| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.7272727272727275, | |
| "global_step": 15, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2e-05, | |
| "loss": 2.1084, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.9749279121818235e-05, | |
| "loss": 1.9844, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.900968867902419e-05, | |
| "loss": 1.6348, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.78183148246803e-05, | |
| "loss": 1.8301, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.6234898018587336e-05, | |
| "loss": 1.7168, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.4338837391175582e-05, | |
| "loss": 1.0171, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.2225209339563144e-05, | |
| "loss": 0.8047, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5701, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 7.774790660436857e-06, | |
| "loss": 0.5728, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 5.66116260882442e-06, | |
| "loss": 0.5532, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3.7651019814126656e-06, | |
| "loss": 0.4041, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 2.1816851753197023e-06, | |
| "loss": 0.1893, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 9.903113209758098e-07, | |
| "loss": 0.1756, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 2.507208781817638e-07, | |
| "loss": 0.1517, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.0, | |
| "loss": 0.1946, | |
| "step": 15 | |
| } | |
| ], | |
| "max_steps": 15, | |
| "num_train_epochs": 3, | |
| "total_flos": 1147989393408.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |