| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 14.996463098325867, | |
| "eval_steps": 2000, | |
| "global_step": 31800, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.00014879999999999998, | |
| "loss": 15.1742, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0002985, | |
| "loss": 3.8658, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0002998088484992574, | |
| "loss": 1.8845, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0002992281499624614, | |
| "loss": 1.54, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "eval_cer": 0.26165714958754477, | |
| "eval_loss": 1.005660891532898, | |
| "eval_runtime": 2601.6674, | |
| "eval_samples_per_second": 6.501, | |
| "eval_steps_per_second": 0.407, | |
| "eval_wer": 0.6134988781327054, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.0002982593949446869, | |
| "loss": 1.3671, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0002969051026198513, | |
| "loss": 1.2849, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00029517264575296907, | |
| "loss": 1.2354, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.00029305958725988945, | |
| "loss": 1.1895, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "eval_cer": 0.20404210489580651, | |
| "eval_loss": 0.7781721949577332, | |
| "eval_runtime": 2606.5317, | |
| "eval_samples_per_second": 6.489, | |
| "eval_steps_per_second": 0.406, | |
| "eval_wer": 0.5034607425730799, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.0002905745131650867, | |
| "loss": 1.1348, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.0002877238857151841, | |
| "loss": 1.0801, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.00028452188731761346, | |
| "loss": 1.0722, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.00028096401365270297, | |
| "loss": 1.0582, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "eval_cer": 0.18262104285084868, | |
| "eval_loss": 0.6766561269760132, | |
| "eval_runtime": 2575.8771, | |
| "eval_samples_per_second": 6.566, | |
| "eval_steps_per_second": 0.411, | |
| "eval_wer": 0.4655154643262057, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 0.0002770655780074579, | |
| "loss": 1.0283, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 0.0002728367179679716, | |
| "loss": 0.9818, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 0.0002682884303639032, | |
| "loss": 0.972, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.0002634425531692912, | |
| "loss": 0.9586, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "eval_cer": 0.1690433917735793, | |
| "eval_loss": 0.6272980570793152, | |
| "eval_runtime": 2527.907, | |
| "eval_samples_per_second": 6.691, | |
| "eval_steps_per_second": 0.419, | |
| "eval_wer": 0.43799601994783116, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 0.00025829226958497964, | |
| "loss": 0.9543, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 0.0002528603801705155, | |
| "loss": 0.9046, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 0.00024716101014216497, | |
| "loss": 0.8973, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 0.00024122112631130689, | |
| "loss": 0.8831, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "eval_cer": 0.15516692331517162, | |
| "eval_loss": 0.5883975028991699, | |
| "eval_runtime": 2528.7031, | |
| "eval_samples_per_second": 6.689, | |
| "eval_steps_per_second": 0.418, | |
| "eval_wer": 0.40706736753641115, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 0.00023503237291623731, | |
| "loss": 0.8808, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 0.00022862249928740867, | |
| "loss": 0.8442, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 0.00022202159516049143, | |
| "loss": 0.8334, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "learning_rate": 0.0002152203749023848, | |
| "loss": 0.8318, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.66, | |
| "eval_cer": 0.14686983340385715, | |
| "eval_loss": 0.5510314106941223, | |
| "eval_runtime": 2595.7595, | |
| "eval_samples_per_second": 6.516, | |
| "eval_steps_per_second": 0.408, | |
| "eval_wer": 0.3897438336940751, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 5.89, | |
| "learning_rate": 0.0002082495540080772, | |
| "loss": 0.8242, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 0.00020112725956837873, | |
| "loss": 0.7965, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 0.00019387201256956072, | |
| "loss": 0.7776, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 0.0001865026797309365, | |
| "loss": 0.7725, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "eval_cer": 0.1407080881358594, | |
| "eval_loss": 0.5327238440513611, | |
| "eval_runtime": 2579.4825, | |
| "eval_samples_per_second": 6.557, | |
| "eval_steps_per_second": 0.41, | |
| "eval_wer": 0.3726026528395532, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 0.0001790534348212884, | |
| "loss": 0.7775, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 0.00017152894049650538, | |
| "loss": 0.7543, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 7.31, | |
| "learning_rate": 0.0001639334517198862, | |
| "loss": 0.7284, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 0.00015630173005929936, | |
| "loss": 0.7254, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "eval_cer": 0.1416372575107858, | |
| "eval_loss": 0.5081394910812378, | |
| "eval_runtime": 2577.7581, | |
| "eval_samples_per_second": 6.562, | |
| "eval_steps_per_second": 0.41, | |
| "eval_wer": 0.3675760530885046, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 0.0001486536212278503, | |
| "loss": 0.7224, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 0.00014100901355223894, | |
| "loss": 0.7136, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 0.00013338778625467495, | |
| "loss": 0.6742, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 0.00012580975775846912, | |
| "loss": 0.6802, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "eval_cer": 0.13128552546054956, | |
| "eval_loss": 0.4845993220806122, | |
| "eval_runtime": 2577.5228, | |
| "eval_samples_per_second": 6.562, | |
| "eval_steps_per_second": 0.41, | |
| "eval_wer": 0.350228734073845, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 0.0001182946341517271, | |
| "loss": 0.6779, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 0.00011087672812556355, | |
| "loss": 0.6727, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 9.2, | |
| "learning_rate": 0.00010354560475282154, | |
| "loss": 0.6437, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "learning_rate": 9.633528251009509e-05, | |
| "loss": 0.6386, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 9.43, | |
| "eval_cer": 0.12406592304034689, | |
| "eval_loss": 0.46761998534202576, | |
| "eval_runtime": 2525.8593, | |
| "eval_samples_per_second": 6.696, | |
| "eval_steps_per_second": 0.419, | |
| "eval_wer": 0.3343878093063451, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "learning_rate": 8.926451129327824e-05, | |
| "loss": 0.6291, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 8.23653341292733e-05, | |
| "loss": 0.6257, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 10.14, | |
| "learning_rate": 7.562804579791497e-05, | |
| "loss": 0.608, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "learning_rate": 6.909703945913943e-05, | |
| "loss": 0.5949, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 10.37, | |
| "eval_cer": 0.11851489942723033, | |
| "eval_loss": 0.45102670788764954, | |
| "eval_runtime": 2545.5284, | |
| "eval_samples_per_second": 6.645, | |
| "eval_steps_per_second": 0.416, | |
| "eval_wer": 0.3250481649739552, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 10.61, | |
| "learning_rate": 6.276312780656023e-05, | |
| "loss": 0.5946, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 10.85, | |
| "learning_rate": 5.665606902083949e-05, | |
| "loss": 0.5947, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 11.08, | |
| "learning_rate": 5.079174404496348e-05, | |
| "loss": 0.5854, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "learning_rate": 4.519634802070792e-05, | |
| "loss": 0.5736, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "eval_cer": 0.11609600544414732, | |
| "eval_loss": 0.4416215717792511, | |
| "eval_runtime": 2548.5938, | |
| "eval_samples_per_second": 6.637, | |
| "eval_steps_per_second": 0.415, | |
| "eval_wer": 0.31890366212369875, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 3.9862009784738855e-05, | |
| "loss": 0.5637, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 11.79, | |
| "learning_rate": 3.4814077035373634e-05, | |
| "loss": 0.5681, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 12.03, | |
| "learning_rate": 3.0065676538704393e-05, | |
| "loss": 0.563, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 12.26, | |
| "learning_rate": 2.562915614992792e-05, | |
| "loss": 0.5451, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 12.26, | |
| "eval_cer": 0.1143270937937593, | |
| "eval_loss": 0.43376967310905457, | |
| "eval_runtime": 2588.7798, | |
| "eval_samples_per_second": 6.534, | |
| "eval_steps_per_second": 0.409, | |
| "eval_wer": 0.31443204972686695, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 2.1523948888423446e-05, | |
| "loss": 0.5525, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 12.73, | |
| "learning_rate": 1.7744279871280954e-05, | |
| "loss": 0.5442, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 1.4308531819153024e-05, | |
| "loss": 0.5452, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "learning_rate": 1.1225639134154647e-05, | |
| "loss": 0.5375, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "eval_cer": 0.11264106579653371, | |
| "eval_loss": 0.4286753833293915, | |
| "eval_runtime": 2584.5725, | |
| "eval_samples_per_second": 6.544, | |
| "eval_steps_per_second": 0.409, | |
| "eval_wer": 0.30952437583743625, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 13.44, | |
| "learning_rate": 8.508697672362313e-06, | |
| "loss": 0.5351, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 13.68, | |
| "learning_rate": 6.153885454076635e-06, | |
| "loss": 0.5365, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 13.91, | |
| "learning_rate": 4.173134142042017e-06, | |
| "loss": 0.5308, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "learning_rate": 2.5715945297517193e-06, | |
| "loss": 0.5335, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 14.15, | |
| "eval_cer": 0.11220919834058202, | |
| "eval_loss": 0.42732492089271545, | |
| "eval_runtime": 2538.009, | |
| "eval_samples_per_second": 6.664, | |
| "eval_steps_per_second": 0.417, | |
| "eval_wer": 0.3079228409009823, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 14.38, | |
| "learning_rate": 1.3534312994473606e-06, | |
| "loss": 0.5315, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 14.62, | |
| "learning_rate": 5.218121921786278e-07, | |
| "loss": 0.5312, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 14.85, | |
| "learning_rate": 7.939673168200766e-08, | |
| "loss": 0.5325, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "step": 31800, | |
| "total_flos": 2.135574616282843e+20, | |
| "train_loss": 1.0645792571253747, | |
| "train_runtime": 75054.0862, | |
| "train_samples_per_second": 27.122, | |
| "train_steps_per_second": 0.424 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 31800, | |
| "num_train_epochs": 15, | |
| "save_steps": 2000, | |
| "total_flos": 2.135574616282843e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |