{ "best_metric": 15.734731594103234, "best_model_checkpoint": "/scratch/project_465000370/asr_children/models/whisper-large-et-clinic/checkpoint-4000", "epoch": 32.25612806403202, "global_step": 4000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.2, "learning_rate": 1.2000000000000002e-06, "loss": 0.4973, "step": 25 }, { "epoch": 0.4, "learning_rate": 2.4500000000000003e-06, "loss": 0.2667, "step": 50 }, { "epoch": 0.6, "learning_rate": 3.7e-06, "loss": 0.2056, "step": 75 }, { "epoch": 0.8, "learning_rate": 4.95e-06, "loss": 0.2174, "step": 100 }, { "epoch": 1.01, "learning_rate": 6.200000000000001e-06, "loss": 0.198, "step": 125 }, { "epoch": 1.21, "learning_rate": 7.4e-06, "loss": 0.1203, "step": 150 }, { "epoch": 1.41, "learning_rate": 8.65e-06, "loss": 0.1191, "step": 175 }, { "epoch": 1.61, "learning_rate": 9.9e-06, "loss": 0.1183, "step": 200 }, { "epoch": 1.81, "learning_rate": 9.939473684210526e-06, "loss": 0.1172, "step": 225 }, { "epoch": 2.02, "learning_rate": 9.873684210526317e-06, "loss": 0.1264, "step": 250 }, { "epoch": 2.22, "learning_rate": 9.807894736842106e-06, "loss": 0.0478, "step": 275 }, { "epoch": 2.42, "learning_rate": 9.742105263157897e-06, "loss": 0.0544, "step": 300 }, { "epoch": 2.62, "learning_rate": 9.678947368421053e-06, "loss": 0.0605, "step": 325 }, { "epoch": 2.82, "learning_rate": 9.613157894736844e-06, "loss": 0.0575, "step": 350 }, { "epoch": 3.02, "learning_rate": 9.547368421052631e-06, "loss": 0.0494, "step": 375 }, { "epoch": 3.22, "learning_rate": 9.481578947368422e-06, "loss": 0.0262, "step": 400 }, { "epoch": 3.42, "learning_rate": 9.415789473684211e-06, "loss": 0.0237, "step": 425 }, { "epoch": 3.62, "learning_rate": 9.350000000000002e-06, "loss": 0.0293, "step": 450 }, { "epoch": 3.82, "learning_rate": 9.28421052631579e-06, "loss": 0.0284, "step": 475 }, { "epoch": 4.03, "learning_rate": 9.21842105263158e-06, "loss": 0.0297, "step": 500 }, { "epoch": 4.03, "eval_loss": 0.3037151098251343, "eval_runtime": 2297.7169, "eval_samples_per_second": 0.56, "eval_steps_per_second": 0.56, "eval_wer": 16.71466024842051, "step": 500 }, { "epoch": 4.23, "learning_rate": 9.15263157894737e-06, "loss": 0.0151, "step": 525 }, { "epoch": 4.43, "learning_rate": 9.086842105263158e-06, "loss": 0.0161, "step": 550 }, { "epoch": 4.63, "learning_rate": 9.021052631578948e-06, "loss": 0.0165, "step": 575 }, { "epoch": 4.83, "learning_rate": 8.955263157894738e-06, "loss": 0.017, "step": 600 }, { "epoch": 5.04, "learning_rate": 8.889473684210528e-06, "loss": 0.0154, "step": 625 }, { "epoch": 5.24, "learning_rate": 8.823684210526317e-06, "loss": 0.0093, "step": 650 }, { "epoch": 5.44, "learning_rate": 8.757894736842106e-06, "loss": 0.0109, "step": 675 }, { "epoch": 5.64, "learning_rate": 8.692105263157895e-06, "loss": 0.009, "step": 700 }, { "epoch": 5.84, "learning_rate": 8.626315789473686e-06, "loss": 0.0109, "step": 725 }, { "epoch": 6.05, "learning_rate": 8.560526315789475e-06, "loss": 0.0105, "step": 750 }, { "epoch": 6.25, "learning_rate": 8.494736842105264e-06, "loss": 0.0077, "step": 775 }, { "epoch": 6.45, "learning_rate": 8.428947368421053e-06, "loss": 0.0063, "step": 800 }, { "epoch": 6.65, "learning_rate": 8.363157894736842e-06, "loss": 0.0093, "step": 825 }, { "epoch": 6.85, "learning_rate": 8.297368421052631e-06, "loss": 0.0072, "step": 850 }, { "epoch": 7.06, "learning_rate": 8.231578947368422e-06, "loss": 0.0081, "step": 875 }, { "epoch": 7.26, "learning_rate": 8.165789473684211e-06, "loss": 0.0047, "step": 900 }, { "epoch": 7.46, "learning_rate": 8.1e-06, "loss": 0.0058, "step": 925 }, { "epoch": 7.66, "learning_rate": 8.03421052631579e-06, "loss": 0.0051, "step": 950 }, { "epoch": 7.86, "learning_rate": 7.96842105263158e-06, "loss": 0.0056, "step": 975 }, { "epoch": 8.06, "learning_rate": 7.90263157894737e-06, "loss": 0.0057, "step": 1000 }, { "epoch": 8.06, "eval_loss": 0.3434995114803314, "eval_runtime": 2295.9725, "eval_samples_per_second": 0.56, "eval_steps_per_second": 0.56, "eval_wer": 15.949628228821936, "step": 1000 }, { "epoch": 8.26, "learning_rate": 7.836842105263159e-06, "loss": 0.0031, "step": 1025 }, { "epoch": 8.46, "learning_rate": 7.771052631578948e-06, "loss": 0.0055, "step": 1050 }, { "epoch": 8.66, "learning_rate": 7.705263157894738e-06, "loss": 0.0056, "step": 1075 }, { "epoch": 8.86, "learning_rate": 7.639473684210526e-06, "loss": 0.0049, "step": 1100 }, { "epoch": 9.07, "learning_rate": 7.573684210526317e-06, "loss": 0.004, "step": 1125 }, { "epoch": 9.27, "learning_rate": 7.507894736842106e-06, "loss": 0.0036, "step": 1150 }, { "epoch": 9.47, "learning_rate": 7.442105263157895e-06, "loss": 0.0028, "step": 1175 }, { "epoch": 9.67, "learning_rate": 7.376315789473685e-06, "loss": 0.0034, "step": 1200 }, { "epoch": 9.87, "learning_rate": 7.310526315789475e-06, "loss": 0.0027, "step": 1225 }, { "epoch": 10.08, "learning_rate": 7.244736842105263e-06, "loss": 0.0027, "step": 1250 }, { "epoch": 10.28, "learning_rate": 7.178947368421053e-06, "loss": 0.0022, "step": 1275 }, { "epoch": 10.48, "learning_rate": 7.113157894736843e-06, "loss": 0.0033, "step": 1300 }, { "epoch": 10.68, "learning_rate": 7.047368421052631e-06, "loss": 0.0033, "step": 1325 }, { "epoch": 10.88, "learning_rate": 6.981578947368421e-06, "loss": 0.005, "step": 1350 }, { "epoch": 11.09, "learning_rate": 6.915789473684211e-06, "loss": 0.0034, "step": 1375 }, { "epoch": 11.29, "learning_rate": 6.850000000000001e-06, "loss": 0.0035, "step": 1400 }, { "epoch": 11.49, "learning_rate": 6.78421052631579e-06, "loss": 0.0033, "step": 1425 }, { "epoch": 11.69, "learning_rate": 6.71842105263158e-06, "loss": 0.0041, "step": 1450 }, { "epoch": 11.89, "learning_rate": 6.6526315789473695e-06, "loss": 0.0047, "step": 1475 }, { "epoch": 12.1, "learning_rate": 6.586842105263158e-06, "loss": 0.0049, "step": 1500 }, { "epoch": 12.1, "eval_loss": 0.3610152006149292, "eval_runtime": 2310.3916, "eval_samples_per_second": 0.557, "eval_steps_per_second": 0.557, "eval_wer": 16.87368375811235, "step": 1500 }, { "epoch": 12.3, "learning_rate": 6.521052631578948e-06, "loss": 0.0035, "step": 1525 }, { "epoch": 12.5, "learning_rate": 6.455263157894738e-06, "loss": 0.0036, "step": 1550 }, { "epoch": 12.7, "learning_rate": 6.389473684210527e-06, "loss": 0.0038, "step": 1575 }, { "epoch": 12.9, "learning_rate": 6.323684210526316e-06, "loss": 0.0033, "step": 1600 }, { "epoch": 13.1, "learning_rate": 6.257894736842106e-06, "loss": 0.0028, "step": 1625 }, { "epoch": 13.3, "learning_rate": 6.192105263157895e-06, "loss": 0.0029, "step": 1650 }, { "epoch": 13.5, "learning_rate": 6.126315789473685e-06, "loss": 0.0033, "step": 1675 }, { "epoch": 13.7, "learning_rate": 6.060526315789474e-06, "loss": 0.003, "step": 1700 }, { "epoch": 13.9, "learning_rate": 5.994736842105263e-06, "loss": 0.0036, "step": 1725 }, { "epoch": 14.11, "learning_rate": 5.928947368421053e-06, "loss": 0.0029, "step": 1750 }, { "epoch": 14.31, "learning_rate": 5.863157894736842e-06, "loss": 0.0027, "step": 1775 }, { "epoch": 14.51, "learning_rate": 5.7973684210526315e-06, "loss": 0.0023, "step": 1800 }, { "epoch": 14.71, "learning_rate": 5.7315789473684215e-06, "loss": 0.0023, "step": 1825 }, { "epoch": 14.91, "learning_rate": 5.6657894736842115e-06, "loss": 0.0033, "step": 1850 }, { "epoch": 15.12, "learning_rate": 5.600000000000001e-06, "loss": 0.0033, "step": 1875 }, { "epoch": 15.32, "learning_rate": 5.53421052631579e-06, "loss": 0.0037, "step": 1900 }, { "epoch": 15.52, "learning_rate": 5.46842105263158e-06, "loss": 0.0035, "step": 1925 }, { "epoch": 15.72, "learning_rate": 5.40263157894737e-06, "loss": 0.0037, "step": 1950 }, { "epoch": 15.92, "learning_rate": 5.336842105263158e-06, "loss": 0.0024, "step": 1975 }, { "epoch": 16.13, "learning_rate": 5.271052631578948e-06, "loss": 0.0015, "step": 2000 }, { "epoch": 16.13, "eval_loss": 0.36096739768981934, "eval_runtime": 2288.7771, "eval_samples_per_second": 0.562, "eval_steps_per_second": 0.562, "eval_wer": 16.125843469291272, "step": 2000 }, { "epoch": 16.33, "learning_rate": 5.205263157894738e-06, "loss": 0.0017, "step": 2025 }, { "epoch": 16.53, "learning_rate": 5.139473684210526e-06, "loss": 0.0017, "step": 2050 }, { "epoch": 16.73, "learning_rate": 5.073684210526316e-06, "loss": 0.0017, "step": 2075 }, { "epoch": 16.93, "learning_rate": 5.007894736842106e-06, "loss": 0.003, "step": 2100 }, { "epoch": 17.14, "learning_rate": 4.942105263157895e-06, "loss": 0.0017, "step": 2125 }, { "epoch": 17.34, "learning_rate": 4.876315789473684e-06, "loss": 0.0014, "step": 2150 }, { "epoch": 17.54, "learning_rate": 4.8105263157894735e-06, "loss": 0.0022, "step": 2175 }, { "epoch": 17.74, "learning_rate": 4.7447368421052634e-06, "loss": 0.0023, "step": 2200 }, { "epoch": 17.94, "learning_rate": 4.6789473684210525e-06, "loss": 0.0026, "step": 2225 }, { "epoch": 18.14, "learning_rate": 4.6131578947368425e-06, "loss": 0.0013, "step": 2250 }, { "epoch": 18.34, "learning_rate": 4.547368421052632e-06, "loss": 0.0012, "step": 2275 }, { "epoch": 18.54, "learning_rate": 4.481578947368422e-06, "loss": 0.0014, "step": 2300 }, { "epoch": 18.74, "learning_rate": 4.415789473684211e-06, "loss": 0.0017, "step": 2325 }, { "epoch": 18.94, "learning_rate": 4.350000000000001e-06, "loss": 0.002, "step": 2350 }, { "epoch": 19.15, "learning_rate": 4.28421052631579e-06, "loss": 0.0015, "step": 2375 }, { "epoch": 19.35, "learning_rate": 4.21842105263158e-06, "loss": 0.0016, "step": 2400 }, { "epoch": 19.55, "learning_rate": 4.152631578947369e-06, "loss": 0.0013, "step": 2425 }, { "epoch": 19.75, "learning_rate": 4.086842105263159e-06, "loss": 0.0014, "step": 2450 }, { "epoch": 19.95, "learning_rate": 4.021052631578948e-06, "loss": 0.0011, "step": 2475 }, { "epoch": 20.16, "learning_rate": 3.955263157894737e-06, "loss": 0.001, "step": 2500 }, { "epoch": 20.16, "eval_loss": 0.38679206371307373, "eval_runtime": 2297.8177, "eval_samples_per_second": 0.56, "eval_steps_per_second": 0.56, "eval_wer": 16.035586882709417, "step": 2500 }, { "epoch": 20.36, "learning_rate": 3.889473684210527e-06, "loss": 0.0013, "step": 2525 }, { "epoch": 20.56, "learning_rate": 3.823684210526316e-06, "loss": 0.0012, "step": 2550 }, { "epoch": 20.76, "learning_rate": 3.7578947368421053e-06, "loss": 0.001, "step": 2575 }, { "epoch": 20.96, "learning_rate": 3.6921052631578953e-06, "loss": 0.0012, "step": 2600 }, { "epoch": 21.17, "learning_rate": 3.6263157894736844e-06, "loss": 0.0013, "step": 2625 }, { "epoch": 21.37, "learning_rate": 3.5605263157894736e-06, "loss": 0.0008, "step": 2650 }, { "epoch": 21.57, "learning_rate": 3.4947368421052635e-06, "loss": 0.0013, "step": 2675 }, { "epoch": 21.77, "learning_rate": 3.4289473684210527e-06, "loss": 0.001, "step": 2700 }, { "epoch": 21.97, "learning_rate": 3.3631578947368426e-06, "loss": 0.0014, "step": 2725 }, { "epoch": 22.18, "learning_rate": 3.2973684210526318e-06, "loss": 0.0012, "step": 2750 }, { "epoch": 22.38, "learning_rate": 3.2315789473684213e-06, "loss": 0.0008, "step": 2775 }, { "epoch": 22.58, "learning_rate": 3.165789473684211e-06, "loss": 0.001, "step": 2800 }, { "epoch": 22.78, "learning_rate": 3.1000000000000004e-06, "loss": 0.0012, "step": 2825 }, { "epoch": 22.98, "learning_rate": 3.0342105263157895e-06, "loss": 0.001, "step": 2850 }, { "epoch": 23.18, "learning_rate": 2.9684210526315795e-06, "loss": 0.0008, "step": 2875 }, { "epoch": 23.38, "learning_rate": 2.9026315789473686e-06, "loss": 0.0007, "step": 2900 }, { "epoch": 23.58, "learning_rate": 2.8368421052631586e-06, "loss": 0.0011, "step": 2925 }, { "epoch": 23.78, "learning_rate": 2.7710526315789477e-06, "loss": 0.0015, "step": 2950 }, { "epoch": 23.98, "learning_rate": 2.705263157894737e-06, "loss": 0.0011, "step": 2975 }, { "epoch": 24.19, "learning_rate": 2.639473684210527e-06, "loss": 0.0011, "step": 3000 }, { "epoch": 24.19, "eval_loss": 0.394901841878891, "eval_runtime": 2297.2378, "eval_samples_per_second": 0.56, "eval_steps_per_second": 0.56, "eval_wer": 15.829286113379466, "step": 3000 }, { "epoch": 24.39, "learning_rate": 2.573684210526316e-06, "loss": 0.0007, "step": 3025 }, { "epoch": 24.59, "learning_rate": 2.5078947368421055e-06, "loss": 0.0012, "step": 3050 }, { "epoch": 24.79, "learning_rate": 2.442105263157895e-06, "loss": 0.0011, "step": 3075 }, { "epoch": 24.99, "learning_rate": 2.3763157894736846e-06, "loss": 0.0009, "step": 3100 }, { "epoch": 25.2, "learning_rate": 2.310526315789474e-06, "loss": 0.0009, "step": 3125 }, { "epoch": 25.4, "learning_rate": 2.2447368421052636e-06, "loss": 0.0008, "step": 3150 }, { "epoch": 25.6, "learning_rate": 2.1789473684210528e-06, "loss": 0.0009, "step": 3175 }, { "epoch": 25.8, "learning_rate": 2.1131578947368423e-06, "loss": 0.0009, "step": 3200 }, { "epoch": 26.01, "learning_rate": 2.047368421052632e-06, "loss": 0.0011, "step": 3225 }, { "epoch": 26.21, "learning_rate": 1.9815789473684214e-06, "loss": 0.0007, "step": 3250 }, { "epoch": 26.41, "learning_rate": 1.9157894736842105e-06, "loss": 0.001, "step": 3275 }, { "epoch": 26.61, "learning_rate": 1.85e-06, "loss": 0.0008, "step": 3300 }, { "epoch": 26.81, "learning_rate": 1.7842105263157896e-06, "loss": 0.0008, "step": 3325 }, { "epoch": 27.02, "learning_rate": 1.7184210526315792e-06, "loss": 0.0012, "step": 3350 }, { "epoch": 27.22, "learning_rate": 1.6526315789473685e-06, "loss": 0.0009, "step": 3375 }, { "epoch": 27.42, "learning_rate": 1.586842105263158e-06, "loss": 0.0009, "step": 3400 }, { "epoch": 27.62, "learning_rate": 1.5210526315789476e-06, "loss": 0.0009, "step": 3425 }, { "epoch": 27.82, "learning_rate": 1.4552631578947371e-06, "loss": 0.0009, "step": 3450 }, { "epoch": 28.02, "learning_rate": 1.3894736842105263e-06, "loss": 0.001, "step": 3475 }, { "epoch": 28.22, "learning_rate": 1.3236842105263158e-06, "loss": 0.0006, "step": 3500 }, { "epoch": 28.22, "eval_loss": 0.4028264284133911, "eval_runtime": 2295.0076, "eval_samples_per_second": 0.56, "eval_steps_per_second": 0.56, "eval_wer": 15.74762539218636, "step": 3500 }, { "epoch": 28.42, "learning_rate": 1.2578947368421054e-06, "loss": 0.0005, "step": 3525 }, { "epoch": 28.62, "learning_rate": 1.192105263157895e-06, "loss": 0.0006, "step": 3550 }, { "epoch": 28.82, "learning_rate": 1.1263157894736842e-06, "loss": 0.0011, "step": 3575 }, { "epoch": 29.03, "learning_rate": 1.0605263157894738e-06, "loss": 0.0012, "step": 3600 }, { "epoch": 29.23, "learning_rate": 9.947368421052631e-07, "loss": 0.0006, "step": 3625 }, { "epoch": 29.43, "learning_rate": 9.289473684210528e-07, "loss": 0.0008, "step": 3650 }, { "epoch": 29.63, "learning_rate": 8.631578947368421e-07, "loss": 0.0009, "step": 3675 }, { "epoch": 29.83, "learning_rate": 7.973684210526317e-07, "loss": 0.0008, "step": 3700 }, { "epoch": 30.04, "learning_rate": 7.315789473684211e-07, "loss": 0.0008, "step": 3725 }, { "epoch": 30.24, "learning_rate": 6.657894736842106e-07, "loss": 0.0007, "step": 3750 }, { "epoch": 30.44, "learning_rate": 6.000000000000001e-07, "loss": 0.0008, "step": 3775 }, { "epoch": 30.64, "learning_rate": 5.342105263157895e-07, "loss": 0.0009, "step": 3800 }, { "epoch": 30.84, "learning_rate": 4.6842105263157896e-07, "loss": 0.0008, "step": 3825 }, { "epoch": 31.05, "learning_rate": 4.0263157894736845e-07, "loss": 0.0007, "step": 3850 }, { "epoch": 31.25, "learning_rate": 3.368421052631579e-07, "loss": 0.0006, "step": 3875 }, { "epoch": 31.45, "learning_rate": 2.710526315789474e-07, "loss": 0.0008, "step": 3900 }, { "epoch": 31.65, "learning_rate": 2.0526315789473685e-07, "loss": 0.0007, "step": 3925 }, { "epoch": 31.85, "learning_rate": 1.3947368421052632e-07, "loss": 0.001, "step": 3950 }, { "epoch": 32.06, "learning_rate": 7.368421052631579e-08, "loss": 0.0006, "step": 3975 }, { "epoch": 32.26, "learning_rate": 7.894736842105264e-09, "loss": 0.0008, "step": 4000 }, { "epoch": 32.26, "eval_loss": 0.4131947159767151, "eval_runtime": 2384.3215, "eval_samples_per_second": 0.539, "eval_steps_per_second": 0.539, "eval_wer": 15.734731594103234, "step": 4000 }, { "epoch": 32.26, "step": 4000, "total_flos": 2.73804134842368e+20, "train_loss": 0.017603627420263366, "train_runtime": 74040.5343, "train_samples_per_second": 1.729, "train_steps_per_second": 0.054 } ], "max_steps": 4000, "num_train_epochs": 33, "total_flos": 2.73804134842368e+20, "trial_name": null, "trial_params": null }