{ "best_metric": 0.3494756519794464, "best_model_checkpoint": "whisper-medium-v3-ha-4/checkpoint-4000", "epoch": 8.223684210526315, "eval_steps": 2000, "global_step": 10000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.41118421052631576, "grad_norm": 8.568055152893066, "learning_rate": 9.827745664739885e-06, "loss": 0.9437, "step": 500 }, { "epoch": 0.8223684210526315, "grad_norm": 9.2145357131958, "learning_rate": 9.635067437379577e-06, "loss": 0.514, "step": 1000 }, { "epoch": 1.2335526315789473, "grad_norm": 7.131311893463135, "learning_rate": 9.442389210019268e-06, "loss": 0.3292, "step": 1500 }, { "epoch": 1.6447368421052633, "grad_norm": 7.76482629776001, "learning_rate": 9.249710982658961e-06, "loss": 0.2577, "step": 2000 }, { "epoch": 1.6447368421052633, "eval_loss": 0.3743760883808136, "eval_runtime": 317.1777, "eval_samples_per_second": 3.862, "eval_steps_per_second": 0.486, "eval_wer": 0.28840193180999063, "step": 2000 }, { "epoch": 2.0559210526315788, "grad_norm": 5.391658782958984, "learning_rate": 9.057032755298651e-06, "loss": 0.2277, "step": 2500 }, { "epoch": 2.4671052631578947, "grad_norm": 4.684976100921631, "learning_rate": 8.864354527938343e-06, "loss": 0.1206, "step": 3000 }, { "epoch": 2.8782894736842106, "grad_norm": 6.3501877784729, "learning_rate": 8.671676300578035e-06, "loss": 0.1193, "step": 3500 }, { "epoch": 3.2894736842105265, "grad_norm": 3.3425328731536865, "learning_rate": 8.478998073217727e-06, "loss": 0.0717, "step": 4000 }, { "epoch": 3.2894736842105265, "eval_loss": 0.3494756519794464, "eval_runtime": 317.0862, "eval_samples_per_second": 3.863, "eval_steps_per_second": 0.486, "eval_wer": 0.24580119656887478, "step": 4000 }, { "epoch": 3.700657894736842, "grad_norm": 5.446820259094238, "learning_rate": 8.286705202312139e-06, "loss": 0.0541, "step": 4500 }, { "epoch": 4.1118421052631575, "grad_norm": 3.8512370586395264, "learning_rate": 8.094026974951831e-06, "loss": 0.0468, "step": 5000 }, { "epoch": 4.5230263157894735, "grad_norm": 3.269277811050415, "learning_rate": 7.901348747591523e-06, "loss": 0.0247, "step": 5500 }, { "epoch": 4.934210526315789, "grad_norm": 4.412689208984375, "learning_rate": 7.708670520231215e-06, "loss": 0.0274, "step": 6000 }, { "epoch": 4.934210526315789, "eval_loss": 0.35736316442489624, "eval_runtime": 312.254, "eval_samples_per_second": 3.923, "eval_steps_per_second": 0.493, "eval_wer": 0.22150940676133496, "step": 6000 }, { "epoch": 5.345394736842105, "grad_norm": 0.7462264895439148, "learning_rate": 7.516377649325627e-06, "loss": 0.0147, "step": 6500 }, { "epoch": 5.756578947368421, "grad_norm": 1.4490907192230225, "learning_rate": 7.323699421965319e-06, "loss": 0.0131, "step": 7000 }, { "epoch": 6.167763157894737, "grad_norm": 0.5171042084693909, "learning_rate": 7.13102119460501e-06, "loss": 0.0108, "step": 7500 }, { "epoch": 6.578947368421053, "grad_norm": 3.7838356494903564, "learning_rate": 6.9383429672447026e-06, "loss": 0.0073, "step": 8000 }, { "epoch": 6.578947368421053, "eval_loss": 0.38202571868896484, "eval_runtime": 312.9667, "eval_samples_per_second": 3.914, "eval_steps_per_second": 0.492, "eval_wer": 0.21552656238737114, "step": 8000 }, { "epoch": 6.990131578947368, "grad_norm": 0.41568490862846375, "learning_rate": 6.745664739884394e-06, "loss": 0.0073, "step": 8500 }, { "epoch": 7.401315789473684, "grad_norm": 0.18531061708927155, "learning_rate": 6.5529865125240855e-06, "loss": 0.0038, "step": 9000 }, { "epoch": 7.8125, "grad_norm": 0.08499964326620102, "learning_rate": 6.360693641618498e-06, "loss": 0.005, "step": 9500 }, { "epoch": 8.223684210526315, "grad_norm": 0.25189268589019775, "learning_rate": 6.1680154142581895e-06, "loss": 0.0036, "step": 10000 }, { "epoch": 8.223684210526315, "eval_loss": 0.4093479812145233, "eval_runtime": 316.7106, "eval_samples_per_second": 3.868, "eval_steps_per_second": 0.486, "eval_wer": 0.21884235565486918, "step": 10000 }, { "epoch": 8.223684210526315, "step": 10000, "total_flos": 2.307759729278976e+19, "train_loss": 0.14012403349876404, "train_runtime": 14667.5247, "train_samples_per_second": 14.181, "train_steps_per_second": 1.773 } ], "logging_steps": 500, "max_steps": 26000, "num_input_tokens_seen": 0, "num_train_epochs": 22, "save_steps": 2000, "stateful_callbacks": { "EarlyStoppingCallback": { "args": { "early_stopping_patience": 3, "early_stopping_threshold": 0.0 }, "attributes": { "early_stopping_patience_counter": 3 } }, "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 2.307759729278976e+19, "train_batch_size": 8, "trial_name": null, "trial_params": null }