{ "best_metric": 63.62742130750605, "best_model_checkpoint": "./checkpoint-600", "epoch": 85.71428571428571, "global_step": 600, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 3.57, "learning_rate": 2.875862068965517e-07, "loss": 2.755, "step": 25 }, { "epoch": 7.14, "learning_rate": 2.6172413793103444e-07, "loss": 2.0968, "step": 50 }, { "epoch": 10.71, "learning_rate": 2.3586206896551725e-07, "loss": 1.703, "step": 75 }, { "epoch": 14.29, "learning_rate": 2.0999999999999997e-07, "loss": 1.4728, "step": 100 }, { "epoch": 14.29, "eval_loss": 1.4428775310516357, "eval_runtime": 1406.9013, "eval_samples_per_second": 0.364, "eval_steps_per_second": 0.023, "eval_wer": 83.59423434593924, "step": 100 }, { "epoch": 17.86, "learning_rate": 1.8413793103448276e-07, "loss": 1.2819, "step": 125 }, { "epoch": 21.43, "learning_rate": 1.5827586206896554e-07, "loss": 1.1744, "step": 150 }, { "epoch": 25.0, "learning_rate": 1.3241379310344827e-07, "loss": 1.0885, "step": 175 }, { "epoch": 28.57, "learning_rate": 1.0655172413793102e-07, "loss": 1.0344, "step": 200 }, { "epoch": 28.57, "eval_loss": 1.158418893814087, "eval_runtime": 1566.2048, "eval_samples_per_second": 0.327, "eval_steps_per_second": 0.02, "eval_wer": 69.97055176689399, "step": 200 }, { "epoch": 32.14, "learning_rate": 8.068965517241379e-08, "loss": 0.9889, "step": 225 }, { "epoch": 35.71, "learning_rate": 5.482758620689655e-08, "loss": 0.9647, "step": 250 }, { "epoch": 39.29, "learning_rate": 2.8965517241379308e-08, "loss": 0.9435, "step": 275 }, { "epoch": 42.86, "learning_rate": 3.103448275862069e-09, "loss": 0.9318, "step": 300 }, { "epoch": 42.86, "eval_loss": 1.1061279773712158, "eval_runtime": 1528.8642, "eval_samples_per_second": 0.335, "eval_steps_per_second": 0.021, "eval_wer": 67.83942963422194, "step": 300 }, { "epoch": 46.43, "learning_rate": 6e-08, "loss": 0.919, "step": 325 }, { "epoch": 50.0, "learning_rate": 4.076923076923077e-08, "loss": 0.9097, "step": 350 }, { "epoch": 53.57, "learning_rate": 2.1538461538461537e-08, "loss": 0.8935, "step": 375 }, { "epoch": 57.14, "learning_rate": 2.3076923076923076e-09, "loss": 0.8882, "step": 400 }, { "epoch": 57.14, "eval_loss": 1.0769362449645996, "eval_runtime": 1533.1785, "eval_samples_per_second": 0.334, "eval_steps_per_second": 0.021, "eval_wer": 66.42901425914445, "step": 400 }, { "epoch": 60.71, "learning_rate": 4.775510204081632e-08, "loss": 0.8804, "step": 425 }, { "epoch": 64.29, "learning_rate": 3.2448979591836735e-08, "loss": 0.8655, "step": 450 }, { "epoch": 67.86, "learning_rate": 1.714285714285714e-08, "loss": 0.8605, "step": 475 }, { "epoch": 71.43, "learning_rate": 1.836734693877551e-09, "loss": 0.8609, "step": 500 }, { "epoch": 71.43, "eval_loss": 1.0574754476547241, "eval_runtime": 1547.0272, "eval_samples_per_second": 0.331, "eval_steps_per_second": 0.021, "eval_wer": 66.1965282083075, "step": 500 }, { "epoch": 75.0, "learning_rate": 3.966101694915254e-08, "loss": 0.8491, "step": 525 }, { "epoch": 78.57, "learning_rate": 2.694915254237288e-08, "loss": 0.8417, "step": 550 }, { "epoch": 82.14, "learning_rate": 1.423728813559322e-08, "loss": 0.8278, "step": 575 }, { "epoch": 85.71, "learning_rate": 1.525423728813559e-09, "loss": 0.8262, "step": 600 }, { "epoch": 85.71, "eval_loss": 1.021410584449768, "eval_runtime": 1545.4583, "eval_samples_per_second": 0.331, "eval_steps_per_second": 0.021, "eval_wer": 63.62742130750605, "step": 600 }, { "epoch": 85.71, "step": 600, "total_flos": 3.77164543574016e+19, "train_loss": 0.1393694019317627, "train_runtime": 2431.6943, "train_samples_per_second": 15.791, "train_steps_per_second": 0.247 } ], "max_steps": 600, "num_train_epochs": 86, "total_flos": 3.77164543574016e+19, "trial_name": null, "trial_params": null }