| { | |
| "best_metric": 6.433092380590046, | |
| "best_model_checkpoint": "./checkpoint-9000", | |
| "epoch": 58.13953488372093, | |
| "eval_steps": 1000, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.2000000000000006e-07, | |
| "loss": 8.0695, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.200000000000001e-07, | |
| "loss": 5.4689, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.42e-06, | |
| "loss": 2.9498, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.9200000000000003e-06, | |
| "loss": 1.7344, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.42e-06, | |
| "loss": 1.1958, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.92e-06, | |
| "loss": 0.8685, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.4200000000000007e-06, | |
| "loss": 0.7085, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.920000000000001e-06, | |
| "loss": 0.5764, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 4.42e-06, | |
| "loss": 0.5645, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.92e-06, | |
| "loss": 0.4629, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 5.420000000000001e-06, | |
| "loss": 0.4344, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 5.92e-06, | |
| "loss": 0.4064, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 6.42e-06, | |
| "loss": 0.371, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.92e-06, | |
| "loss": 0.3631, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 7.420000000000001e-06, | |
| "loss": 0.2891, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 7.92e-06, | |
| "loss": 0.2875, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.42e-06, | |
| "loss": 0.2741, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 8.920000000000001e-06, | |
| "loss": 0.258, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 9.42e-06, | |
| "loss": 0.2627, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 9.920000000000002e-06, | |
| "loss": 0.2376, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 9.977894736842106e-06, | |
| "loss": 0.2263, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 9.951578947368423e-06, | |
| "loss": 0.1841, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 9.925263157894738e-06, | |
| "loss": 0.1725, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 9.898947368421054e-06, | |
| "loss": 0.1716, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 9.87263157894737e-06, | |
| "loss": 0.1799, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 9.846315789473684e-06, | |
| "loss": 0.1617, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 9.820000000000001e-06, | |
| "loss": 0.1695, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 9.793684210526316e-06, | |
| "loss": 0.1509, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 9.767368421052632e-06, | |
| "loss": 0.1202, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 9.741052631578947e-06, | |
| "loss": 0.1149, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 9.714736842105264e-06, | |
| "loss": 0.1147, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 9.68842105263158e-06, | |
| "loss": 0.1155, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 9.662105263157896e-06, | |
| "loss": 0.1103, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 9.635789473684212e-06, | |
| "loss": 0.115, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 9.609473684210527e-06, | |
| "loss": 0.0933, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 9.583157894736842e-06, | |
| "loss": 0.0763, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 9.556842105263159e-06, | |
| "loss": 0.0779, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 5.52, | |
| "learning_rate": 9.530526315789474e-06, | |
| "loss": 0.0842, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 9.50421052631579e-06, | |
| "loss": 0.0808, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 9.477894736842106e-06, | |
| "loss": 0.0825, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "eval_loss": 0.18628132343292236, | |
| "eval_runtime": 666.9578, | |
| "eval_samples_per_second": 9.591, | |
| "eval_steps_per_second": 0.3, | |
| "eval_wer": 7.014145885933663, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 9.451578947368422e-06, | |
| "loss": 0.0815, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 9.425263157894737e-06, | |
| "loss": 0.0645, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 9.398947368421052e-06, | |
| "loss": 0.0576, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 9.372631578947369e-06, | |
| "loss": 0.0537, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 9.346315789473684e-06, | |
| "loss": 0.0604, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 6.69, | |
| "learning_rate": 9.32e-06, | |
| "loss": 0.0568, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 6.83, | |
| "learning_rate": 9.293684210526317e-06, | |
| "loss": 0.0599, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "learning_rate": 9.267368421052632e-06, | |
| "loss": 0.0601, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 9.241052631578949e-06, | |
| "loss": 0.0431, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 9.214736842105264e-06, | |
| "loss": 0.0399, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 7.41, | |
| "learning_rate": 9.18842105263158e-06, | |
| "loss": 0.0442, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "learning_rate": 9.162105263157895e-06, | |
| "loss": 0.0412, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 9.13578947368421e-06, | |
| "loss": 0.0447, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 9.109473684210527e-06, | |
| "loss": 0.0415, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 9.083157894736842e-06, | |
| "loss": 0.0417, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 8.14, | |
| "learning_rate": 9.056842105263159e-06, | |
| "loss": 0.0293, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 8.28, | |
| "learning_rate": 9.030526315789474e-06, | |
| "loss": 0.0302, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 9.00421052631579e-06, | |
| "loss": 0.0343, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 8.58, | |
| "learning_rate": 8.977894736842107e-06, | |
| "loss": 0.0324, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 8.951578947368422e-06, | |
| "loss": 0.0314, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 8.925263157894739e-06, | |
| "loss": 0.0303, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 8.898947368421054e-06, | |
| "loss": 0.0303, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "learning_rate": 8.872631578947369e-06, | |
| "loss": 0.0223, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "learning_rate": 8.846315789473685e-06, | |
| "loss": 0.0238, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 8.82e-06, | |
| "loss": 0.0236, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 8.793684210526317e-06, | |
| "loss": 0.023, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 9.74, | |
| "learning_rate": 8.767368421052632e-06, | |
| "loss": 0.0252, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 9.88, | |
| "learning_rate": 8.741052631578949e-06, | |
| "loss": 0.0251, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 10.03, | |
| "learning_rate": 8.714736842105264e-06, | |
| "loss": 0.022, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 10.17, | |
| "learning_rate": 8.688421052631579e-06, | |
| "loss": 0.0213, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 10.32, | |
| "learning_rate": 8.662105263157895e-06, | |
| "loss": 0.0181, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 10.47, | |
| "learning_rate": 8.63578947368421e-06, | |
| "loss": 0.0191, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 10.61, | |
| "learning_rate": 8.609473684210527e-06, | |
| "loss": 0.0202, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 10.76, | |
| "learning_rate": 8.583157894736843e-06, | |
| "loss": 0.0199, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 10.9, | |
| "learning_rate": 8.556842105263158e-06, | |
| "loss": 0.0182, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 8.530526315789475e-06, | |
| "loss": 0.0188, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 11.19, | |
| "learning_rate": 8.50421052631579e-06, | |
| "loss": 0.0145, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 11.34, | |
| "learning_rate": 8.477894736842107e-06, | |
| "loss": 0.0156, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 11.48, | |
| "learning_rate": 8.451578947368422e-06, | |
| "loss": 0.014, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 11.63, | |
| "learning_rate": 8.425263157894737e-06, | |
| "loss": 0.0138, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 11.63, | |
| "eval_loss": 0.21974040567874908, | |
| "eval_runtime": 667.6098, | |
| "eval_samples_per_second": 9.582, | |
| "eval_steps_per_second": 0.3, | |
| "eval_wer": 6.687303289177879, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 11.77, | |
| "learning_rate": 8.398947368421053e-06, | |
| "loss": 0.015, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 11.92, | |
| "learning_rate": 8.372631578947368e-06, | |
| "loss": 0.0142, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 12.06, | |
| "learning_rate": 8.346315789473685e-06, | |
| "loss": 0.0127, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 12.21, | |
| "learning_rate": 8.32e-06, | |
| "loss": 0.0112, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 12.35, | |
| "learning_rate": 8.293684210526317e-06, | |
| "loss": 0.0114, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 8.267368421052632e-06, | |
| "loss": 0.0127, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 12.65, | |
| "learning_rate": 8.241052631578948e-06, | |
| "loss": 0.0121, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 12.79, | |
| "learning_rate": 8.214736842105265e-06, | |
| "loss": 0.0124, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 12.94, | |
| "learning_rate": 8.18842105263158e-06, | |
| "loss": 0.0113, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 13.08, | |
| "learning_rate": 8.162105263157895e-06, | |
| "loss": 0.0102, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 13.23, | |
| "learning_rate": 8.135789473684212e-06, | |
| "loss": 0.0105, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 13.37, | |
| "learning_rate": 8.109473684210527e-06, | |
| "loss": 0.0092, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 13.52, | |
| "learning_rate": 8.083157894736843e-06, | |
| "loss": 0.0097, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 13.66, | |
| "learning_rate": 8.056842105263158e-06, | |
| "loss": 0.0099, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 13.81, | |
| "learning_rate": 8.030526315789475e-06, | |
| "loss": 0.0105, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 13.95, | |
| "learning_rate": 8.00421052631579e-06, | |
| "loss": 0.0103, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "learning_rate": 7.977894736842105e-06, | |
| "loss": 0.0101, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 14.24, | |
| "learning_rate": 7.951578947368421e-06, | |
| "loss": 0.0084, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 14.39, | |
| "learning_rate": 7.925263157894736e-06, | |
| "loss": 0.0093, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 14.53, | |
| "learning_rate": 7.898947368421053e-06, | |
| "loss": 0.0085, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 14.68, | |
| "learning_rate": 7.87263157894737e-06, | |
| "loss": 0.0086, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 14.83, | |
| "learning_rate": 7.846315789473685e-06, | |
| "loss": 0.0092, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 14.97, | |
| "learning_rate": 7.820000000000001e-06, | |
| "loss": 0.0089, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 15.12, | |
| "learning_rate": 7.793684210526316e-06, | |
| "loss": 0.0073, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 15.26, | |
| "learning_rate": 7.767368421052633e-06, | |
| "loss": 0.007, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 15.41, | |
| "learning_rate": 7.741052631578948e-06, | |
| "loss": 0.0071, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 15.55, | |
| "learning_rate": 7.714736842105263e-06, | |
| "loss": 0.008, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 15.7, | |
| "learning_rate": 7.68842105263158e-06, | |
| "loss": 0.0074, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 15.84, | |
| "learning_rate": 7.662105263157895e-06, | |
| "loss": 0.008, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 15.99, | |
| "learning_rate": 7.635789473684211e-06, | |
| "loss": 0.0075, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 16.13, | |
| "learning_rate": 7.609473684210526e-06, | |
| "loss": 0.007, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 16.28, | |
| "learning_rate": 7.583157894736842e-06, | |
| "loss": 0.0051, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 16.42, | |
| "learning_rate": 7.556842105263158e-06, | |
| "loss": 0.0064, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 16.57, | |
| "learning_rate": 7.5305263157894745e-06, | |
| "loss": 0.0067, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 16.72, | |
| "learning_rate": 7.50421052631579e-06, | |
| "loss": 0.0061, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 16.86, | |
| "learning_rate": 7.477894736842106e-06, | |
| "loss": 0.0065, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 17.01, | |
| "learning_rate": 7.451578947368422e-06, | |
| "loss": 0.0055, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 17.15, | |
| "learning_rate": 7.425263157894738e-06, | |
| "loss": 0.0054, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 17.3, | |
| "learning_rate": 7.398947368421054e-06, | |
| "loss": 0.0044, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 17.44, | |
| "learning_rate": 7.3726315789473694e-06, | |
| "loss": 0.0059, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 17.44, | |
| "eval_loss": 0.24355579912662506, | |
| "eval_runtime": 669.778, | |
| "eval_samples_per_second": 9.551, | |
| "eval_steps_per_second": 0.299, | |
| "eval_wer": 6.592190364196036, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 17.59, | |
| "learning_rate": 7.346315789473684e-06, | |
| "loss": 0.0057, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 17.73, | |
| "learning_rate": 7.32e-06, | |
| "loss": 0.0059, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 17.88, | |
| "learning_rate": 7.293684210526316e-06, | |
| "loss": 0.0049, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 18.02, | |
| "learning_rate": 7.267368421052632e-06, | |
| "loss": 0.006, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 18.17, | |
| "learning_rate": 7.241052631578948e-06, | |
| "loss": 0.0052, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 18.31, | |
| "learning_rate": 7.2147368421052635e-06, | |
| "loss": 0.0058, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 18.46, | |
| "learning_rate": 7.18842105263158e-06, | |
| "loss": 0.005, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 18.6, | |
| "learning_rate": 7.162105263157896e-06, | |
| "loss": 0.0057, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 7.135789473684212e-06, | |
| "loss": 0.0053, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 18.9, | |
| "learning_rate": 7.109473684210528e-06, | |
| "loss": 0.005, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 7.083157894736843e-06, | |
| "loss": 0.0049, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 19.19, | |
| "learning_rate": 7.056842105263158e-06, | |
| "loss": 0.0058, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 19.33, | |
| "learning_rate": 7.030526315789474e-06, | |
| "loss": 0.005, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 19.48, | |
| "learning_rate": 7.00421052631579e-06, | |
| "loss": 0.0062, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 19.62, | |
| "learning_rate": 6.977894736842106e-06, | |
| "loss": 0.0057, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 19.77, | |
| "learning_rate": 6.951578947368422e-06, | |
| "loss": 0.0045, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 19.91, | |
| "learning_rate": 6.9252631578947375e-06, | |
| "loss": 0.0047, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 20.06, | |
| "learning_rate": 6.8989473684210525e-06, | |
| "loss": 0.0044, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 20.2, | |
| "learning_rate": 6.872631578947368e-06, | |
| "loss": 0.0048, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 20.35, | |
| "learning_rate": 6.846315789473684e-06, | |
| "loss": 0.004, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 20.49, | |
| "learning_rate": 6.820000000000001e-06, | |
| "loss": 0.0049, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 20.64, | |
| "learning_rate": 6.793684210526317e-06, | |
| "loss": 0.0049, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 20.78, | |
| "learning_rate": 6.767368421052632e-06, | |
| "loss": 0.005, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 20.93, | |
| "learning_rate": 6.741052631578948e-06, | |
| "loss": 0.0037, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 21.08, | |
| "learning_rate": 6.714736842105264e-06, | |
| "loss": 0.0044, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 21.22, | |
| "learning_rate": 6.68842105263158e-06, | |
| "loss": 0.0037, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 21.37, | |
| "learning_rate": 6.662105263157896e-06, | |
| "loss": 0.0043, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 21.51, | |
| "learning_rate": 6.635789473684211e-06, | |
| "loss": 0.0035, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 21.66, | |
| "learning_rate": 6.6094736842105265e-06, | |
| "loss": 0.0041, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 21.8, | |
| "learning_rate": 6.583157894736842e-06, | |
| "loss": 0.0039, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 21.95, | |
| "learning_rate": 6.556842105263158e-06, | |
| "loss": 0.0038, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 22.09, | |
| "learning_rate": 6.530526315789474e-06, | |
| "loss": 0.0036, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 22.24, | |
| "learning_rate": 6.50421052631579e-06, | |
| "loss": 0.0036, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 22.38, | |
| "learning_rate": 6.477894736842106e-06, | |
| "loss": 0.0027, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 22.53, | |
| "learning_rate": 6.451578947368422e-06, | |
| "loss": 0.0037, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 22.67, | |
| "learning_rate": 6.425263157894738e-06, | |
| "loss": 0.0032, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 22.82, | |
| "learning_rate": 6.398947368421054e-06, | |
| "loss": 0.0044, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 22.97, | |
| "learning_rate": 6.372631578947369e-06, | |
| "loss": 0.0035, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 23.11, | |
| "learning_rate": 6.346315789473685e-06, | |
| "loss": 0.0034, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 23.26, | |
| "learning_rate": 6.3200000000000005e-06, | |
| "loss": 0.0033, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 23.26, | |
| "eval_loss": 0.2535954415798187, | |
| "eval_runtime": 662.8905, | |
| "eval_samples_per_second": 9.65, | |
| "eval_steps_per_second": 0.302, | |
| "eval_wer": 6.503994742849237, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 23.4, | |
| "learning_rate": 6.293684210526316e-06, | |
| "loss": 0.0031, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 23.55, | |
| "learning_rate": 6.267368421052632e-06, | |
| "loss": 0.0031, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 23.69, | |
| "learning_rate": 6.241052631578948e-06, | |
| "loss": 0.0026, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 23.84, | |
| "learning_rate": 6.214736842105264e-06, | |
| "loss": 0.0032, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 23.98, | |
| "learning_rate": 6.188421052631579e-06, | |
| "loss": 0.0035, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 24.13, | |
| "learning_rate": 6.1621052631578945e-06, | |
| "loss": 0.003, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 24.27, | |
| "learning_rate": 6.13578947368421e-06, | |
| "loss": 0.0027, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 24.42, | |
| "learning_rate": 6.109473684210527e-06, | |
| "loss": 0.003, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 24.56, | |
| "learning_rate": 6.083157894736843e-06, | |
| "loss": 0.0031, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 24.71, | |
| "learning_rate": 6.056842105263159e-06, | |
| "loss": 0.0028, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 24.85, | |
| "learning_rate": 6.0305263157894745e-06, | |
| "loss": 0.0025, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 6.00421052631579e-06, | |
| "loss": 0.002, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 25.15, | |
| "learning_rate": 5.977894736842106e-06, | |
| "loss": 0.003, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 25.29, | |
| "learning_rate": 5.951578947368422e-06, | |
| "loss": 0.0038, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 25.44, | |
| "learning_rate": 5.925263157894737e-06, | |
| "loss": 0.0025, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 25.58, | |
| "learning_rate": 5.898947368421053e-06, | |
| "loss": 0.0027, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 25.73, | |
| "learning_rate": 5.8726315789473685e-06, | |
| "loss": 0.0025, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 25.87, | |
| "learning_rate": 5.846315789473684e-06, | |
| "loss": 0.0027, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 5.82e-06, | |
| "loss": 0.0037, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 26.16, | |
| "learning_rate": 5.793684210526316e-06, | |
| "loss": 0.0028, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 26.31, | |
| "learning_rate": 5.767368421052632e-06, | |
| "loss": 0.0028, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 26.45, | |
| "learning_rate": 5.7410526315789485e-06, | |
| "loss": 0.0027, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 26.6, | |
| "learning_rate": 5.714736842105264e-06, | |
| "loss": 0.0023, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 26.74, | |
| "learning_rate": 5.68842105263158e-06, | |
| "loss": 0.0026, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 26.89, | |
| "learning_rate": 5.662105263157895e-06, | |
| "loss": 0.0035, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 27.03, | |
| "learning_rate": 5.635789473684211e-06, | |
| "loss": 0.0033, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 27.18, | |
| "learning_rate": 5.609473684210527e-06, | |
| "loss": 0.0034, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 27.33, | |
| "learning_rate": 5.5831578947368425e-06, | |
| "loss": 0.0029, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 27.47, | |
| "learning_rate": 5.556842105263158e-06, | |
| "loss": 0.0032, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 27.62, | |
| "learning_rate": 5.530526315789474e-06, | |
| "loss": 0.0035, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 27.76, | |
| "learning_rate": 5.50421052631579e-06, | |
| "loss": 0.0027, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 27.91, | |
| "learning_rate": 5.477894736842105e-06, | |
| "loss": 0.0026, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 28.05, | |
| "learning_rate": 5.451578947368421e-06, | |
| "loss": 0.0023, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 28.2, | |
| "learning_rate": 5.425263157894737e-06, | |
| "loss": 0.0022, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 28.34, | |
| "learning_rate": 5.398947368421053e-06, | |
| "loss": 0.0024, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 28.49, | |
| "learning_rate": 5.372631578947369e-06, | |
| "loss": 0.0024, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 28.63, | |
| "learning_rate": 5.346315789473685e-06, | |
| "loss": 0.0023, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 28.78, | |
| "learning_rate": 5.320000000000001e-06, | |
| "loss": 0.003, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 28.92, | |
| "learning_rate": 5.2936842105263166e-06, | |
| "loss": 0.0029, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 29.07, | |
| "learning_rate": 5.267368421052632e-06, | |
| "loss": 0.002, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 29.07, | |
| "eval_loss": 0.26064327359199524, | |
| "eval_runtime": 660.6329, | |
| "eval_samples_per_second": 9.683, | |
| "eval_steps_per_second": 0.303, | |
| "eval_wer": 6.6924912669041605, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 29.22, | |
| "learning_rate": 5.241052631578948e-06, | |
| "loss": 0.0026, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 29.36, | |
| "learning_rate": 5.214736842105263e-06, | |
| "loss": 0.0021, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 29.51, | |
| "learning_rate": 5.188421052631579e-06, | |
| "loss": 0.0021, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 29.65, | |
| "learning_rate": 5.162105263157895e-06, | |
| "loss": 0.0021, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 29.8, | |
| "learning_rate": 5.135789473684211e-06, | |
| "loss": 0.0019, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 29.94, | |
| "learning_rate": 5.1094736842105264e-06, | |
| "loss": 0.0023, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 30.09, | |
| "learning_rate": 5.083157894736842e-06, | |
| "loss": 0.002, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 30.23, | |
| "learning_rate": 5.056842105263158e-06, | |
| "loss": 0.0021, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 30.38, | |
| "learning_rate": 5.030526315789475e-06, | |
| "loss": 0.0019, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 30.52, | |
| "learning_rate": 5.0042105263157906e-06, | |
| "loss": 0.0022, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 30.67, | |
| "learning_rate": 4.9778947368421055e-06, | |
| "loss": 0.0018, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 30.81, | |
| "learning_rate": 4.951578947368421e-06, | |
| "loss": 0.0014, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 30.96, | |
| "learning_rate": 4.925263157894737e-06, | |
| "loss": 0.0016, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 31.1, | |
| "learning_rate": 4.898947368421053e-06, | |
| "loss": 0.0014, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 31.25, | |
| "learning_rate": 4.872631578947369e-06, | |
| "loss": 0.0016, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 31.4, | |
| "learning_rate": 4.846315789473685e-06, | |
| "loss": 0.0017, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 31.54, | |
| "learning_rate": 4.8200000000000004e-06, | |
| "loss": 0.002, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 31.69, | |
| "learning_rate": 4.793684210526316e-06, | |
| "loss": 0.0018, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 31.83, | |
| "learning_rate": 4.767368421052632e-06, | |
| "loss": 0.0023, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 31.98, | |
| "learning_rate": 4.741052631578948e-06, | |
| "loss": 0.0018, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 32.12, | |
| "learning_rate": 4.714736842105264e-06, | |
| "loss": 0.0019, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 32.27, | |
| "learning_rate": 4.6884210526315795e-06, | |
| "loss": 0.002, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 32.41, | |
| "learning_rate": 4.662105263157895e-06, | |
| "loss": 0.0014, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 32.56, | |
| "learning_rate": 4.63578947368421e-06, | |
| "loss": 0.0021, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 32.7, | |
| "learning_rate": 4.609473684210526e-06, | |
| "loss": 0.0018, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 32.85, | |
| "learning_rate": 4.583157894736843e-06, | |
| "loss": 0.0016, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 32.99, | |
| "learning_rate": 4.556842105263159e-06, | |
| "loss": 0.0015, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 33.14, | |
| "learning_rate": 4.5305263157894744e-06, | |
| "loss": 0.0015, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 33.28, | |
| "learning_rate": 4.504210526315789e-06, | |
| "loss": 0.0016, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 33.43, | |
| "learning_rate": 4.477894736842105e-06, | |
| "loss": 0.0015, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 33.58, | |
| "learning_rate": 4.451578947368421e-06, | |
| "loss": 0.0018, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 33.72, | |
| "learning_rate": 4.425263157894737e-06, | |
| "loss": 0.0015, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 33.87, | |
| "learning_rate": 4.3989473684210535e-06, | |
| "loss": 0.0019, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 34.01, | |
| "learning_rate": 4.3726315789473685e-06, | |
| "loss": 0.0017, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 34.16, | |
| "learning_rate": 4.346315789473684e-06, | |
| "loss": 0.0015, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 34.3, | |
| "learning_rate": 4.32e-06, | |
| "loss": 0.0027, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 34.45, | |
| "learning_rate": 4.293684210526316e-06, | |
| "loss": 0.0025, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 34.59, | |
| "learning_rate": 4.267368421052632e-06, | |
| "loss": 0.0016, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 34.74, | |
| "learning_rate": 4.241052631578948e-06, | |
| "loss": 0.0017, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 34.88, | |
| "learning_rate": 4.214736842105263e-06, | |
| "loss": 0.002, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 34.88, | |
| "eval_loss": 0.26901113986968994, | |
| "eval_runtime": 662.8358, | |
| "eval_samples_per_second": 9.651, | |
| "eval_steps_per_second": 0.302, | |
| "eval_wer": 6.689032615086639, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 35.03, | |
| "learning_rate": 4.188421052631579e-06, | |
| "loss": 0.0014, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 35.17, | |
| "learning_rate": 4.162105263157895e-06, | |
| "loss": 0.0023, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 35.32, | |
| "learning_rate": 4.135789473684211e-06, | |
| "loss": 0.0012, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 35.47, | |
| "learning_rate": 4.109473684210527e-06, | |
| "loss": 0.0023, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 35.61, | |
| "learning_rate": 4.0831578947368425e-06, | |
| "loss": 0.0021, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 35.76, | |
| "learning_rate": 4.056842105263158e-06, | |
| "loss": 0.0014, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 35.9, | |
| "learning_rate": 4.030526315789474e-06, | |
| "loss": 0.0022, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 36.05, | |
| "learning_rate": 4.00421052631579e-06, | |
| "loss": 0.0014, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 36.19, | |
| "learning_rate": 3.977894736842106e-06, | |
| "loss": 0.0012, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 36.34, | |
| "learning_rate": 3.951578947368422e-06, | |
| "loss": 0.0013, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 36.48, | |
| "learning_rate": 3.9252631578947366e-06, | |
| "loss": 0.0012, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 36.63, | |
| "learning_rate": 3.898947368421052e-06, | |
| "loss": 0.0012, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 36.77, | |
| "learning_rate": 3.872631578947369e-06, | |
| "loss": 0.0011, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 36.92, | |
| "learning_rate": 3.846315789473685e-06, | |
| "loss": 0.0015, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 37.06, | |
| "learning_rate": 3.820000000000001e-06, | |
| "loss": 0.0015, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 37.21, | |
| "learning_rate": 3.793684210526316e-06, | |
| "loss": 0.0012, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 37.35, | |
| "learning_rate": 3.767368421052632e-06, | |
| "loss": 0.0012, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 37.5, | |
| "learning_rate": 3.7410526315789473e-06, | |
| "loss": 0.0018, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 37.65, | |
| "learning_rate": 3.714736842105263e-06, | |
| "loss": 0.0012, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 37.79, | |
| "learning_rate": 3.6884210526315794e-06, | |
| "loss": 0.0016, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 37.94, | |
| "learning_rate": 3.662105263157895e-06, | |
| "loss": 0.0011, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 38.08, | |
| "learning_rate": 3.635789473684211e-06, | |
| "loss": 0.0017, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 38.23, | |
| "learning_rate": 3.6094736842105264e-06, | |
| "loss": 0.0015, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 38.37, | |
| "learning_rate": 3.5831578947368422e-06, | |
| "loss": 0.0017, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 38.52, | |
| "learning_rate": 3.556842105263158e-06, | |
| "loss": 0.0015, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 38.66, | |
| "learning_rate": 3.5305263157894743e-06, | |
| "loss": 0.0012, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 38.81, | |
| "learning_rate": 3.50421052631579e-06, | |
| "loss": 0.0015, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 38.95, | |
| "learning_rate": 3.4778947368421055e-06, | |
| "loss": 0.001, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 39.1, | |
| "learning_rate": 3.4515789473684213e-06, | |
| "loss": 0.0012, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 39.24, | |
| "learning_rate": 3.425263157894737e-06, | |
| "loss": 0.001, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 39.39, | |
| "learning_rate": 3.398947368421053e-06, | |
| "loss": 0.0014, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 39.53, | |
| "learning_rate": 3.3726315789473683e-06, | |
| "loss": 0.0011, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 39.68, | |
| "learning_rate": 3.3463157894736846e-06, | |
| "loss": 0.0008, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 39.83, | |
| "learning_rate": 3.3200000000000004e-06, | |
| "loss": 0.001, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 39.97, | |
| "learning_rate": 3.2936842105263162e-06, | |
| "loss": 0.0008, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 40.12, | |
| "learning_rate": 3.267368421052632e-06, | |
| "loss": 0.0012, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 40.26, | |
| "learning_rate": 3.2410526315789474e-06, | |
| "loss": 0.0011, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 40.41, | |
| "learning_rate": 3.2147368421052633e-06, | |
| "loss": 0.001, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 40.55, | |
| "learning_rate": 3.188421052631579e-06, | |
| "loss": 0.0015, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 40.7, | |
| "learning_rate": 3.1621052631578953e-06, | |
| "loss": 0.001, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 40.7, | |
| "eval_loss": 0.2734251320362091, | |
| "eval_runtime": 662.975, | |
| "eval_samples_per_second": 9.649, | |
| "eval_steps_per_second": 0.302, | |
| "eval_wer": 6.5852730605609935, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 40.84, | |
| "learning_rate": 3.135789473684211e-06, | |
| "loss": 0.0014, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 40.99, | |
| "learning_rate": 3.1094736842105265e-06, | |
| "loss": 0.001, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 41.13, | |
| "learning_rate": 3.0831578947368423e-06, | |
| "loss": 0.0011, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 41.28, | |
| "learning_rate": 3.056842105263158e-06, | |
| "loss": 0.001, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 41.42, | |
| "learning_rate": 3.0305263157894736e-06, | |
| "loss": 0.001, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 41.57, | |
| "learning_rate": 3.0042105263157894e-06, | |
| "loss": 0.0009, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 41.72, | |
| "learning_rate": 2.9778947368421056e-06, | |
| "loss": 0.0009, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 41.86, | |
| "learning_rate": 2.9515789473684214e-06, | |
| "loss": 0.0009, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 42.01, | |
| "learning_rate": 2.9252631578947373e-06, | |
| "loss": 0.0009, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 42.15, | |
| "learning_rate": 2.8989473684210526e-06, | |
| "loss": 0.001, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 42.3, | |
| "learning_rate": 2.8726315789473685e-06, | |
| "loss": 0.0012, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 42.44, | |
| "learning_rate": 2.8463157894736843e-06, | |
| "loss": 0.001, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 42.59, | |
| "learning_rate": 2.82e-06, | |
| "loss": 0.0009, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 42.73, | |
| "learning_rate": 2.7936842105263163e-06, | |
| "loss": 0.0012, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 42.88, | |
| "learning_rate": 2.7673684210526317e-06, | |
| "loss": 0.0009, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 43.02, | |
| "learning_rate": 2.7410526315789476e-06, | |
| "loss": 0.0011, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 43.17, | |
| "learning_rate": 2.7147368421052634e-06, | |
| "loss": 0.0012, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 43.31, | |
| "learning_rate": 2.688421052631579e-06, | |
| "loss": 0.0016, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 43.46, | |
| "learning_rate": 2.6621052631578946e-06, | |
| "loss": 0.0016, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 43.6, | |
| "learning_rate": 2.635789473684211e-06, | |
| "loss": 0.0011, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 43.75, | |
| "learning_rate": 2.6094736842105267e-06, | |
| "loss": 0.001, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 43.9, | |
| "learning_rate": 2.5831578947368425e-06, | |
| "loss": 0.0009, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 44.04, | |
| "learning_rate": 2.5568421052631583e-06, | |
| "loss": 0.0008, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 44.19, | |
| "learning_rate": 2.5305263157894737e-06, | |
| "loss": 0.0006, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 44.33, | |
| "learning_rate": 2.5042105263157895e-06, | |
| "loss": 0.0007, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 44.48, | |
| "learning_rate": 2.4778947368421053e-06, | |
| "loss": 0.0006, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 44.62, | |
| "learning_rate": 2.451578947368421e-06, | |
| "loss": 0.0006, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 44.77, | |
| "learning_rate": 2.425263157894737e-06, | |
| "loss": 0.0012, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 44.91, | |
| "learning_rate": 2.3989473684210528e-06, | |
| "loss": 0.0009, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 45.06, | |
| "learning_rate": 2.3726315789473686e-06, | |
| "loss": 0.0006, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 45.2, | |
| "learning_rate": 2.3463157894736844e-06, | |
| "loss": 0.0007, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 45.35, | |
| "learning_rate": 2.3200000000000002e-06, | |
| "loss": 0.0007, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 45.49, | |
| "learning_rate": 2.293684210526316e-06, | |
| "loss": 0.0009, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 45.64, | |
| "learning_rate": 2.267368421052632e-06, | |
| "loss": 0.0013, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 45.78, | |
| "learning_rate": 2.2410526315789473e-06, | |
| "loss": 0.0008, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 45.93, | |
| "learning_rate": 2.2147368421052635e-06, | |
| "loss": 0.0006, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 46.08, | |
| "learning_rate": 2.188421052631579e-06, | |
| "loss": 0.001, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 46.22, | |
| "learning_rate": 2.1621052631578947e-06, | |
| "loss": 0.0008, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 46.37, | |
| "learning_rate": 2.135789473684211e-06, | |
| "loss": 0.0008, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 46.51, | |
| "learning_rate": 2.1094736842105264e-06, | |
| "loss": 0.0007, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 46.51, | |
| "eval_loss": 0.2820800542831421, | |
| "eval_runtime": 661.693, | |
| "eval_samples_per_second": 9.668, | |
| "eval_steps_per_second": 0.302, | |
| "eval_wer": 6.479784180126587, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 46.66, | |
| "learning_rate": 2.083157894736842e-06, | |
| "loss": 0.0009, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 46.8, | |
| "learning_rate": 2.056842105263158e-06, | |
| "loss": 0.0011, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 46.95, | |
| "learning_rate": 2.030526315789474e-06, | |
| "loss": 0.001, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 47.09, | |
| "learning_rate": 2.0042105263157896e-06, | |
| "loss": 0.0016, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 47.24, | |
| "learning_rate": 1.9778947368421055e-06, | |
| "loss": 0.0021, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 47.38, | |
| "learning_rate": 1.9515789473684213e-06, | |
| "loss": 0.0018, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 47.53, | |
| "learning_rate": 1.925263157894737e-06, | |
| "loss": 0.0009, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 47.67, | |
| "learning_rate": 1.8989473684210527e-06, | |
| "loss": 0.001, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 47.82, | |
| "learning_rate": 1.8726315789473687e-06, | |
| "loss": 0.0007, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 47.97, | |
| "learning_rate": 1.8463157894736843e-06, | |
| "loss": 0.0006, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 48.11, | |
| "learning_rate": 1.8200000000000002e-06, | |
| "loss": 0.0007, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 48.26, | |
| "learning_rate": 1.7936842105263158e-06, | |
| "loss": 0.0006, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 48.4, | |
| "learning_rate": 1.7673684210526318e-06, | |
| "loss": 0.0009, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 48.55, | |
| "learning_rate": 1.7410526315789474e-06, | |
| "loss": 0.0007, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 48.69, | |
| "learning_rate": 1.7147368421052632e-06, | |
| "loss": 0.0006, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 48.84, | |
| "learning_rate": 1.6884210526315792e-06, | |
| "loss": 0.0006, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 48.98, | |
| "learning_rate": 1.6621052631578948e-06, | |
| "loss": 0.0007, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 49.13, | |
| "learning_rate": 1.6357894736842107e-06, | |
| "loss": 0.0007, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 49.27, | |
| "learning_rate": 1.6094736842105265e-06, | |
| "loss": 0.0006, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 49.42, | |
| "learning_rate": 1.5831578947368423e-06, | |
| "loss": 0.001, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 49.56, | |
| "learning_rate": 1.556842105263158e-06, | |
| "loss": 0.0006, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 49.71, | |
| "learning_rate": 1.5305263157894737e-06, | |
| "loss": 0.0007, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 49.85, | |
| "learning_rate": 1.5042105263157898e-06, | |
| "loss": 0.0007, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "learning_rate": 1.4778947368421054e-06, | |
| "loss": 0.0008, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 50.15, | |
| "learning_rate": 1.4515789473684212e-06, | |
| "loss": 0.0006, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 50.29, | |
| "learning_rate": 1.425263157894737e-06, | |
| "loss": 0.0007, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 50.44, | |
| "learning_rate": 1.3989473684210528e-06, | |
| "loss": 0.0007, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 50.58, | |
| "learning_rate": 1.3726315789473684e-06, | |
| "loss": 0.0006, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 50.73, | |
| "learning_rate": 1.3463157894736842e-06, | |
| "loss": 0.0006, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 50.87, | |
| "learning_rate": 1.32e-06, | |
| "loss": 0.0005, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 51.02, | |
| "learning_rate": 1.2936842105263159e-06, | |
| "loss": 0.0006, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 51.16, | |
| "learning_rate": 1.2673684210526315e-06, | |
| "loss": 0.0005, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 51.31, | |
| "learning_rate": 1.2410526315789475e-06, | |
| "loss": 0.0006, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 51.45, | |
| "learning_rate": 1.2147368421052633e-06, | |
| "loss": 0.0006, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 51.6, | |
| "learning_rate": 1.188421052631579e-06, | |
| "loss": 0.001, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 51.74, | |
| "learning_rate": 1.1621052631578948e-06, | |
| "loss": 0.0006, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 51.89, | |
| "learning_rate": 1.1357894736842106e-06, | |
| "loss": 0.0008, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 52.03, | |
| "learning_rate": 1.1094736842105264e-06, | |
| "loss": 0.0005, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 52.18, | |
| "learning_rate": 1.0831578947368422e-06, | |
| "loss": 0.0008, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 52.33, | |
| "learning_rate": 1.0568421052631578e-06, | |
| "loss": 0.0006, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 52.33, | |
| "eval_loss": 0.2829355001449585, | |
| "eval_runtime": 660.3086, | |
| "eval_samples_per_second": 9.688, | |
| "eval_steps_per_second": 0.303, | |
| "eval_wer": 6.433092380590046, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 52.47, | |
| "learning_rate": 1.0305263157894739e-06, | |
| "loss": 0.0006, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 52.62, | |
| "learning_rate": 1.0042105263157897e-06, | |
| "loss": 0.0005, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 52.76, | |
| "learning_rate": 9.778947368421053e-07, | |
| "loss": 0.0012, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 52.91, | |
| "learning_rate": 9.515789473684212e-07, | |
| "loss": 0.0007, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 53.05, | |
| "learning_rate": 9.252631578947368e-07, | |
| "loss": 0.0006, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 53.2, | |
| "learning_rate": 8.989473684210527e-07, | |
| "loss": 0.001, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 53.34, | |
| "learning_rate": 8.726315789473686e-07, | |
| "loss": 0.0006, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 53.49, | |
| "learning_rate": 8.463157894736843e-07, | |
| "loss": 0.0007, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 53.63, | |
| "learning_rate": 8.200000000000001e-07, | |
| "loss": 0.0005, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 53.78, | |
| "learning_rate": 7.936842105263158e-07, | |
| "loss": 0.0008, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 53.92, | |
| "learning_rate": 7.673684210526316e-07, | |
| "loss": 0.0005, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 54.07, | |
| "learning_rate": 7.410526315789475e-07, | |
| "loss": 0.0005, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 54.22, | |
| "learning_rate": 7.147368421052632e-07, | |
| "loss": 0.0007, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 54.36, | |
| "learning_rate": 6.884210526315791e-07, | |
| "loss": 0.001, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 54.51, | |
| "learning_rate": 6.621052631578948e-07, | |
| "loss": 0.0005, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 54.65, | |
| "learning_rate": 6.357894736842106e-07, | |
| "loss": 0.0006, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 54.8, | |
| "learning_rate": 6.094736842105263e-07, | |
| "loss": 0.0006, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 54.94, | |
| "learning_rate": 5.831578947368421e-07, | |
| "loss": 0.0007, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 55.09, | |
| "learning_rate": 5.56842105263158e-07, | |
| "loss": 0.0006, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 55.23, | |
| "learning_rate": 5.305263157894737e-07, | |
| "loss": 0.0005, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 55.38, | |
| "learning_rate": 5.042105263157895e-07, | |
| "loss": 0.0005, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 55.52, | |
| "learning_rate": 4.778947368421053e-07, | |
| "loss": 0.0009, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 55.67, | |
| "learning_rate": 4.5157894736842107e-07, | |
| "loss": 0.0005, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 55.81, | |
| "learning_rate": 4.2526315789473684e-07, | |
| "loss": 0.0006, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 55.96, | |
| "learning_rate": 3.9894736842105266e-07, | |
| "loss": 0.0005, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 56.1, | |
| "learning_rate": 3.726315789473685e-07, | |
| "loss": 0.0005, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 56.25, | |
| "learning_rate": 3.4631578947368424e-07, | |
| "loss": 0.0005, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 56.4, | |
| "learning_rate": 3.2e-07, | |
| "loss": 0.0004, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 56.54, | |
| "learning_rate": 2.936842105263158e-07, | |
| "loss": 0.0006, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 56.69, | |
| "learning_rate": 2.6736842105263164e-07, | |
| "loss": 0.0005, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 56.83, | |
| "learning_rate": 2.410526315789474e-07, | |
| "loss": 0.0005, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 56.98, | |
| "learning_rate": 2.1473684210526317e-07, | |
| "loss": 0.0005, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 57.12, | |
| "learning_rate": 1.8842105263157897e-07, | |
| "loss": 0.0005, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 57.27, | |
| "learning_rate": 1.6210526315789476e-07, | |
| "loss": 0.0005, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 57.41, | |
| "learning_rate": 1.3578947368421055e-07, | |
| "loss": 0.0005, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 57.56, | |
| "learning_rate": 1.0947368421052632e-07, | |
| "loss": 0.0004, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 57.7, | |
| "learning_rate": 8.315789473684211e-08, | |
| "loss": 0.0005, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 57.85, | |
| "learning_rate": 5.68421052631579e-08, | |
| "loss": 0.0007, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 57.99, | |
| "learning_rate": 3.0526315789473686e-08, | |
| "loss": 0.0004, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 58.14, | |
| "learning_rate": 4.210526315789474e-09, | |
| "loss": 0.0007, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 58.14, | |
| "eval_loss": 0.28468087315559387, | |
| "eval_runtime": 662.3319, | |
| "eval_samples_per_second": 9.658, | |
| "eval_steps_per_second": 0.302, | |
| "eval_wer": 6.445197661951371, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 58.14, | |
| "step": 10000, | |
| "total_flos": 6.498130209885389e+20, | |
| "train_loss": 0.07657152335755527, | |
| "train_runtime": 36606.1146, | |
| "train_samples_per_second": 17.483, | |
| "train_steps_per_second": 0.273 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 59, | |
| "save_steps": 1000, | |
| "total_flos": 6.498130209885389e+20, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |