tmalsayed/hscode-classifier

1{2  "best_metric": null,3  "best_model_checkpoint": null,4  "epoch": 4.0,5  "eval_steps": 500,6  "global_step": 124,7  "is_hyper_param_search": false,8  "is_local_process_zero": true,9  "is_world_process_zero": true,10  "log_history": [11    {12      "epoch": 1.6129032258064515,13      "grad_norm": 0.660424530506134,14      "learning_rate": 3.387096774193548e-05,15      "loss": 9.0013,16      "step": 5017    },18    {19      "epoch": 3.225806451612903,20      "grad_norm": 0.7054967284202576,21      "learning_rate": 1.774193548387097e-05,22      "loss": 8.9754,23      "step": 10024    }25  ],26  "logging_steps": 50,27  "max_steps": 155,28  "num_input_tokens_seen": 0,29  "num_train_epochs": 5,30  "save_steps": 500,31  "stateful_callbacks": {32    "TrainerControl": {33      "args": {34        "should_epoch_stop": false,35        "should_evaluate": false,36        "should_log": false,37        "should_save": true,38        "should_training_stop": false39      },40      "attributes": {}41    }42  },43  "total_flos": 8915546083368960.0,44  "train_batch_size": 256,45  "trial_name": null,46  "trial_params": null47}48