| { | |
| "best_metric": 19.79221609526093, | |
| "best_model_checkpoint": "./checkpoint-1000", | |
| "epoch": 4.044, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.6000000000000004e-07, | |
| "loss": 1.4914, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.600000000000001e-07, | |
| "loss": 1.0083, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.46e-06, | |
| "loss": 0.5977, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.9600000000000003e-06, | |
| "loss": 0.3947, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 2.46e-06, | |
| "loss": 0.3114, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 2.96e-06, | |
| "loss": 0.2963, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 3.46e-06, | |
| "loss": 0.1983, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 3.96e-06, | |
| "loss": 0.1535, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.4600000000000005e-06, | |
| "loss": 0.1561, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 4.960000000000001e-06, | |
| "loss": 0.1553, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 5.460000000000001e-06, | |
| "loss": 0.1518, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 5.9600000000000005e-06, | |
| "loss": 0.17, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 6.460000000000001e-06, | |
| "loss": 0.1565, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 6.96e-06, | |
| "loss": 0.1514, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 7.4600000000000006e-06, | |
| "loss": 0.1364, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 7.960000000000002e-06, | |
| "loss": 0.1035, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 8.46e-06, | |
| "loss": 0.0687, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 8.96e-06, | |
| "loss": 0.0601, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 9.460000000000001e-06, | |
| "loss": 0.0586, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 9.960000000000001e-06, | |
| "loss": 0.0687, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 9.54e-06, | |
| "loss": 0.0654, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 9.040000000000002e-06, | |
| "loss": 0.0694, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 8.540000000000001e-06, | |
| "loss": 0.0633, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 8.040000000000001e-06, | |
| "loss": 0.0553, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 7.540000000000001e-06, | |
| "loss": 0.0467, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 7.04e-06, | |
| "loss": 0.0263, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 6.540000000000001e-06, | |
| "loss": 0.0211, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 6.040000000000001e-06, | |
| "loss": 0.0205, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 5.540000000000001e-06, | |
| "loss": 0.0198, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 5.04e-06, | |
| "loss": 0.0194, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 4.540000000000001e-06, | |
| "loss": 0.0227, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 4.04e-06, | |
| "loss": 0.02, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 3.54e-06, | |
| "loss": 0.0193, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 3.04e-06, | |
| "loss": 0.0161, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 2.5400000000000002e-06, | |
| "loss": 0.0118, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 2.04e-06, | |
| "loss": 0.0076, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.54e-06, | |
| "loss": 0.0062, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 1.04e-06, | |
| "loss": 0.0061, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 5.4e-07, | |
| "loss": 0.0073, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 4e-08, | |
| "loss": 0.0068, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "eval_loss": 0.3235916197299957, | |
| "eval_runtime": 8424.7, | |
| "eval_samples_per_second": 1.099, | |
| "eval_steps_per_second": 0.034, | |
| "eval_wer": 19.79221609526093, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "step": 1000, | |
| "total_flos": 6.519215907274752e+19, | |
| "train_loss": 0.1604953267276287, | |
| "train_runtime": 25527.9444, | |
| "train_samples_per_second": 2.507, | |
| "train_steps_per_second": 0.039 | |
| } | |
| ], | |
| "max_steps": 1000, | |
| "num_train_epochs": 9223372036854775807, | |
| "total_flos": 6.519215907274752e+19, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |