| { | |
| "best_metric": 18.28368532693904, | |
| "best_model_checkpoint": "./whisper-tiny-id/checkpoint-10000", | |
| "epoch": 6.618133686300463, | |
| "global_step": 10000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.800000000000001e-07, | |
| "loss": 0.706, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.800000000000001e-07, | |
| "loss": 0.6876, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.48e-06, | |
| "loss": 0.641, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.98e-06, | |
| "loss": 0.622, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 2.4800000000000004e-06, | |
| "loss": 0.6178, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 2.9800000000000003e-06, | |
| "loss": 0.6123, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 3.48e-06, | |
| "loss": 0.5628, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.5787, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.5556, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.5128, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.5205, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.5354, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.5321, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.5282, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.5283, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.4944, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.4952, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.5032, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.4818, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.4781, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 9.975789473684211e-06, | |
| "loss": 0.4579, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 9.949473684210526e-06, | |
| "loss": 0.4633, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 9.923157894736844e-06, | |
| "loss": 0.4467, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 9.89684210526316e-06, | |
| "loss": 0.4338, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 9.870526315789474e-06, | |
| "loss": 0.449, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 9.84421052631579e-06, | |
| "loss": 0.422, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 9.817894736842106e-06, | |
| "loss": 0.421, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 9.791578947368422e-06, | |
| "loss": 0.4303, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 9.765263157894737e-06, | |
| "loss": 0.4425, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 9.738947368421054e-06, | |
| "loss": 0.4119, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 9.712631578947369e-06, | |
| "loss": 0.396, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 9.686315789473684e-06, | |
| "loss": 0.4307, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 9.66e-06, | |
| "loss": 0.4086, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 9.633684210526316e-06, | |
| "loss": 0.4253, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 9.607368421052632e-06, | |
| "loss": 0.4262, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 9.581052631578947e-06, | |
| "loss": 0.4054, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 9.554736842105264e-06, | |
| "loss": 0.3982, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 9.52842105263158e-06, | |
| "loss": 0.3704, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 9.502105263157896e-06, | |
| "loss": 0.3882, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 9.475789473684212e-06, | |
| "loss": 0.4103, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "eval_loss": 0.38016477227211, | |
| "eval_runtime": 292.4375, | |
| "eval_samples_per_second": 12.372, | |
| "eval_steps_per_second": 0.776, | |
| "eval_wer": 27.049709489993546, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 9.449473684210527e-06, | |
| "loss": 0.389, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 9.423157894736842e-06, | |
| "loss": 0.4011, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 9.396842105263159e-06, | |
| "loss": 0.3932, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 9.370526315789474e-06, | |
| "loss": 0.3766, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 9.34421052631579e-06, | |
| "loss": 0.3609, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 9.317894736842105e-06, | |
| "loss": 0.4152, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 9.291578947368422e-06, | |
| "loss": 0.3786, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 9.265263157894737e-06, | |
| "loss": 0.3966, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.238947368421052e-06, | |
| "loss": 0.3547, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 9.21263157894737e-06, | |
| "loss": 0.3748, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 9.186315789473685e-06, | |
| "loss": 0.353, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 9.16e-06, | |
| "loss": 0.3616, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 9.133684210526317e-06, | |
| "loss": 0.3832, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 9.107368421052632e-06, | |
| "loss": 0.3899, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 9.081052631578949e-06, | |
| "loss": 0.3435, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 9.054736842105264e-06, | |
| "loss": 0.3485, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.02842105263158e-06, | |
| "loss": 0.3614, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 9.002105263157895e-06, | |
| "loss": 0.3417, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 8.97578947368421e-06, | |
| "loss": 0.3358, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 8.949473684210527e-06, | |
| "loss": 0.3364, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 8.923157894736842e-06, | |
| "loss": 0.3311, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 8.896842105263159e-06, | |
| "loss": 0.283, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 8.870526315789474e-06, | |
| "loss": 0.2718, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 8.84421052631579e-06, | |
| "loss": 0.2745, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 8.817894736842107e-06, | |
| "loss": 0.2743, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 8.791578947368422e-06, | |
| "loss": 0.2698, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 8.765263157894739e-06, | |
| "loss": 0.2858, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 8.738947368421053e-06, | |
| "loss": 0.2841, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 8.712631578947368e-06, | |
| "loss": 0.3, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 8.686315789473685e-06, | |
| "loss": 0.2691, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 8.66e-06, | |
| "loss": 0.2502, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.633684210526317e-06, | |
| "loss": 0.261, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 8.607368421052632e-06, | |
| "loss": 0.2669, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 8.581052631578948e-06, | |
| "loss": 0.2621, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 8.554736842105263e-06, | |
| "loss": 0.2664, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 8.528421052631578e-06, | |
| "loss": 0.2688, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 8.502105263157897e-06, | |
| "loss": 0.2766, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 8.475789473684212e-06, | |
| "loss": 0.2829, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 8.449473684210527e-06, | |
| "loss": 0.2497, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 8.423157894736843e-06, | |
| "loss": 0.2682, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "eval_loss": 0.32230183482170105, | |
| "eval_runtime": 273.1698, | |
| "eval_samples_per_second": 13.245, | |
| "eval_steps_per_second": 0.831, | |
| "eval_wer": 22.936456700175228, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 8.396842105263158e-06, | |
| "loss": 0.2666, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 8.370526315789475e-06, | |
| "loss": 0.2527, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 8.34421052631579e-06, | |
| "loss": 0.2583, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 8.317894736842107e-06, | |
| "loss": 0.2878, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 8.291578947368422e-06, | |
| "loss": 0.2557, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 8.265263157894737e-06, | |
| "loss": 0.2634, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 8.238947368421053e-06, | |
| "loss": 0.2642, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 8.212631578947368e-06, | |
| "loss": 0.2561, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 8.186315789473685e-06, | |
| "loss": 0.273, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 8.16e-06, | |
| "loss": 0.2583, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 8.133684210526316e-06, | |
| "loss": 0.2457, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 8.107368421052633e-06, | |
| "loss": 0.2511, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 8.081052631578948e-06, | |
| "loss": 0.2552, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 8.054736842105265e-06, | |
| "loss": 0.2588, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 8.02842105263158e-06, | |
| "loss": 0.2628, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 8.002105263157895e-06, | |
| "loss": 0.2393, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 7.975789473684211e-06, | |
| "loss": 0.2415, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 7.949473684210526e-06, | |
| "loss": 0.2442, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 7.923157894736843e-06, | |
| "loss": 0.2702, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 7.896842105263158e-06, | |
| "loss": 0.2467, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 7.870526315789475e-06, | |
| "loss": 0.2554, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 7.84421052631579e-06, | |
| "loss": 0.2486, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 7.817894736842105e-06, | |
| "loss": 0.2536, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 7.791578947368423e-06, | |
| "loss": 0.2543, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 7.765263157894738e-06, | |
| "loss": 0.261, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 7.738947368421053e-06, | |
| "loss": 0.2595, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 7.71263157894737e-06, | |
| "loss": 0.2421, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 7.686315789473685e-06, | |
| "loss": 0.2423, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.660000000000001e-06, | |
| "loss": 0.2518, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.633684210526316e-06, | |
| "loss": 0.2496, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 7.607368421052632e-06, | |
| "loss": 0.2724, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 7.581052631578948e-06, | |
| "loss": 0.2235, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 7.554736842105264e-06, | |
| "loss": 0.2406, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 7.5284210526315794e-06, | |
| "loss": 0.2412, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 7.502105263157895e-06, | |
| "loss": 0.2392, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.475789473684211e-06, | |
| "loss": 0.2432, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 7.449473684210526e-06, | |
| "loss": 0.247, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 7.4231578947368436e-06, | |
| "loss": 0.2447, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 7.3968421052631585e-06, | |
| "loss": 0.274, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 7.370526315789474e-06, | |
| "loss": 0.2381, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "eval_loss": 0.28843218088150024, | |
| "eval_runtime": 269.9427, | |
| "eval_samples_per_second": 13.403, | |
| "eval_steps_per_second": 0.841, | |
| "eval_wer": 20.82449506594116, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.34421052631579e-06, | |
| "loss": 0.2326, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 7.317894736842106e-06, | |
| "loss": 0.1808, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 7.291578947368422e-06, | |
| "loss": 0.1783, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 7.265263157894738e-06, | |
| "loss": 0.1744, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 7.2389473684210534e-06, | |
| "loss": 0.1965, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 7.212631578947369e-06, | |
| "loss": 0.1741, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 7.186315789473684e-06, | |
| "loss": 0.1768, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 7.16e-06, | |
| "loss": 0.1797, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 7.133684210526316e-06, | |
| "loss": 0.1796, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 7.107368421052632e-06, | |
| "loss": 0.1789, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 7.0810526315789475e-06, | |
| "loss": 0.1964, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 7.054736842105264e-06, | |
| "loss": 0.1947, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 7.02842105263158e-06, | |
| "loss": 0.1899, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 7.002105263157896e-06, | |
| "loss": 0.1878, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 6.975789473684212e-06, | |
| "loss": 0.1914, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 6.9494736842105275e-06, | |
| "loss": 0.2036, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 6.9231578947368424e-06, | |
| "loss": 0.1905, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 6.896842105263158e-06, | |
| "loss": 0.1835, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 6.870526315789474e-06, | |
| "loss": 0.1941, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 6.84421052631579e-06, | |
| "loss": 0.184, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 6.817894736842106e-06, | |
| "loss": 0.1809, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 6.7915789473684215e-06, | |
| "loss": 0.1947, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 6.765263157894737e-06, | |
| "loss": 0.1796, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 6.738947368421052e-06, | |
| "loss": 0.1711, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 6.71263157894737e-06, | |
| "loss": 0.1833, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 6.686315789473685e-06, | |
| "loss": 0.174, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 0.1778, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 6.6336842105263164e-06, | |
| "loss": 0.1559, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 6.607368421052632e-06, | |
| "loss": 0.1761, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 6.581052631578948e-06, | |
| "loss": 0.185, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 6.554736842105264e-06, | |
| "loss": 0.1885, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 6.52842105263158e-06, | |
| "loss": 0.19, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 6.5021052631578955e-06, | |
| "loss": 0.1814, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 6.4757894736842105e-06, | |
| "loss": 0.1898, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 6.449473684210526e-06, | |
| "loss": 0.1856, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 6.423157894736842e-06, | |
| "loss": 0.1833, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.396842105263158e-06, | |
| "loss": 0.1803, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.370526315789474e-06, | |
| "loss": 0.1632, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.3442105263157904e-06, | |
| "loss": 0.1933, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 6.317894736842106e-06, | |
| "loss": 0.1606, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "eval_loss": 0.27274927496910095, | |
| "eval_runtime": 278.2051, | |
| "eval_samples_per_second": 13.005, | |
| "eval_steps_per_second": 0.816, | |
| "eval_wer": 20.192751083648435, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 6.291578947368422e-06, | |
| "loss": 0.1872, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 6.265263157894738e-06, | |
| "loss": 0.2002, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 6.238947368421054e-06, | |
| "loss": 0.1811, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 6.212631578947369e-06, | |
| "loss": 0.1695, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 6.1863157894736845e-06, | |
| "loss": 0.1916, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 6.16e-06, | |
| "loss": 0.17, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 6.133684210526316e-06, | |
| "loss": 0.1902, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 6.107368421052632e-06, | |
| "loss": 0.1937, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 6.081052631578948e-06, | |
| "loss": 0.1708, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 6.054736842105264e-06, | |
| "loss": 0.1911, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 6.0284210526315786e-06, | |
| "loss": 0.1966, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 6.002105263157896e-06, | |
| "loss": 0.1764, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 5.975789473684212e-06, | |
| "loss": 0.1773, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 5.950526315789474e-06, | |
| "loss": 0.183, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 5.92421052631579e-06, | |
| "loss": 0.1772, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 5.897894736842106e-06, | |
| "loss": 0.1817, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 5.871578947368422e-06, | |
| "loss": 0.1818, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 5.845263157894737e-06, | |
| "loss": 0.1784, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 5.818947368421053e-06, | |
| "loss": 0.1664, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 5.7926315789473685e-06, | |
| "loss": 0.1595, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 5.766315789473684e-06, | |
| "loss": 0.1624, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 5.74e-06, | |
| "loss": 0.1602, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.713684210526316e-06, | |
| "loss": 0.133, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 5.687368421052633e-06, | |
| "loss": 0.1421, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 5.661052631578948e-06, | |
| "loss": 0.1342, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 5.634736842105264e-06, | |
| "loss": 0.1416, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 5.60842105263158e-06, | |
| "loss": 0.1193, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 5.582105263157895e-06, | |
| "loss": 0.1272, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 5.555789473684211e-06, | |
| "loss": 0.1283, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 5.529473684210527e-06, | |
| "loss": 0.1286, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 5.5031578947368425e-06, | |
| "loss": 0.1365, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 5.476842105263158e-06, | |
| "loss": 0.1195, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 5.450526315789474e-06, | |
| "loss": 0.1376, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 5.42421052631579e-06, | |
| "loss": 0.1309, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 5.397894736842105e-06, | |
| "loss": 0.1285, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 5.371578947368421e-06, | |
| "loss": 0.133, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 5.345263157894738e-06, | |
| "loss": 0.1274, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 5.318947368421053e-06, | |
| "loss": 0.1392, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 5.292631578947369e-06, | |
| "loss": 0.145, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 5.266315789473685e-06, | |
| "loss": 0.1246, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "eval_loss": 0.25959450006484985, | |
| "eval_runtime": 273.2524, | |
| "eval_samples_per_second": 13.241, | |
| "eval_steps_per_second": 0.831, | |
| "eval_wer": 18.99843216821913, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 5.240000000000001e-06, | |
| "loss": 0.1401, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 5.2136842105263165e-06, | |
| "loss": 0.1397, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 5.187368421052632e-06, | |
| "loss": 0.1434, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 5.161052631578948e-06, | |
| "loss": 0.1511, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 5.134736842105263e-06, | |
| "loss": 0.1419, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 5.108421052631579e-06, | |
| "loss": 0.1472, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 5.082105263157895e-06, | |
| "loss": 0.1279, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 5.0557894736842105e-06, | |
| "loss": 0.1388, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 5.029473684210526e-06, | |
| "loss": 0.1329, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 5.003157894736842e-06, | |
| "loss": 0.131, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 4.976842105263158e-06, | |
| "loss": 0.1418, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 4.950526315789474e-06, | |
| "loss": 0.1436, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 4.92421052631579e-06, | |
| "loss": 0.1364, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 4.897894736842106e-06, | |
| "loss": 0.1371, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 4.871578947368421e-06, | |
| "loss": 0.1318, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 4.845263157894737e-06, | |
| "loss": 0.1338, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 4.818947368421053e-06, | |
| "loss": 0.1328, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 4.792631578947369e-06, | |
| "loss": 0.1218, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 4.7663157894736845e-06, | |
| "loss": 0.1435, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 4.74e-06, | |
| "loss": 0.1245, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 4.713684210526316e-06, | |
| "loss": 0.1283, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 4.687368421052632e-06, | |
| "loss": 0.142, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 4.661052631578948e-06, | |
| "loss": 0.1538, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 4.634736842105264e-06, | |
| "loss": 0.1399, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 4.6084210526315794e-06, | |
| "loss": 0.1399, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 4.582105263157895e-06, | |
| "loss": 0.1457, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 4.55578947368421e-06, | |
| "loss": 0.1352, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 4.529473684210527e-06, | |
| "loss": 0.1436, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 4.503157894736843e-06, | |
| "loss": 0.1481, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 4.4768421052631585e-06, | |
| "loss": 0.1272, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 4.450526315789474e-06, | |
| "loss": 0.1317, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 4.424210526315789e-06, | |
| "loss": 0.1354, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 4.397894736842105e-06, | |
| "loss": 0.1319, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 4.371578947368421e-06, | |
| "loss": 0.127, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 4.345263157894738e-06, | |
| "loss": 0.1376, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 4.3189473684210535e-06, | |
| "loss": 0.1363, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 4.2926315789473684e-06, | |
| "loss": 0.1483, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 4.266315789473684e-06, | |
| "loss": 0.1461, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 4.24e-06, | |
| "loss": 0.124, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 4.213684210526316e-06, | |
| "loss": 0.1344, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "eval_loss": 0.2481694519519806, | |
| "eval_runtime": 273.3201, | |
| "eval_samples_per_second": 13.237, | |
| "eval_steps_per_second": 0.831, | |
| "eval_wer": 18.754034861200775, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 4.1873684210526325e-06, | |
| "loss": 0.1364, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 4.1610526315789475e-06, | |
| "loss": 0.1289, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 4.134736842105263e-06, | |
| "loss": 0.1012, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4.108421052631579e-06, | |
| "loss": 0.1069, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 4.082105263157895e-06, | |
| "loss": 0.1126, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 4.055789473684211e-06, | |
| "loss": 0.1, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 4.029473684210527e-06, | |
| "loss": 0.0985, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 4.0031578947368424e-06, | |
| "loss": 0.1205, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 3.976842105263158e-06, | |
| "loss": 0.1049, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 3.950526315789474e-06, | |
| "loss": 0.1129, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 3.92421052631579e-06, | |
| "loss": 0.0972, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 3.897894736842106e-06, | |
| "loss": 0.1121, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 3.8715789473684215e-06, | |
| "loss": 0.1038, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 3.845263157894737e-06, | |
| "loss": 0.1077, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 3.818947368421053e-06, | |
| "loss": 0.1065, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 3.792631578947369e-06, | |
| "loss": 0.1116, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 3.767368421052632e-06, | |
| "loss": 0.1128, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 3.7410526315789473e-06, | |
| "loss": 0.1009, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 3.714736842105263e-06, | |
| "loss": 0.1039, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 3.6884210526315794e-06, | |
| "loss": 0.0984, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 3.662105263157895e-06, | |
| "loss": 0.1269, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 3.635789473684211e-06, | |
| "loss": 0.1067, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 3.6094736842105264e-06, | |
| "loss": 0.0989, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 3.5831578947368422e-06, | |
| "loss": 0.1084, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 3.556842105263158e-06, | |
| "loss": 0.094, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 3.5305263157894743e-06, | |
| "loss": 0.1013, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 3.50421052631579e-06, | |
| "loss": 0.1135, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 3.4778947368421055e-06, | |
| "loss": 0.1154, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 3.4515789473684213e-06, | |
| "loss": 0.0954, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 3.425263157894737e-06, | |
| "loss": 0.0996, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 3.398947368421053e-06, | |
| "loss": 0.108, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 3.3726315789473683e-06, | |
| "loss": 0.0987, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 3.3463157894736846e-06, | |
| "loss": 0.11, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 3.3200000000000004e-06, | |
| "loss": 0.1126, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 3.2936842105263162e-06, | |
| "loss": 0.1092, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 3.267368421052632e-06, | |
| "loss": 0.1191, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 3.2410526315789474e-06, | |
| "loss": 0.1086, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 3.2147368421052633e-06, | |
| "loss": 0.1079, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 3.188421052631579e-06, | |
| "loss": 0.1045, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 3.1621052631578953e-06, | |
| "loss": 0.0975, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "eval_loss": 0.2470633089542389, | |
| "eval_runtime": 276.0791, | |
| "eval_samples_per_second": 13.105, | |
| "eval_steps_per_second": 0.822, | |
| "eval_wer": 18.638753112607212, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.135789473684211e-06, | |
| "loss": 0.106, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 3.1094736842105265e-06, | |
| "loss": 0.1056, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 3.0831578947368423e-06, | |
| "loss": 0.1067, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 3.056842105263158e-06, | |
| "loss": 0.1051, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 3.0305263157894736e-06, | |
| "loss": 0.1097, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 3.0042105263157894e-06, | |
| "loss": 0.1095, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 2.9778947368421056e-06, | |
| "loss": 0.0976, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 2.9515789473684214e-06, | |
| "loss": 0.0967, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 2.9252631578947373e-06, | |
| "loss": 0.0999, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 2.8989473684210526e-06, | |
| "loss": 0.0957, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 2.8726315789473685e-06, | |
| "loss": 0.1107, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 2.8463157894736843e-06, | |
| "loss": 0.0988, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 2.82e-06, | |
| "loss": 0.1095, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 2.7936842105263163e-06, | |
| "loss": 0.1064, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 2.7673684210526317e-06, | |
| "loss": 0.106, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 2.7410526315789476e-06, | |
| "loss": 0.1043, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 2.7147368421052634e-06, | |
| "loss": 0.1086, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 2.6894736842105267e-06, | |
| "loss": 0.1032, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 2.6631578947368426e-06, | |
| "loss": 0.1074, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 2.6368421052631584e-06, | |
| "loss": 0.1005, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 2.6105263157894738e-06, | |
| "loss": 0.1072, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 2.5842105263157896e-06, | |
| "loss": 0.0957, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 2.5578947368421054e-06, | |
| "loss": 0.0908, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 2.5315789473684212e-06, | |
| "loss": 0.0878, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 2.5052631578947375e-06, | |
| "loss": 0.0842, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 5.06, | |
| "learning_rate": 2.478947368421053e-06, | |
| "loss": 0.0829, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 2.4526315789473687e-06, | |
| "loss": 0.0859, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 2.4263157894736845e-06, | |
| "loss": 0.0886, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 5.11, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 0.0844, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 5.13, | |
| "learning_rate": 2.373684210526316e-06, | |
| "loss": 0.0861, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 2.347368421052632e-06, | |
| "loss": 0.0761, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 5.16, | |
| "learning_rate": 2.3210526315789473e-06, | |
| "loss": 0.0894, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 2.294736842105263e-06, | |
| "loss": 0.0801, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 5.2, | |
| "learning_rate": 2.268421052631579e-06, | |
| "loss": 0.0826, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 2.242105263157895e-06, | |
| "loss": 0.0848, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 2.2157894736842106e-06, | |
| "loss": 0.0887, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 2.1894736842105264e-06, | |
| "loss": 0.0848, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 2.1631578947368423e-06, | |
| "loss": 0.0838, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 2.136842105263158e-06, | |
| "loss": 0.0907, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "learning_rate": 2.110526315789474e-06, | |
| "loss": 0.0916, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 5.29, | |
| "eval_loss": 0.24362976849079132, | |
| "eval_runtime": 278.4941, | |
| "eval_samples_per_second": 12.991, | |
| "eval_steps_per_second": 0.815, | |
| "eval_wer": 18.961542008669188, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 2.0842105263157897e-06, | |
| "loss": 0.0934, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 2.0578947368421055e-06, | |
| "loss": 0.0975, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 5.34, | |
| "learning_rate": 2.031578947368421e-06, | |
| "loss": 0.0842, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 2.005263157894737e-06, | |
| "loss": 0.0879, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 1.978947368421053e-06, | |
| "loss": 0.0806, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 1.9526315789473684e-06, | |
| "loss": 0.0842, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 1.9263157894736846e-06, | |
| "loss": 0.0802, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 5.43, | |
| "learning_rate": 1.9000000000000002e-06, | |
| "loss": 0.0879, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 1.8736842105263158e-06, | |
| "loss": 0.088, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 1.8473684210526319e-06, | |
| "loss": 0.0824, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 1.8210526315789475e-06, | |
| "loss": 0.0962, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 1.7947368421052633e-06, | |
| "loss": 0.0861, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 1.768421052631579e-06, | |
| "loss": 0.0817, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 1.742105263157895e-06, | |
| "loss": 0.1009, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 1.7157894736842107e-06, | |
| "loss": 0.0846, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 1.6894736842105264e-06, | |
| "loss": 0.0903, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 1.6631578947368424e-06, | |
| "loss": 0.0938, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 1.636842105263158e-06, | |
| "loss": 0.0869, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 5.61, | |
| "learning_rate": 1.6105263157894738e-06, | |
| "loss": 0.0856, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 1.5842105263157894e-06, | |
| "loss": 0.0922, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 5.64, | |
| "learning_rate": 1.5578947368421054e-06, | |
| "loss": 0.0815, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 1.5315789473684213e-06, | |
| "loss": 0.0847, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 1.5052631578947369e-06, | |
| "loss": 0.0802, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 1.478947368421053e-06, | |
| "loss": 0.074, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 5.71, | |
| "learning_rate": 1.4526315789473685e-06, | |
| "loss": 0.0836, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 1.4263157894736843e-06, | |
| "loss": 0.0858, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 0.0878, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 1.373684210526316e-06, | |
| "loss": 0.0924, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 1.3473684210526316e-06, | |
| "loss": 0.0941, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 1.3210526315789474e-06, | |
| "loss": 0.089, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 1.2947368421052634e-06, | |
| "loss": 0.0855, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 5.82, | |
| "learning_rate": 1.268421052631579e-06, | |
| "loss": 0.092, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 5.84, | |
| "learning_rate": 1.2421052631578948e-06, | |
| "loss": 0.0835, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 1.2157894736842107e-06, | |
| "loss": 0.0864, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 5.87, | |
| "learning_rate": 1.1894736842105265e-06, | |
| "loss": 0.093, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 1.163157894736842e-06, | |
| "loss": 0.0889, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 5.91, | |
| "learning_rate": 1.136842105263158e-06, | |
| "loss": 0.0862, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 1.110526315789474e-06, | |
| "loss": 0.0925, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 5.94, | |
| "learning_rate": 1.0842105263157895e-06, | |
| "loss": 0.0766, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 1.0578947368421054e-06, | |
| "loss": 0.0854, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "eval_loss": 0.2413194477558136, | |
| "eval_runtime": 270.4383, | |
| "eval_samples_per_second": 13.378, | |
| "eval_steps_per_second": 0.839, | |
| "eval_wer": 18.311352946601495, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 1.0315789473684212e-06, | |
| "loss": 0.0876, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 1.005263157894737e-06, | |
| "loss": 0.0907, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 9.789473684210526e-07, | |
| "loss": 0.0832, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 9.526315789473685e-07, | |
| "loss": 0.078, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 9.263157894736844e-07, | |
| "loss": 0.0821, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 9.000000000000001e-07, | |
| "loss": 0.0703, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 8.736842105263159e-07, | |
| "loss": 0.0795, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 8.473684210526316e-07, | |
| "loss": 0.0782, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 8.210526315789474e-07, | |
| "loss": 0.0839, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 6.12, | |
| "learning_rate": 7.947368421052632e-07, | |
| "loss": 0.0686, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 6.14, | |
| "learning_rate": 7.684210526315789e-07, | |
| "loss": 0.0888, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 7.421052631578949e-07, | |
| "loss": 0.0852, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 7.157894736842106e-07, | |
| "loss": 0.0696, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 6.19, | |
| "learning_rate": 6.894736842105264e-07, | |
| "loss": 0.0748, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 6.631578947368422e-07, | |
| "loss": 0.0716, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 6.368421052631579e-07, | |
| "loss": 0.0759, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 6.105263157894738e-07, | |
| "loss": 0.0744, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 5.842105263157896e-07, | |
| "loss": 0.0779, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 5.578947368421053e-07, | |
| "loss": 0.0699, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 5.315789473684211e-07, | |
| "loss": 0.0644, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 5.052631578947369e-07, | |
| "loss": 0.0726, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 4.789473684210526e-07, | |
| "loss": 0.0704, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 4.526315789473685e-07, | |
| "loss": 0.0717, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 4.2631578947368427e-07, | |
| "loss": 0.0714, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 0.0789, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 3.736842105263158e-07, | |
| "loss": 0.0699, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 3.4736842105263157e-07, | |
| "loss": 0.0795, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 6.42, | |
| "learning_rate": 3.210526315789474e-07, | |
| "loss": 0.0751, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 2.9473684210526315e-07, | |
| "loss": 0.0697, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 2.6842105263157897e-07, | |
| "loss": 0.0819, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 2.4210526315789473e-07, | |
| "loss": 0.0834, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 2.1578947368421053e-07, | |
| "loss": 0.0822, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 1.8947368421052634e-07, | |
| "loss": 0.0752, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 1.631578947368421e-07, | |
| "loss": 0.0733, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 1.368421052631579e-07, | |
| "loss": 0.0811, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "learning_rate": 1.1052631578947368e-07, | |
| "loss": 0.0745, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 8.421052631578947e-08, | |
| "loss": 0.0688, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 5.789473684210527e-08, | |
| "loss": 0.0801, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 3.157894736842106e-08, | |
| "loss": 0.0861, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 5.263157894736842e-09, | |
| "loss": 0.0812, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "eval_loss": 0.24088862538337708, | |
| "eval_runtime": 274.913, | |
| "eval_samples_per_second": 13.161, | |
| "eval_steps_per_second": 0.826, | |
| "eval_wer": 18.28368532693904, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "step": 10000, | |
| "total_flos": 7.87567982542848e+18, | |
| "train_loss": 0.19139070073366166, | |
| "train_runtime": 13859.2793, | |
| "train_samples_per_second": 23.089, | |
| "train_steps_per_second": 0.722 | |
| } | |
| ], | |
| "max_steps": 10000, | |
| "num_train_epochs": 7, | |
| "total_flos": 7.87567982542848e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |