| { | |
| "best_metric": 7.663424968719482, | |
| "best_model_checkpoint": "/mmfs1/gscratch/stf/abhinavp/corpus-filtering/lm-training/src/outputs/2023-11-15/09-23-00/checkpoints/checkpoint-60", | |
| "epoch": 0.48, | |
| "eval_steps": 10, | |
| "global_step": 60, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.866666666666667e-05, | |
| "loss": 9.9717, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "eval_loss": 9.324381828308105, | |
| "eval_runtime": 2.0042, | |
| "eval_samples_per_second": 498.949, | |
| "eval_steps_per_second": 62.369, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7333333333333336e-05, | |
| "loss": 9.1002, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "eval_loss": 8.892333984375, | |
| "eval_runtime": 1.9424, | |
| "eval_samples_per_second": 514.829, | |
| "eval_steps_per_second": 64.354, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.600000000000001e-05, | |
| "loss": 8.7625, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "eval_loss": 8.54922103881836, | |
| "eval_runtime": 1.9052, | |
| "eval_samples_per_second": 524.878, | |
| "eval_steps_per_second": 65.61, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.466666666666667e-05, | |
| "loss": 8.4987, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "eval_loss": 8.234408378601074, | |
| "eval_runtime": 2.0895, | |
| "eval_samples_per_second": 478.585, | |
| "eval_steps_per_second": 59.823, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.3333333333333334e-05, | |
| "loss": 8.2485, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 7.941398620605469, | |
| "eval_runtime": 2.1108, | |
| "eval_samples_per_second": 473.76, | |
| "eval_steps_per_second": 59.22, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.2e-05, | |
| "loss": 8.0802, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "eval_loss": 7.663424968719482, | |
| "eval_runtime": 1.991, | |
| "eval_samples_per_second": 502.251, | |
| "eval_steps_per_second": 62.781, | |
| "step": 60 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 375, | |
| "num_train_epochs": 3, | |
| "save_steps": 10, | |
| "total_flos": 7109521735680.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |