| { |
| "best_metric": 0.8726145625114441, |
| "best_model_checkpoint": "runs/codeparrot/codeparrot-small-multi-ecore_line-512/checkpoint-800", |
| "epoch": 4.998829496683574, |
| "global_step": 800, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 1.0, |
| "learning_rate": 4.5e-05, |
| "loss": 1.0255, |
| "step": 160 |
| }, |
| { |
| "epoch": 1.0, |
| "eval_loss": 0.9181002378463745, |
| "eval_runtime": 18.9044, |
| "eval_samples_per_second": 78.923, |
| "eval_steps_per_second": 19.731, |
| "step": 160 |
| }, |
| { |
| "epoch": 2.0, |
| "learning_rate": 4e-05, |
| "loss": 0.8865, |
| "step": 320 |
| }, |
| { |
| "epoch": 2.0, |
| "eval_loss": 0.8927440643310547, |
| "eval_runtime": 18.8288, |
| "eval_samples_per_second": 79.24, |
| "eval_steps_per_second": 19.81, |
| "step": 320 |
| }, |
| { |
| "epoch": 3.0, |
| "learning_rate": 3.5e-05, |
| "loss": 0.8313, |
| "step": 480 |
| }, |
| { |
| "epoch": 3.0, |
| "eval_loss": 0.8809273838996887, |
| "eval_runtime": 18.7583, |
| "eval_samples_per_second": 79.538, |
| "eval_steps_per_second": 19.885, |
| "step": 480 |
| }, |
| { |
| "epoch": 4.0, |
| "learning_rate": 3e-05, |
| "loss": 0.7884, |
| "step": 640 |
| }, |
| { |
| "epoch": 4.0, |
| "eval_loss": 0.8755712509155273, |
| "eval_runtime": 19.3412, |
| "eval_samples_per_second": 77.141, |
| "eval_steps_per_second": 19.285, |
| "step": 640 |
| }, |
| { |
| "epoch": 5.0, |
| "learning_rate": 2.5e-05, |
| "loss": 0.7539, |
| "step": 800 |
| }, |
| { |
| "epoch": 5.0, |
| "eval_loss": 0.8726145625114441, |
| "eval_runtime": 18.9056, |
| "eval_samples_per_second": 78.918, |
| "eval_steps_per_second": 19.73, |
| "step": 800 |
| } |
| ], |
| "max_steps": 1600, |
| "num_train_epochs": 10, |
| "total_flos": 1.339121664e+16, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
