| { |
| "best_metric": 0.48831769824028015, |
| "best_model_checkpoint": "./mistral/07-03-24-Weni-pipeline_test4_Zeroshot-2_max_steps-3224_batch_64_2024-03-07_ppid_7/checkpoint-100", |
| "epoch": 0.24798512089274644, |
| "eval_steps": 100, |
| "global_step": 100, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.05, |
| "grad_norm": 1.670975685119629, |
| "learning_rate": 1.0559006211180125e-05, |
| "loss": 1.9061, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.1, |
| "grad_norm": 1.0817188024520874, |
| "learning_rate": 2.2981366459627328e-05, |
| "loss": 1.6745, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.15, |
| "grad_norm": 1.4779456853866577, |
| "learning_rate": 3.5403726708074535e-05, |
| "loss": 1.2739, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.2, |
| "grad_norm": 0.6264486908912659, |
| "learning_rate": 4.782608695652174e-05, |
| "loss": 0.7589, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.25, |
| "grad_norm": 0.21179942786693573, |
| "learning_rate": 6.024844720496895e-05, |
| "loss": 0.5363, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.25, |
| "eval_loss": 0.48831769824028015, |
| "eval_runtime": 181.0058, |
| "eval_samples_per_second": 15.839, |
| "eval_steps_per_second": 0.994, |
| "step": 100 |
| } |
| ], |
| "logging_steps": 20, |
| "max_steps": 3224, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 8, |
| "save_steps": 100, |
| "total_flos": 2.057848754274304e+17, |
| "train_batch_size": 8, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|