| { | |
| "best_metric": 8.260101318359375, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/outputs/2023-11-20/16-19-50/checkpoints/checkpoint-70", | |
| "epoch": 0.56, | |
| "eval_steps": 10, | |
| "global_step": 70, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 10.8117, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "eval_loss": 10.801591873168945, | |
| "eval_runtime": 2.6865, | |
| "eval_samples_per_second": 372.226, | |
| "eval_steps_per_second": 46.528, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.2e-05, | |
| "loss": 10.7909, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "eval_loss": 10.775726318359375, | |
| "eval_runtime": 2.779, | |
| "eval_samples_per_second": 359.838, | |
| "eval_steps_per_second": 44.98, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 3.8e-05, | |
| "loss": 10.7524, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "eval_loss": 10.704446792602539, | |
| "eval_runtime": 2.7821, | |
| "eval_samples_per_second": 359.435, | |
| "eval_steps_per_second": 44.929, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.4000000000000007e-05, | |
| "loss": 10.5629, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_loss": 10.172320365905762, | |
| "eval_runtime": 2.7542, | |
| "eval_samples_per_second": 363.084, | |
| "eval_steps_per_second": 45.385, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 3e-05, | |
| "loss": 9.7608, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 9.128562927246094, | |
| "eval_runtime": 2.7812, | |
| "eval_samples_per_second": 359.553, | |
| "eval_steps_per_second": 44.944, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 2.6000000000000002e-05, | |
| "loss": 8.9735, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "eval_loss": 8.557924270629883, | |
| "eval_runtime": 2.6727, | |
| "eval_samples_per_second": 374.147, | |
| "eval_steps_per_second": 46.768, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 2.2000000000000003e-05, | |
| "loss": 8.5033, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "eval_loss": 8.260101318359375, | |
| "eval_runtime": 2.7328, | |
| "eval_samples_per_second": 365.926, | |
| "eval_steps_per_second": 45.741, | |
| "step": 70 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 125, | |
| "num_train_epochs": 1, | |
| "save_steps": 10, | |
| "total_flos": 4923721736640.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |