dzungpham/graphcodebert-code-classification

1{2  "train_config": {3    "TEST": false,4    "model_name": "/kaggle/input/models/dzung271828/microsoft-graphcodebert-base/transformers/default/1",5    "output_dir": "training/fourier-spectral-norm-classifier/",6    "num_epochs": 5,7    "max_steps": -1,8    "batch_size": 512,9    "learning_rate": 1e-06,10    "max_length": 512,11    "num_labels": 2,12    "use_wandb": false,13    "freeze_base": true,14    "loss_type": "ce",15    "focal_alpha": 1.0,16    "focal_gamma": 2.0,17    "r_drop_alpha": 6.0,18    "infonce_temperature": 0.07,19    "infonce_weight": 0.5,20    "seed": 42,21    "resume_from_checkpoint": null,22    "save_steps": 500,23    "eval_steps": 500,24    "logging_steps": 5,25    "label_smoothing": 0.5,26    "adversarial_epsilon": 0.5,27    "use_swa": true,28    "swa_start_epoch": 0,29    "swa_lr": 1e-06,30    "data_augmentation": true,31    "aug_rename_prob": 0.7,32    "aug_format_prob": 0.7,33    "weight_decay": 0.1,34    "mixup_alpha": 1.0,35    "low_pass_keep_ratio": 0.5,36    "freq_consistency_weight": 0.2,37    "use_mixcode": true,38    "use_fgm": true,39    "fgm_freq": 5,40    "use_r_drop": true,41    "use_freq_consistency_loss": true,42    "use_attn_spectral": false,43    "attn_spectral_weight": 0.1,44    "attn_spectral_cutoff_ratio": 0.25,45    "hidden_dropout_prob": 0.3,46    "attention_probs_dropout_prob": 0.3,47    "classifier_dropout": 0.4,48    "device": "cuda",49    "torch_compile": true,50    "cache_dir": "./tokenized_cache",51    "use_swa_actual": true,52    "use_fgm_actual": true,53    "use_r_drop_actual": true,54    "use_mixcode_actual": true,55    "use_attn_spectral_actual": false,56    "use_freq_consistency_loss_actual": true,57    "use_spectral_norm": true58  },59  "training_arguments": {60    "output_dir": "training/fourier-spectral-norm-classifier/",61    "num_train_epochs": 5,62    "per_device_train_batch_size": 512,63    "per_device_eval_batch_size": 1024,64    "learning_rate": 1e-06,65    "warmup_steps": 488,66    "weight_decay": 0.1,67    "logging_steps": 5,68    "eval_steps": 500,69    "save_steps": 500,70    "metric_for_best_model": "macro_f1",71    "greater_is_better": true,72    "save_total_limit": 5,73    "fp16": false,74    "seed": 4275  },76  "training_state": {77    "global_step": 1500,78    "epoch": 1.5353121801432958,79    "best_metric": 0.6724504812400831,80    "best_model_checkpoint": "training/fourier-spectral-norm-classifier/checkpoint-1000"81  }82}