| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 49.99819168173599, | |
| "global_step": 6900, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.485e-05, | |
| "loss": 10.7673, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.985e-05, | |
| "loss": 4.4331, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 4.484999999999999e-05, | |
| "loss": 3.4707, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 5.985e-05, | |
| "loss": 3.0807, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 7.484999999999999e-05, | |
| "loss": 2.9736, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "eval_cer": 1.0, | |
| "eval_loss": 2.9508235454559326, | |
| "eval_runtime": 114.1579, | |
| "eval_samples_per_second": 33.664, | |
| "eval_steps_per_second": 4.213, | |
| "eval_wer": 1.0, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 8.984999999999999e-05, | |
| "loss": 2.9229, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 0.00010484999999999999, | |
| "loss": 2.8971, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 5.8, | |
| "learning_rate": 0.00011985, | |
| "loss": 2.7156, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 0.00013485, | |
| "loss": 1.8057, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 0.00014984999999999998, | |
| "loss": 1.3293, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "eval_cer": 0.08615393306828052, | |
| "eval_loss": 0.33302000164985657, | |
| "eval_runtime": 114.4981, | |
| "eval_samples_per_second": 33.564, | |
| "eval_steps_per_second": 4.201, | |
| "eval_wer": 0.8407494145199064, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 0.00016485, | |
| "loss": 1.1212, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 0.00017984999999999998, | |
| "loss": 1.0403, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 0.00019484999999999997, | |
| "loss": 0.9903, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 0.00020984999999999998, | |
| "loss": 0.9766, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "learning_rate": 0.00022485, | |
| "loss": 0.956, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "eval_cer": 0.06022301157428658, | |
| "eval_loss": 0.2042141854763031, | |
| "eval_runtime": 114.0353, | |
| "eval_samples_per_second": 33.7, | |
| "eval_steps_per_second": 4.218, | |
| "eval_wer": 0.6872235232890971, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 11.59, | |
| "learning_rate": 0.00023984999999999998, | |
| "loss": 0.9616, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 12.32, | |
| "learning_rate": 0.00025485, | |
| "loss": 0.9653, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 13.04, | |
| "learning_rate": 0.00026984999999999997, | |
| "loss": 0.9653, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 13.77, | |
| "learning_rate": 0.00028484999999999996, | |
| "loss": 0.9486, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 14.49, | |
| "learning_rate": 0.00029985, | |
| "loss": 0.9509, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 14.49, | |
| "eval_cer": 0.06515214643343903, | |
| "eval_loss": 0.21843121945858002, | |
| "eval_runtime": 115.7242, | |
| "eval_samples_per_second": 33.208, | |
| "eval_steps_per_second": 4.156, | |
| "eval_wer": 0.7088212334113974, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 15.22, | |
| "learning_rate": 0.00029393877551020406, | |
| "loss": 0.9402, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 15.94, | |
| "learning_rate": 0.0002878163265306122, | |
| "loss": 0.9357, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 16.67, | |
| "learning_rate": 0.0002816938775510204, | |
| "loss": 0.9301, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 17.39, | |
| "learning_rate": 0.00027563265306122445, | |
| "loss": 0.9268, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 18.12, | |
| "learning_rate": 0.00026951020408163266, | |
| "loss": 0.9272, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 18.12, | |
| "eval_cer": 0.07027901163078096, | |
| "eval_loss": 0.2312462031841278, | |
| "eval_runtime": 107.6714, | |
| "eval_samples_per_second": 35.692, | |
| "eval_steps_per_second": 4.467, | |
| "eval_wer": 0.7210512620348686, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 18.84, | |
| "learning_rate": 0.0002633877551020408, | |
| "loss": 0.9191, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 19.56, | |
| "learning_rate": 0.00025726530612244896, | |
| "loss": 0.9016, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 20.29, | |
| "learning_rate": 0.0002511428571428571, | |
| "loss": 0.9105, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 21.01, | |
| "learning_rate": 0.00024502040816326527, | |
| "loss": 0.8923, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 21.74, | |
| "learning_rate": 0.00023889795918367345, | |
| "loss": 0.8561, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 21.74, | |
| "eval_cer": 0.06309716327582676, | |
| "eval_loss": 0.21578675508499146, | |
| "eval_runtime": 108.0601, | |
| "eval_samples_per_second": 35.564, | |
| "eval_steps_per_second": 4.451, | |
| "eval_wer": 0.6838407494145199, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 22.46, | |
| "learning_rate": 0.0002327755102040816, | |
| "loss": 0.8562, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 23.19, | |
| "learning_rate": 0.00022665306122448975, | |
| "loss": 0.855, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 23.91, | |
| "learning_rate": 0.00022053061224489796, | |
| "loss": 0.8398, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 24.64, | |
| "learning_rate": 0.0002144081632653061, | |
| "loss": 0.827, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 25.36, | |
| "learning_rate": 0.00020828571428571426, | |
| "loss": 0.8258, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 25.36, | |
| "eval_cer": 0.060060590225059496, | |
| "eval_loss": 0.19697847962379456, | |
| "eval_runtime": 107.078, | |
| "eval_samples_per_second": 35.89, | |
| "eval_steps_per_second": 4.492, | |
| "eval_wer": 0.6843611761644548, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 26.09, | |
| "learning_rate": 0.00020216326530612242, | |
| "loss": 0.8283, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 26.81, | |
| "learning_rate": 0.00019604081632653057, | |
| "loss": 0.8212, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 27.54, | |
| "learning_rate": 0.00018991836734693878, | |
| "loss": 0.8332, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 28.26, | |
| "learning_rate": 0.00018379591836734693, | |
| "loss": 0.8324, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 28.98, | |
| "learning_rate": 0.00017767346938775508, | |
| "loss": 0.7993, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 28.98, | |
| "eval_cer": 0.05770901156016298, | |
| "eval_loss": 0.18949392437934875, | |
| "eval_runtime": 106.6069, | |
| "eval_samples_per_second": 36.048, | |
| "eval_steps_per_second": 4.512, | |
| "eval_wer": 0.6697892271662763, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 29.71, | |
| "learning_rate": 0.00017155102040816324, | |
| "loss": 0.7777, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 30.43, | |
| "learning_rate": 0.0001654285714285714, | |
| "loss": 0.7764, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 31.16, | |
| "learning_rate": 0.0001593061224489796, | |
| "loss": 0.7748, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 31.88, | |
| "learning_rate": 0.00015318367346938775, | |
| "loss": 0.7587, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 32.61, | |
| "learning_rate": 0.00014712244897959183, | |
| "loss": 0.7525, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 32.61, | |
| "eval_cer": 0.05496197221888748, | |
| "eval_loss": 0.18448850512504578, | |
| "eval_runtime": 106.7665, | |
| "eval_samples_per_second": 35.994, | |
| "eval_steps_per_second": 4.505, | |
| "eval_wer": 0.6453291699193339, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 33.33, | |
| "learning_rate": 0.00014099999999999998, | |
| "loss": 0.7558, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 34.06, | |
| "learning_rate": 0.00013487755102040816, | |
| "loss": 0.7444, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 34.78, | |
| "learning_rate": 0.00012875510204081632, | |
| "loss": 0.7302, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 35.51, | |
| "learning_rate": 0.0001226326530612245, | |
| "loss": 0.7391, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 36.23, | |
| "learning_rate": 0.00011657142857142856, | |
| "loss": 0.7211, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 36.23, | |
| "eval_cer": 0.05309765760167223, | |
| "eval_loss": 0.17814987897872925, | |
| "eval_runtime": 106.0316, | |
| "eval_samples_per_second": 36.244, | |
| "eval_steps_per_second": 4.536, | |
| "eval_wer": 0.6273744470465782, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 36.95, | |
| "learning_rate": 0.00011044897959183672, | |
| "loss": 0.7074, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 37.68, | |
| "learning_rate": 0.00010432653061224489, | |
| "loss": 0.6976, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 38.41, | |
| "learning_rate": 9.820408163265305e-05, | |
| "loss": 0.6933, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 39.13, | |
| "learning_rate": 9.214285714285714e-05, | |
| "loss": 0.6845, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 39.85, | |
| "learning_rate": 8.602040816326529e-05, | |
| "loss": 0.677, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 39.85, | |
| "eval_cer": 0.05139576433368407, | |
| "eval_loss": 0.17315863072872162, | |
| "eval_runtime": 106.2239, | |
| "eval_samples_per_second": 36.178, | |
| "eval_steps_per_second": 4.528, | |
| "eval_wer": 0.6187874056726516, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 40.58, | |
| "learning_rate": 7.989795918367346e-05, | |
| "loss": 0.6951, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 41.3, | |
| "learning_rate": 7.377551020408162e-05, | |
| "loss": 0.6814, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 42.03, | |
| "learning_rate": 6.765306122448979e-05, | |
| "loss": 0.6822, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 42.75, | |
| "learning_rate": 6.153061224489796e-05, | |
| "loss": 0.6658, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 43.48, | |
| "learning_rate": 5.5408163265306116e-05, | |
| "loss": 0.6517, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 43.48, | |
| "eval_cer": 0.05030824747364184, | |
| "eval_loss": 0.1690707802772522, | |
| "eval_runtime": 105.9054, | |
| "eval_samples_per_second": 36.287, | |
| "eval_steps_per_second": 4.542, | |
| "eval_wer": 0.6177465521727816, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 44.2, | |
| "learning_rate": 4.928571428571428e-05, | |
| "loss": 0.6327, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 44.93, | |
| "learning_rate": 4.316326530612245e-05, | |
| "loss": 0.6355, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 45.65, | |
| "learning_rate": 3.704081632653061e-05, | |
| "loss": 0.6328, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 46.38, | |
| "learning_rate": 3.0918367346938774e-05, | |
| "loss": 0.6388, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 47.1, | |
| "learning_rate": 2.4795918367346937e-05, | |
| "loss": 0.6326, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 47.1, | |
| "eval_cer": 0.0478507418418581, | |
| "eval_loss": 0.1618689000606537, | |
| "eval_runtime": 108.5769, | |
| "eval_samples_per_second": 35.394, | |
| "eval_steps_per_second": 4.43, | |
| "eval_wer": 0.6044756700494406, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 47.82, | |
| "learning_rate": 1.86734693877551e-05, | |
| "loss": 0.6231, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 48.55, | |
| "learning_rate": 1.2551020408163265e-05, | |
| "loss": 0.6309, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 49.27, | |
| "learning_rate": 6.428571428571428e-06, | |
| "loss": 0.6341, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "learning_rate": 3.0612244897959183e-07, | |
| "loss": 0.6141, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "step": 6900, | |
| "total_flos": 4.523197640315366e+19, | |
| "train_loss": 1.2140741569408473, | |
| "train_runtime": 17169.0647, | |
| "train_samples_per_second": 25.747, | |
| "train_steps_per_second": 0.402 | |
| } | |
| ], | |
| "max_steps": 6900, | |
| "num_train_epochs": 50, | |
| "total_flos": 4.523197640315366e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |