| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9984, | |
| "eval_steps": 390, | |
| "global_step": 390, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0128, | |
| "grad_norm": 161.61465524426922, | |
| "learning_rate": 7.500000000000001e-08, | |
| "loss": 2.4411, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0256, | |
| "grad_norm": 154.06818191195163, | |
| "learning_rate": 1.5000000000000002e-07, | |
| "loss": 2.414, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0384, | |
| "grad_norm": 117.19018117409365, | |
| "learning_rate": 2.25e-07, | |
| "loss": 2.2681, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0512, | |
| "grad_norm": 75.19062918097424, | |
| "learning_rate": 3.0000000000000004e-07, | |
| "loss": 1.7582, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.064, | |
| "grad_norm": 58.25969834757995, | |
| "learning_rate": 3.75e-07, | |
| "loss": 1.1475, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0768, | |
| "grad_norm": 17.7044048384125, | |
| "learning_rate": 4.5e-07, | |
| "loss": 0.606, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0896, | |
| "grad_norm": 16.186509390955344, | |
| "learning_rate": 5.25e-07, | |
| "loss": 0.4462, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1024, | |
| "grad_norm": 28.948881179049923, | |
| "learning_rate": 6.000000000000001e-07, | |
| "loss": 0.3074, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.1152, | |
| "grad_norm": 9.08435498039366, | |
| "learning_rate": 6.75e-07, | |
| "loss": 0.2532, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.128, | |
| "grad_norm": 6.5840402479180575, | |
| "learning_rate": 7.5e-07, | |
| "loss": 0.2201, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1408, | |
| "grad_norm": 7.327177616981191, | |
| "learning_rate": 8.25e-07, | |
| "loss": 0.1959, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.1536, | |
| "grad_norm": 6.172579048424751, | |
| "learning_rate": 9e-07, | |
| "loss": 0.1806, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.1664, | |
| "grad_norm": 6.085525518741895, | |
| "learning_rate": 9.75e-07, | |
| "loss": 0.169, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.1792, | |
| "grad_norm": 16.06939591089291, | |
| "learning_rate": 1.05e-06, | |
| "loss": 0.1498, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.192, | |
| "grad_norm": 8.11523279664268, | |
| "learning_rate": 1.125e-06, | |
| "loss": 0.1537, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.2048, | |
| "grad_norm": 7.906944055310749, | |
| "learning_rate": 1.2000000000000002e-06, | |
| "loss": 0.1439, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2176, | |
| "grad_norm": 6.43033129109838, | |
| "learning_rate": 1.275e-06, | |
| "loss": 0.1326, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.2304, | |
| "grad_norm": 8.02640200042406, | |
| "learning_rate": 1.35e-06, | |
| "loss": 0.1227, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.2432, | |
| "grad_norm": 6.69740377832319, | |
| "learning_rate": 1.425e-06, | |
| "loss": 0.1158, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.256, | |
| "grad_norm": 5.197915018046733, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.1094, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.2688, | |
| "grad_norm": 5.060029864642565, | |
| "learning_rate": 1.5750000000000002e-06, | |
| "loss": 0.1003, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.2816, | |
| "grad_norm": 7.497811681915685, | |
| "learning_rate": 1.65e-06, | |
| "loss": 0.091, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2944, | |
| "grad_norm": 7.034577528708689, | |
| "learning_rate": 1.725e-06, | |
| "loss": 0.0861, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.3072, | |
| "grad_norm": 11.080391678804938, | |
| "learning_rate": 1.8e-06, | |
| "loss": 0.0893, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 19.426631066876524, | |
| "learning_rate": 1.875e-06, | |
| "loss": 0.0827, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.3328, | |
| "grad_norm": 4.952420510103284, | |
| "learning_rate": 1.95e-06, | |
| "loss": 0.0831, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.3456, | |
| "grad_norm": 10.227216835328893, | |
| "learning_rate": 2.025e-06, | |
| "loss": 0.0773, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.3584, | |
| "grad_norm": 7.807787341242278, | |
| "learning_rate": 2.1e-06, | |
| "loss": 0.0761, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.3712, | |
| "grad_norm": 5.774196535784976, | |
| "learning_rate": 2.175e-06, | |
| "loss": 0.0752, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.384, | |
| "grad_norm": 3.9786936164168414, | |
| "learning_rate": 2.25e-06, | |
| "loss": 0.0722, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.3968, | |
| "grad_norm": 4.6447288095056996, | |
| "learning_rate": 2.325e-06, | |
| "loss": 0.0606, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.4096, | |
| "grad_norm": 5.149169735072777, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.0713, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.4224, | |
| "grad_norm": 11.516436146930202, | |
| "learning_rate": 2.475e-06, | |
| "loss": 0.064, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.4352, | |
| "grad_norm": 12.631792384481054, | |
| "learning_rate": 2.55e-06, | |
| "loss": 0.061, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.448, | |
| "grad_norm": 12.412923039035704, | |
| "learning_rate": 2.6250000000000003e-06, | |
| "loss": 0.0621, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.4608, | |
| "grad_norm": 6.029127426855161, | |
| "learning_rate": 2.7e-06, | |
| "loss": 0.0674, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.4736, | |
| "grad_norm": 4.2613952463850415, | |
| "learning_rate": 2.775e-06, | |
| "loss": 0.068, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.4864, | |
| "grad_norm": 6.372179389423735, | |
| "learning_rate": 2.85e-06, | |
| "loss": 0.0628, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.4992, | |
| "grad_norm": 6.41276579707489, | |
| "learning_rate": 2.925e-06, | |
| "loss": 0.0612, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.512, | |
| "grad_norm": 5.9771618570220655, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0496, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.5248, | |
| "grad_norm": 3.7882498321692775, | |
| "learning_rate": 2.9210526315789475e-06, | |
| "loss": 0.0579, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.5376, | |
| "grad_norm": 3.2901697453585323, | |
| "learning_rate": 2.8421052631578946e-06, | |
| "loss": 0.0603, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.5504, | |
| "grad_norm": 4.418781826576313, | |
| "learning_rate": 2.763157894736842e-06, | |
| "loss": 0.0596, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.5632, | |
| "grad_norm": 6.086889232894656, | |
| "learning_rate": 2.6842105263157895e-06, | |
| "loss": 0.0585, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.576, | |
| "grad_norm": 4.6861307205960925, | |
| "learning_rate": 2.605263157894737e-06, | |
| "loss": 0.0557, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.5888, | |
| "grad_norm": 4.745714589263688, | |
| "learning_rate": 2.526315789473684e-06, | |
| "loss": 0.0562, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.6016, | |
| "grad_norm": 4.697181466590033, | |
| "learning_rate": 2.4473684210526314e-06, | |
| "loss": 0.052, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.6144, | |
| "grad_norm": 5.676309664606183, | |
| "learning_rate": 2.368421052631579e-06, | |
| "loss": 0.0581, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.6272, | |
| "grad_norm": 6.187909188215041, | |
| "learning_rate": 2.2894736842105263e-06, | |
| "loss": 0.0596, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 4.216720589170728, | |
| "learning_rate": 2.2105263157894738e-06, | |
| "loss": 0.0496, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.6528, | |
| "grad_norm": 5.196715826539319, | |
| "learning_rate": 2.1315789473684212e-06, | |
| "loss": 0.0515, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.6656, | |
| "grad_norm": 11.353541751433445, | |
| "learning_rate": 2.0526315789473687e-06, | |
| "loss": 0.0407, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.6784, | |
| "grad_norm": 5.177415075589649, | |
| "learning_rate": 1.973684210526316e-06, | |
| "loss": 0.0417, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.6912, | |
| "grad_norm": 8.946108487101375, | |
| "learning_rate": 1.8947368421052632e-06, | |
| "loss": 0.0488, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.704, | |
| "grad_norm": 4.475569997515576, | |
| "learning_rate": 1.8157894736842106e-06, | |
| "loss": 0.0521, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.7168, | |
| "grad_norm": 6.784366889150163, | |
| "learning_rate": 1.736842105263158e-06, | |
| "loss": 0.0473, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.7296, | |
| "grad_norm": 12.414020629521627, | |
| "learning_rate": 1.6578947368421056e-06, | |
| "loss": 0.046, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.7424, | |
| "grad_norm": 9.884779353467142, | |
| "learning_rate": 1.5789473684210526e-06, | |
| "loss": 0.0502, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.7552, | |
| "grad_norm": 6.678072565724174, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0473, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.768, | |
| "grad_norm": 5.712356237287952, | |
| "learning_rate": 1.4210526315789473e-06, | |
| "loss": 0.0431, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.7808, | |
| "grad_norm": 4.9395960174524145, | |
| "learning_rate": 1.3421052631578947e-06, | |
| "loss": 0.0401, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.7936, | |
| "grad_norm": 6.834755086014392, | |
| "learning_rate": 1.263157894736842e-06, | |
| "loss": 0.0397, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.8064, | |
| "grad_norm": 2.775196760101048, | |
| "learning_rate": 1.1842105263157894e-06, | |
| "loss": 0.0377, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.8192, | |
| "grad_norm": 4.786995996855356, | |
| "learning_rate": 1.1052631578947369e-06, | |
| "loss": 0.0398, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.832, | |
| "grad_norm": 14.331978713606828, | |
| "learning_rate": 1.0263157894736843e-06, | |
| "loss": 0.0387, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.8448, | |
| "grad_norm": 5.110049394601092, | |
| "learning_rate": 9.473684210526316e-07, | |
| "loss": 0.0412, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.8576, | |
| "grad_norm": 2.3797196581438, | |
| "learning_rate": 8.68421052631579e-07, | |
| "loss": 0.0407, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.8704, | |
| "grad_norm": 3.1978316895686003, | |
| "learning_rate": 7.894736842105263e-07, | |
| "loss": 0.0397, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.8832, | |
| "grad_norm": 4.6674024529684885, | |
| "learning_rate": 7.105263157894736e-07, | |
| "loss": 0.0399, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.896, | |
| "grad_norm": 2.5803866776162767, | |
| "learning_rate": 6.31578947368421e-07, | |
| "loss": 0.0336, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.9088, | |
| "grad_norm": 4.604776602886812, | |
| "learning_rate": 5.526315789473684e-07, | |
| "loss": 0.0383, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.9216, | |
| "grad_norm": 11.254341619231537, | |
| "learning_rate": 4.736842105263158e-07, | |
| "loss": 0.0437, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.9344, | |
| "grad_norm": 4.658978706509525, | |
| "learning_rate": 3.9473684210526315e-07, | |
| "loss": 0.0402, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.9472, | |
| "grad_norm": 5.509226127844381, | |
| "learning_rate": 3.157894736842105e-07, | |
| "loss": 0.0422, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 2.556812465734739, | |
| "learning_rate": 2.368421052631579e-07, | |
| "loss": 0.0354, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.9728, | |
| "grad_norm": 1.847642831707115, | |
| "learning_rate": 1.5789473684210525e-07, | |
| "loss": 0.0392, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.9856, | |
| "grad_norm": 3.1108025008097293, | |
| "learning_rate": 7.894736842105262e-08, | |
| "loss": 0.0367, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.9984, | |
| "grad_norm": 4.407140308538649, | |
| "learning_rate": 0.0, | |
| "loss": 0.0361, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.9984, | |
| "eval_accuracy": 0.007941866145578897, | |
| "eval_loss": 0.031122559681534767, | |
| "eval_runtime": 44.2812, | |
| "eval_samples_per_second": 22.583, | |
| "eval_steps_per_second": 0.723, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.9984, | |
| "logic_eval_extr_stps.D-0.answer_accuracy": 0.3333333333333333, | |
| "logic_eval_extr_stps.D-0.proof_accuracy.zero_one": 0.3333333333333333, | |
| "logic_eval_extr_stps.D-0.rouge1": 35.245819999999995, | |
| "logic_eval_extr_stps.D-0.rouge2": 0.3415933333333333, | |
| "logic_eval_extr_stps.D-0.rougeL": 35.245819999999995, | |
| "logic_eval_extr_stps.D-0.rougeLsum": 35.245819999999995, | |
| "logic_eval_extr_stps.D-1.answer_accuracy": 0.391304347826087, | |
| "logic_eval_extr_stps.D-1.proof_accuracy.zero_one": 0.30434782608695654, | |
| "logic_eval_extr_stps.D-1.rouge1": 30.34304782608695, | |
| "logic_eval_extr_stps.D-1.rouge2": 4.933721739130434, | |
| "logic_eval_extr_stps.D-1.rougeL": 30.34304782608695, | |
| "logic_eval_extr_stps.D-1.rougeLsum": 30.34304782608695, | |
| "logic_eval_extr_stps.D-2.answer_accuracy": 0.3333333333333333, | |
| "logic_eval_extr_stps.D-2.proof_accuracy.zero_one": 0.2222222222222222, | |
| "logic_eval_extr_stps.D-2.rouge1": 33.186255555555555, | |
| "logic_eval_extr_stps.D-2.rouge2": 6.600677777777777, | |
| "logic_eval_extr_stps.D-2.rougeL": 32.34552777777778, | |
| "logic_eval_extr_stps.D-2.rougeLsum": 32.828688888888884, | |
| "logic_eval_extr_stps.D-3.answer_accuracy": 0.21428571428571427, | |
| "logic_eval_extr_stps.D-3.proof_accuracy.zero_one": 0.07142857142857142, | |
| "logic_eval_extr_stps.D-3.rouge1": 11.588842857142854, | |
| "logic_eval_extr_stps.D-3.rouge2": 7.087435714285713, | |
| "logic_eval_extr_stps.D-3.rougeL": 10.851507142857143, | |
| "logic_eval_extr_stps.D-3.rougeLsum": 11.588842857142854, | |
| "logic_eval_extr_stps.D-None.answer_accuracy": 0.12903225806451613, | |
| "logic_eval_extr_stps.D-None.proof_accuracy.zero_one": 0.12903225806451613, | |
| "logic_eval_extr_stps.D-None.rouge1": 14.077829032258066, | |
| "logic_eval_extr_stps.D-None.rouge2": 0.0, | |
| "logic_eval_extr_stps.D-None.rougeL": 14.077829032258066, | |
| "logic_eval_extr_stps.D-None.rougeLsum": 14.077829032258066, | |
| "logic_eval_extr_stps.D-all.answer_accuracy": 0.26732673267326734, | |
| "logic_eval_extr_stps.D-all.proof_accuracy.zero_one": 0.2079207920792079, | |
| "logic_eval_extr_stps.D-all.rouge1": 23.986004950495047, | |
| "logic_eval_extr_stps.D-all.rouge2": 3.3330277227722767, | |
| "logic_eval_extr_stps.D-all.rougeL": 23.73396732673267, | |
| "logic_eval_extr_stps.D-all.rougeLsum": 23.922280198019806, | |
| "logic_eval_gen_len": 331.7734375, | |
| "logic_eval_runtime": 658.0701, | |
| "logic_eval_samples_per_second": 0.153, | |
| "logic_eval_steps_per_second": 0.006, | |
| "logic_eval_strct.D-0.answer_accuracy": 0.3333333333333333, | |
| "logic_eval_strct.D-0.proof_accuracy.zero_one": 0.3333333333333333, | |
| "logic_eval_strct.D-0.rouge1": 35.245819999999995, | |
| "logic_eval_strct.D-0.rouge2": 0.3415933333333333, | |
| "logic_eval_strct.D-0.rougeL": 35.245819999999995, | |
| "logic_eval_strct.D-0.rougeLsum": 35.245819999999995, | |
| "logic_eval_strct.D-1.answer_accuracy": 0.391304347826087, | |
| "logic_eval_strct.D-1.proof_accuracy.zero_one": 0.21739130434782608, | |
| "logic_eval_strct.D-1.rouge1": 30.34304782608695, | |
| "logic_eval_strct.D-1.rouge2": 4.933721739130434, | |
| "logic_eval_strct.D-1.rougeL": 30.34304782608695, | |
| "logic_eval_strct.D-1.rougeLsum": 30.34304782608695, | |
| "logic_eval_strct.D-2.answer_accuracy": 0.3333333333333333, | |
| "logic_eval_strct.D-2.proof_accuracy.zero_one": 0.2222222222222222, | |
| "logic_eval_strct.D-2.rouge1": 33.186255555555555, | |
| "logic_eval_strct.D-2.rouge2": 6.600677777777777, | |
| "logic_eval_strct.D-2.rougeL": 32.34552777777778, | |
| "logic_eval_strct.D-2.rougeLsum": 32.828688888888884, | |
| "logic_eval_strct.D-3.answer_accuracy": 0.21428571428571427, | |
| "logic_eval_strct.D-3.proof_accuracy.zero_one": 0.0, | |
| "logic_eval_strct.D-3.rouge1": 11.588842857142854, | |
| "logic_eval_strct.D-3.rouge2": 7.087435714285713, | |
| "logic_eval_strct.D-3.rougeL": 10.851507142857143, | |
| "logic_eval_strct.D-3.rougeLsum": 11.588842857142854, | |
| "logic_eval_strct.D-None.answer_accuracy": 0.12903225806451613, | |
| "logic_eval_strct.D-None.proof_accuracy.zero_one": 0.12903225806451613, | |
| "logic_eval_strct.D-None.rouge1": 14.077829032258066, | |
| "logic_eval_strct.D-None.rouge2": 0.0, | |
| "logic_eval_strct.D-None.rougeL": 14.077829032258066, | |
| "logic_eval_strct.D-None.rougeLsum": 14.077829032258066, | |
| "logic_eval_strct.D-all.answer_accuracy": 0.26732673267326734, | |
| "logic_eval_strct.D-all.proof_accuracy.zero_one": 0.1782178217821782, | |
| "logic_eval_strct.D-all.rouge1": 23.986004950495047, | |
| "logic_eval_strct.D-all.rouge2": 3.3330277227722767, | |
| "logic_eval_strct.D-all.rougeL": 23.73396732673267, | |
| "logic_eval_strct.D-all.rougeLsum": 23.922280198019806, | |
| "step": 390 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 390, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 390, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1617567361794048.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |