{ "best_metric": 37.5, "best_model_checkpoint": "./whisper-large-shona/checkpoint-1000", "epoch": 83.32, "global_step": 1000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 2.08, "learning_rate": 6.712113404111031e-06, "loss": 1.3237, "step": 25 }, { "epoch": 4.16, "learning_rate": 8.360489289678585e-06, "loss": 0.3887, "step": 50 }, { "epoch": 6.24, "learning_rate": 9.316614300602277e-06, "loss": 0.1151, "step": 75 }, { "epoch": 8.32, "learning_rate": 9.978175972987748e-06, "loss": 0.0305, "step": 100 }, { "epoch": 10.4, "learning_rate": 9.733333333333334e-06, "loss": 0.0107, "step": 125 }, { "epoch": 12.48, "learning_rate": 9.444444444444445e-06, "loss": 0.0059, "step": 150 }, { "epoch": 14.56, "learning_rate": 9.155555555555557e-06, "loss": 0.0036, "step": 175 }, { "epoch": 16.64, "learning_rate": 8.866666666666668e-06, "loss": 0.0028, "step": 200 }, { "epoch": 18.72, "learning_rate": 8.577777777777778e-06, "loss": 0.0018, "step": 225 }, { "epoch": 20.8, "learning_rate": 8.288888888888889e-06, "loss": 0.0013, "step": 250 }, { "epoch": 22.88, "learning_rate": 8.000000000000001e-06, "loss": 0.0011, "step": 275 }, { "epoch": 24.96, "learning_rate": 7.711111111111112e-06, "loss": 0.001, "step": 300 }, { "epoch": 27.08, "learning_rate": 7.422222222222223e-06, "loss": 0.0009, "step": 325 }, { "epoch": 29.16, "learning_rate": 7.133333333333334e-06, "loss": 0.0008, "step": 350 }, { "epoch": 31.24, "learning_rate": 6.844444444444445e-06, "loss": 0.0007, "step": 375 }, { "epoch": 33.32, "learning_rate": 6.555555555555556e-06, "loss": 0.0007, "step": 400 }, { "epoch": 35.4, "learning_rate": 6.266666666666668e-06, "loss": 0.0006, "step": 425 }, { "epoch": 37.48, "learning_rate": 5.977777777777778e-06, "loss": 0.0006, "step": 450 }, { "epoch": 39.56, "learning_rate": 5.688888888888889e-06, "loss": 0.0005, "step": 475 }, { "epoch": 41.64, "learning_rate": 5.400000000000001e-06, "loss": 0.0005, "step": 500 }, { "epoch": 41.64, "eval_loss": 0.87841796875, "eval_runtime": 349.183, "eval_samples_per_second": 2.649, "eval_steps_per_second": 0.043, "eval_wer": 37.525, "step": 500 }, { "epoch": 43.72, "learning_rate": 5.1111111111111115e-06, "loss": 0.0005, "step": 525 }, { "epoch": 45.8, "learning_rate": 4.822222222222222e-06, "loss": 0.0005, "step": 550 }, { "epoch": 47.88, "learning_rate": 4.533333333333334e-06, "loss": 0.0004, "step": 575 }, { "epoch": 49.96, "learning_rate": 4.244444444444445e-06, "loss": 0.0004, "step": 600 }, { "epoch": 52.08, "learning_rate": 3.944444444444445e-06, "loss": 0.0004, "step": 625 }, { "epoch": 54.16, "learning_rate": 3.6555555555555562e-06, "loss": 0.0004, "step": 650 }, { "epoch": 56.24, "learning_rate": 3.366666666666667e-06, "loss": 0.0004, "step": 675 }, { "epoch": 58.32, "learning_rate": 3.077777777777778e-06, "loss": 0.0004, "step": 700 }, { "epoch": 60.4, "learning_rate": 2.788888888888889e-06, "loss": 0.0004, "step": 725 }, { "epoch": 62.48, "learning_rate": 2.5e-06, "loss": 0.0003, "step": 750 }, { "epoch": 64.56, "learning_rate": 2.2111111111111113e-06, "loss": 0.0003, "step": 775 }, { "epoch": 66.64, "learning_rate": 1.9222222222222224e-06, "loss": 0.0003, "step": 800 }, { "epoch": 68.72, "learning_rate": 1.6333333333333335e-06, "loss": 0.0003, "step": 825 }, { "epoch": 70.8, "learning_rate": 1.3444444444444446e-06, "loss": 0.0003, "step": 850 }, { "epoch": 72.88, "learning_rate": 1.0555555555555557e-06, "loss": 0.0003, "step": 875 }, { "epoch": 74.96, "learning_rate": 7.666666666666667e-07, "loss": 0.0003, "step": 900 }, { "epoch": 77.08, "learning_rate": 4.777777777777778e-07, "loss": 0.0003, "step": 925 }, { "epoch": 79.16, "learning_rate": 1.888888888888889e-07, "loss": 0.0003, "step": 950 }, { "epoch": 81.24, "learning_rate": 0.0, "loss": 0.0003, "step": 975 }, { "epoch": 83.32, "learning_rate": 0.0, "loss": 0.0003, "step": 1000 }, { "epoch": 83.32, "eval_loss": 0.9189453125, "eval_runtime": 348.727, "eval_samples_per_second": 2.653, "eval_steps_per_second": 0.043, "eval_wer": 37.5, "step": 1000 }, { "epoch": 83.32, "step": 1000, "total_flos": 1.387025210307332e+20, "train_loss": 0.04746642480790615, "train_runtime": 8896.3772, "train_samples_per_second": 7.194, "train_steps_per_second": 0.112 } ], "max_steps": 1000, "num_train_epochs": 84, "total_flos": 1.387025210307332e+20, "trial_name": null, "trial_params": null }