dzungpham/graphcodebert-code-classification

1{2  "train_config": {3    "model_name": "microsoft/graphcodebert-base",4    "num_epochs": 3,5    "batch_size": 256,6    "learning_rate": 2e-05,7    "max_length": 512,8    "num_labels": 2,9    "loss_type": "ce",10    "focal_alpha": 1.0,11    "focal_gamma": 2.0,12    "r_drop_alpha": 6.0,13    "infonce_temperature": 0.07,14    "infonce_weight": 0.5,15    "label_smoothing": 0,16    "adversarial_epsilon": 0,17    "use_swa": false,18    "swa_start_epoch": 0,19    "swa_lr": 1e-05,20    "data_augmentation": false,21    "aug_rename_prob": 0.0,22    "aug_format_prob": 0.0,23    "freeze_base": true,24    "seed": 42,25    "use_wandb": true,26    "mixup_alpha": 0.0,27    "low_pass_keep_ratio": 0.5,28    "freq_consistency_weight": 0.029  },30  "training_arguments": {31    "output_dir": "output_checkpoints/graphcodebert-vanilla/",32    "num_train_epochs": 3,33    "per_device_train_batch_size": 256,34    "per_device_eval_batch_size": 512,35    "learning_rate": 2e-05,36    "warmup_steps": 612,37    "weight_decay": 0.1,38    "logging_steps": 5,39    "eval_steps": 50,40    "save_steps": 100,41    "metric_for_best_model": "macro_f1",42    "greater_is_better": true,43    "save_total_limit": 5,44    "fp16": true,45    "seed": 4246  },47  "training_state": {48    "global_step": 200,49    "epoch": 0.19569471624266144,50    "best_metric": 0.5962545254252696,51    "best_model_checkpoint": "output_checkpoints/graphcodebert-vanilla/checkpoint-200"52  }53}