| { | |
| "best_metric": 42.76086285863452, | |
| "best_model_checkpoint": "./checkpoint-10000", | |
| "epoch": 1.0, | |
| "eval_steps": 2000, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.4e-07, | |
| "loss": 1.6662, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.400000000000001e-07, | |
| "loss": 1.5882, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.44e-06, | |
| "loss": 1.4859, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.94e-06, | |
| "loss": 1.3801, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.4400000000000004e-06, | |
| "loss": 1.2195, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.9400000000000002e-06, | |
| "loss": 1.2027, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.44e-06, | |
| "loss": 1.1254, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.94e-06, | |
| "loss": 1.0992, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.440000000000001e-06, | |
| "loss": 1.1116, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.94e-06, | |
| "loss": 1.1071, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.4400000000000004e-06, | |
| "loss": 1.0727, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.94e-06, | |
| "loss": 0.9899, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.440000000000001e-06, | |
| "loss": 1.0348, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.9400000000000005e-06, | |
| "loss": 1.0282, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.440000000000001e-06, | |
| "loss": 0.9794, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.94e-06, | |
| "loss": 0.9765, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.44e-06, | |
| "loss": 0.9348, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.94e-06, | |
| "loss": 0.9781, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.440000000000001e-06, | |
| "loss": 0.9395, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.940000000000001e-06, | |
| "loss": 0.9068, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.976842105263158e-06, | |
| "loss": 0.9286, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.950526315789475e-06, | |
| "loss": 0.9343, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.92421052631579e-06, | |
| "loss": 0.9175, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.897894736842107e-06, | |
| "loss": 0.8807, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.871578947368422e-06, | |
| "loss": 0.9134, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.845263157894738e-06, | |
| "loss": 0.8834, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.818947368421053e-06, | |
| "loss": 0.8511, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.79263157894737e-06, | |
| "loss": 0.8868, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.766315789473685e-06, | |
| "loss": 0.8664, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.74e-06, | |
| "loss": 0.8324, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.713684210526317e-06, | |
| "loss": 0.8426, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.687368421052632e-06, | |
| "loss": 0.8381, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.661052631578948e-06, | |
| "loss": 0.8184, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.634736842105265e-06, | |
| "loss": 0.8261, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.60842105263158e-06, | |
| "loss": 0.8248, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.582105263157897e-06, | |
| "loss": 0.803, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.555789473684211e-06, | |
| "loss": 0.7759, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.529473684210528e-06, | |
| "loss": 0.8195, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.503157894736843e-06, | |
| "loss": 0.8211, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.476842105263158e-06, | |
| "loss": 0.8057, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.450526315789475e-06, | |
| "loss": 0.825, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.42421052631579e-06, | |
| "loss": 0.7549, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.397894736842106e-06, | |
| "loss": 0.8096, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.371578947368421e-06, | |
| "loss": 0.7947, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.345263157894738e-06, | |
| "loss": 0.812, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.318947368421053e-06, | |
| "loss": 0.7615, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.292631578947368e-06, | |
| "loss": 0.781, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.266315789473685e-06, | |
| "loss": 0.7546, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.240000000000001e-06, | |
| "loss": 0.7737, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.213684210526316e-06, | |
| "loss": 0.7733, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.187368421052633e-06, | |
| "loss": 0.7421, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.161052631578948e-06, | |
| "loss": 0.7283, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.134736842105265e-06, | |
| "loss": 0.7855, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.10842105263158e-06, | |
| "loss": 0.7727, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.082105263157896e-06, | |
| "loss": 0.7846, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.055789473684211e-06, | |
| "loss": 0.7535, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.029473684210526e-06, | |
| "loss": 0.7695, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.003157894736843e-06, | |
| "loss": 0.7598, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.976842105263158e-06, | |
| "loss": 0.765, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.950526315789474e-06, | |
| "loss": 0.7752, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.92421052631579e-06, | |
| "loss": 0.7359, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.897894736842106e-06, | |
| "loss": 0.8049, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.871578947368423e-06, | |
| "loss": 0.7799, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.845263157894738e-06, | |
| "loss": 0.7506, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.818947368421054e-06, | |
| "loss": 0.7106, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.79263157894737e-06, | |
| "loss": 0.6927, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.766315789473684e-06, | |
| "loss": 0.7267, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.740000000000001e-06, | |
| "loss": 0.7268, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.713684210526316e-06, | |
| "loss": 0.7288, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.687368421052633e-06, | |
| "loss": 0.7342, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.661052631578948e-06, | |
| "loss": 0.7446, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.634736842105264e-06, | |
| "loss": 0.7206, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.60842105263158e-06, | |
| "loss": 0.6896, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.582105263157894e-06, | |
| "loss": 0.6801, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.555789473684213e-06, | |
| "loss": 0.7053, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.529473684210528e-06, | |
| "loss": 0.7464, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.503157894736843e-06, | |
| "loss": 0.7012, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.47684210526316e-06, | |
| "loss": 0.7474, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.450526315789474e-06, | |
| "loss": 0.7307, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.42421052631579e-06, | |
| "loss": 0.7005, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.7135426998138428, | |
| "eval_runtime": 436.6989, | |
| "eval_samples_per_second": 11.461, | |
| "eval_steps_per_second": 0.36, | |
| "eval_wer": 51.536581314221976, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.397894736842106e-06, | |
| "loss": 0.7708, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.371578947368422e-06, | |
| "loss": 0.7349, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.345263157894737e-06, | |
| "loss": 0.7119, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.318947368421052e-06, | |
| "loss": 0.7018, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.292631578947369e-06, | |
| "loss": 0.7114, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.266315789473684e-06, | |
| "loss": 0.701, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.24e-06, | |
| "loss": 0.6855, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.213684210526316e-06, | |
| "loss": 0.7156, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.187368421052632e-06, | |
| "loss": 0.6985, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.161052631578949e-06, | |
| "loss": 0.705, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.134736842105264e-06, | |
| "loss": 0.6979, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.10842105263158e-06, | |
| "loss": 0.6929, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.082105263157896e-06, | |
| "loss": 0.6943, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.05578947368421e-06, | |
| "loss": 0.707, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.029473684210527e-06, | |
| "loss": 0.678, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.003157894736842e-06, | |
| "loss": 0.6805, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.976842105263159e-06, | |
| "loss": 0.6967, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.950526315789474e-06, | |
| "loss": 0.6595, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.92421052631579e-06, | |
| "loss": 0.653, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.897894736842106e-06, | |
| "loss": 0.687, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.87157894736842e-06, | |
| "loss": 0.6772, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.845263157894739e-06, | |
| "loss": 0.6975, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.818947368421054e-06, | |
| "loss": 0.6932, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.792631578947369e-06, | |
| "loss": 0.7012, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.766315789473685e-06, | |
| "loss": 0.686, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.74e-06, | |
| "loss": 0.6454, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.713684210526317e-06, | |
| "loss": 0.6852, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.687368421052632e-06, | |
| "loss": 0.6506, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.661052631578949e-06, | |
| "loss": 0.6917, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.634736842105264e-06, | |
| "loss": 0.6545, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.6084210526315795e-06, | |
| "loss": 0.6586, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.582105263157895e-06, | |
| "loss": 0.679, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.555789473684211e-06, | |
| "loss": 0.6672, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.529473684210526e-06, | |
| "loss": 0.6611, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.503157894736842e-06, | |
| "loss": 0.6907, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.476842105263159e-06, | |
| "loss": 0.6636, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.4505263157894744e-06, | |
| "loss": 0.6438, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.42421052631579e-06, | |
| "loss": 0.6208, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.397894736842106e-06, | |
| "loss": 0.638, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.371578947368422e-06, | |
| "loss": 0.6909, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.345263157894738e-06, | |
| "loss": 0.6642, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.3189473684210535e-06, | |
| "loss": 0.6624, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.292631578947369e-06, | |
| "loss": 0.6324, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.266315789473684e-06, | |
| "loss": 0.6556, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.24e-06, | |
| "loss": 0.7184, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.213684210526316e-06, | |
| "loss": 0.6502, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.187368421052632e-06, | |
| "loss": 0.6639, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.161052631578948e-06, | |
| "loss": 0.639, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.134736842105264e-06, | |
| "loss": 0.6593, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.10842105263158e-06, | |
| "loss": 0.6519, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.082105263157896e-06, | |
| "loss": 0.6465, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.055789473684212e-06, | |
| "loss": 0.6401, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.0294736842105275e-06, | |
| "loss": 0.6329, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.0031578947368425e-06, | |
| "loss": 0.6461, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.976842105263158e-06, | |
| "loss": 0.6286, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.950526315789474e-06, | |
| "loss": 0.646, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.92421052631579e-06, | |
| "loss": 0.6578, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.897894736842106e-06, | |
| "loss": 0.6468, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.871578947368422e-06, | |
| "loss": 0.6224, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.845263157894737e-06, | |
| "loss": 0.6628, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.818947368421052e-06, | |
| "loss": 0.6386, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.792631578947368e-06, | |
| "loss": 0.6612, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.766315789473685e-06, | |
| "loss": 0.6707, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.740000000000001e-06, | |
| "loss": 0.6297, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.7136842105263165e-06, | |
| "loss": 0.626, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.687368421052632e-06, | |
| "loss": 0.6481, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.661052631578948e-06, | |
| "loss": 0.6489, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.634736842105264e-06, | |
| "loss": 0.6169, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.60842105263158e-06, | |
| "loss": 0.6204, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.582105263157896e-06, | |
| "loss": 0.6138, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.5557894736842106e-06, | |
| "loss": 0.609, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.529473684210526e-06, | |
| "loss": 0.6137, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.503157894736842e-06, | |
| "loss": 0.6208, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.476842105263158e-06, | |
| "loss": 0.6497, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.450526315789474e-06, | |
| "loss": 0.6193, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.42421052631579e-06, | |
| "loss": 0.6229, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.397894736842106e-06, | |
| "loss": 0.5991, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.371578947368422e-06, | |
| "loss": 0.6543, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.345263157894738e-06, | |
| "loss": 0.648, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.318947368421054e-06, | |
| "loss": 0.6267, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.6309367418289185, | |
| "eval_runtime": 403.8252, | |
| "eval_samples_per_second": 12.394, | |
| "eval_steps_per_second": 0.389, | |
| "eval_wer": 50.94330667552624, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.292631578947369e-06, | |
| "loss": 0.635, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.267368421052632e-06, | |
| "loss": 0.6179, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.241052631578948e-06, | |
| "loss": 0.6384, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.214736842105264e-06, | |
| "loss": 0.6188, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.188421052631579e-06, | |
| "loss": 0.6474, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.1621052631578945e-06, | |
| "loss": 0.6328, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.13578947368421e-06, | |
| "loss": 0.6264, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.109473684210527e-06, | |
| "loss": 0.5977, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.083157894736843e-06, | |
| "loss": 0.6131, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.056842105263159e-06, | |
| "loss": 0.6194, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.0305263157894745e-06, | |
| "loss": 0.6223, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.00421052631579e-06, | |
| "loss": 0.6101, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.977894736842106e-06, | |
| "loss": 0.6336, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.951578947368422e-06, | |
| "loss": 0.6282, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.925263157894737e-06, | |
| "loss": 0.6566, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.898947368421053e-06, | |
| "loss": 0.6117, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.8726315789473685e-06, | |
| "loss": 0.6015, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.846315789473684e-06, | |
| "loss": 0.6123, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.82e-06, | |
| "loss": 0.5862, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.793684210526316e-06, | |
| "loss": 0.6229, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.767368421052632e-06, | |
| "loss": 0.618, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.7410526315789485e-06, | |
| "loss": 0.5953, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.714736842105264e-06, | |
| "loss": 0.5907, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.68842105263158e-06, | |
| "loss": 0.6315, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.662105263157895e-06, | |
| "loss": 0.6302, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.635789473684211e-06, | |
| "loss": 0.5802, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.609473684210527e-06, | |
| "loss": 0.6067, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.5831578947368425e-06, | |
| "loss": 0.6164, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.556842105263158e-06, | |
| "loss": 0.6171, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.530526315789474e-06, | |
| "loss": 0.6182, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.50421052631579e-06, | |
| "loss": 0.6245, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.477894736842105e-06, | |
| "loss": 0.6152, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.451578947368421e-06, | |
| "loss": 0.6387, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.425263157894737e-06, | |
| "loss": 0.6096, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.398947368421053e-06, | |
| "loss": 0.6575, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.372631578947369e-06, | |
| "loss": 0.5805, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.346315789473685e-06, | |
| "loss": 0.5909, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.320000000000001e-06, | |
| "loss": 0.5884, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.2936842105263166e-06, | |
| "loss": 0.6108, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.267368421052632e-06, | |
| "loss": 0.6104, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.241052631578948e-06, | |
| "loss": 0.6116, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.214736842105263e-06, | |
| "loss": 0.6191, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.188421052631579e-06, | |
| "loss": 0.6403, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.162105263157895e-06, | |
| "loss": 0.6045, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.135789473684211e-06, | |
| "loss": 0.5822, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.1094736842105264e-06, | |
| "loss": 0.5876, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.083157894736842e-06, | |
| "loss": 0.634, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.056842105263158e-06, | |
| "loss": 0.6033, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.030526315789475e-06, | |
| "loss": 0.6072, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.0042105263157906e-06, | |
| "loss": 0.5876, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.9778947368421055e-06, | |
| "loss": 0.6234, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.951578947368421e-06, | |
| "loss": 0.6089, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.925263157894737e-06, | |
| "loss": 0.5988, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.898947368421053e-06, | |
| "loss": 0.6183, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.872631578947369e-06, | |
| "loss": 0.6051, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.846315789473685e-06, | |
| "loss": 0.575, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.8200000000000004e-06, | |
| "loss": 0.6117, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.793684210526316e-06, | |
| "loss": 0.5778, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.767368421052632e-06, | |
| "loss": 0.6341, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.741052631578948e-06, | |
| "loss": 0.6157, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.714736842105264e-06, | |
| "loss": 0.5898, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.6884210526315795e-06, | |
| "loss": 0.575, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.662105263157895e-06, | |
| "loss": 0.6109, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.63578947368421e-06, | |
| "loss": 0.5851, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.609473684210526e-06, | |
| "loss": 0.6109, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.583157894736843e-06, | |
| "loss": 0.6034, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.556842105263159e-06, | |
| "loss": 0.5678, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.5305263157894744e-06, | |
| "loss": 0.6222, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.504210526315789e-06, | |
| "loss": 0.5833, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.477894736842105e-06, | |
| "loss": 0.5756, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.451578947368421e-06, | |
| "loss": 0.6017, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.425263157894737e-06, | |
| "loss": 0.6033, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.3989473684210535e-06, | |
| "loss": 0.5788, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.3726315789473685e-06, | |
| "loss": 0.5869, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.346315789473684e-06, | |
| "loss": 0.5965, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.32e-06, | |
| "loss": 0.5983, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.293684210526316e-06, | |
| "loss": 0.5778, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.267368421052632e-06, | |
| "loss": 0.565, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.241052631578948e-06, | |
| "loss": 0.6058, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.214736842105263e-06, | |
| "loss": 0.5886, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.5892367362976074, | |
| "eval_runtime": 422.3368, | |
| "eval_samples_per_second": 11.851, | |
| "eval_steps_per_second": 0.372, | |
| "eval_wer": 50.02254443627044, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.188421052631579e-06, | |
| "loss": 0.5986, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.162105263157895e-06, | |
| "loss": 0.5945, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.135789473684211e-06, | |
| "loss": 0.591, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.109473684210527e-06, | |
| "loss": 0.5771, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.08421052631579e-06, | |
| "loss": 0.5946, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.057894736842106e-06, | |
| "loss": 0.5638, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.031578947368422e-06, | |
| "loss": 0.597, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.005263157894737e-06, | |
| "loss": 0.5793, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9789473684210525e-06, | |
| "loss": 0.5645, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.952631578947368e-06, | |
| "loss": 0.6107, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.926315789473685e-06, | |
| "loss": 0.5951, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.5811, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.873684210526316e-06, | |
| "loss": 0.5749, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.8473684210526316e-06, | |
| "loss": 0.5914, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.821052631578947e-06, | |
| "loss": 0.5615, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.794736842105263e-06, | |
| "loss": 0.5701, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.768421052631579e-06, | |
| "loss": 0.5854, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.7421052631578953e-06, | |
| "loss": 0.5495, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.715789473684211e-06, | |
| "loss": 0.6006, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.6894736842105265e-06, | |
| "loss": 0.6187, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.6631578947368423e-06, | |
| "loss": 0.5812, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.636842105263158e-06, | |
| "loss": 0.5769, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.610526315789474e-06, | |
| "loss": 0.5981, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.58421052631579e-06, | |
| "loss": 0.6079, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.5578947368421056e-06, | |
| "loss": 0.5721, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.5315789473684214e-06, | |
| "loss": 0.5537, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.505263157894737e-06, | |
| "loss": 0.5549, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.478947368421053e-06, | |
| "loss": 0.5688, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.4526315789473684e-06, | |
| "loss": 0.5795, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.4263157894736842e-06, | |
| "loss": 0.574, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.5672, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.3736842105263163e-06, | |
| "loss": 0.5893, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.347368421052632e-06, | |
| "loss": 0.5996, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.3210526315789475e-06, | |
| "loss": 0.5472, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2947368421052633e-06, | |
| "loss": 0.5832, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.268421052631579e-06, | |
| "loss": 0.5587, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2421052631578945e-06, | |
| "loss": 0.588, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.215789473684211e-06, | |
| "loss": 0.5712, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1894736842105266e-06, | |
| "loss": 0.6115, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1631578947368424e-06, | |
| "loss": 0.57, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1368421052631582e-06, | |
| "loss": 0.5756, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1105263157894736e-06, | |
| "loss": 0.5401, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0842105263157895e-06, | |
| "loss": 0.5939, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0578947368421053e-06, | |
| "loss": 0.5911, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0315789473684215e-06, | |
| "loss": 0.5722, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0052631578947373e-06, | |
| "loss": 0.5766, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9789473684210527e-06, | |
| "loss": 0.5718, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9526315789473685e-06, | |
| "loss": 0.5511, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9263157894736844e-06, | |
| "loss": 0.5551, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.5906, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.8736842105263164e-06, | |
| "loss": 0.5383, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.847368421052632e-06, | |
| "loss": 0.5771, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.8210526315789476e-06, | |
| "loss": 0.5755, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.7947368421052635e-06, | |
| "loss": 0.5551, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7684210526315793e-06, | |
| "loss": 0.564, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7421052631578947e-06, | |
| "loss": 0.5542, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7157894736842105e-06, | |
| "loss": 0.5679, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.6894736842105267e-06, | |
| "loss": 0.5954, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6631578947368426e-06, | |
| "loss": 0.5502, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6368421052631584e-06, | |
| "loss": 0.5569, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6105263157894738e-06, | |
| "loss": 0.5477, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5842105263157896e-06, | |
| "loss": 0.5707, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5578947368421054e-06, | |
| "loss": 0.5519, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5315789473684212e-06, | |
| "loss": 0.5938, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5052631578947375e-06, | |
| "loss": 0.5453, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.478947368421053e-06, | |
| "loss": 0.5627, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.4526315789473687e-06, | |
| "loss": 0.5972, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.4263157894736845e-06, | |
| "loss": 0.573, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.5764, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.373684210526316e-06, | |
| "loss": 0.5714, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.347368421052632e-06, | |
| "loss": 0.5668, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3210526315789473e-06, | |
| "loss": 0.5775, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.294736842105263e-06, | |
| "loss": 0.5845, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.268421052631579e-06, | |
| "loss": 0.5707, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.242105263157895e-06, | |
| "loss": 0.5894, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2157894736842106e-06, | |
| "loss": 0.5615, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.1894736842105264e-06, | |
| "loss": 0.5488, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1631578947368423e-06, | |
| "loss": 0.5586, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.136842105263158e-06, | |
| "loss": 0.5904, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.110526315789474e-06, | |
| "loss": 0.5627, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.567868173122406, | |
| "eval_runtime": 396.2585, | |
| "eval_samples_per_second": 12.631, | |
| "eval_steps_per_second": 0.396, | |
| "eval_wer": 43.945039037471226, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.0842105263157897e-06, | |
| "loss": 0.5628, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.0578947368421055e-06, | |
| "loss": 0.5972, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.031578947368421e-06, | |
| "loss": 0.5498, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.005263157894737e-06, | |
| "loss": 0.5645, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.978947368421053e-06, | |
| "loss": 0.5648, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.9526315789473684e-06, | |
| "loss": 0.5667, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.9263157894736846e-06, | |
| "loss": 0.5696, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.5914, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8736842105263158e-06, | |
| "loss": 0.5556, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8473684210526319e-06, | |
| "loss": 0.5576, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8210526315789475e-06, | |
| "loss": 0.5615, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7947368421052633e-06, | |
| "loss": 0.5495, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.768421052631579e-06, | |
| "loss": 0.5661, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.742105263157895e-06, | |
| "loss": 0.544, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.716842105263158e-06, | |
| "loss": 0.5326, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6905263157894739e-06, | |
| "loss": 0.5353, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6642105263157895e-06, | |
| "loss": 0.5401, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6378947368421053e-06, | |
| "loss": 0.5504, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6115789473684211e-06, | |
| "loss": 0.5611, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.585263157894737e-06, | |
| "loss": 0.5457, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5589473684210526e-06, | |
| "loss": 0.5333, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5326315789473686e-06, | |
| "loss": 0.5699, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.5063157894736844e-06, | |
| "loss": 0.5316, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.5738, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.453684210526316e-06, | |
| "loss": 0.565, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4273684210526317e-06, | |
| "loss": 0.571, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.4010526315789475e-06, | |
| "loss": 0.5665, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.374736842105263e-06, | |
| "loss": 0.5527, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.3484210526315791e-06, | |
| "loss": 0.5277, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.322105263157895e-06, | |
| "loss": 0.522, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2957894736842105e-06, | |
| "loss": 0.5784, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2694736842105266e-06, | |
| "loss": 0.5685, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2431578947368422e-06, | |
| "loss": 0.5707, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.216842105263158e-06, | |
| "loss": 0.5582, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1905263157894738e-06, | |
| "loss": 0.5629, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1642105263157896e-06, | |
| "loss": 0.5391, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1378947368421052e-06, | |
| "loss": 0.5384, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1115789473684213e-06, | |
| "loss": 0.5458, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0852631578947369e-06, | |
| "loss": 0.5788, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0589473684210527e-06, | |
| "loss": 0.5698, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0326315789473685e-06, | |
| "loss": 0.5349, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0063157894736843e-06, | |
| "loss": 0.5231, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 0.5363, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.536842105263158e-07, | |
| "loss": 0.566, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.273684210526317e-07, | |
| "loss": 0.5928, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.010526315789474e-07, | |
| "loss": 0.5697, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.747368421052632e-07, | |
| "loss": 0.5603, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.48421052631579e-07, | |
| "loss": 0.5458, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.221052631578947e-07, | |
| "loss": 0.5733, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.957894736842107e-07, | |
| "loss": 0.5626, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.694736842105263e-07, | |
| "loss": 0.5243, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.431578947368422e-07, | |
| "loss": 0.5582, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.16842105263158e-07, | |
| "loss": 0.5822, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.905263157894737e-07, | |
| "loss": 0.543, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.642105263157895e-07, | |
| "loss": 0.5528, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.378947368421053e-07, | |
| "loss": 0.5618, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.115789473684211e-07, | |
| "loss": 0.5527, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.852631578947369e-07, | |
| "loss": 0.5681, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.589473684210526e-07, | |
| "loss": 0.5638, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.326315789473684e-07, | |
| "loss": 0.568, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.063157894736842e-07, | |
| "loss": 0.5424, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 0.5734, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.5368421052631583e-07, | |
| "loss": 0.5366, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.273684210526316e-07, | |
| "loss": 0.5485, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.0105263157894736e-07, | |
| "loss": 0.5489, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.7473684210526323e-07, | |
| "loss": 0.5682, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.48421052631579e-07, | |
| "loss": 0.5699, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.2210526315789476e-07, | |
| "loss": 0.5595, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.9578947368421053e-07, | |
| "loss": 0.5514, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.6947368421052635e-07, | |
| "loss": 0.5848, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.431578947368421e-07, | |
| "loss": 0.5723, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.168421052631579e-07, | |
| "loss": 0.5605, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.9052631578947372e-07, | |
| "loss": 0.5862, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.642105263157895e-07, | |
| "loss": 0.5483, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.3789473684210528e-07, | |
| "loss": 0.5528, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.1157894736842106e-07, | |
| "loss": 0.5322, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.526315789473685e-08, | |
| "loss": 0.5614, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.8947368421052637e-08, | |
| "loss": 0.5457, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.263157894736842e-08, | |
| "loss": 0.5877, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 6.315789473684211e-09, | |
| "loss": 0.5694, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.5582889914512634, | |
| "eval_runtime": 400.4487, | |
| "eval_samples_per_second": 12.498, | |
| "eval_steps_per_second": 0.392, | |
| "eval_wer": 42.76086285863452, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 10000, | |
| "total_flos": 7.8780432384e+18, | |
| "train_loss": 0.6556177062988281, | |
| "train_runtime": 23200.9159, | |
| "train_samples_per_second": 13.793, | |
| "train_steps_per_second": 0.431 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9223372036854775807, | |
| "save_steps": 2000, | |
| "total_flos": 7.8780432384e+18, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |