| { | |
| "best_metric": 19.56338265908963, | |
| "best_model_checkpoint": "./checkpoint-10000", | |
| "epoch": 1.0, | |
| "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": 4.2000000000000006e-07, | |
| "loss": 2.7063, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 2.2141, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.42e-06, | |
| "loss": 1.7227, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 1.3071, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.42e-06, | |
| "loss": 1.1018, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.9452, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.8143, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.6824, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.6915, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.6743, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.6472, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.5266, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.4592, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.5424, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.4654, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.54, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.4821, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.4119, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.4349, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.499, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.977894736842106e-06, | |
| "loss": 0.4124, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.951578947368423e-06, | |
| "loss": 0.3877, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.925263157894738e-06, | |
| "loss": 0.4366, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.898947368421054e-06, | |
| "loss": 0.4779, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.87263157894737e-06, | |
| "loss": 0.3866, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.846315789473684e-06, | |
| "loss": 0.373, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.820000000000001e-06, | |
| "loss": 0.3673, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.793684210526316e-06, | |
| "loss": 0.3992, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.767368421052632e-06, | |
| "loss": 0.4059, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.741052631578947e-06, | |
| "loss": 0.408, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.714736842105264e-06, | |
| "loss": 0.3547, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.68842105263158e-06, | |
| "loss": 0.3968, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.662105263157896e-06, | |
| "loss": 0.3506, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.635789473684212e-06, | |
| "loss": 0.2993, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.609473684210527e-06, | |
| "loss": 0.2815, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.583157894736842e-06, | |
| "loss": 0.3597, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.556842105263159e-06, | |
| "loss": 0.3249, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.530526315789474e-06, | |
| "loss": 0.3299, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.50421052631579e-06, | |
| "loss": 0.4214, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.477894736842106e-06, | |
| "loss": 0.3624, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "eval_loss": 0.46733734011650085, | |
| "eval_runtime": 16108.2955, | |
| "eval_samples_per_second": 0.903, | |
| "eval_steps_per_second": 0.903, | |
| "eval_wer": 43.614471389185844, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.451578947368422e-06, | |
| "loss": 0.3118, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.425263157894737e-06, | |
| "loss": 0.3262, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.398947368421052e-06, | |
| "loss": 0.253, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.372631578947369e-06, | |
| "loss": 0.2956, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.346315789473684e-06, | |
| "loss": 0.2814, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.32e-06, | |
| "loss": 0.2864, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.293684210526317e-06, | |
| "loss": 0.3649, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.267368421052632e-06, | |
| "loss": 0.2587, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.241052631578949e-06, | |
| "loss": 0.3107, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.214736842105264e-06, | |
| "loss": 0.2769, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.18842105263158e-06, | |
| "loss": 0.2529, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.162105263157895e-06, | |
| "loss": 0.2327, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.13578947368421e-06, | |
| "loss": 0.3611, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.109473684210527e-06, | |
| "loss": 0.31, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.083157894736842e-06, | |
| "loss": 0.3208, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.056842105263159e-06, | |
| "loss": 0.3195, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.030526315789474e-06, | |
| "loss": 0.2415, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.00421052631579e-06, | |
| "loss": 0.3314, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.977894736842107e-06, | |
| "loss": 0.2909, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.951578947368422e-06, | |
| "loss": 0.2933, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.925263157894739e-06, | |
| "loss": 0.2859, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 8.898947368421054e-06, | |
| "loss": 0.2198, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.872631578947369e-06, | |
| "loss": 0.2457, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.846315789473685e-06, | |
| "loss": 0.307, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 8.82e-06, | |
| "loss": 0.3081, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.793684210526317e-06, | |
| "loss": 0.2923, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.767368421052632e-06, | |
| "loss": 0.2635, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.741052631578949e-06, | |
| "loss": 0.2473, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.714736842105264e-06, | |
| "loss": 0.2758, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 8.688421052631579e-06, | |
| "loss": 0.2352, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.662105263157895e-06, | |
| "loss": 0.2723, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.63578947368421e-06, | |
| "loss": 0.2984, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.609473684210527e-06, | |
| "loss": 0.2674, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 8.583157894736843e-06, | |
| "loss": 0.2726, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.556842105263158e-06, | |
| "loss": 0.2737, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.530526315789475e-06, | |
| "loss": 0.2864, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 8.50421052631579e-06, | |
| "loss": 0.2703, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.477894736842107e-06, | |
| "loss": 0.2434, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.451578947368422e-06, | |
| "loss": 0.2455, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.425263157894737e-06, | |
| "loss": 0.2545, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "eval_loss": 0.40201425552368164, | |
| "eval_runtime": 15893.8779, | |
| "eval_samples_per_second": 0.915, | |
| "eval_steps_per_second": 0.915, | |
| "eval_wer": 36.79063748382378, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.398947368421053e-06, | |
| "loss": 0.249, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 8.372631578947368e-06, | |
| "loss": 0.2145, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.346315789473685e-06, | |
| "loss": 0.22, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.32e-06, | |
| "loss": 0.2295, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.293684210526317e-06, | |
| "loss": 0.2681, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 8.267368421052632e-06, | |
| "loss": 0.2511, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.241052631578948e-06, | |
| "loss": 0.2246, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.214736842105265e-06, | |
| "loss": 0.2637, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 8.18842105263158e-06, | |
| "loss": 0.2347, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.162105263157895e-06, | |
| "loss": 0.3043, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.135789473684212e-06, | |
| "loss": 0.2546, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.109473684210527e-06, | |
| "loss": 0.2846, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.083157894736843e-06, | |
| "loss": 0.2308, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.056842105263158e-06, | |
| "loss": 0.2462, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.030526315789475e-06, | |
| "loss": 0.2716, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.00421052631579e-06, | |
| "loss": 0.2759, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.977894736842105e-06, | |
| "loss": 0.2312, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 7.951578947368421e-06, | |
| "loss": 0.3181, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.925263157894736e-06, | |
| "loss": 0.2586, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.898947368421053e-06, | |
| "loss": 0.3121, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.87263157894737e-06, | |
| "loss": 0.3096, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.846315789473685e-06, | |
| "loss": 0.3347, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.820000000000001e-06, | |
| "loss": 0.313, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.793684210526316e-06, | |
| "loss": 0.2248, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.767368421052633e-06, | |
| "loss": 0.2801, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.741052631578948e-06, | |
| "loss": 0.2529, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.714736842105263e-06, | |
| "loss": 0.2754, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.68842105263158e-06, | |
| "loss": 0.2193, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 7.662105263157895e-06, | |
| "loss": 0.2425, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.635789473684211e-06, | |
| "loss": 0.2891, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.609473684210526e-06, | |
| "loss": 0.2889, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.583157894736842e-06, | |
| "loss": 0.2657, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.556842105263158e-06, | |
| "loss": 0.2662, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 7.5305263157894745e-06, | |
| "loss": 0.2419, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.50421052631579e-06, | |
| "loss": 0.2233, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.477894736842106e-06, | |
| "loss": 0.251, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.451578947368422e-06, | |
| "loss": 0.215, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 7.425263157894738e-06, | |
| "loss": 0.2424, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.398947368421054e-06, | |
| "loss": 0.2173, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.373684210526316e-06, | |
| "loss": 0.2277, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "eval_loss": 0.3431873023509979, | |
| "eval_runtime": 15984.0037, | |
| "eval_samples_per_second": 0.91, | |
| "eval_steps_per_second": 0.91, | |
| "eval_wer": 33.054633432735045, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.347368421052632e-06, | |
| "loss": 0.243, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 7.321052631578948e-06, | |
| "loss": 0.2002, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.2947368421052636e-06, | |
| "loss": 0.1748, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.268421052631579e-06, | |
| "loss": 0.1771, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 7.242105263157896e-06, | |
| "loss": 0.1603, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.215789473684212e-06, | |
| "loss": 0.1948, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.189473684210527e-06, | |
| "loss": 0.2245, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.163157894736843e-06, | |
| "loss": 0.1586, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.1368421052631585e-06, | |
| "loss": 0.172, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.110526315789474e-06, | |
| "loss": 0.2503, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.08421052631579e-06, | |
| "loss": 0.1755, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.057894736842106e-06, | |
| "loss": 0.1914, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.031578947368422e-06, | |
| "loss": 0.1853, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.005263157894738e-06, | |
| "loss": 0.194, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.9789473684210525e-06, | |
| "loss": 0.2195, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.952631578947368e-06, | |
| "loss": 0.2203, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.926315789473684e-06, | |
| "loss": 0.2099, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.2671, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.873684210526317e-06, | |
| "loss": 0.2283, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.8473684210526325e-06, | |
| "loss": 0.2256, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.821052631578948e-06, | |
| "loss": 0.2214, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 6.794736842105264e-06, | |
| "loss": 0.1956, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.76842105263158e-06, | |
| "loss": 0.1863, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.742105263157896e-06, | |
| "loss": 0.2448, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.715789473684211e-06, | |
| "loss": 0.276, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 6.6894736842105266e-06, | |
| "loss": 0.241, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.663157894736842e-06, | |
| "loss": 0.2317, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.636842105263158e-06, | |
| "loss": 0.2393, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 6.610526315789474e-06, | |
| "loss": 0.2278, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.58421052631579e-06, | |
| "loss": 0.294, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.557894736842106e-06, | |
| "loss": 0.2329, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.531578947368422e-06, | |
| "loss": 0.1795, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 6.505263157894738e-06, | |
| "loss": 0.2842, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.478947368421054e-06, | |
| "loss": 0.2002, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.452631578947369e-06, | |
| "loss": 0.2134, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.426315789473685e-06, | |
| "loss": 0.3012, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.2445, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.373684210526316e-06, | |
| "loss": 0.2097, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.347368421052632e-06, | |
| "loss": 0.2946, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.321052631578948e-06, | |
| "loss": 0.2705, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.3101558983325958, | |
| "eval_runtime": 16090.3734, | |
| "eval_samples_per_second": 0.904, | |
| "eval_steps_per_second": 0.904, | |
| "eval_wer": 29.53018623755134, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.294736842105264e-06, | |
| "loss": 0.2786, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.268421052631579e-06, | |
| "loss": 0.2752, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.242105263157895e-06, | |
| "loss": 0.2662, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.2157894736842104e-06, | |
| "loss": 0.2624, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.189473684210526e-06, | |
| "loss": 0.267, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.163157894736843e-06, | |
| "loss": 0.2599, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.136842105263159e-06, | |
| "loss": 0.2551, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.1105263157894746e-06, | |
| "loss": 0.3011, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.08421052631579e-06, | |
| "loss": 0.2521, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.057894736842106e-06, | |
| "loss": 0.2332, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.031578947368422e-06, | |
| "loss": 0.2634, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.005263157894737e-06, | |
| "loss": 0.2626, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.978947368421053e-06, | |
| "loss": 0.2938, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 5.952631578947369e-06, | |
| "loss": 0.3338, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.9263157894736844e-06, | |
| "loss": 0.2677, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.2715, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 5.873684210526316e-06, | |
| "loss": 0.2896, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.847368421052632e-06, | |
| "loss": 0.2297, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.8210526315789486e-06, | |
| "loss": 0.328, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.794736842105264e-06, | |
| "loss": 0.2461, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.76842105263158e-06, | |
| "loss": 0.2583, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.742105263157895e-06, | |
| "loss": 0.3263, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.715789473684211e-06, | |
| "loss": 0.2214, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.689473684210527e-06, | |
| "loss": 0.2239, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.663157894736843e-06, | |
| "loss": 0.2198, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.6368421052631584e-06, | |
| "loss": 0.1888, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.610526315789474e-06, | |
| "loss": 0.2148, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.58421052631579e-06, | |
| "loss": 0.2461, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.557894736842105e-06, | |
| "loss": 0.237, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.531578947368421e-06, | |
| "loss": 0.2502, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.505263157894737e-06, | |
| "loss": 0.2344, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.4789473684210525e-06, | |
| "loss": 0.2393, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.452631578947369e-06, | |
| "loss": 0.3315, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.426315789473685e-06, | |
| "loss": 0.2186, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.25, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.373684210526317e-06, | |
| "loss": 0.2442, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.3473684210526325e-06, | |
| "loss": 0.2324, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.321052631578948e-06, | |
| "loss": 0.2339, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.294736842105263e-06, | |
| "loss": 0.2998, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.268421052631579e-06, | |
| "loss": 0.3037, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "eval_loss": 0.27908819913864136, | |
| "eval_runtime": 16017.6918, | |
| "eval_samples_per_second": 0.908, | |
| "eval_steps_per_second": 0.908, | |
| "eval_wer": 26.013616159342824, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.242105263157895e-06, | |
| "loss": 0.2304, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.215789473684211e-06, | |
| "loss": 0.2299, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.1894736842105265e-06, | |
| "loss": 0.2663, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.163157894736842e-06, | |
| "loss": 0.2903, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.136842105263158e-06, | |
| "loss": 0.2498, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.110526315789474e-06, | |
| "loss": 0.1992, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.084210526315791e-06, | |
| "loss": 0.2307, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.0578947368421065e-06, | |
| "loss": 0.2184, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5.0315789473684214e-06, | |
| "loss": 0.2322, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 5.005263157894737e-06, | |
| "loss": 0.2221, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.978947368421053e-06, | |
| "loss": 0.1857, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.952631578947369e-06, | |
| "loss": 0.2228, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.926315789473685e-06, | |
| "loss": 0.2142, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.213, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.873684210526316e-06, | |
| "loss": 0.2028, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.847368421052631e-06, | |
| "loss": 0.2526, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.821052631578948e-06, | |
| "loss": 0.2017, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.794736842105264e-06, | |
| "loss": 0.2112, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.76842105263158e-06, | |
| "loss": 0.2116, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.7421052631578954e-06, | |
| "loss": 0.2518, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.71578947368421e-06, | |
| "loss": 0.24, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.689473684210526e-06, | |
| "loss": 0.2086, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.663157894736842e-06, | |
| "loss": 0.2712, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.636842105263159e-06, | |
| "loss": 0.2205, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.6105263157894745e-06, | |
| "loss": 0.2198, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.5842105263157895e-06, | |
| "loss": 0.2666, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.557894736842105e-06, | |
| "loss": 0.227, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.531578947368421e-06, | |
| "loss": 0.2003, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.505263157894737e-06, | |
| "loss": 0.2248, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.478947368421054e-06, | |
| "loss": 0.261, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.452631578947369e-06, | |
| "loss": 0.2562, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.426315789473684e-06, | |
| "loss": 0.2355, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.2401, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.373684210526316e-06, | |
| "loss": 0.2463, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.347368421052632e-06, | |
| "loss": 0.2237, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.321052631578948e-06, | |
| "loss": 0.231, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.2947368421052635e-06, | |
| "loss": 0.1764, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.268421052631579e-06, | |
| "loss": 0.1842, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.242105263157895e-06, | |
| "loss": 0.2062, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.215789473684211e-06, | |
| "loss": 0.2305, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "eval_loss": 0.25806787610054016, | |
| "eval_runtime": 16029.9634, | |
| "eval_samples_per_second": 0.907, | |
| "eval_steps_per_second": 0.907, | |
| "eval_wer": 23.4400495133067, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.189473684210527e-06, | |
| "loss": 0.2037, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.163157894736843e-06, | |
| "loss": 0.2035, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.136842105263158e-06, | |
| "loss": 0.1744, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.110526315789474e-06, | |
| "loss": 0.221, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.08421052631579e-06, | |
| "loss": 0.1453, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.057894736842106e-06, | |
| "loss": 0.3219, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.031578947368422e-06, | |
| "loss": 0.2429, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.005263157894737e-06, | |
| "loss": 0.1918, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9789473684210525e-06, | |
| "loss": 0.1868, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.952631578947368e-06, | |
| "loss": 0.1869, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.926315789473685e-06, | |
| "loss": 0.233, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.1965, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.873684210526316e-06, | |
| "loss": 0.2425, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.8473684210526316e-06, | |
| "loss": 0.19, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.821052631578947e-06, | |
| "loss": 0.1333, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.794736842105263e-06, | |
| "loss": 0.1947, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.768421052631579e-06, | |
| "loss": 0.1997, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.7421052631578953e-06, | |
| "loss": 0.2169, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.715789473684211e-06, | |
| "loss": 0.1955, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.6894736842105265e-06, | |
| "loss": 0.1971, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.6631578947368423e-06, | |
| "loss": 0.2377, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.636842105263158e-06, | |
| "loss": 0.217, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.610526315789474e-06, | |
| "loss": 0.1776, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.58421052631579e-06, | |
| "loss": 0.2029, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.5578947368421056e-06, | |
| "loss": 0.1875, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.5315789473684214e-06, | |
| "loss": 0.1315, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.505263157894737e-06, | |
| "loss": 0.2163, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.478947368421053e-06, | |
| "loss": 0.2003, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.4526315789473684e-06, | |
| "loss": 0.1764, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.4263157894736842e-06, | |
| "loss": 0.1842, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.1277, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.3736842105263163e-06, | |
| "loss": 0.2011, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.347368421052632e-06, | |
| "loss": 0.1866, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.3210526315789475e-06, | |
| "loss": 0.1197, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2947368421052633e-06, | |
| "loss": 0.1777, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.268421052631579e-06, | |
| "loss": 0.2, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.2421052631578945e-06, | |
| "loss": 0.1808, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.215789473684211e-06, | |
| "loss": 0.2083, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1894736842105266e-06, | |
| "loss": 0.1743, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1631578947368424e-06, | |
| "loss": 0.1279, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "eval_loss": 0.2451580911874771, | |
| "eval_runtime": 16136.9321, | |
| "eval_samples_per_second": 0.901, | |
| "eval_steps_per_second": 0.901, | |
| "eval_wer": 22.222472289427785, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1368421052631582e-06, | |
| "loss": 0.136, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.1105263157894736e-06, | |
| "loss": 0.1664, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0842105263157895e-06, | |
| "loss": 0.1344, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.058947368421053e-06, | |
| "loss": 0.1894, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0326315789473686e-06, | |
| "loss": 0.1335, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.0063157894736844e-06, | |
| "loss": 0.1622, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 0.1872, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9536842105263157e-06, | |
| "loss": 0.138, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.927368421052632e-06, | |
| "loss": 0.1726, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 2.9010526315789477e-06, | |
| "loss": 0.1876, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.8747368421052635e-06, | |
| "loss": 0.1599, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.8484210526315794e-06, | |
| "loss": 0.1572, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.8221052631578948e-06, | |
| "loss": 0.1549, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.7957894736842106e-06, | |
| "loss": 0.1333, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7694736842105264e-06, | |
| "loss": 0.1411, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7431578947368426e-06, | |
| "loss": 0.226, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.7168421052631585e-06, | |
| "loss": 0.1715, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 2.690526315789474e-06, | |
| "loss": 0.1366, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6642105263157897e-06, | |
| "loss": 0.1373, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6378947368421055e-06, | |
| "loss": 0.1591, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 2.6115789473684213e-06, | |
| "loss": 0.173, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5852631578947367e-06, | |
| "loss": 0.1492, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.558947368421053e-06, | |
| "loss": 0.1349, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5326315789473688e-06, | |
| "loss": 0.148, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 2.5063157894736846e-06, | |
| "loss": 0.1191, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 0.1395, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.453684210526316e-06, | |
| "loss": 0.1689, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.427368421052632e-06, | |
| "loss": 0.1229, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 2.4010526315789474e-06, | |
| "loss": 0.1734, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.3747368421052632e-06, | |
| "loss": 0.1519, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.348421052631579e-06, | |
| "loss": 0.1707, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.322105263157895e-06, | |
| "loss": 0.2123, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 2.2957894736842107e-06, | |
| "loss": 0.1287, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2694736842105265e-06, | |
| "loss": 0.1522, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.2431578947368423e-06, | |
| "loss": 0.1744, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.216842105263158e-06, | |
| "loss": 0.1423, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 2.190526315789474e-06, | |
| "loss": 0.1823, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.16421052631579e-06, | |
| "loss": 0.1429, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.1378947368421056e-06, | |
| "loss": 0.1877, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.111578947368421e-06, | |
| "loss": 0.1415, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.23196841776371002, | |
| "eval_runtime": 16106.3232, | |
| "eval_samples_per_second": 0.903, | |
| "eval_steps_per_second": 0.903, | |
| "eval_wer": 21.881505654644688, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 2.085263157894737e-06, | |
| "loss": 0.1879, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.06e-06, | |
| "loss": 0.1574, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.033684210526316e-06, | |
| "loss": 0.1164, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 2.007368421052632e-06, | |
| "loss": 0.1221, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.9810526315789476e-06, | |
| "loss": 0.1813, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.9547368421052635e-06, | |
| "loss": 0.1535, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.9284210526315793e-06, | |
| "loss": 0.1235, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.9021052631578949e-06, | |
| "loss": 0.1424, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8757894736842105e-06, | |
| "loss": 0.1686, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.8494736842105265e-06, | |
| "loss": 0.1372, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.8231578947368423e-06, | |
| "loss": 0.1773, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.796842105263158e-06, | |
| "loss": 0.1633, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.770526315789474e-06, | |
| "loss": 0.1447, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7442105263157896e-06, | |
| "loss": 0.1649, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.7178947368421054e-06, | |
| "loss": 0.1108, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6915789473684212e-06, | |
| "loss": 0.1117, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.665263157894737e-06, | |
| "loss": 0.1402, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.6389473684210526e-06, | |
| "loss": 0.1874, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.6126315789473685e-06, | |
| "loss": 0.1468, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.5863157894736845e-06, | |
| "loss": 0.1296, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.56e-06, | |
| "loss": 0.1429, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.533684210526316e-06, | |
| "loss": 0.1477, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.5073684210526317e-06, | |
| "loss": 0.1457, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4810526315789476e-06, | |
| "loss": 0.0825, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4547368421052632e-06, | |
| "loss": 0.1377, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.428421052631579e-06, | |
| "loss": 0.1466, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.402105263157895e-06, | |
| "loss": 0.1307, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.3757894736842106e-06, | |
| "loss": 0.1306, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.3494736842105262e-06, | |
| "loss": 0.1019, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.3231578947368423e-06, | |
| "loss": 0.1158, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.296842105263158e-06, | |
| "loss": 0.126, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2705263157894737e-06, | |
| "loss": 0.1015, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.2442105263157895e-06, | |
| "loss": 0.1506, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.2178947368421053e-06, | |
| "loss": 0.1688, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1915789473684211e-06, | |
| "loss": 0.1712, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.165263157894737e-06, | |
| "loss": 0.1219, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.1389473684210526e-06, | |
| "loss": 0.1043, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.1126315789473686e-06, | |
| "loss": 0.1148, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0863157894736844e-06, | |
| "loss": 0.1352, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.06e-06, | |
| "loss": 0.1374, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "eval_loss": 0.2300652116537094, | |
| "eval_runtime": 16107.3691, | |
| "eval_samples_per_second": 0.903, | |
| "eval_steps_per_second": 0.903, | |
| "eval_wer": 20.530017442187585, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.0336842105263158e-06, | |
| "loss": 0.1394, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.0073684210526317e-06, | |
| "loss": 0.2105, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.810526315789475e-07, | |
| "loss": 0.1421, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.547368421052633e-07, | |
| "loss": 0.1785, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.28421052631579e-07, | |
| "loss": 0.1647, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 9.021052631578948e-07, | |
| "loss": 0.1976, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.757894736842105e-07, | |
| "loss": 0.1684, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.494736842105264e-07, | |
| "loss": 0.1476, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 8.231578947368423e-07, | |
| "loss": 0.1707, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.96842105263158e-07, | |
| "loss": 0.1937, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.705263157894738e-07, | |
| "loss": 0.2084, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.442105263157895e-07, | |
| "loss": 0.1405, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 7.178947368421053e-07, | |
| "loss": 0.1661, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.91578947368421e-07, | |
| "loss": 0.1448, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.652631578947369e-07, | |
| "loss": 0.2067, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.389473684210527e-07, | |
| "loss": 0.1522, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 6.126315789473684e-07, | |
| "loss": 0.1458, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 5.863157894736843e-07, | |
| "loss": 0.1628, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.6e-07, | |
| "loss": 0.1406, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.336842105263159e-07, | |
| "loss": 0.1617, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.073684210526316e-07, | |
| "loss": 0.1693, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.810526315789474e-07, | |
| "loss": 0.1711, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.547368421052632e-07, | |
| "loss": 0.1394, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.28421052631579e-07, | |
| "loss": 0.1424, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.021052631578948e-07, | |
| "loss": 0.1303, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.7578947368421056e-07, | |
| "loss": 0.1634, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.494736842105263e-07, | |
| "loss": 0.1465, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.231578947368421e-07, | |
| "loss": 0.1736, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.968421052631579e-07, | |
| "loss": 0.152, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.705263157894737e-07, | |
| "loss": 0.1749, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.442105263157895e-07, | |
| "loss": 0.1609, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.1789473684210528e-07, | |
| "loss": 0.1698, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.9157894736842107e-07, | |
| "loss": 0.1646, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.6526315789473687e-07, | |
| "loss": 0.1317, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.3894736842105266e-07, | |
| "loss": 0.1417, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.1263157894736842e-07, | |
| "loss": 0.1737, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.631578947368422e-08, | |
| "loss": 0.1027, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 6.000000000000001e-08, | |
| "loss": 0.1327, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.3684210526315793e-08, | |
| "loss": 0.1623, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 7.3684210526315796e-09, | |
| "loss": 0.1794, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.2195526361465454, | |
| "eval_runtime": 16059.0756, | |
| "eval_samples_per_second": 0.906, | |
| "eval_steps_per_second": 0.906, | |
| "eval_wer": 19.56338265908963, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 10000, | |
| "total_flos": 2.04120981504e+19, | |
| "train_loss": 0.25352880897521973, | |
| "train_runtime": 168931.2225, | |
| "train_samples_per_second": 0.118, | |
| "train_steps_per_second": 0.059 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 2.04120981504e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |