| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.006072874493927126, | |
| "eval_steps": 9, | |
| "global_step": 27, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0002249212775528565, | |
| "eval_loss": 1.1927239894866943, | |
| "eval_runtime": 453.4202, | |
| "eval_samples_per_second": 16.514, | |
| "eval_steps_per_second": 2.064, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0006747638326585695, | |
| "grad_norm": 6.131251811981201, | |
| "learning_rate": 3e-05, | |
| "loss": 4.7417, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.001349527665317139, | |
| "grad_norm": 4.951833724975586, | |
| "learning_rate": 6e-05, | |
| "loss": 4.453, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0020242914979757085, | |
| "grad_norm": 4.731398105621338, | |
| "learning_rate": 9e-05, | |
| "loss": 3.9574, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0020242914979757085, | |
| "eval_loss": 0.9269639849662781, | |
| "eval_runtime": 974.5185, | |
| "eval_samples_per_second": 7.684, | |
| "eval_steps_per_second": 0.96, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.002699055330634278, | |
| "grad_norm": 3.335261821746826, | |
| "learning_rate": 9.987820251299122e-05, | |
| "loss": 3.6606, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0033738191632928477, | |
| "grad_norm": 3.0242624282836914, | |
| "learning_rate": 9.924038765061042e-05, | |
| "loss": 3.377, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.004048582995951417, | |
| "grad_norm": 2.4873292446136475, | |
| "learning_rate": 9.806308479691595e-05, | |
| "loss": 3.1263, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.004048582995951417, | |
| "eval_loss": 0.7899854779243469, | |
| "eval_runtime": 873.2265, | |
| "eval_samples_per_second": 8.575, | |
| "eval_steps_per_second": 1.072, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.004723346828609987, | |
| "grad_norm": 2.649564266204834, | |
| "learning_rate": 9.635919272833938e-05, | |
| "loss": 3.1013, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.005398110661268556, | |
| "grad_norm": 2.411900520324707, | |
| "learning_rate": 9.414737964294636e-05, | |
| "loss": 3.0944, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.006072874493927126, | |
| "grad_norm": 2.7092297077178955, | |
| "learning_rate": 9.145187862775209e-05, | |
| "loss": 2.9518, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.006072874493927126, | |
| "eval_loss": 0.7562000751495361, | |
| "eval_runtime": 970.9794, | |
| "eval_samples_per_second": 7.712, | |
| "eval_steps_per_second": 0.964, | |
| "step": 27 | |
| } | |
| ], | |
| "logging_steps": 3, | |
| "max_steps": 100, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 9, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.796899292879258e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |