| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 320, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 71.75464630126953, | |
| "learning_rate": 4.0000000000000003e-07, | |
| "loss": 3.61, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 45.94756317138672, | |
| "learning_rate": 1.4000000000000001e-06, | |
| "loss": 3.3547, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 29.31285858154297, | |
| "learning_rate": 2.4000000000000003e-06, | |
| "loss": 2.7402, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 19.44002914428711, | |
| "learning_rate": 3.4000000000000005e-06, | |
| "loss": 2.12, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 13.159948348999023, | |
| "learning_rate": 4.4e-06, | |
| "loss": 1.5923, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 8.778341293334961, | |
| "learning_rate": 5.400000000000001e-06, | |
| "loss": 1.2181, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "grad_norm": 5.181447982788086, | |
| "learning_rate": 6.4000000000000006e-06, | |
| "loss": 0.9039, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 5.2995686531066895, | |
| "learning_rate": 7.4e-06, | |
| "loss": 0.6798, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_cer": 0.1788136402712771, | |
| "eval_loss": 0.5512658953666687, | |
| "eval_runtime": 167.9142, | |
| "eval_samples_per_second": 7.522, | |
| "eval_steps_per_second": 0.06, | |
| "eval_wer": 0.6854535695115406, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.125, | |
| "grad_norm": 3.869173288345337, | |
| "learning_rate": 8.400000000000001e-06, | |
| "loss": 0.5485, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "grad_norm": 3.1130523681640625, | |
| "learning_rate": 9.4e-06, | |
| "loss": 0.4481, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.375, | |
| "grad_norm": 2.2851545810699463, | |
| "learning_rate": 9.81818181818182e-06, | |
| "loss": 0.4103, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "grad_norm": 2.8455662727355957, | |
| "learning_rate": 9.363636363636365e-06, | |
| "loss": 0.37, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.625, | |
| "grad_norm": 2.575657367706299, | |
| "learning_rate": 8.90909090909091e-06, | |
| "loss": 0.3513, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "grad_norm": 2.5928118228912354, | |
| "learning_rate": 8.454545454545455e-06, | |
| "loss": 0.3504, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.875, | |
| "grad_norm": 2.5617828369140625, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.324, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "grad_norm": 3.5175580978393555, | |
| "learning_rate": 7.545454545454546e-06, | |
| "loss": 0.3095, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_cer": 0.09716305282261917, | |
| "eval_loss": 0.2984148859977722, | |
| "eval_runtime": 164.9586, | |
| "eval_samples_per_second": 7.656, | |
| "eval_steps_per_second": 0.061, | |
| "eval_wer": 0.4486312399355878, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.125, | |
| "grad_norm": 2.2015202045440674, | |
| "learning_rate": 7.0909090909090916e-06, | |
| "loss": 0.2926, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "grad_norm": 3.0814766883850098, | |
| "learning_rate": 6.6363636363636375e-06, | |
| "loss": 0.2836, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.375, | |
| "grad_norm": 1.8548959493637085, | |
| "learning_rate": 6.181818181818182e-06, | |
| "loss": 0.2743, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "grad_norm": 2.3138978481292725, | |
| "learning_rate": 5.727272727272728e-06, | |
| "loss": 0.2707, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.625, | |
| "grad_norm": 2.2470438480377197, | |
| "learning_rate": 5.272727272727273e-06, | |
| "loss": 0.2799, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "grad_norm": 2.5589115619659424, | |
| "learning_rate": 4.818181818181819e-06, | |
| "loss": 0.2669, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.875, | |
| "grad_norm": 2.1574714183807373, | |
| "learning_rate": 4.363636363636364e-06, | |
| "loss": 0.2615, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "grad_norm": 3.022969961166382, | |
| "learning_rate": 3.90909090909091e-06, | |
| "loss": 0.2673, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_cer": 0.0882032667876588, | |
| "eval_loss": 0.2675623595714569, | |
| "eval_runtime": 162.9283, | |
| "eval_samples_per_second": 7.752, | |
| "eval_steps_per_second": 0.061, | |
| "eval_wer": 0.4142780461621041, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 3.125, | |
| "grad_norm": 1.893800973892212, | |
| "learning_rate": 3.454545454545455e-06, | |
| "loss": 0.2601, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "grad_norm": 2.2726192474365234, | |
| "learning_rate": 3e-06, | |
| "loss": 0.2575, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 3.375, | |
| "grad_norm": 1.9179538488388062, | |
| "learning_rate": 2.5454545454545456e-06, | |
| "loss": 0.2375, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "grad_norm": 1.734628677368164, | |
| "learning_rate": 2.090909090909091e-06, | |
| "loss": 0.238, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 3.625, | |
| "grad_norm": 1.9225579500198364, | |
| "learning_rate": 1.6363636363636365e-06, | |
| "loss": 0.237, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "grad_norm": 1.7918205261230469, | |
| "learning_rate": 1.181818181818182e-06, | |
| "loss": 0.2364, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 3.875, | |
| "grad_norm": 2.021205186843872, | |
| "learning_rate": 7.272727272727273e-07, | |
| "loss": 0.2422, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "grad_norm": 2.901258707046509, | |
| "learning_rate": 2.7272727272727274e-07, | |
| "loss": 0.2428, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_cer": 0.08564332792052727, | |
| "eval_loss": 0.2611912190914154, | |
| "eval_runtime": 163.7635, | |
| "eval_samples_per_second": 7.712, | |
| "eval_steps_per_second": 0.061, | |
| "eval_wer": 0.40128824476650565, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "step": 320, | |
| "total_flos": 2.6372074438656e+18, | |
| "train_loss": 0.7337436556816102, | |
| "train_runtime": 5396.4455, | |
| "train_samples_per_second": 7.535, | |
| "train_steps_per_second": 0.059 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 320, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 4, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.6372074438656e+18, | |
| "train_batch_size": 128, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |