| { | |
| "best_metric": 0.028508992865681648, | |
| "best_model_checkpoint": "checkpoints/instrucode/low-shot-task-specific/word-sorting/checkpoint-100", | |
| "epoch": 4.0, | |
| "eval_steps": 500, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 5.9999999999999995e-05, | |
| "loss": 0.3822, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00011999999999999999, | |
| "loss": 0.1419, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 0.04200100898742676, | |
| "eval_runtime": 16.0265, | |
| "eval_samples_per_second": 12.479, | |
| "eval_steps_per_second": 1.56, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00017999999999999998, | |
| "loss": 0.0359, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00023999999999999998, | |
| "loss": 0.0356, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0003, | |
| "loss": 0.0303, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 0.031496673822402954, | |
| "eval_runtime": 15.9936, | |
| "eval_samples_per_second": 12.505, | |
| "eval_steps_per_second": 1.563, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 0.000285, | |
| "loss": 0.0203, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.00027, | |
| "loss": 0.0191, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 0.029132427647709846, | |
| "eval_runtime": 15.9682, | |
| "eval_samples_per_second": 12.525, | |
| "eval_steps_per_second": 1.566, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 0.00025499999999999996, | |
| "loss": 0.0161, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 0.00023999999999999998, | |
| "loss": 0.0116, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.000225, | |
| "loss": 0.0116, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 0.028508992865681648, | |
| "eval_runtime": 15.9929, | |
| "eval_samples_per_second": 12.506, | |
| "eval_steps_per_second": 1.563, | |
| "step": 100 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 250, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "total_flos": 4.016032383172608e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |