 1.0617283950617282e-05, | |
| "loss": 0.5522, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 2.9066666666666667, | |
| "grad_norm": 0.062039174139499664, | |
| "learning_rate": 1.0370370370370368e-05, | |
| "loss": 0.4738, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 2.908888888888889, | |
| "grad_norm": 0.07000809162855148, | |
| "learning_rate": 1.0123456790123457e-05, | |
| "loss": 0.6082, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 2.911111111111111, | |
| "grad_norm": 0.05195184424519539, | |
| "learning_rate": 9.876543209876543e-06, | |
| "loss": 0.6155, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 2.913333333333333, | |
| "grad_norm": 0.08509308099746704, | |
| "learning_rate": 9.62962962962963e-06, | |
| "loss": 0.5573, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 2.9155555555555557, | |
| "grad_norm": 0.08439867198467255, | |
| "learning_rate": 9.382716049382717e-06, | |
| "loss": 0.4368, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 2.917777777777778, | |
| "grad_norm": 0.054084569215774536, | |
| "learning_rate": 9.135802469135802e-06, | |
| "loss": 0.4594, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "grad_norm": 0.07232722640037537, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.5347, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 2.9222222222222225, | |
| "grad_norm": 0.06299246847629547, | |
| "learning_rate": 8.641975308641975e-06, | |
| "loss": 0.4886, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 2.924444444444444, | |
| "grad_norm": 0.06266815960407257, | |
| "learning_rate": 8.395061728395062e-06, | |
| "loss": 0.6174, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 2.9266666666666667, | |
| "grad_norm": 0.06049675866961479, | |
| "learning_rate": 8.148148148148147e-06, | |
| "loss": 0.6329, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 2.928888888888889, | |
| "grad_norm": 0.05617294833064079, | |
| "learning_rate": 7.901234567901233e-06, | |
| "loss": 0.5556, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 2.931111111111111, | |
| "grad_norm": 0.06655992567539215, | |
| "learning_rate": 7.65432098765432e-06, | |
| "loss": 0.5181, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 2.9333333333333336, | |
| "grad_norm": 0.06388793885707855, | |
| "learning_rate": 7.407407407407407e-06, | |
| "loss": 0.6094, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 2.9355555555555557, | |
| "grad_norm": 0.04894081875681877, | |
| "learning_rate": 7.160493827160493e-06, | |
| "loss": 0.4628, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 2.937777777777778, | |
| "grad_norm": 0.06200474873185158, | |
| "learning_rate": 6.91358024691358e-06, | |
| "loss": 0.7136, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "grad_norm": 0.07943463325500488, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.7333, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 2.942222222222222, | |
| "grad_norm": 0.051468003541231155, | |
| "learning_rate": 6.4197530864197525e-06, | |
| "loss": 0.5301, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 2.9444444444444446, | |
| "grad_norm": 0.0468410961329937, | |
| "learning_rate": 6.172839506172839e-06, | |
| "loss": 0.4637, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 2.9466666666666668, | |
| "grad_norm": 0.09125997126102448, | |
| "learning_rate": 5.925925925925925e-06, | |
| "loss": 0.6117, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 2.948888888888889, | |
| "grad_norm": 0.0682133361697197, | |
| "learning_rate": 5.679012345679012e-06, | |
| "loss": 0.6474, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 2.951111111111111, | |
| "grad_norm": 0.05989382416009903, | |
| "learning_rate": 5.432098765432098e-06, | |
| "loss": 0.6313, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 2.953333333333333, | |
| "grad_norm": 0.07264363765716553, | |
| "learning_rate": 5.185185185185184e-06, | |
| "loss": 0.5645, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 2.9555555555555557, | |
| "grad_norm": 0.057875052094459534, | |
| "learning_rate": 4.938271604938272e-06, | |
| "loss": 0.4804, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 2.957777777777778, | |
| "grad_norm": 0.07367154955863953, | |
| "learning_rate": 4.691358024691358e-06, | |
| "loss": 0.6391, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "grad_norm": 0.05268028378486633, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.492, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 2.962222222222222, | |
| "grad_norm": 0.060173213481903076, | |
| "learning_rate": 4.197530864197531e-06, | |
| "loss": 0.5008, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 2.964444444444444, | |
| "grad_norm": 0.06758608669042587, | |
| "learning_rate": 3.950617283950617e-06, | |
| "loss": 0.5625, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 2.966666666666667, | |
| "grad_norm": 0.04485061764717102, | |
| "learning_rate": 3.7037037037037033e-06, | |
| "loss": 0.6103, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 2.968888888888889, | |
| "grad_norm": 0.05087360367178917, | |
| "learning_rate": 3.45679012345679e-06, | |
| "loss": 0.4286, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 2.971111111111111, | |
| "grad_norm": 0.05051387846469879, | |
| "learning_rate": 3.2098765432098762e-06, | |
| "loss": 0.4035, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 2.9733333333333336, | |
| "grad_norm": 0.06164296716451645, | |
| "learning_rate": 2.9629629629629625e-06, | |
| "loss": 0.4906, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 2.9755555555555553, | |
| "grad_norm": 0.05128742754459381, | |
| "learning_rate": 2.716049382716049e-06, | |
| "loss": 0.6804, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 2.977777777777778, | |
| "grad_norm": 0.07346834242343903, | |
| "learning_rate": 2.469135802469136e-06, | |
| "loss": 0.5991, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "grad_norm": 0.15875644981861115, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.5792, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 2.982222222222222, | |
| "grad_norm": 0.05861335247755051, | |
| "learning_rate": 1.9753086419753083e-06, | |
| "loss": 0.4663, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 2.9844444444444447, | |
| "grad_norm": 0.05954683944582939, | |
| "learning_rate": 1.728395061728395e-06, | |
| "loss": 0.435, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 2.986666666666667, | |
| "grad_norm": 0.07032537460327148, | |
| "learning_rate": 1.4814814814814812e-06, | |
| "loss": 0.4384, | |
| "step": 1344 | |
| }, | |
| { | |
| "epoch": 2.988888888888889, | |
| "grad_norm": 0.07222862541675568, | |
| "learning_rate": 1.234567901234568e-06, | |
| "loss": 0.5334, | |
| "step": 1345 | |
| }, | |
| { | |
| "epoch": 2.991111111111111, | |
| "grad_norm": 0.07875492423772812, | |
| "learning_rate": 9.876543209876542e-07, | |
| "loss": 0.7617, | |
| "step": 1346 | |
| }, | |
| { | |
| "epoch": 2.993333333333333, | |
| "grad_norm": 0.07582969963550568, | |
| "learning_rate": 7.407407407407406e-07, | |
| "loss": 0.5529, | |
| "step": 1347 | |
| }, | |
| { | |
| "epoch": 2.9955555555555557, | |
| "grad_norm": 0.05928955227136612, | |
| "learning_rate": 4.938271604938271e-07, | |
| "loss": 0.6101, | |
| "step": 1348 | |
| }, | |
| { | |
| "epoch": 2.997777777777778, | |
| "grad_norm": 0.07385487854480743, | |
| "learning_rate": 2.4691358024691354e-07, | |
| "loss": 0.3858, | |
| "step": 1349 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 0.06644037365913391, | |
| "learning_rate": 0.0, | |
| "loss": 0.4738, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 1350, | |
| "total_flos": 5.088308053724365e+17, | |
| "train_loss": 0.5877564937980087, | |
| "train_runtime": 3779.9459, | |
| "train_samples_per_second": 5.714, | |
| "train_steps_per_second": 0.357 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 1350, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.088308053724365e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |