| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 20.0, | |
| "eval_steps": 500, | |
| "global_step": 3280, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 9.6805, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 9.3474, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3e-06, | |
| "loss": 8.6058, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_cer": 0.986131055853677, | |
| "eval_loss": 6.288315296173096, | |
| "eval_runtime": 21.6292, | |
| "eval_samples_per_second": 121.318, | |
| "eval_steps_per_second": 7.582, | |
| "eval_wer": 0.9742561721882254, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 6.5168, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 5e-06, | |
| "loss": 5.178, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 6e-06, | |
| "loss": 4.6616, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_cer": 0.986131055853677, | |
| "eval_loss": 4.045103549957275, | |
| "eval_runtime": 21.856, | |
| "eval_samples_per_second": 120.059, | |
| "eval_steps_per_second": 7.504, | |
| "eval_wer": 0.9742561721882254, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 7e-06, | |
| "loss": 4.2963, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 4.0823, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 9e-06, | |
| "loss": 3.8526, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_cer": 0.986131055853677, | |
| "eval_loss": 3.5416758060455322, | |
| "eval_runtime": 21.8041, | |
| "eval_samples_per_second": 120.344, | |
| "eval_steps_per_second": 7.522, | |
| "eval_wer": 0.9742561721882254, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1e-05, | |
| "loss": 3.7408, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.1000000000000001e-05, | |
| "loss": 3.5235, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 1.2e-05, | |
| "loss": 3.3617, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 1.3000000000000001e-05, | |
| "loss": 3.2384, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_cer": 0.986131055853677, | |
| "eval_loss": 3.0504775047302246, | |
| "eval_runtime": 22.0029, | |
| "eval_samples_per_second": 119.257, | |
| "eval_steps_per_second": 7.454, | |
| "eval_wer": 0.9742561721882254, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 1.4e-05, | |
| "loss": 3.0665, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 1.5000000000000002e-05, | |
| "loss": 2.9046, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 1.6000000000000003e-05, | |
| "loss": 2.7948, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_cer": 0.986131055853677, | |
| "eval_loss": 2.670631170272827, | |
| "eval_runtime": 21.7032, | |
| "eval_samples_per_second": 120.904, | |
| "eval_steps_per_second": 7.556, | |
| "eval_wer": 0.9742561721882254, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 1.7e-05, | |
| "loss": 2.6959, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 1.8e-05, | |
| "loss": 2.6047, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 1.9e-05, | |
| "loss": 2.549, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_cer": 0.986131055853677, | |
| "eval_loss": 2.426814317703247, | |
| "eval_runtime": 21.9773, | |
| "eval_samples_per_second": 119.396, | |
| "eval_steps_per_second": 7.462, | |
| "eval_wer": 0.9742561721882254, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 2e-05, | |
| "loss": 2.4746, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 1.9561403508771933e-05, | |
| "loss": 2.353, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 1.912280701754386e-05, | |
| "loss": 2.1808, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_cer": 0.986131055853677, | |
| "eval_loss": 1.8554401397705078, | |
| "eval_runtime": 21.7818, | |
| "eval_samples_per_second": 120.467, | |
| "eval_steps_per_second": 7.529, | |
| "eval_wer": 0.9742561721882254, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 7.01, | |
| "learning_rate": 1.868421052631579e-05, | |
| "loss": 2.028, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 1.824561403508772e-05, | |
| "loss": 1.8932, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 1.780701754385965e-05, | |
| "loss": 1.7557, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 1.736842105263158e-05, | |
| "loss": 1.6069, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_cer": 0.6230821555437986, | |
| "eval_loss": 1.2551288604736328, | |
| "eval_runtime": 21.5405, | |
| "eval_samples_per_second": 121.817, | |
| "eval_steps_per_second": 7.614, | |
| "eval_wer": 0.6822114370120278, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 8.23, | |
| "learning_rate": 1.692982456140351e-05, | |
| "loss": 1.4826, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 1.649122807017544e-05, | |
| "loss": 1.3003, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 1.605263157894737e-05, | |
| "loss": 1.1916, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_cer": 0.22420829869246467, | |
| "eval_loss": 0.7985360026359558, | |
| "eval_runtime": 22.1233, | |
| "eval_samples_per_second": 118.608, | |
| "eval_steps_per_second": 7.413, | |
| "eval_wer": 0.3679397903917845, | |
| "step": 1476 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 1.56140350877193e-05, | |
| "loss": 1.1239, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 1.517543859649123e-05, | |
| "loss": 1.0425, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 1.4736842105263159e-05, | |
| "loss": 0.9977, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_cer": 0.18267704633058726, | |
| "eval_loss": 0.6234104633331299, | |
| "eval_runtime": 21.4356, | |
| "eval_samples_per_second": 122.413, | |
| "eval_steps_per_second": 7.651, | |
| "eval_wer": 0.31180980516283324, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 10.06, | |
| "learning_rate": 1.429824561403509e-05, | |
| "loss": 0.9625, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "learning_rate": 1.385964912280702e-05, | |
| "loss": 0.9316, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 10.67, | |
| "learning_rate": 1.3421052631578948e-05, | |
| "loss": 0.8949, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 1.2982456140350879e-05, | |
| "loss": 0.836, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 11.0, | |
| "eval_cer": 0.16431108759730934, | |
| "eval_loss": 0.510288417339325, | |
| "eval_runtime": 21.4424, | |
| "eval_samples_per_second": 122.374, | |
| "eval_steps_per_second": 7.648, | |
| "eval_wer": 0.2800872195259197, | |
| "step": 1804 | |
| }, | |
| { | |
| "epoch": 11.28, | |
| "learning_rate": 1.2543859649122808e-05, | |
| "loss": 0.821, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 11.59, | |
| "learning_rate": 1.2105263157894737e-05, | |
| "loss": 0.7918, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 11.89, | |
| "learning_rate": 1.1666666666666668e-05, | |
| "loss": 0.7515, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "eval_cer": 0.1549391580379412, | |
| "eval_loss": 0.43052738904953003, | |
| "eval_runtime": 21.9755, | |
| "eval_samples_per_second": 119.406, | |
| "eval_steps_per_second": 7.463, | |
| "eval_wer": 0.26630090736442286, | |
| "step": 1968 | |
| }, | |
| { | |
| "epoch": 12.2, | |
| "learning_rate": 1.1228070175438597e-05, | |
| "loss": 0.7517, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 1.0789473684210528e-05, | |
| "loss": 0.7086, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 12.8, | |
| "learning_rate": 1.0350877192982459e-05, | |
| "loss": 0.7045, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 13.0, | |
| "eval_cer": 0.14133474416143904, | |
| "eval_loss": 0.3688310384750366, | |
| "eval_runtime": 21.9834, | |
| "eval_samples_per_second": 119.363, | |
| "eval_steps_per_second": 7.46, | |
| "eval_wer": 0.24892734050784274, | |
| "step": 2132 | |
| }, | |
| { | |
| "epoch": 13.11, | |
| "learning_rate": 9.912280701754386e-06, | |
| "loss": 0.6799, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 13.41, | |
| "learning_rate": 9.473684210526315e-06, | |
| "loss": 0.6578, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 13.72, | |
| "learning_rate": 9.035087719298246e-06, | |
| "loss": 0.6533, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 14.0, | |
| "eval_cer": 0.1339656866450004, | |
| "eval_loss": 0.3257971704006195, | |
| "eval_runtime": 21.7706, | |
| "eval_samples_per_second": 120.529, | |
| "eval_steps_per_second": 7.533, | |
| "eval_wer": 0.2398536962791025, | |
| "step": 2296 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 8.596491228070176e-06, | |
| "loss": 0.6267, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 14.33, | |
| "learning_rate": 8.157894736842106e-06, | |
| "loss": 0.6625, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 14.63, | |
| "learning_rate": 7.719298245614036e-06, | |
| "loss": 0.6162, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 14.94, | |
| "learning_rate": 7.280701754385966e-06, | |
| "loss": 0.5906, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "eval_cer": 0.1287884513642204, | |
| "eval_loss": 0.29405683279037476, | |
| "eval_runtime": 21.5389, | |
| "eval_samples_per_second": 121.826, | |
| "eval_steps_per_second": 7.614, | |
| "eval_wer": 0.23183512696068087, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 15.24, | |
| "learning_rate": 6.842105263157896e-06, | |
| "loss": 0.6021, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 15.55, | |
| "learning_rate": 6.403508771929825e-06, | |
| "loss": 0.5722, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 15.85, | |
| "learning_rate": 5.964912280701755e-06, | |
| "loss": 0.5746, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 16.0, | |
| "eval_cer": 0.12780591036202857, | |
| "eval_loss": 0.2748357355594635, | |
| "eval_runtime": 21.719, | |
| "eval_samples_per_second": 120.816, | |
| "eval_steps_per_second": 7.551, | |
| "eval_wer": 0.23000633044946192, | |
| "step": 2624 | |
| }, | |
| { | |
| "epoch": 16.16, | |
| "learning_rate": 5.526315789473685e-06, | |
| "loss": 0.559, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 16.46, | |
| "learning_rate": 5.087719298245615e-06, | |
| "loss": 0.5676, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 16.77, | |
| "learning_rate": 4.649122807017544e-06, | |
| "loss": 0.5169, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 17.0, | |
| "eval_cer": 0.12417806666162799, | |
| "eval_loss": 0.25731268525123596, | |
| "eval_runtime": 21.7044, | |
| "eval_samples_per_second": 120.897, | |
| "eval_steps_per_second": 7.556, | |
| "eval_wer": 0.224027572624323, | |
| "step": 2788 | |
| }, | |
| { | |
| "epoch": 17.07, | |
| "learning_rate": 4.210526315789474e-06, | |
| "loss": 0.5603, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 17.38, | |
| "learning_rate": 3.7719298245614037e-06, | |
| "loss": 0.5362, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 17.68, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.5187, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 17.99, | |
| "learning_rate": 2.8947368421052634e-06, | |
| "loss": 0.5511, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 18.0, | |
| "eval_cer": 0.12277983523543194, | |
| "eval_loss": 0.24789051711559296, | |
| "eval_runtime": 21.566, | |
| "eval_samples_per_second": 121.673, | |
| "eval_steps_per_second": 7.605, | |
| "eval_wer": 0.22107336287543083, | |
| "step": 2952 | |
| }, | |
| { | |
| "epoch": 18.29, | |
| "learning_rate": 2.456140350877193e-06, | |
| "loss": 0.541, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 18.6, | |
| "learning_rate": 2.017543859649123e-06, | |
| "loss": 0.5095, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 18.9, | |
| "learning_rate": 1.5789473684210526e-06, | |
| "loss": 0.5318, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 19.0, | |
| "eval_cer": 0.12100370342377749, | |
| "eval_loss": 0.2409585565328598, | |
| "eval_runtime": 21.8589, | |
| "eval_samples_per_second": 120.043, | |
| "eval_steps_per_second": 7.503, | |
| "eval_wer": 0.21861152141802068, | |
| "step": 3116 | |
| }, | |
| { | |
| "epoch": 19.21, | |
| "learning_rate": 1.1403508771929824e-06, | |
| "loss": 0.5156, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 19.51, | |
| "learning_rate": 7.017543859649123e-07, | |
| "loss": 0.4971, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 19.82, | |
| "learning_rate": 2.6315789473684213e-07, | |
| "loss": 0.5174, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "eval_cer": 0.12100370342377749, | |
| "eval_loss": 0.23930250108242035, | |
| "eval_runtime": 21.6599, | |
| "eval_samples_per_second": 121.146, | |
| "eval_steps_per_second": 7.572, | |
| "eval_wer": 0.21868185974537527, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 20.0, | |
| "step": 3280, | |
| "total_flos": 1.8383341927882616e+18, | |
| "train_loss": 1.9778692466456715, | |
| "train_runtime": 7257.7038, | |
| "train_samples_per_second": 28.924, | |
| "train_steps_per_second": 0.452 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 3280, | |
| "num_train_epochs": 20, | |
| "save_steps": 500, | |
| "total_flos": 1.8383341927882616e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |