thangcd/thread-intelligence-classifier

1{2  "best_global_step": 338,3  "best_metric": 0.9908485856905158,4  "best_model_checkpoint": "./classifier_model_refined/checkpoint-338",5  "epoch": 1.0,6  "eval_steps": 500,7  "global_step": 338,8  "is_hyper_param_search": false,9  "is_local_process_zero": true,10  "is_world_process_zero": true,11  "log_history": [12    {13      "epoch": 0.2958579881656805,14      "grad_norm": 14.180212020874023,15      "learning_rate": 1.7071005917159763e-06,16      "loss": 0.053766274452209474,17      "step": 10018    },19    {20      "epoch": 0.591715976331361,21      "grad_norm": 2.814095973968506,22      "learning_rate": 1.4112426035502957e-06,23      "loss": 0.03928844690322876,24      "step": 20025    },26    {27      "epoch": 0.8875739644970414,28      "grad_norm": 0.014567424543201923,29      "learning_rate": 1.1153846153846154e-06,30      "loss": 0.03940390110015869,31      "step": 30032    },33    {34      "epoch": 1.0,35      "eval_accuracy": 0.9908485856905158,36      "eval_f1_macro": 0.9934468211839772,37      "eval_loss": 0.06776593625545502,38      "eval_runtime": 2.8974,39      "eval_samples_per_second": 414.854,40      "eval_steps_per_second": 13.115,41      "step": 33842    }43  ],44  "logging_steps": 100,45  "max_steps": 676,46  "num_input_tokens_seen": 0,47  "num_train_epochs": 2,48  "save_steps": 500,49  "stateful_callbacks": {50    "TrainerControl": {51      "args": {52        "should_epoch_stop": false,53        "should_evaluate": false,54        "should_log": false,55        "should_save": true,56        "should_training_stop": false57      },58      "attributes": {}59    }60  },61  "total_flos": 358030248816384.0,62  "train_batch_size": 32,63  "trial_name": null,64  "trial_params": null65}66