| { | |
| "best_metric": 23.35348393254852, | |
| "best_model_checkpoint": "whisper3/checkpoint-240", | |
| "epoch": 8.333333333333334, | |
| "eval_steps": 10, | |
| "global_step": 300, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.1388888888888889, | |
| "grad_norm": 46.06148147583008, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 3.9402, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "grad_norm": 43.4765625, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 3.8281, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.2777777777777778, | |
| "eval_loss": 3.7929115295410156, | |
| "eval_runtime": 253.0403, | |
| "eval_samples_per_second": 1.976, | |
| "eval_steps_per_second": 0.249, | |
| "eval_wer": 80.40089086859689, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.4166666666666667, | |
| "grad_norm": 40.57815933227539, | |
| "learning_rate": 3e-06, | |
| "loss": 3.5929, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "grad_norm": 39.72583770751953, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 3.209, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.5555555555555556, | |
| "eval_loss": 3.0014312267303467, | |
| "eval_runtime": 246.2101, | |
| "eval_samples_per_second": 2.031, | |
| "eval_steps_per_second": 0.256, | |
| "eval_wer": 68.37416481069042, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.6944444444444444, | |
| "grad_norm": 39.53627395629883, | |
| "learning_rate": 5e-06, | |
| "loss": 2.7486, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 30.079750061035156, | |
| "learning_rate": 6e-06, | |
| "loss": 2.1066, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "eval_loss": 1.761271595954895, | |
| "eval_runtime": 245.5315, | |
| "eval_samples_per_second": 2.036, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 63.91982182628062, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.9722222222222222, | |
| "grad_norm": 19.831071853637695, | |
| "learning_rate": 7.000000000000001e-06, | |
| "loss": 1.5134, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "grad_norm": 9.755999565124512, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.9963, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.1111111111111112, | |
| "eval_loss": 0.8740884065628052, | |
| "eval_runtime": 246.6146, | |
| "eval_samples_per_second": 2.027, | |
| "eval_steps_per_second": 0.255, | |
| "eval_wer": 52.43398027362392, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 6.842897891998291, | |
| "learning_rate": 9e-06, | |
| "loss": 0.786, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "grad_norm": 5.720729351043701, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6922, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.3888888888888888, | |
| "eval_loss": 0.7008740901947021, | |
| "eval_runtime": 245.5713, | |
| "eval_samples_per_second": 2.036, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 35.82564428889596, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 1.5277777777777777, | |
| "grad_norm": 4.806775093078613, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 0.6427, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "grad_norm": 5.128376483917236, | |
| "learning_rate": 1.2e-05, | |
| "loss": 0.5816, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.6666666666666665, | |
| "eval_loss": 0.6238442659378052, | |
| "eval_runtime": 245.679, | |
| "eval_samples_per_second": 2.035, | |
| "eval_steps_per_second": 0.256, | |
| "eval_wer": 31.148584155265667, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.8055555555555556, | |
| "grad_norm": 4.993675231933594, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 0.5805, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 1.9444444444444444, | |
| "grad_norm": 4.856825351715088, | |
| "learning_rate": 1.4000000000000001e-05, | |
| "loss": 0.5684, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.9444444444444444, | |
| "eval_loss": 0.5697694420814514, | |
| "eval_runtime": 245.5413, | |
| "eval_samples_per_second": 2.036, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 35.47566019726376, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 2.0833333333333335, | |
| "grad_norm": 4.464582443237305, | |
| "learning_rate": 1.5e-05, | |
| "loss": 0.4534, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 4.251033306121826, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 0.427, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "eval_loss": 0.5380394458770752, | |
| "eval_runtime": 244.4819, | |
| "eval_samples_per_second": 2.045, | |
| "eval_steps_per_second": 0.258, | |
| "eval_wer": 27.266942411708563, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 2.361111111111111, | |
| "grad_norm": 4.489510536193848, | |
| "learning_rate": 1.7000000000000003e-05, | |
| "loss": 0.3929, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 4.552371025085449, | |
| "learning_rate": 1.8e-05, | |
| "loss": 0.4395, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "eval_loss": 0.5162410140037537, | |
| "eval_runtime": 245.2373, | |
| "eval_samples_per_second": 2.039, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 32.73942093541203, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 2.638888888888889, | |
| "grad_norm": 4.691618919372559, | |
| "learning_rate": 1.9e-05, | |
| "loss": 0.3825, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "grad_norm": 4.219367027282715, | |
| "learning_rate": 2e-05, | |
| "loss": 0.3861, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.7777777777777777, | |
| "eval_loss": 0.495292991399765, | |
| "eval_runtime": 243.4193, | |
| "eval_samples_per_second": 2.054, | |
| "eval_steps_per_second": 0.259, | |
| "eval_wer": 24.530703149856826, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 2.9166666666666665, | |
| "grad_norm": 4.323045253753662, | |
| "learning_rate": 2.1e-05, | |
| "loss": 0.3669, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 3.0555555555555554, | |
| "grad_norm": 3.2159509658813477, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 0.3745, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 3.0555555555555554, | |
| "eval_loss": 0.4837464392185211, | |
| "eval_runtime": 244.5759, | |
| "eval_samples_per_second": 2.044, | |
| "eval_steps_per_second": 0.258, | |
| "eval_wer": 24.626153356665608, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 3.1944444444444446, | |
| "grad_norm": 3.675457000732422, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 0.257, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "grad_norm": 2.8939876556396484, | |
| "learning_rate": 2.4e-05, | |
| "loss": 0.2487, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 3.3333333333333335, | |
| "eval_loss": 0.4732927978038788, | |
| "eval_runtime": 244.6891, | |
| "eval_samples_per_second": 2.043, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 23.57620108176901, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 3.4722222222222223, | |
| "grad_norm": 3.4589827060699463, | |
| "learning_rate": 2.5e-05, | |
| "loss": 0.253, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 3.611111111111111, | |
| "grad_norm": 3.1798577308654785, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 0.2343, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 3.611111111111111, | |
| "eval_loss": 0.46519017219543457, | |
| "eval_runtime": 244.3925, | |
| "eval_samples_per_second": 2.046, | |
| "eval_steps_per_second": 0.258, | |
| "eval_wer": 24.94432071269488, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 4.061887741088867, | |
| "learning_rate": 2.7000000000000002e-05, | |
| "loss": 0.2354, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 3.888888888888889, | |
| "grad_norm": 4.474591255187988, | |
| "learning_rate": 2.8000000000000003e-05, | |
| "loss": 0.2429, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 3.888888888888889, | |
| "eval_loss": 0.4581267833709717, | |
| "eval_runtime": 244.836, | |
| "eval_samples_per_second": 2.042, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 24.085268851415844, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 4.027777777777778, | |
| "grad_norm": 2.3235318660736084, | |
| "learning_rate": 2.9e-05, | |
| "loss": 0.2728, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 4.166666666666667, | |
| "grad_norm": 2.3824808597564697, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1286, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 4.166666666666667, | |
| "eval_loss": 0.46725359559059143, | |
| "eval_runtime": 245.6982, | |
| "eval_samples_per_second": 2.035, | |
| "eval_steps_per_second": 0.256, | |
| "eval_wer": 24.276169265033406, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 4.305555555555555, | |
| "grad_norm": 2.5686404705047607, | |
| "learning_rate": 3.1e-05, | |
| "loss": 0.1301, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 4.444444444444445, | |
| "grad_norm": 2.7436068058013916, | |
| "learning_rate": 3.2000000000000005e-05, | |
| "loss": 0.1304, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 4.444444444444445, | |
| "eval_loss": 0.46984970569610596, | |
| "eval_runtime": 245.0991, | |
| "eval_samples_per_second": 2.04, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 31.72128539611836, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 4.583333333333333, | |
| "grad_norm": 2.83823823928833, | |
| "learning_rate": 3.3e-05, | |
| "loss": 0.1408, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 4.722222222222222, | |
| "grad_norm": 2.7204811573028564, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 0.1361, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 4.722222222222222, | |
| "eval_loss": 0.4690161943435669, | |
| "eval_runtime": 246.5232, | |
| "eval_samples_per_second": 2.028, | |
| "eval_steps_per_second": 0.256, | |
| "eval_wer": 33.08940502704423, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 4.861111111111111, | |
| "grad_norm": 3.671097993850708, | |
| "learning_rate": 3.5e-05, | |
| "loss": 0.1511, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "grad_norm": 6.484060764312744, | |
| "learning_rate": 3.6e-05, | |
| "loss": 0.1447, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 0.4811546802520752, | |
| "eval_runtime": 244.9356, | |
| "eval_samples_per_second": 2.041, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 24.657970092268535, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 5.138888888888889, | |
| "grad_norm": 1.9667352437973022, | |
| "learning_rate": 3.7e-05, | |
| "loss": 0.063, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 5.277777777777778, | |
| "grad_norm": 2.1828482151031494, | |
| "learning_rate": 3.8e-05, | |
| "loss": 0.0617, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 5.277777777777778, | |
| "eval_loss": 0.48713362216949463, | |
| "eval_runtime": 244.9851, | |
| "eval_samples_per_second": 2.041, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 29.939548202354437, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 5.416666666666667, | |
| "grad_norm": 1.8774911165237427, | |
| "learning_rate": 3.9000000000000006e-05, | |
| "loss": 0.0606, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 5.555555555555555, | |
| "grad_norm": 1.8562583923339844, | |
| "learning_rate": 4e-05, | |
| "loss": 0.0617, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 5.555555555555555, | |
| "eval_loss": 0.488438218832016, | |
| "eval_runtime": 244.9014, | |
| "eval_samples_per_second": 2.042, | |
| "eval_steps_per_second": 0.257, | |
| "eval_wer": 24.848870505886094, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 5.694444444444445, | |
| "grad_norm": 1.9106348752975464, | |
| "learning_rate": 4.1e-05, | |
| "loss": 0.0617, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 5.833333333333333, | |
| "grad_norm": 1.8114972114562988, | |
| "learning_rate": 4.2e-05, | |
| "loss": 0.0577, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 5.833333333333333, | |
| "eval_loss": 0.4998014569282532, | |
| "eval_runtime": 244.1029, | |
| "eval_samples_per_second": 2.048, | |
| "eval_steps_per_second": 0.258, | |
| "eval_wer": 26.853324848870507, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 5.972222222222222, | |
| "grad_norm": 5.00437593460083, | |
| "learning_rate": 4.3e-05, | |
| "loss": 0.078, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 6.111111111111111, | |
| "grad_norm": 1.4013047218322754, | |
| "learning_rate": 4.4000000000000006e-05, | |
| "loss": 0.038, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 6.111111111111111, | |
| "eval_loss": 0.500673770904541, | |
| "eval_runtime": 247.5538, | |
| "eval_samples_per_second": 2.02, | |
| "eval_steps_per_second": 0.254, | |
| "eval_wer": 24.848870505886094, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "grad_norm": 1.4778488874435425, | |
| "learning_rate": 4.5e-05, | |
| "loss": 0.0243, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 6.388888888888889, | |
| "grad_norm": 1.3681198358535767, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 0.0269, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 6.388888888888889, | |
| "eval_loss": 0.5122880935668945, | |
| "eval_runtime": 243.6648, | |
| "eval_samples_per_second": 2.052, | |
| "eval_steps_per_second": 0.259, | |
| "eval_wer": 27.139675469296847, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 6.527777777777778, | |
| "grad_norm": 1.450726866722107, | |
| "learning_rate": 4.7e-05, | |
| "loss": 0.0297, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 6.666666666666667, | |
| "grad_norm": 1.4052125215530396, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.0321, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 6.666666666666667, | |
| "eval_loss": 0.500522792339325, | |
| "eval_runtime": 247.602, | |
| "eval_samples_per_second": 2.019, | |
| "eval_steps_per_second": 0.254, | |
| "eval_wer": 23.35348393254852, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 6.805555555555555, | |
| "grad_norm": 1.2223644256591797, | |
| "learning_rate": 4.9e-05, | |
| "loss": 0.0291, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 6.944444444444445, | |
| "grad_norm": 1.463398814201355, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0296, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 6.944444444444445, | |
| "eval_loss": 0.5332342386245728, | |
| "eval_runtime": 246.3422, | |
| "eval_samples_per_second": 2.03, | |
| "eval_steps_per_second": 0.256, | |
| "eval_wer": 31.880369074132993, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 7.083333333333333, | |
| "grad_norm": 4.257472991943359, | |
| "learning_rate": 5.1000000000000006e-05, | |
| "loss": 0.027, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 7.222222222222222, | |
| "grad_norm": 2.294562339782715, | |
| "learning_rate": 5.2000000000000004e-05, | |
| "loss": 0.0207, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 7.222222222222222, | |
| "eval_loss": 0.5236981511116028, | |
| "eval_runtime": 244.1894, | |
| "eval_samples_per_second": 2.048, | |
| "eval_steps_per_second": 0.258, | |
| "eval_wer": 30.066815144766146, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 7.361111111111111, | |
| "grad_norm": 1.2468712329864502, | |
| "learning_rate": 5.300000000000001e-05, | |
| "loss": 0.0228, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "grad_norm": 1.8487240076065063, | |
| "learning_rate": 5.4000000000000005e-05, | |
| "loss": 0.0215, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "eval_loss": 0.5222529768943787, | |
| "eval_runtime": 243.6778, | |
| "eval_samples_per_second": 2.052, | |
| "eval_steps_per_second": 0.259, | |
| "eval_wer": 25.548838689150493, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 7.638888888888889, | |
| "grad_norm": 1.1909741163253784, | |
| "learning_rate": 5.500000000000001e-05, | |
| "loss": 0.0201, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 7.777777777777778, | |
| "grad_norm": 1.6141778230667114, | |
| "learning_rate": 5.6000000000000006e-05, | |
| "loss": 0.0198, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 7.777777777777778, | |
| "eval_loss": 0.5157026648521423, | |
| "eval_runtime": 244.0734, | |
| "eval_samples_per_second": 2.049, | |
| "eval_steps_per_second": 0.258, | |
| "eval_wer": 30.194082087177854, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 7.916666666666667, | |
| "grad_norm": 1.1372332572937012, | |
| "learning_rate": 5.6999999999999996e-05, | |
| "loss": 0.0193, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 8.055555555555555, | |
| "grad_norm": 2.210016965866089, | |
| "learning_rate": 5.8e-05, | |
| "loss": 0.0273, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 8.055555555555555, | |
| "eval_loss": 0.5289562940597534, | |
| "eval_runtime": 243.9152, | |
| "eval_samples_per_second": 2.05, | |
| "eval_steps_per_second": 0.258, | |
| "eval_wer": 27.553293032134903, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 8.194444444444445, | |
| "grad_norm": 1.942575454711914, | |
| "learning_rate": 5.9e-05, | |
| "loss": 0.0201, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "grad_norm": 1.3640440702438354, | |
| "learning_rate": 6e-05, | |
| "loss": 0.0197, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "eval_loss": 0.5509196519851685, | |
| "eval_runtime": 243.9508, | |
| "eval_samples_per_second": 2.05, | |
| "eval_steps_per_second": 0.258, | |
| "eval_wer": 26.948775055679285, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 8.333333333333334, | |
| "step": 300, | |
| "total_flos": 9.2409447186432e+17, | |
| "train_loss": 0.5431244759509961, | |
| "train_runtime": 10016.0212, | |
| "train_samples_per_second": 3.834, | |
| "train_steps_per_second": 0.03 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 300, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9, | |
| "save_steps": 10, | |
| "total_flos": 9.2409447186432e+17, | |
| "train_batch_size": 128, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |