| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 0, | |
| "global_step": 246, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0040650406504065045, | |
| "grad_norm": 1.2443748712539673, | |
| "learning_rate": 1e-05, | |
| "loss": 2.1704, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008130081300813009, | |
| "grad_norm": 1.198485255241394, | |
| "learning_rate": 9.959349593495936e-06, | |
| "loss": 2.1382, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.012195121951219513, | |
| "grad_norm": 1.1655025482177734, | |
| "learning_rate": 9.91869918699187e-06, | |
| "loss": 2.1257, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.016260162601626018, | |
| "grad_norm": 1.1087367534637451, | |
| "learning_rate": 9.878048780487805e-06, | |
| "loss": 2.1008, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.02032520325203252, | |
| "grad_norm": 1.0223109722137451, | |
| "learning_rate": 9.837398373983741e-06, | |
| "loss": 2.0118, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.024390243902439025, | |
| "grad_norm": 1.0011451244354248, | |
| "learning_rate": 9.796747967479675e-06, | |
| "loss": 2.0316, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.028455284552845527, | |
| "grad_norm": 0.968566358089447, | |
| "learning_rate": 9.756097560975611e-06, | |
| "loss": 2.0904, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.032520325203252036, | |
| "grad_norm": 0.9195019602775574, | |
| "learning_rate": 9.715447154471546e-06, | |
| "loss": 2.0712, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.036585365853658534, | |
| "grad_norm": 0.8058457970619202, | |
| "learning_rate": 9.67479674796748e-06, | |
| "loss": 1.8876, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04065040650406504, | |
| "grad_norm": 0.7549053430557251, | |
| "learning_rate": 9.634146341463415e-06, | |
| "loss": 1.8909, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.044715447154471545, | |
| "grad_norm": 0.7223576307296753, | |
| "learning_rate": 9.59349593495935e-06, | |
| "loss": 2.0057, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04878048780487805, | |
| "grad_norm": 0.6774242520332336, | |
| "learning_rate": 9.552845528455286e-06, | |
| "loss": 1.9098, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.052845528455284556, | |
| "grad_norm": 0.6018033027648926, | |
| "learning_rate": 9.51219512195122e-06, | |
| "loss": 1.8596, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.056910569105691054, | |
| "grad_norm": 0.5510868430137634, | |
| "learning_rate": 9.471544715447156e-06, | |
| "loss": 1.8775, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06097560975609756, | |
| "grad_norm": 0.5308274030685425, | |
| "learning_rate": 9.43089430894309e-06, | |
| "loss": 1.853, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06504065040650407, | |
| "grad_norm": 0.47880470752716064, | |
| "learning_rate": 9.390243902439025e-06, | |
| "loss": 1.8277, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.06910569105691057, | |
| "grad_norm": 0.49379265308380127, | |
| "learning_rate": 9.34959349593496e-06, | |
| "loss": 1.9113, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07317073170731707, | |
| "grad_norm": 0.462348073720932, | |
| "learning_rate": 9.308943089430895e-06, | |
| "loss": 1.7872, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.07723577235772358, | |
| "grad_norm": 0.45618218183517456, | |
| "learning_rate": 9.268292682926831e-06, | |
| "loss": 1.8817, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08130081300813008, | |
| "grad_norm": 0.4135752022266388, | |
| "learning_rate": 9.227642276422764e-06, | |
| "loss": 1.7765, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08536585365853659, | |
| "grad_norm": 0.4315033555030823, | |
| "learning_rate": 9.1869918699187e-06, | |
| "loss": 1.7997, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.08943089430894309, | |
| "grad_norm": 0.41308435797691345, | |
| "learning_rate": 9.146341463414635e-06, | |
| "loss": 1.7839, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.09349593495934959, | |
| "grad_norm": 0.38933464884757996, | |
| "learning_rate": 9.10569105691057e-06, | |
| "loss": 1.7688, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0975609756097561, | |
| "grad_norm": 0.37282490730285645, | |
| "learning_rate": 9.065040650406505e-06, | |
| "loss": 1.7134, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.1016260162601626, | |
| "grad_norm": 0.37599578499794006, | |
| "learning_rate": 9.02439024390244e-06, | |
| "loss": 1.7569, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.10569105691056911, | |
| "grad_norm": 0.36915072798728943, | |
| "learning_rate": 8.983739837398374e-06, | |
| "loss": 1.7513, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.10975609756097561, | |
| "grad_norm": 0.37054121494293213, | |
| "learning_rate": 8.94308943089431e-06, | |
| "loss": 1.7617, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11382113821138211, | |
| "grad_norm": 0.3536517918109894, | |
| "learning_rate": 8.902439024390244e-06, | |
| "loss": 1.7369, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.11788617886178862, | |
| "grad_norm": 0.34083041548728943, | |
| "learning_rate": 8.86178861788618e-06, | |
| "loss": 1.6592, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.12195121951219512, | |
| "grad_norm": 0.358023464679718, | |
| "learning_rate": 8.821138211382113e-06, | |
| "loss": 1.7287, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.12601626016260162, | |
| "grad_norm": 0.3317272961139679, | |
| "learning_rate": 8.78048780487805e-06, | |
| "loss": 1.6652, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.13008130081300814, | |
| "grad_norm": 0.3286559581756592, | |
| "learning_rate": 8.739837398373985e-06, | |
| "loss": 1.6648, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.13414634146341464, | |
| "grad_norm": 0.3171916902065277, | |
| "learning_rate": 8.69918699186992e-06, | |
| "loss": 1.6781, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.13821138211382114, | |
| "grad_norm": 0.3212682008743286, | |
| "learning_rate": 8.658536585365854e-06, | |
| "loss": 1.706, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.14227642276422764, | |
| "grad_norm": 0.321341335773468, | |
| "learning_rate": 8.617886178861789e-06, | |
| "loss": 1.6633, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.14634146341463414, | |
| "grad_norm": 0.3067283034324646, | |
| "learning_rate": 8.577235772357724e-06, | |
| "loss": 1.6671, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.15040650406504066, | |
| "grad_norm": 0.2952353358268738, | |
| "learning_rate": 8.536585365853658e-06, | |
| "loss": 1.6429, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.15447154471544716, | |
| "grad_norm": 0.28897207975387573, | |
| "learning_rate": 8.495934959349595e-06, | |
| "loss": 1.5994, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.15853658536585366, | |
| "grad_norm": 0.3044757544994354, | |
| "learning_rate": 8.45528455284553e-06, | |
| "loss": 1.6484, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.16260162601626016, | |
| "grad_norm": 0.2715882658958435, | |
| "learning_rate": 8.414634146341464e-06, | |
| "loss": 1.5489, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.262287437915802, | |
| "learning_rate": 8.373983739837399e-06, | |
| "loss": 1.5362, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.17073170731707318, | |
| "grad_norm": 0.2884230315685272, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 1.6417, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.17479674796747968, | |
| "grad_norm": 0.2722843289375305, | |
| "learning_rate": 8.292682926829268e-06, | |
| "loss": 1.5869, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.17886178861788618, | |
| "grad_norm": 0.25501325726509094, | |
| "learning_rate": 8.252032520325203e-06, | |
| "loss": 1.4989, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.18292682926829268, | |
| "grad_norm": 0.2949991822242737, | |
| "learning_rate": 8.21138211382114e-06, | |
| "loss": 1.6111, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.18699186991869918, | |
| "grad_norm": 0.26580971479415894, | |
| "learning_rate": 8.170731707317073e-06, | |
| "loss": 1.573, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.1910569105691057, | |
| "grad_norm": 0.28367504477500916, | |
| "learning_rate": 8.130081300813009e-06, | |
| "loss": 1.6192, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.1951219512195122, | |
| "grad_norm": 0.2633240222930908, | |
| "learning_rate": 8.089430894308944e-06, | |
| "loss": 1.5603, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.1991869918699187, | |
| "grad_norm": 0.24888679385185242, | |
| "learning_rate": 8.048780487804879e-06, | |
| "loss": 1.5013, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.2032520325203252, | |
| "grad_norm": 0.25869449973106384, | |
| "learning_rate": 8.008130081300813e-06, | |
| "loss": 1.5718, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2073170731707317, | |
| "grad_norm": 0.26135310530662537, | |
| "learning_rate": 7.967479674796748e-06, | |
| "loss": 1.5476, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.21138211382113822, | |
| "grad_norm": 0.2613206207752228, | |
| "learning_rate": 7.926829268292685e-06, | |
| "loss": 1.5685, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.21544715447154472, | |
| "grad_norm": 0.24914291501045227, | |
| "learning_rate": 7.886178861788618e-06, | |
| "loss": 1.522, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.21951219512195122, | |
| "grad_norm": 0.2416287362575531, | |
| "learning_rate": 7.845528455284554e-06, | |
| "loss": 1.5258, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.22357723577235772, | |
| "grad_norm": 0.23537378013134003, | |
| "learning_rate": 7.804878048780489e-06, | |
| "loss": 1.5002, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.22764227642276422, | |
| "grad_norm": 0.24182282388210297, | |
| "learning_rate": 7.764227642276424e-06, | |
| "loss": 1.5264, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.23170731707317074, | |
| "grad_norm": 0.24550394713878632, | |
| "learning_rate": 7.723577235772358e-06, | |
| "loss": 1.5361, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.23577235772357724, | |
| "grad_norm": 0.22000114619731903, | |
| "learning_rate": 7.682926829268293e-06, | |
| "loss": 1.4482, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.23983739837398374, | |
| "grad_norm": 0.21993432939052582, | |
| "learning_rate": 7.64227642276423e-06, | |
| "loss": 1.452, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.24390243902439024, | |
| "grad_norm": 0.23607592284679413, | |
| "learning_rate": 7.601626016260163e-06, | |
| "loss": 1.4749, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.24796747967479674, | |
| "grad_norm": 0.22953565418720245, | |
| "learning_rate": 7.560975609756098e-06, | |
| "loss": 1.4828, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.25203252032520324, | |
| "grad_norm": 0.21641074120998383, | |
| "learning_rate": 7.520325203252034e-06, | |
| "loss": 1.4558, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.25609756097560976, | |
| "grad_norm": 0.23295767605304718, | |
| "learning_rate": 7.4796747967479676e-06, | |
| "loss": 1.5053, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.2601626016260163, | |
| "grad_norm": 0.21931757032871246, | |
| "learning_rate": 7.439024390243903e-06, | |
| "loss": 1.4599, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.26422764227642276, | |
| "grad_norm": 0.21457414329051971, | |
| "learning_rate": 7.398373983739838e-06, | |
| "loss": 1.4686, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.2682926829268293, | |
| "grad_norm": 0.2142302244901657, | |
| "learning_rate": 7.357723577235773e-06, | |
| "loss": 1.4362, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.27235772357723576, | |
| "grad_norm": 0.21409818530082703, | |
| "learning_rate": 7.317073170731707e-06, | |
| "loss": 1.4607, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.2764227642276423, | |
| "grad_norm": 0.211210697889328, | |
| "learning_rate": 7.276422764227643e-06, | |
| "loss": 1.4587, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2804878048780488, | |
| "grad_norm": 0.21644359827041626, | |
| "learning_rate": 7.2357723577235786e-06, | |
| "loss": 1.4481, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2845528455284553, | |
| "grad_norm": 0.2180495411157608, | |
| "learning_rate": 7.1951219512195125e-06, | |
| "loss": 1.416, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2886178861788618, | |
| "grad_norm": 0.20745021104812622, | |
| "learning_rate": 7.154471544715448e-06, | |
| "loss": 1.4157, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.2926829268292683, | |
| "grad_norm": 0.20739568769931793, | |
| "learning_rate": 7.113821138211383e-06, | |
| "loss": 1.4276, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2967479674796748, | |
| "grad_norm": 0.20237509906291962, | |
| "learning_rate": 7.0731707317073175e-06, | |
| "loss": 1.4011, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.3008130081300813, | |
| "grad_norm": 0.19157394766807556, | |
| "learning_rate": 7.032520325203252e-06, | |
| "loss": 1.3681, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.3048780487804878, | |
| "grad_norm": 0.2131585031747818, | |
| "learning_rate": 6.991869918699188e-06, | |
| "loss": 1.4443, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3089430894308943, | |
| "grad_norm": 0.208163782954216, | |
| "learning_rate": 6.951219512195122e-06, | |
| "loss": 1.4089, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3130081300813008, | |
| "grad_norm": 0.1992938071489334, | |
| "learning_rate": 6.910569105691057e-06, | |
| "loss": 1.3887, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3170731707317073, | |
| "grad_norm": 0.20457559823989868, | |
| "learning_rate": 6.869918699186993e-06, | |
| "loss": 1.3853, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.32113821138211385, | |
| "grad_norm": 0.20216362178325653, | |
| "learning_rate": 6.829268292682928e-06, | |
| "loss": 1.3893, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3252032520325203, | |
| "grad_norm": 0.19666500389575958, | |
| "learning_rate": 6.788617886178862e-06, | |
| "loss": 1.4147, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.32926829268292684, | |
| "grad_norm": 0.2000611275434494, | |
| "learning_rate": 6.747967479674797e-06, | |
| "loss": 1.3686, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.21209818124771118, | |
| "learning_rate": 6.707317073170733e-06, | |
| "loss": 1.3857, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.33739837398373984, | |
| "grad_norm": 0.21749112010002136, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 1.415, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.34146341463414637, | |
| "grad_norm": 0.2626613676548004, | |
| "learning_rate": 6.626016260162602e-06, | |
| "loss": 1.3677, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.34552845528455284, | |
| "grad_norm": 0.19924424588680267, | |
| "learning_rate": 6.585365853658538e-06, | |
| "loss": 1.3835, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.34959349593495936, | |
| "grad_norm": 0.19534048438072205, | |
| "learning_rate": 6.544715447154472e-06, | |
| "loss": 1.3563, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.35365853658536583, | |
| "grad_norm": 0.2014574259519577, | |
| "learning_rate": 6.504065040650407e-06, | |
| "loss": 1.3704, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.35772357723577236, | |
| "grad_norm": 0.2033248394727707, | |
| "learning_rate": 6.463414634146342e-06, | |
| "loss": 1.4016, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3617886178861789, | |
| "grad_norm": 0.2129010260105133, | |
| "learning_rate": 6.422764227642278e-06, | |
| "loss": 1.4318, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.36585365853658536, | |
| "grad_norm": 0.19626440107822418, | |
| "learning_rate": 6.3821138211382115e-06, | |
| "loss": 1.3912, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3699186991869919, | |
| "grad_norm": 0.205119788646698, | |
| "learning_rate": 6.341463414634147e-06, | |
| "loss": 1.4114, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.37398373983739835, | |
| "grad_norm": 0.20082072913646698, | |
| "learning_rate": 6.300813008130082e-06, | |
| "loss": 1.3829, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.3780487804878049, | |
| "grad_norm": 0.20235207676887512, | |
| "learning_rate": 6.260162601626017e-06, | |
| "loss": 1.3832, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3821138211382114, | |
| "grad_norm": 0.2006131112575531, | |
| "learning_rate": 6.219512195121951e-06, | |
| "loss": 1.376, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.3861788617886179, | |
| "grad_norm": 0.19920463860034943, | |
| "learning_rate": 6.178861788617887e-06, | |
| "loss": 1.3868, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.3902439024390244, | |
| "grad_norm": 0.19277149438858032, | |
| "learning_rate": 6.138211382113821e-06, | |
| "loss": 1.3526, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.3943089430894309, | |
| "grad_norm": 0.1967436820268631, | |
| "learning_rate": 6.0975609756097564e-06, | |
| "loss": 1.3311, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.3983739837398374, | |
| "grad_norm": 0.20269839465618134, | |
| "learning_rate": 6.056910569105692e-06, | |
| "loss": 1.3427, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4024390243902439, | |
| "grad_norm": 0.2014787793159485, | |
| "learning_rate": 6.016260162601627e-06, | |
| "loss": 1.3508, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4065040650406504, | |
| "grad_norm": 0.20438070595264435, | |
| "learning_rate": 5.9756097560975615e-06, | |
| "loss": 1.3559, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4105691056910569, | |
| "grad_norm": 0.19471648335456848, | |
| "learning_rate": 5.934959349593496e-06, | |
| "loss": 1.3263, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4146341463414634, | |
| "grad_norm": 0.1897473782300949, | |
| "learning_rate": 5.894308943089432e-06, | |
| "loss": 1.3599, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.4186991869918699, | |
| "grad_norm": 0.19831465184688568, | |
| "learning_rate": 5.853658536585366e-06, | |
| "loss": 1.356, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.42276422764227645, | |
| "grad_norm": 0.19808410108089447, | |
| "learning_rate": 5.813008130081301e-06, | |
| "loss": 1.3559, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.4268292682926829, | |
| "grad_norm": 0.1782761812210083, | |
| "learning_rate": 5.772357723577237e-06, | |
| "loss": 1.31, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.43089430894308944, | |
| "grad_norm": 0.2128615379333496, | |
| "learning_rate": 5.731707317073171e-06, | |
| "loss": 1.402, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4349593495934959, | |
| "grad_norm": 0.2091735452413559, | |
| "learning_rate": 5.691056910569106e-06, | |
| "loss": 1.3964, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.43902439024390244, | |
| "grad_norm": 0.19047264754772186, | |
| "learning_rate": 5.650406504065041e-06, | |
| "loss": 1.3781, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.44308943089430897, | |
| "grad_norm": 0.201967254281044, | |
| "learning_rate": 5.609756097560977e-06, | |
| "loss": 1.3574, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.44715447154471544, | |
| "grad_norm": 0.17995291948318481, | |
| "learning_rate": 5.569105691056911e-06, | |
| "loss": 1.3217, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.45121951219512196, | |
| "grad_norm": 0.2009151726961136, | |
| "learning_rate": 5.528455284552846e-06, | |
| "loss": 1.372, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.45528455284552843, | |
| "grad_norm": 0.18182797729969025, | |
| "learning_rate": 5.487804878048781e-06, | |
| "loss": 1.2898, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.45934959349593496, | |
| "grad_norm": 0.18322426080703735, | |
| "learning_rate": 5.447154471544716e-06, | |
| "loss": 1.3016, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.4634146341463415, | |
| "grad_norm": 0.18239295482635498, | |
| "learning_rate": 5.4065040650406504e-06, | |
| "loss": 1.3583, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.46747967479674796, | |
| "grad_norm": 0.19691058993339539, | |
| "learning_rate": 5.365853658536586e-06, | |
| "loss": 1.313, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.4715447154471545, | |
| "grad_norm": 0.18001240491867065, | |
| "learning_rate": 5.32520325203252e-06, | |
| "loss": 1.3059, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.47560975609756095, | |
| "grad_norm": 0.1999204009771347, | |
| "learning_rate": 5.2845528455284555e-06, | |
| "loss": 1.324, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4796747967479675, | |
| "grad_norm": 0.19896095991134644, | |
| "learning_rate": 5.243902439024391e-06, | |
| "loss": 1.3648, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.483739837398374, | |
| "grad_norm": 0.18776021897792816, | |
| "learning_rate": 5.203252032520326e-06, | |
| "loss": 1.2972, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.4878048780487805, | |
| "grad_norm": 0.19416946172714233, | |
| "learning_rate": 5.162601626016261e-06, | |
| "loss": 1.3373, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.491869918699187, | |
| "grad_norm": 0.19433578848838806, | |
| "learning_rate": 5.121951219512195e-06, | |
| "loss": 1.3076, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.4959349593495935, | |
| "grad_norm": 0.18256975710391998, | |
| "learning_rate": 5.081300813008131e-06, | |
| "loss": 1.2832, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.18513993918895721, | |
| "learning_rate": 5.040650406504065e-06, | |
| "loss": 1.3172, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5040650406504065, | |
| "grad_norm": 0.19009090960025787, | |
| "learning_rate": 5e-06, | |
| "loss": 1.3169, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.508130081300813, | |
| "grad_norm": 0.1833588182926178, | |
| "learning_rate": 4.959349593495935e-06, | |
| "loss": 1.2932, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5121951219512195, | |
| "grad_norm": 0.17829759418964386, | |
| "learning_rate": 4.918699186991871e-06, | |
| "loss": 1.3326, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.516260162601626, | |
| "grad_norm": 0.18041731417179108, | |
| "learning_rate": 4.8780487804878055e-06, | |
| "loss": 1.2987, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5203252032520326, | |
| "grad_norm": 0.19885776937007904, | |
| "learning_rate": 4.83739837398374e-06, | |
| "loss": 1.3069, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.524390243902439, | |
| "grad_norm": 0.17805083096027374, | |
| "learning_rate": 4.796747967479675e-06, | |
| "loss": 1.3226, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5284552845528455, | |
| "grad_norm": 0.176824152469635, | |
| "learning_rate": 4.75609756097561e-06, | |
| "loss": 1.2887, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.532520325203252, | |
| "grad_norm": 0.19132983684539795, | |
| "learning_rate": 4.715447154471545e-06, | |
| "loss": 1.3114, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5365853658536586, | |
| "grad_norm": 0.18644371628761292, | |
| "learning_rate": 4.67479674796748e-06, | |
| "loss": 1.3038, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.540650406504065, | |
| "grad_norm": 0.19085392355918884, | |
| "learning_rate": 4.634146341463416e-06, | |
| "loss": 1.3141, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5447154471544715, | |
| "grad_norm": 0.17294111847877502, | |
| "learning_rate": 4.59349593495935e-06, | |
| "loss": 1.2709, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5487804878048781, | |
| "grad_norm": 0.21765175461769104, | |
| "learning_rate": 4.552845528455285e-06, | |
| "loss": 1.3192, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5528455284552846, | |
| "grad_norm": 0.1797131896018982, | |
| "learning_rate": 4.51219512195122e-06, | |
| "loss": 1.2781, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.556910569105691, | |
| "grad_norm": 0.1938735544681549, | |
| "learning_rate": 4.471544715447155e-06, | |
| "loss": 1.33, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5609756097560976, | |
| "grad_norm": 0.17687402665615082, | |
| "learning_rate": 4.43089430894309e-06, | |
| "loss": 1.2757, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5650406504065041, | |
| "grad_norm": 0.181070938706398, | |
| "learning_rate": 4.390243902439025e-06, | |
| "loss": 1.2786, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5691056910569106, | |
| "grad_norm": 0.22047542035579681, | |
| "learning_rate": 4.34959349593496e-06, | |
| "loss": 1.337, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.573170731707317, | |
| "grad_norm": 0.1789373755455017, | |
| "learning_rate": 4.308943089430894e-06, | |
| "loss": 1.2926, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5772357723577236, | |
| "grad_norm": 0.192376047372818, | |
| "learning_rate": 4.268292682926829e-06, | |
| "loss": 1.3014, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.5813008130081301, | |
| "grad_norm": 0.18353502452373505, | |
| "learning_rate": 4.227642276422765e-06, | |
| "loss": 1.3375, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.5853658536585366, | |
| "grad_norm": 0.18353086709976196, | |
| "learning_rate": 4.1869918699186995e-06, | |
| "loss": 1.2781, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5894308943089431, | |
| "grad_norm": 0.19154737889766693, | |
| "learning_rate": 4.146341463414634e-06, | |
| "loss": 1.2515, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5934959349593496, | |
| "grad_norm": 0.19711637496948242, | |
| "learning_rate": 4.10569105691057e-06, | |
| "loss": 1.325, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5975609756097561, | |
| "grad_norm": 0.19635708630084991, | |
| "learning_rate": 4.0650406504065046e-06, | |
| "loss": 1.3318, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6016260162601627, | |
| "grad_norm": 0.19847285747528076, | |
| "learning_rate": 4.024390243902439e-06, | |
| "loss": 1.3464, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6056910569105691, | |
| "grad_norm": 0.20473280549049377, | |
| "learning_rate": 3.983739837398374e-06, | |
| "loss": 1.3094, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6097560975609756, | |
| "grad_norm": 0.1827707588672638, | |
| "learning_rate": 3.943089430894309e-06, | |
| "loss": 1.3238, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6138211382113821, | |
| "grad_norm": 0.18579614162445068, | |
| "learning_rate": 3.902439024390244e-06, | |
| "loss": 1.3175, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6178861788617886, | |
| "grad_norm": 0.17664667963981628, | |
| "learning_rate": 3.861788617886179e-06, | |
| "loss": 1.2808, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6219512195121951, | |
| "grad_norm": 0.18311426043510437, | |
| "learning_rate": 3.821138211382115e-06, | |
| "loss": 1.2676, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6260162601626016, | |
| "grad_norm": 0.18462157249450684, | |
| "learning_rate": 3.780487804878049e-06, | |
| "loss": 1.3112, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6300813008130082, | |
| "grad_norm": 0.1831246167421341, | |
| "learning_rate": 3.7398373983739838e-06, | |
| "loss": 1.2657, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6341463414634146, | |
| "grad_norm": 0.17907500267028809, | |
| "learning_rate": 3.699186991869919e-06, | |
| "loss": 1.2794, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6382113821138211, | |
| "grad_norm": 0.18612568080425262, | |
| "learning_rate": 3.6585365853658537e-06, | |
| "loss": 1.3098, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.6422764227642277, | |
| "grad_norm": 0.19356365501880646, | |
| "learning_rate": 3.6178861788617893e-06, | |
| "loss": 1.2861, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6463414634146342, | |
| "grad_norm": 0.18286974728107452, | |
| "learning_rate": 3.577235772357724e-06, | |
| "loss": 1.2853, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6504065040650406, | |
| "grad_norm": 0.19886158406734467, | |
| "learning_rate": 3.5365853658536588e-06, | |
| "loss": 1.2977, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6544715447154471, | |
| "grad_norm": 0.17624559998512268, | |
| "learning_rate": 3.495934959349594e-06, | |
| "loss": 1.2371, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6585365853658537, | |
| "grad_norm": 0.18079955875873566, | |
| "learning_rate": 3.4552845528455287e-06, | |
| "loss": 1.2884, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.6626016260162602, | |
| "grad_norm": 0.18184345960617065, | |
| "learning_rate": 3.414634146341464e-06, | |
| "loss": 1.2872, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.18814067542552948, | |
| "learning_rate": 3.3739837398373986e-06, | |
| "loss": 1.3077, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.6707317073170732, | |
| "grad_norm": 0.18188297748565674, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 1.252, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.6747967479674797, | |
| "grad_norm": 0.18605956435203552, | |
| "learning_rate": 3.292682926829269e-06, | |
| "loss": 1.2462, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6788617886178862, | |
| "grad_norm": 0.18796762824058533, | |
| "learning_rate": 3.2520325203252037e-06, | |
| "loss": 1.207, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.6829268292682927, | |
| "grad_norm": 0.17487967014312744, | |
| "learning_rate": 3.211382113821139e-06, | |
| "loss": 1.2696, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.6869918699186992, | |
| "grad_norm": 0.1884680986404419, | |
| "learning_rate": 3.1707317073170736e-06, | |
| "loss": 1.2916, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6910569105691057, | |
| "grad_norm": 0.18868300318717957, | |
| "learning_rate": 3.1300813008130083e-06, | |
| "loss": 1.259, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6951219512195121, | |
| "grad_norm": 0.1876087635755539, | |
| "learning_rate": 3.0894308943089435e-06, | |
| "loss": 1.287, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6991869918699187, | |
| "grad_norm": 0.19446611404418945, | |
| "learning_rate": 3.0487804878048782e-06, | |
| "loss": 1.2429, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7032520325203252, | |
| "grad_norm": 0.1855759173631668, | |
| "learning_rate": 3.0081300813008134e-06, | |
| "loss": 1.2698, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7073170731707317, | |
| "grad_norm": 0.18855759501457214, | |
| "learning_rate": 2.967479674796748e-06, | |
| "loss": 1.2977, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7113821138211383, | |
| "grad_norm": 0.20293879508972168, | |
| "learning_rate": 2.926829268292683e-06, | |
| "loss": 1.2701, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7154471544715447, | |
| "grad_norm": 0.1948600560426712, | |
| "learning_rate": 2.8861788617886185e-06, | |
| "loss": 1.3133, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7195121951219512, | |
| "grad_norm": 0.1887495517730713, | |
| "learning_rate": 2.845528455284553e-06, | |
| "loss": 1.26, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7235772357723578, | |
| "grad_norm": 0.19558636844158173, | |
| "learning_rate": 2.8048780487804884e-06, | |
| "loss": 1.2934, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7276422764227642, | |
| "grad_norm": 0.1895134001970291, | |
| "learning_rate": 2.764227642276423e-06, | |
| "loss": 1.3002, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7317073170731707, | |
| "grad_norm": 0.18442754447460175, | |
| "learning_rate": 2.723577235772358e-06, | |
| "loss": 1.2515, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7357723577235772, | |
| "grad_norm": 0.17319710552692413, | |
| "learning_rate": 2.682926829268293e-06, | |
| "loss": 1.2455, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7398373983739838, | |
| "grad_norm": 0.17397530376911163, | |
| "learning_rate": 2.6422764227642278e-06, | |
| "loss": 1.2526, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7439024390243902, | |
| "grad_norm": 0.18584556877613068, | |
| "learning_rate": 2.601626016260163e-06, | |
| "loss": 1.2593, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.7479674796747967, | |
| "grad_norm": 0.1924012452363968, | |
| "learning_rate": 2.5609756097560977e-06, | |
| "loss": 1.2845, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7520325203252033, | |
| "grad_norm": 0.18047532439231873, | |
| "learning_rate": 2.5203252032520324e-06, | |
| "loss": 1.3022, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7560975609756098, | |
| "grad_norm": 0.19239181280136108, | |
| "learning_rate": 2.4796747967479676e-06, | |
| "loss": 1.3515, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.7601626016260162, | |
| "grad_norm": 0.18099461495876312, | |
| "learning_rate": 2.4390243902439027e-06, | |
| "loss": 1.2365, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.7642276422764228, | |
| "grad_norm": 0.19709230959415436, | |
| "learning_rate": 2.3983739837398375e-06, | |
| "loss": 1.2553, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.7682926829268293, | |
| "grad_norm": 0.18170712888240814, | |
| "learning_rate": 2.3577235772357727e-06, | |
| "loss": 1.2782, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.7723577235772358, | |
| "grad_norm": 0.18511487543582916, | |
| "learning_rate": 2.317073170731708e-06, | |
| "loss": 1.289, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.7764227642276422, | |
| "grad_norm": 0.1844373643398285, | |
| "learning_rate": 2.2764227642276426e-06, | |
| "loss": 1.2878, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.7804878048780488, | |
| "grad_norm": 0.175090953707695, | |
| "learning_rate": 2.2357723577235773e-06, | |
| "loss": 1.2684, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.7845528455284553, | |
| "grad_norm": 0.1956668645143509, | |
| "learning_rate": 2.1951219512195125e-06, | |
| "loss": 1.2691, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7886178861788617, | |
| "grad_norm": 0.17475838959217072, | |
| "learning_rate": 2.154471544715447e-06, | |
| "loss": 1.2333, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.7926829268292683, | |
| "grad_norm": 0.1879858672618866, | |
| "learning_rate": 2.1138211382113824e-06, | |
| "loss": 1.2953, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.7967479674796748, | |
| "grad_norm": 0.1937568187713623, | |
| "learning_rate": 2.073170731707317e-06, | |
| "loss": 1.2671, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.8008130081300813, | |
| "grad_norm": 0.191900834441185, | |
| "learning_rate": 2.0325203252032523e-06, | |
| "loss": 1.2859, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8048780487804879, | |
| "grad_norm": 0.17896786332130432, | |
| "learning_rate": 1.991869918699187e-06, | |
| "loss": 1.2723, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8089430894308943, | |
| "grad_norm": 0.21629469096660614, | |
| "learning_rate": 1.951219512195122e-06, | |
| "loss": 1.2974, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8130081300813008, | |
| "grad_norm": 0.1890254020690918, | |
| "learning_rate": 1.9105691056910574e-06, | |
| "loss": 1.2321, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8170731707317073, | |
| "grad_norm": 0.19379396736621857, | |
| "learning_rate": 1.8699186991869919e-06, | |
| "loss": 1.2474, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.8211382113821138, | |
| "grad_norm": 0.17745374143123627, | |
| "learning_rate": 1.8292682926829268e-06, | |
| "loss": 1.2523, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.8252032520325203, | |
| "grad_norm": 0.19853124022483826, | |
| "learning_rate": 1.788617886178862e-06, | |
| "loss": 1.2544, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.8292682926829268, | |
| "grad_norm": 0.19184249639511108, | |
| "learning_rate": 1.747967479674797e-06, | |
| "loss": 1.2272, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 0.18143822252750397, | |
| "learning_rate": 1.707317073170732e-06, | |
| "loss": 1.237, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.8373983739837398, | |
| "grad_norm": 0.18249163031578064, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 1.2222, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.8414634146341463, | |
| "grad_norm": 0.1959405392408371, | |
| "learning_rate": 1.6260162601626018e-06, | |
| "loss": 1.2525, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.8455284552845529, | |
| "grad_norm": 0.25541865825653076, | |
| "learning_rate": 1.5853658536585368e-06, | |
| "loss": 1.3476, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.8495934959349594, | |
| "grad_norm": 0.18992851674556732, | |
| "learning_rate": 1.5447154471544717e-06, | |
| "loss": 1.277, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.8536585365853658, | |
| "grad_norm": 0.21258093416690826, | |
| "learning_rate": 1.5040650406504067e-06, | |
| "loss": 1.2651, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.8577235772357723, | |
| "grad_norm": 0.19474169611930847, | |
| "learning_rate": 1.4634146341463414e-06, | |
| "loss": 1.2357, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.8617886178861789, | |
| "grad_norm": 0.1917145848274231, | |
| "learning_rate": 1.4227642276422766e-06, | |
| "loss": 1.2572, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.8658536585365854, | |
| "grad_norm": 0.19641472399234772, | |
| "learning_rate": 1.3821138211382116e-06, | |
| "loss": 1.2907, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.8699186991869918, | |
| "grad_norm": 0.19125299155712128, | |
| "learning_rate": 1.3414634146341465e-06, | |
| "loss": 1.2274, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.8739837398373984, | |
| "grad_norm": 0.18693210184574127, | |
| "learning_rate": 1.3008130081300815e-06, | |
| "loss": 1.2848, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.8780487804878049, | |
| "grad_norm": 0.1859700083732605, | |
| "learning_rate": 1.2601626016260162e-06, | |
| "loss": 1.2359, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.8821138211382114, | |
| "grad_norm": 0.20786939561367035, | |
| "learning_rate": 1.2195121951219514e-06, | |
| "loss": 1.2777, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.8861788617886179, | |
| "grad_norm": 0.18669180572032928, | |
| "learning_rate": 1.1788617886178863e-06, | |
| "loss": 1.2823, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.8902439024390244, | |
| "grad_norm": 0.2070901244878769, | |
| "learning_rate": 1.1382113821138213e-06, | |
| "loss": 1.3024, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.8943089430894309, | |
| "grad_norm": 0.19596239924430847, | |
| "learning_rate": 1.0975609756097562e-06, | |
| "loss": 1.2324, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.8983739837398373, | |
| "grad_norm": 0.19302377104759216, | |
| "learning_rate": 1.0569105691056912e-06, | |
| "loss": 1.2666, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.9024390243902439, | |
| "grad_norm": 0.19474874436855316, | |
| "learning_rate": 1.0162601626016261e-06, | |
| "loss": 1.2566, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.9065040650406504, | |
| "grad_norm": 0.18354295194149017, | |
| "learning_rate": 9.75609756097561e-07, | |
| "loss": 1.2615, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.9105691056910569, | |
| "grad_norm": 0.19459068775177002, | |
| "learning_rate": 9.349593495934959e-07, | |
| "loss": 1.2776, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.9146341463414634, | |
| "grad_norm": 0.20152311027050018, | |
| "learning_rate": 8.94308943089431e-07, | |
| "loss": 1.2719, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.9186991869918699, | |
| "grad_norm": 0.18795745074748993, | |
| "learning_rate": 8.53658536585366e-07, | |
| "loss": 1.2685, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.9227642276422764, | |
| "grad_norm": 0.1888560950756073, | |
| "learning_rate": 8.130081300813009e-07, | |
| "loss": 1.2476, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.926829268292683, | |
| "grad_norm": 0.189235657453537, | |
| "learning_rate": 7.723577235772359e-07, | |
| "loss": 1.2692, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.9308943089430894, | |
| "grad_norm": 0.17088742554187775, | |
| "learning_rate": 7.317073170731707e-07, | |
| "loss": 1.2531, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.9349593495934959, | |
| "grad_norm": 0.18722310662269592, | |
| "learning_rate": 6.910569105691058e-07, | |
| "loss": 1.2428, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.9390243902439024, | |
| "grad_norm": 0.17706981301307678, | |
| "learning_rate": 6.504065040650407e-07, | |
| "loss": 1.2566, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.943089430894309, | |
| "grad_norm": 0.20467019081115723, | |
| "learning_rate": 6.097560975609757e-07, | |
| "loss": 1.2928, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.9471544715447154, | |
| "grad_norm": 0.17970900237560272, | |
| "learning_rate": 5.691056910569106e-07, | |
| "loss": 1.2513, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.9512195121951219, | |
| "grad_norm": 0.17920483648777008, | |
| "learning_rate": 5.284552845528456e-07, | |
| "loss": 1.2004, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.9552845528455285, | |
| "grad_norm": 0.18399786949157715, | |
| "learning_rate": 4.878048780487805e-07, | |
| "loss": 1.2692, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.959349593495935, | |
| "grad_norm": 0.17873434722423553, | |
| "learning_rate": 4.471544715447155e-07, | |
| "loss": 1.2468, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.9634146341463414, | |
| "grad_norm": 0.18509528040885925, | |
| "learning_rate": 4.0650406504065046e-07, | |
| "loss": 1.271, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.967479674796748, | |
| "grad_norm": 0.19402359426021576, | |
| "learning_rate": 3.6585365853658536e-07, | |
| "loss": 1.2472, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.9715447154471545, | |
| "grad_norm": 0.19081974029541016, | |
| "learning_rate": 3.2520325203252037e-07, | |
| "loss": 1.2416, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.975609756097561, | |
| "grad_norm": 0.17355622351169586, | |
| "learning_rate": 2.845528455284553e-07, | |
| "loss": 1.2045, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.9796747967479674, | |
| "grad_norm": 0.20138268172740936, | |
| "learning_rate": 2.439024390243903e-07, | |
| "loss": 1.2959, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.983739837398374, | |
| "grad_norm": 0.18235686421394348, | |
| "learning_rate": 2.0325203252032523e-07, | |
| "loss": 1.2448, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.9878048780487805, | |
| "grad_norm": 0.17484009265899658, | |
| "learning_rate": 1.6260162601626018e-07, | |
| "loss": 1.2441, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.991869918699187, | |
| "grad_norm": 0.18030861020088196, | |
| "learning_rate": 1.2195121951219514e-07, | |
| "loss": 1.2433, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.9959349593495935, | |
| "grad_norm": 0.1873464286327362, | |
| "learning_rate": 8.130081300813009e-08, | |
| "loss": 1.2672, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.17657622694969177, | |
| "learning_rate": 4.0650406504065046e-08, | |
| "loss": 1.2245, | |
| "step": 246 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 246, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 0, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8.599595463265485e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |