| { | |
| "best_metric": 14.112716355356747, | |
| "best_model_checkpoint": "./checkpoint-10000", | |
| "epoch": 58.47953216374269, | |
| "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": 7.8458, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 5.3933, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 3.1983, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 2.0944, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 1.4986, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.9e-06, | |
| "loss": 1.204, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 1.0132, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.8502, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.7544, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.6692, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.6239, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.5859, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.5357, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.48, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.4385, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.4022, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.3822, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.3729, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.3506, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.3446, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 9.978947368421053e-06, | |
| "loss": 0.29, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 9.95263157894737e-06, | |
| "loss": 0.26, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 9.926315789473685e-06, | |
| "loss": 0.2491, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 9.9e-06, | |
| "loss": 0.25, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 9.873684210526317e-06, | |
| "loss": 0.2447, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 9.847368421052632e-06, | |
| "loss": 0.2293, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 9.821052631578948e-06, | |
| "loss": 0.231, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 9.794736842105263e-06, | |
| "loss": 0.1991, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 9.76842105263158e-06, | |
| "loss": 0.1723, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 9.742105263157897e-06, | |
| "loss": 0.1718, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 9.715789473684212e-06, | |
| "loss": 0.1707, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 9.689473684210527e-06, | |
| "loss": 0.1656, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 9.663157894736843e-06, | |
| "loss": 0.1591, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 9.636842105263158e-06, | |
| "loss": 0.1637, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 9.610526315789475e-06, | |
| "loss": 0.1245, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 9.58421052631579e-06, | |
| "loss": 0.1219, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 9.557894736842107e-06, | |
| "loss": 0.1178, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 9.531578947368422e-06, | |
| "loss": 0.1155, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 9.505263157894738e-06, | |
| "loss": 0.1175, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 9.478947368421053e-06, | |
| "loss": 0.1127, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "eval_loss": 0.2775653302669525, | |
| "eval_runtime": 718.7064, | |
| "eval_samples_per_second": 9.171, | |
| "eval_steps_per_second": 0.287, | |
| "eval_wer": 17.56231235769409, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 9.452631578947368e-06, | |
| "loss": 0.1172, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 9.426315789473685e-06, | |
| "loss": 0.0831, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.0851, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 9.373684210526316e-06, | |
| "loss": 0.0841, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 9.347368421052633e-06, | |
| "loss": 0.0843, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "learning_rate": 9.321052631578948e-06, | |
| "loss": 0.0843, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "learning_rate": 9.294736842105265e-06, | |
| "loss": 0.0855, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 9.26842105263158e-06, | |
| "loss": 0.0836, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 9.242105263157896e-06, | |
| "loss": 0.0594, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 9.215789473684211e-06, | |
| "loss": 0.0564, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 9.189473684210526e-06, | |
| "loss": 0.0638, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 9.163157894736843e-06, | |
| "loss": 0.0625, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 9.136842105263158e-06, | |
| "loss": 0.0649, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 9.110526315789475e-06, | |
| "loss": 0.061, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 9.08421052631579e-06, | |
| "loss": 0.0583, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 8.19, | |
| "learning_rate": 9.057894736842106e-06, | |
| "loss": 0.0454, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 9.031578947368423e-06, | |
| "loss": 0.045, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 8.48, | |
| "learning_rate": 9.005263157894738e-06, | |
| "loss": 0.0425, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 8.978947368421055e-06, | |
| "loss": 0.0481, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 8.95263157894737e-06, | |
| "loss": 0.0475, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 8.926315789473685e-06, | |
| "loss": 0.0475, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 9.06, | |
| "learning_rate": 8.900000000000001e-06, | |
| "loss": 0.0384, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 9.21, | |
| "learning_rate": 8.873684210526316e-06, | |
| "loss": 0.0343, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 9.36, | |
| "learning_rate": 8.847368421052633e-06, | |
| "loss": 0.0351, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 8.821052631578948e-06, | |
| "loss": 0.0352, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 9.65, | |
| "learning_rate": 8.794736842105264e-06, | |
| "loss": 0.0366, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 8.76842105263158e-06, | |
| "loss": 0.0335, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 9.94, | |
| "learning_rate": 8.742105263157894e-06, | |
| "loss": 0.0349, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 10.09, | |
| "learning_rate": 8.715789473684211e-06, | |
| "loss": 0.0292, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 10.23, | |
| "learning_rate": 8.689473684210526e-06, | |
| "loss": 0.0243, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 10.38, | |
| "learning_rate": 8.663157894736843e-06, | |
| "loss": 0.0255, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 10.53, | |
| "learning_rate": 8.63684210526316e-06, | |
| "loss": 0.0249, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 10.67, | |
| "learning_rate": 8.610526315789474e-06, | |
| "loss": 0.0281, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 10.82, | |
| "learning_rate": 8.584210526315791e-06, | |
| "loss": 0.0281, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 10.96, | |
| "learning_rate": 8.557894736842106e-06, | |
| "loss": 0.0281, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 11.11, | |
| "learning_rate": 8.531578947368423e-06, | |
| "loss": 0.0215, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 11.26, | |
| "learning_rate": 8.505263157894738e-06, | |
| "loss": 0.0211, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 8.478947368421053e-06, | |
| "loss": 0.0199, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 8.45263157894737e-06, | |
| "loss": 0.0209, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 8.426315789473684e-06, | |
| "loss": 0.0225, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "eval_loss": 0.31292760372161865, | |
| "eval_runtime": 713.6603, | |
| "eval_samples_per_second": 9.235, | |
| "eval_steps_per_second": 0.289, | |
| "eval_wer": 15.631989361059059, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 11.84, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.0235, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 11.99, | |
| "learning_rate": 8.373684210526316e-06, | |
| "loss": 0.0223, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 12.13, | |
| "learning_rate": 8.347368421052633e-06, | |
| "loss": 0.0174, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 12.28, | |
| "learning_rate": 8.32105263157895e-06, | |
| "loss": 0.0171, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 12.43, | |
| "learning_rate": 8.294736842105264e-06, | |
| "loss": 0.0173, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 12.57, | |
| "learning_rate": 8.26842105263158e-06, | |
| "loss": 0.0175, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 12.72, | |
| "learning_rate": 8.242105263157896e-06, | |
| "loss": 0.0164, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 12.87, | |
| "learning_rate": 8.21578947368421e-06, | |
| "loss": 0.0176, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 13.01, | |
| "learning_rate": 8.189473684210527e-06, | |
| "loss": 0.0165, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 13.16, | |
| "learning_rate": 8.163157894736842e-06, | |
| "loss": 0.0129, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 13.3, | |
| "learning_rate": 8.136842105263159e-06, | |
| "loss": 0.0134, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 13.45, | |
| "learning_rate": 8.110526315789474e-06, | |
| "loss": 0.014, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 13.6, | |
| "learning_rate": 8.08421052631579e-06, | |
| "loss": 0.0132, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 13.74, | |
| "learning_rate": 8.057894736842106e-06, | |
| "loss": 0.014, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 13.89, | |
| "learning_rate": 8.03157894736842e-06, | |
| "loss": 0.0158, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 14.04, | |
| "learning_rate": 8.005263157894737e-06, | |
| "loss": 0.0129, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 14.18, | |
| "learning_rate": 7.978947368421052e-06, | |
| "loss": 0.0122, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 14.33, | |
| "learning_rate": 7.952631578947369e-06, | |
| "loss": 0.0124, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 14.47, | |
| "learning_rate": 7.926315789473686e-06, | |
| "loss": 0.0107, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 14.62, | |
| "learning_rate": 7.9e-06, | |
| "loss": 0.0126, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 14.77, | |
| "learning_rate": 7.873684210526317e-06, | |
| "loss": 0.0112, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 14.91, | |
| "learning_rate": 7.847368421052632e-06, | |
| "loss": 0.0113, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 15.06, | |
| "learning_rate": 7.821052631578949e-06, | |
| "loss": 0.0098, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 15.2, | |
| "learning_rate": 7.794736842105264e-06, | |
| "loss": 0.0088, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 15.35, | |
| "learning_rate": 7.768421052631579e-06, | |
| "loss": 0.0098, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 15.5, | |
| "learning_rate": 7.742105263157896e-06, | |
| "loss": 0.0095, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 15.64, | |
| "learning_rate": 7.71578947368421e-06, | |
| "loss": 0.0098, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 15.79, | |
| "learning_rate": 7.689473684210527e-06, | |
| "loss": 0.0087, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 15.94, | |
| "learning_rate": 7.663157894736842e-06, | |
| "loss": 0.0103, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 16.08, | |
| "learning_rate": 7.636842105263159e-06, | |
| "loss": 0.0086, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 16.23, | |
| "learning_rate": 7.610526315789474e-06, | |
| "loss": 0.0078, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 16.37, | |
| "learning_rate": 7.58421052631579e-06, | |
| "loss": 0.0079, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 16.52, | |
| "learning_rate": 7.557894736842106e-06, | |
| "loss": 0.0088, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 7.531578947368422e-06, | |
| "loss": 0.0074, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 16.81, | |
| "learning_rate": 7.505263157894738e-06, | |
| "loss": 0.0082, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 16.96, | |
| "learning_rate": 7.478947368421054e-06, | |
| "loss": 0.0086, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 17.11, | |
| "learning_rate": 7.4526315789473695e-06, | |
| "loss": 0.0073, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 17.25, | |
| "learning_rate": 7.4263157894736845e-06, | |
| "loss": 0.0077, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 17.4, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.0087, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 17.54, | |
| "learning_rate": 7.373684210526316e-06, | |
| "loss": 0.0074, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 17.54, | |
| "eval_loss": 0.32766830921173096, | |
| "eval_runtime": 714.5938, | |
| "eval_samples_per_second": 9.223, | |
| "eval_steps_per_second": 0.288, | |
| "eval_wer": 14.952950895645692, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 17.69, | |
| "learning_rate": 7.347368421052632e-06, | |
| "loss": 0.0074, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 17.84, | |
| "learning_rate": 7.321052631578948e-06, | |
| "loss": 0.0075, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 17.98, | |
| "learning_rate": 7.2947368421052636e-06, | |
| "loss": 0.0081, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 18.13, | |
| "learning_rate": 7.268421052631579e-06, | |
| "loss": 0.0078, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 18.27, | |
| "learning_rate": 7.242105263157896e-06, | |
| "loss": 0.0065, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 18.42, | |
| "learning_rate": 7.215789473684212e-06, | |
| "loss": 0.0068, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 18.57, | |
| "learning_rate": 7.189473684210527e-06, | |
| "loss": 0.0071, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 18.71, | |
| "learning_rate": 7.163157894736843e-06, | |
| "loss": 0.007, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 18.86, | |
| "learning_rate": 7.1368421052631585e-06, | |
| "loss": 0.0065, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 19.01, | |
| "learning_rate": 7.110526315789474e-06, | |
| "loss": 0.0062, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 19.15, | |
| "learning_rate": 7.08421052631579e-06, | |
| "loss": 0.0055, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 19.3, | |
| "learning_rate": 7.057894736842106e-06, | |
| "loss": 0.0051, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 19.44, | |
| "learning_rate": 7.031578947368422e-06, | |
| "loss": 0.0054, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 19.59, | |
| "learning_rate": 7.005263157894738e-06, | |
| "loss": 0.0058, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 19.74, | |
| "learning_rate": 6.9789473684210525e-06, | |
| "loss": 0.0065, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 19.88, | |
| "learning_rate": 6.952631578947368e-06, | |
| "loss": 0.0059, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 20.03, | |
| "learning_rate": 6.926315789473684e-06, | |
| "loss": 0.0045, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 20.18, | |
| "learning_rate": 6.9e-06, | |
| "loss": 0.0052, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 20.32, | |
| "learning_rate": 6.873684210526317e-06, | |
| "loss": 0.0047, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 20.47, | |
| "learning_rate": 6.8473684210526325e-06, | |
| "loss": 0.0043, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 20.61, | |
| "learning_rate": 6.821052631578948e-06, | |
| "loss": 0.005, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 20.76, | |
| "learning_rate": 6.794736842105264e-06, | |
| "loss": 0.0056, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 20.91, | |
| "learning_rate": 6.76842105263158e-06, | |
| "loss": 0.0056, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 21.05, | |
| "learning_rate": 6.742105263157896e-06, | |
| "loss": 0.0048, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 21.2, | |
| "learning_rate": 6.715789473684211e-06, | |
| "loss": 0.005, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 21.35, | |
| "learning_rate": 6.6894736842105266e-06, | |
| "loss": 0.0047, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 21.49, | |
| "learning_rate": 6.663157894736842e-06, | |
| "loss": 0.0055, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 21.64, | |
| "learning_rate": 6.636842105263158e-06, | |
| "loss": 0.0053, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 21.78, | |
| "learning_rate": 6.610526315789474e-06, | |
| "loss": 0.0046, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 21.93, | |
| "learning_rate": 6.58421052631579e-06, | |
| "loss": 0.0053, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 22.08, | |
| "learning_rate": 6.557894736842106e-06, | |
| "loss": 0.0053, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 22.22, | |
| "learning_rate": 6.531578947368422e-06, | |
| "loss": 0.0049, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 22.37, | |
| "learning_rate": 6.505263157894738e-06, | |
| "loss": 0.0043, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 22.51, | |
| "learning_rate": 6.478947368421054e-06, | |
| "loss": 0.0045, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 22.66, | |
| "learning_rate": 6.452631578947369e-06, | |
| "loss": 0.0035, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 22.81, | |
| "learning_rate": 6.426315789473685e-06, | |
| "loss": 0.0044, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 22.95, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.0041, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 23.1, | |
| "learning_rate": 6.373684210526316e-06, | |
| "loss": 0.0037, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 23.25, | |
| "learning_rate": 6.347368421052632e-06, | |
| "loss": 0.0037, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 23.39, | |
| "learning_rate": 6.321052631578948e-06, | |
| "loss": 0.0041, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 23.39, | |
| "eval_loss": 0.35511845350265503, | |
| "eval_runtime": 714.454, | |
| "eval_samples_per_second": 9.225, | |
| "eval_steps_per_second": 0.288, | |
| "eval_wer": 14.801829575449837, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 23.54, | |
| "learning_rate": 6.294736842105264e-06, | |
| "loss": 0.0041, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 23.68, | |
| "learning_rate": 6.268421052631579e-06, | |
| "loss": 0.0036, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 23.83, | |
| "learning_rate": 6.242105263157895e-06, | |
| "loss": 0.0036, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 23.98, | |
| "learning_rate": 6.2157894736842104e-06, | |
| "loss": 0.0041, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 24.12, | |
| "learning_rate": 6.189473684210526e-06, | |
| "loss": 0.0035, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 24.27, | |
| "learning_rate": 6.163157894736843e-06, | |
| "loss": 0.0035, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 24.42, | |
| "learning_rate": 6.136842105263159e-06, | |
| "loss": 0.0037, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 24.56, | |
| "learning_rate": 6.1105263157894746e-06, | |
| "loss": 0.0031, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 24.71, | |
| "learning_rate": 6.08421052631579e-06, | |
| "loss": 0.0034, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 24.85, | |
| "learning_rate": 6.057894736842106e-06, | |
| "loss": 0.004, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 6.031578947368422e-06, | |
| "loss": 0.0039, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 25.15, | |
| "learning_rate": 6.005263157894737e-06, | |
| "loss": 0.0034, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 25.29, | |
| "learning_rate": 5.978947368421053e-06, | |
| "loss": 0.0024, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 25.44, | |
| "learning_rate": 5.952631578947369e-06, | |
| "loss": 0.0033, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 25.58, | |
| "learning_rate": 5.9263157894736844e-06, | |
| "loss": 0.0035, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 25.73, | |
| "learning_rate": 5.9e-06, | |
| "loss": 0.0029, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 25.88, | |
| "learning_rate": 5.873684210526316e-06, | |
| "loss": 0.0036, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 5.847368421052632e-06, | |
| "loss": 0.0032, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 26.17, | |
| "learning_rate": 5.8210526315789486e-06, | |
| "loss": 0.0032, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 26.32, | |
| "learning_rate": 5.794736842105264e-06, | |
| "loss": 0.0032, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 26.46, | |
| "learning_rate": 5.76842105263158e-06, | |
| "loss": 0.0032, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 26.61, | |
| "learning_rate": 5.742105263157895e-06, | |
| "loss": 0.0039, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 26.75, | |
| "learning_rate": 5.715789473684211e-06, | |
| "loss": 0.0029, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 26.9, | |
| "learning_rate": 5.689473684210527e-06, | |
| "loss": 0.0034, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 27.05, | |
| "learning_rate": 5.663157894736843e-06, | |
| "loss": 0.0027, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 27.19, | |
| "learning_rate": 5.6368421052631584e-06, | |
| "loss": 0.0031, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 27.34, | |
| "learning_rate": 5.610526315789474e-06, | |
| "loss": 0.0031, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 27.49, | |
| "learning_rate": 5.58421052631579e-06, | |
| "loss": 0.0032, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 27.63, | |
| "learning_rate": 5.557894736842105e-06, | |
| "loss": 0.0033, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 27.78, | |
| "learning_rate": 5.531578947368421e-06, | |
| "loss": 0.0035, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 27.92, | |
| "learning_rate": 5.505263157894737e-06, | |
| "loss": 0.0029, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 28.07, | |
| "learning_rate": 5.4789473684210525e-06, | |
| "loss": 0.0034, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 28.22, | |
| "learning_rate": 5.452631578947369e-06, | |
| "loss": 0.0022, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 28.36, | |
| "learning_rate": 5.426315789473685e-06, | |
| "loss": 0.0025, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 28.51, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 0.0028, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 28.65, | |
| "learning_rate": 5.373684210526317e-06, | |
| "loss": 0.0036, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 28.8, | |
| "learning_rate": 5.3473684210526325e-06, | |
| "loss": 0.0031, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 28.95, | |
| "learning_rate": 5.321052631578948e-06, | |
| "loss": 0.0031, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 29.09, | |
| "learning_rate": 5.294736842105263e-06, | |
| "loss": 0.0032, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 29.24, | |
| "learning_rate": 5.268421052631579e-06, | |
| "loss": 0.0032, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 29.24, | |
| "eval_loss": 0.3697735667228699, | |
| "eval_runtime": 714.2425, | |
| "eval_samples_per_second": 9.228, | |
| "eval_steps_per_second": 0.288, | |
| "eval_wer": 14.624513893086705, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 29.39, | |
| "learning_rate": 5.242105263157895e-06, | |
| "loss": 0.0025, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 29.53, | |
| "learning_rate": 5.215789473684211e-06, | |
| "loss": 0.0022, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 29.68, | |
| "learning_rate": 5.1894736842105265e-06, | |
| "loss": 0.0027, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 29.82, | |
| "learning_rate": 5.163157894736842e-06, | |
| "loss": 0.0029, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 29.97, | |
| "learning_rate": 5.136842105263158e-06, | |
| "loss": 0.0026, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 30.12, | |
| "learning_rate": 5.110526315789474e-06, | |
| "loss": 0.0025, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 30.26, | |
| "learning_rate": 5.084210526315791e-06, | |
| "loss": 0.0026, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 30.41, | |
| "learning_rate": 5.0578947368421065e-06, | |
| "loss": 0.0022, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 30.56, | |
| "learning_rate": 5.0315789473684214e-06, | |
| "loss": 0.0022, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 30.7, | |
| "learning_rate": 5.005263157894737e-06, | |
| "loss": 0.0023, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 30.85, | |
| "learning_rate": 4.978947368421053e-06, | |
| "loss": 0.002, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 30.99, | |
| "learning_rate": 4.952631578947369e-06, | |
| "loss": 0.0021, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 31.14, | |
| "learning_rate": 4.926315789473685e-06, | |
| "loss": 0.0023, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 31.29, | |
| "learning_rate": 4.9000000000000005e-06, | |
| "loss": 0.0024, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 31.43, | |
| "learning_rate": 4.873684210526316e-06, | |
| "loss": 0.0023, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 31.58, | |
| "learning_rate": 4.847368421052631e-06, | |
| "loss": 0.0017, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 31.73, | |
| "learning_rate": 4.821052631578948e-06, | |
| "loss": 0.0016, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 31.87, | |
| "learning_rate": 4.794736842105264e-06, | |
| "loss": 0.0024, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 32.02, | |
| "learning_rate": 4.76842105263158e-06, | |
| "loss": 0.002, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 32.16, | |
| "learning_rate": 4.7421052631578954e-06, | |
| "loss": 0.0019, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 32.31, | |
| "learning_rate": 4.71578947368421e-06, | |
| "loss": 0.0018, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 32.46, | |
| "learning_rate": 4.689473684210526e-06, | |
| "loss": 0.002, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 32.6, | |
| "learning_rate": 4.663157894736842e-06, | |
| "loss": 0.0027, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 32.75, | |
| "learning_rate": 4.636842105263159e-06, | |
| "loss": 0.0019, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 32.89, | |
| "learning_rate": 4.6105263157894745e-06, | |
| "loss": 0.0028, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 33.04, | |
| "learning_rate": 4.5842105263157895e-06, | |
| "loss": 0.0022, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 33.19, | |
| "learning_rate": 4.557894736842105e-06, | |
| "loss": 0.002, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 33.33, | |
| "learning_rate": 4.531578947368421e-06, | |
| "loss": 0.0017, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 33.48, | |
| "learning_rate": 4.505263157894737e-06, | |
| "loss": 0.0018, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 33.63, | |
| "learning_rate": 4.478947368421054e-06, | |
| "loss": 0.002, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 33.77, | |
| "learning_rate": 4.452631578947369e-06, | |
| "loss": 0.0022, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 33.92, | |
| "learning_rate": 4.426315789473684e-06, | |
| "loss": 0.0023, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 34.06, | |
| "learning_rate": 4.4e-06, | |
| "loss": 0.0018, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 34.21, | |
| "learning_rate": 4.373684210526316e-06, | |
| "loss": 0.0014, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 34.36, | |
| "learning_rate": 4.347368421052632e-06, | |
| "loss": 0.0013, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 34.5, | |
| "learning_rate": 4.321052631578948e-06, | |
| "loss": 0.0019, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 34.65, | |
| "learning_rate": 4.2947368421052635e-06, | |
| "loss": 0.0026, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 34.8, | |
| "learning_rate": 4.268421052631579e-06, | |
| "loss": 0.002, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 34.94, | |
| "learning_rate": 4.242105263157895e-06, | |
| "loss": 0.0021, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 35.09, | |
| "learning_rate": 4.215789473684211e-06, | |
| "loss": 0.0019, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 35.09, | |
| "eval_loss": 0.38771745562553406, | |
| "eval_runtime": 715.1183, | |
| "eval_samples_per_second": 9.217, | |
| "eval_steps_per_second": 0.288, | |
| "eval_wer": 14.608394285599147, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 35.23, | |
| "learning_rate": 4.189473684210527e-06, | |
| "loss": 0.0021, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 35.38, | |
| "learning_rate": 4.163157894736843e-06, | |
| "loss": 0.0018, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 35.53, | |
| "learning_rate": 4.136842105263158e-06, | |
| "loss": 0.0015, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 35.67, | |
| "learning_rate": 4.110526315789474e-06, | |
| "loss": 0.0016, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 35.82, | |
| "learning_rate": 4.08421052631579e-06, | |
| "loss": 0.0019, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 35.96, | |
| "learning_rate": 4.057894736842106e-06, | |
| "loss": 0.002, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 36.11, | |
| "learning_rate": 4.031578947368422e-06, | |
| "loss": 0.0015, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 36.26, | |
| "learning_rate": 4.005263157894737e-06, | |
| "loss": 0.0019, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 36.4, | |
| "learning_rate": 3.9789473684210525e-06, | |
| "loss": 0.0017, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 36.55, | |
| "learning_rate": 3.952631578947368e-06, | |
| "loss": 0.0013, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 36.7, | |
| "learning_rate": 3.926315789473685e-06, | |
| "loss": 0.0014, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 36.84, | |
| "learning_rate": 3.900000000000001e-06, | |
| "loss": 0.0013, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 36.99, | |
| "learning_rate": 3.873684210526316e-06, | |
| "loss": 0.0013, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 37.13, | |
| "learning_rate": 3.8473684210526316e-06, | |
| "loss": 0.0014, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 37.28, | |
| "learning_rate": 3.821052631578947e-06, | |
| "loss": 0.0012, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 37.43, | |
| "learning_rate": 3.794736842105263e-06, | |
| "loss": 0.0013, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 37.57, | |
| "learning_rate": 3.768421052631579e-06, | |
| "loss": 0.0013, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 37.72, | |
| "learning_rate": 3.7421052631578953e-06, | |
| "loss": 0.0012, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 37.87, | |
| "learning_rate": 3.715789473684211e-06, | |
| "loss": 0.0014, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 38.01, | |
| "learning_rate": 3.6894736842105265e-06, | |
| "loss": 0.002, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 38.16, | |
| "learning_rate": 3.6631578947368423e-06, | |
| "loss": 0.0011, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 38.3, | |
| "learning_rate": 3.636842105263158e-06, | |
| "loss": 0.0014, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 38.45, | |
| "learning_rate": 3.610526315789474e-06, | |
| "loss": 0.0015, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 38.6, | |
| "learning_rate": 3.58421052631579e-06, | |
| "loss": 0.0014, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 38.74, | |
| "learning_rate": 3.5578947368421056e-06, | |
| "loss": 0.0014, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 38.89, | |
| "learning_rate": 3.5315789473684214e-06, | |
| "loss": 0.0017, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 39.04, | |
| "learning_rate": 3.505263157894737e-06, | |
| "loss": 0.0015, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 39.18, | |
| "learning_rate": 3.478947368421053e-06, | |
| "loss": 0.0014, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 39.33, | |
| "learning_rate": 3.4526315789473684e-06, | |
| "loss": 0.0013, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 39.47, | |
| "learning_rate": 3.4263157894736842e-06, | |
| "loss": 0.0012, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 39.62, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 0.0017, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 39.77, | |
| "learning_rate": 3.3736842105263163e-06, | |
| "loss": 0.0014, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 39.91, | |
| "learning_rate": 3.347368421052632e-06, | |
| "loss": 0.0014, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 40.06, | |
| "learning_rate": 3.3210526315789475e-06, | |
| "loss": 0.0011, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 40.2, | |
| "learning_rate": 3.2947368421052633e-06, | |
| "loss": 0.0012, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 40.35, | |
| "learning_rate": 3.268421052631579e-06, | |
| "loss": 0.0011, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 40.5, | |
| "learning_rate": 3.2421052631578945e-06, | |
| "loss": 0.0013, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 40.64, | |
| "learning_rate": 3.215789473684211e-06, | |
| "loss": 0.0012, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 40.79, | |
| "learning_rate": 3.1894736842105266e-06, | |
| "loss": 0.0017, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 40.94, | |
| "learning_rate": 3.1631578947368424e-06, | |
| "loss": 0.0014, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 40.94, | |
| "eval_loss": 0.38908982276916504, | |
| "eval_runtime": 714.3256, | |
| "eval_samples_per_second": 9.227, | |
| "eval_steps_per_second": 0.288, | |
| "eval_wer": 14.497571984122187, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 41.08, | |
| "learning_rate": 3.1368421052631582e-06, | |
| "loss": 0.0014, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 41.23, | |
| "learning_rate": 3.1105263157894736e-06, | |
| "loss": 0.0014, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 41.37, | |
| "learning_rate": 3.0842105263157895e-06, | |
| "loss": 0.0013, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 41.52, | |
| "learning_rate": 3.0578947368421053e-06, | |
| "loss": 0.0013, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 41.67, | |
| "learning_rate": 3.0315789473684215e-06, | |
| "loss": 0.0015, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 41.81, | |
| "learning_rate": 3.0052631578947373e-06, | |
| "loss": 0.0015, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 41.96, | |
| "learning_rate": 2.9789473684210527e-06, | |
| "loss": 0.0012, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 42.11, | |
| "learning_rate": 2.9526315789473685e-06, | |
| "loss": 0.0011, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 42.25, | |
| "learning_rate": 2.9263157894736844e-06, | |
| "loss": 0.0009, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 42.4, | |
| "learning_rate": 2.9e-06, | |
| "loss": 0.0009, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 42.54, | |
| "learning_rate": 2.8736842105263164e-06, | |
| "loss": 0.0012, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 42.69, | |
| "learning_rate": 2.847368421052632e-06, | |
| "loss": 0.0012, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 42.84, | |
| "learning_rate": 2.8210526315789476e-06, | |
| "loss": 0.001, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 42.98, | |
| "learning_rate": 2.7947368421052635e-06, | |
| "loss": 0.0018, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 43.13, | |
| "learning_rate": 2.7684210526315793e-06, | |
| "loss": 0.0013, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 43.27, | |
| "learning_rate": 2.7421052631578947e-06, | |
| "loss": 0.0011, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 43.42, | |
| "learning_rate": 2.7157894736842105e-06, | |
| "loss": 0.0014, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 43.57, | |
| "learning_rate": 2.6894736842105267e-06, | |
| "loss": 0.001, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 43.71, | |
| "learning_rate": 2.6631578947368426e-06, | |
| "loss": 0.0011, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 43.86, | |
| "learning_rate": 2.6368421052631584e-06, | |
| "loss": 0.001, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 44.01, | |
| "learning_rate": 2.6105263157894738e-06, | |
| "loss": 0.0013, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 44.15, | |
| "learning_rate": 2.5842105263157896e-06, | |
| "loss": 0.0015, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 44.3, | |
| "learning_rate": 2.5578947368421054e-06, | |
| "loss": 0.0012, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 44.44, | |
| "learning_rate": 2.5315789473684212e-06, | |
| "loss": 0.0013, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 44.59, | |
| "learning_rate": 2.5052631578947375e-06, | |
| "loss": 0.0012, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 44.74, | |
| "learning_rate": 2.478947368421053e-06, | |
| "loss": 0.0009, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 44.88, | |
| "learning_rate": 2.4526315789473687e-06, | |
| "loss": 0.0011, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 45.03, | |
| "learning_rate": 2.4263157894736845e-06, | |
| "loss": 0.0011, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 45.18, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.0012, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 45.32, | |
| "learning_rate": 2.373684210526316e-06, | |
| "loss": 0.0008, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 45.47, | |
| "learning_rate": 2.347368421052632e-06, | |
| "loss": 0.0009, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 45.61, | |
| "learning_rate": 2.3210526315789473e-06, | |
| "loss": 0.0008, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 45.76, | |
| "learning_rate": 2.294736842105263e-06, | |
| "loss": 0.0009, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 45.91, | |
| "learning_rate": 2.268421052631579e-06, | |
| "loss": 0.0011, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 46.05, | |
| "learning_rate": 2.242105263157895e-06, | |
| "loss": 0.0009, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 46.2, | |
| "learning_rate": 2.2157894736842106e-06, | |
| "loss": 0.0009, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 46.35, | |
| "learning_rate": 2.1894736842105264e-06, | |
| "loss": 0.0008, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 46.49, | |
| "learning_rate": 2.1631578947368423e-06, | |
| "loss": 0.001, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 46.64, | |
| "learning_rate": 2.136842105263158e-06, | |
| "loss": 0.0008, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 46.78, | |
| "learning_rate": 2.110526315789474e-06, | |
| "loss": 0.0008, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 46.78, | |
| "eval_loss": 0.39463770389556885, | |
| "eval_runtime": 713.9647, | |
| "eval_samples_per_second": 9.232, | |
| "eval_steps_per_second": 0.289, | |
| "eval_wer": 14.275927381168268, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 46.93, | |
| "learning_rate": 2.0842105263157897e-06, | |
| "loss": 0.0008, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 47.08, | |
| "learning_rate": 2.0578947368421055e-06, | |
| "loss": 0.0012, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 47.22, | |
| "learning_rate": 2.031578947368421e-06, | |
| "loss": 0.001, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 47.37, | |
| "learning_rate": 2.005263157894737e-06, | |
| "loss": 0.001, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 47.51, | |
| "learning_rate": 1.978947368421053e-06, | |
| "loss": 0.001, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 47.66, | |
| "learning_rate": 1.9526315789473684e-06, | |
| "loss": 0.0012, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 47.81, | |
| "learning_rate": 1.9263157894736846e-06, | |
| "loss": 0.0009, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 47.95, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.0011, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 48.1, | |
| "learning_rate": 1.8736842105263158e-06, | |
| "loss": 0.001, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 48.25, | |
| "learning_rate": 1.8473684210526319e-06, | |
| "loss": 0.0009, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 48.39, | |
| "learning_rate": 1.8210526315789475e-06, | |
| "loss": 0.0011, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 48.54, | |
| "learning_rate": 1.7947368421052633e-06, | |
| "loss": 0.0009, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 48.68, | |
| "learning_rate": 1.768421052631579e-06, | |
| "loss": 0.0007, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 48.83, | |
| "learning_rate": 1.742105263157895e-06, | |
| "loss": 0.0009, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 48.98, | |
| "learning_rate": 1.7157894736842107e-06, | |
| "loss": 0.001, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 49.12, | |
| "learning_rate": 1.6894736842105264e-06, | |
| "loss": 0.0007, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 49.27, | |
| "learning_rate": 1.6631578947368424e-06, | |
| "loss": 0.0009, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 49.42, | |
| "learning_rate": 1.636842105263158e-06, | |
| "loss": 0.0007, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 49.56, | |
| "learning_rate": 1.6105263157894738e-06, | |
| "loss": 0.0008, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 49.71, | |
| "learning_rate": 1.5842105263157894e-06, | |
| "loss": 0.0008, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 49.85, | |
| "learning_rate": 1.5578947368421054e-06, | |
| "loss": 0.0009, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "learning_rate": 1.5315789473684213e-06, | |
| "loss": 0.0007, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 50.15, | |
| "learning_rate": 1.5052631578947369e-06, | |
| "loss": 0.0006, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 50.29, | |
| "learning_rate": 1.478947368421053e-06, | |
| "loss": 0.0009, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 50.44, | |
| "learning_rate": 1.4526315789473685e-06, | |
| "loss": 0.0007, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 50.58, | |
| "learning_rate": 1.4263157894736843e-06, | |
| "loss": 0.0008, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 50.73, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.0006, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 50.88, | |
| "learning_rate": 1.373684210526316e-06, | |
| "loss": 0.001, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 51.02, | |
| "learning_rate": 1.3473684210526316e-06, | |
| "loss": 0.001, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 51.17, | |
| "learning_rate": 1.3210526315789474e-06, | |
| "loss": 0.0007, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 51.32, | |
| "learning_rate": 1.2947368421052634e-06, | |
| "loss": 0.0008, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 51.46, | |
| "learning_rate": 1.268421052631579e-06, | |
| "loss": 0.0007, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 51.61, | |
| "learning_rate": 1.2421052631578948e-06, | |
| "loss": 0.001, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 51.75, | |
| "learning_rate": 1.2157894736842107e-06, | |
| "loss": 0.0009, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 51.9, | |
| "learning_rate": 1.1894736842105265e-06, | |
| "loss": 0.0007, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 52.05, | |
| "learning_rate": 1.163157894736842e-06, | |
| "loss": 0.0006, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 52.19, | |
| "learning_rate": 1.136842105263158e-06, | |
| "loss": 0.0007, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 52.34, | |
| "learning_rate": 1.110526315789474e-06, | |
| "loss": 0.0009, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 52.49, | |
| "learning_rate": 1.0842105263157895e-06, | |
| "loss": 0.0007, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 52.63, | |
| "learning_rate": 1.0578947368421054e-06, | |
| "loss": 0.0007, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 52.63, | |
| "eval_loss": 0.398745596408844, | |
| "eval_runtime": 712.7726, | |
| "eval_samples_per_second": 9.247, | |
| "eval_steps_per_second": 0.289, | |
| "eval_wer": 14.318241350823108, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 52.78, | |
| "learning_rate": 1.0315789473684212e-06, | |
| "loss": 0.0007, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 52.92, | |
| "learning_rate": 1.005263157894737e-06, | |
| "loss": 0.0006, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 53.07, | |
| "learning_rate": 9.789473684210526e-07, | |
| "loss": 0.0005, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 53.22, | |
| "learning_rate": 9.526315789473685e-07, | |
| "loss": 0.0006, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 53.36, | |
| "learning_rate": 9.263157894736844e-07, | |
| "loss": 0.0008, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 53.51, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.0006, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 53.65, | |
| "learning_rate": 8.736842105263159e-07, | |
| "loss": 0.0006, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 53.8, | |
| "learning_rate": 8.473684210526316e-07, | |
| "loss": 0.0006, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 53.95, | |
| "learning_rate": 8.210526315789474e-07, | |
| "loss": 0.0008, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 54.09, | |
| "learning_rate": 7.947368421052632e-07, | |
| "loss": 0.0006, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 54.24, | |
| "learning_rate": 7.684210526315789e-07, | |
| "loss": 0.0005, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 54.39, | |
| "learning_rate": 7.421052631578949e-07, | |
| "loss": 0.0006, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 54.53, | |
| "learning_rate": 7.157894736842106e-07, | |
| "loss": 0.0006, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 54.68, | |
| "learning_rate": 6.894736842105264e-07, | |
| "loss": 0.0008, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 54.82, | |
| "learning_rate": 6.631578947368422e-07, | |
| "loss": 0.0007, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 54.97, | |
| "learning_rate": 6.368421052631579e-07, | |
| "loss": 0.0009, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 55.12, | |
| "learning_rate": 6.105263157894738e-07, | |
| "loss": 0.0005, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 55.26, | |
| "learning_rate": 5.842105263157896e-07, | |
| "loss": 0.0005, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 55.41, | |
| "learning_rate": 5.578947368421053e-07, | |
| "loss": 0.0008, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 55.56, | |
| "learning_rate": 5.315789473684211e-07, | |
| "loss": 0.0006, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 55.7, | |
| "learning_rate": 5.052631578947369e-07, | |
| "loss": 0.0006, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 55.85, | |
| "learning_rate": 4.789473684210526e-07, | |
| "loss": 0.0008, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 55.99, | |
| "learning_rate": 4.526315789473685e-07, | |
| "loss": 0.0006, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 56.14, | |
| "learning_rate": 4.2631578947368427e-07, | |
| "loss": 0.0006, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 56.29, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.0005, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 56.43, | |
| "learning_rate": 3.736842105263158e-07, | |
| "loss": 0.0005, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 56.58, | |
| "learning_rate": 3.4736842105263157e-07, | |
| "loss": 0.0007, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 56.73, | |
| "learning_rate": 3.210526315789474e-07, | |
| "loss": 0.0005, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 56.87, | |
| "learning_rate": 2.9473684210526315e-07, | |
| "loss": 0.0006, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 57.02, | |
| "learning_rate": 2.6842105263157897e-07, | |
| "loss": 0.0005, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 57.16, | |
| "learning_rate": 2.4210526315789473e-07, | |
| "loss": 0.0005, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 57.31, | |
| "learning_rate": 2.1578947368421053e-07, | |
| "loss": 0.0007, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 57.46, | |
| "learning_rate": 1.8947368421052634e-07, | |
| "loss": 0.0008, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 57.6, | |
| "learning_rate": 1.631578947368421e-07, | |
| "loss": 0.0006, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 57.75, | |
| "learning_rate": 1.368421052631579e-07, | |
| "loss": 0.0006, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 57.89, | |
| "learning_rate": 1.1052631578947368e-07, | |
| "loss": 0.0005, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 58.04, | |
| "learning_rate": 8.421052631578947e-08, | |
| "loss": 0.0005, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 58.19, | |
| "learning_rate": 5.789473684210527e-08, | |
| "loss": 0.0006, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 58.33, | |
| "learning_rate": 3.157894736842106e-08, | |
| "loss": 0.0005, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 58.48, | |
| "learning_rate": 5.263157894736842e-09, | |
| "loss": 0.0005, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 58.48, | |
| "eval_loss": 0.3985374867916107, | |
| "eval_runtime": 714.3173, | |
| "eval_samples_per_second": 9.227, | |
| "eval_steps_per_second": 0.288, | |
| "eval_wer": 14.112716355356747, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 58.48, | |
| "step": 10000, | |
| "total_flos": 6.516480686122598e+20, | |
| "train_loss": 0.08935045070629567, | |
| "train_runtime": 37315.2457, | |
| "train_samples_per_second": 17.151, | |
| "train_steps_per_second": 0.268 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 10000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 59, | |
| "save_steps": 1000, | |
| "total_flos": 6.516480686122598e+20, | |
| "train_batch_size": 64, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |