dzungpham/graphcodebert-code-classification

12026-04-20 11:06:55,458 - INFO - ===== Training Configuration =====22026-04-20 11:06:55,460 - INFO - model_name           : microsoft/graphcodebert-base32026-04-20 11:06:55,464 - INFO - output_dir           : output_checkpoints/graphcodebert-mixcode-fft42026-04-20 11:06:55,466 - INFO - num_epochs           : 452026-04-20 11:06:55,467 - INFO - batch_size           : 6462026-04-20 11:06:55,468 - INFO - learning_rate        : 5e-0572026-04-20 11:06:55,471 - INFO - max_length           : 51282026-04-20 11:06:55,472 - INFO - num_labels           : 292026-04-20 11:06:55,474 - INFO - use_wandb            : True102026-04-20 11:06:55,475 - INFO - freeze_base          : True112026-04-20 11:06:55,476 - INFO - loss_type            : r-drop122026-04-20 11:06:55,478 - INFO - focal_alpha          : 1.0132026-04-20 11:06:55,480 - INFO - focal_gamma          : 2.0142026-04-20 11:06:55,482 - INFO - r_drop_alpha         : 6.0152026-04-20 11:06:55,483 - INFO - infonce_temperature  : 0.07162026-04-20 11:06:55,485 - INFO - infonce_weight       : 0.5172026-04-20 11:06:55,487 - INFO - seed                 : 42182026-04-20 11:06:55,488 - INFO - resume_from_checkpoint : None192026-04-20 11:06:55,490 - INFO - save_steps           : 50202026-04-20 11:06:55,491 - INFO - eval_steps           : 1000212026-04-20 11:06:55,494 - INFO - label_smoothing      : 0.3222026-04-20 11:06:55,499 - INFO - adversarial_epsilon  : 0.5232026-04-20 11:06:55,504 - INFO - use_swa              : True242026-04-20 11:06:55,508 - INFO - swa_start_epoch      : 2252026-04-20 11:06:55,510 - INFO - swa_lr               : 1e-05262026-04-20 11:06:55,511 - INFO - data_augmentation    : True272026-04-20 11:06:55,512 - INFO - aug_rename_prob      : 0.6282026-04-20 11:06:55,514 - INFO - aug_format_prob      : 0.6292026-04-20 11:06:55,516 - INFO - hidden_dropout_prob  : 0.3302026-04-20 11:06:55,518 - INFO - attention_probs_dropout_prob : 0.3312026-04-20 11:06:55,519 - INFO - classifier_dropout   : 0.3322026-04-20 11:06:55,521 - INFO - mixup_alpha          : 1.0332026-04-20 11:06:55,523 - INFO - low_pass_keep_ratio  : 0.5342026-04-20 11:06:55,524 - INFO - freq_consistency_weight : 0.2352026-04-20 11:06:55,526 - INFO - =================================362026-04-20 11:06:59,772 - INFO - Model placed on cuda372026-04-20 11:06:59,789 - INFO - ===== Model Architecture =====382026-04-20 11:06:59,794 - INFO - 39RobertaForSequenceClassification(40  (roberta): RobertaModel(41    (embeddings): RobertaEmbeddings(42      (word_embeddings): Embedding(50265, 768, padding_idx=1)43      (position_embeddings): Embedding(514, 768, padding_idx=1)44      (token_type_embeddings): Embedding(1, 768)45      (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)46      (dropout): Dropout(p=0.1, inplace=False)47    )48    (encoder): RobertaEncoder(49      (layer): ModuleList(50        (0-11): 12 x RobertaLayer(51          (attention): RobertaAttention(52            (self): RobertaSdpaSelfAttention(53              (query): Linear(in_features=768, out_features=768, bias=True)54              (key): Linear(in_features=768, out_features=768, bias=True)55              (value): Linear(in_features=768, out_features=768, bias=True)56              (dropout): Dropout(p=0.1, inplace=False)57            )58            (output): RobertaSelfOutput(59              (dense): Linear(in_features=768, out_features=768, bias=True)60              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)61              (dropout): Dropout(p=0.1, inplace=False)62            )63          )64          (intermediate): RobertaIntermediate(65            (dense): Linear(in_features=768, out_features=3072, bias=True)66            (intermediate_act_fn): GELUActivation()67          )68          (output): RobertaOutput(69            (dense): Linear(in_features=3072, out_features=768, bias=True)70            (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)71            (dropout): Dropout(p=0.1, inplace=False)72          )73        )74      )75    )76  )77  (classifier): RobertaClassificationHead(78    (dense): Linear(in_features=768, out_features=768, bias=True)79    (dropout): Dropout(p=0.3, inplace=False)80    (out_proj): Linear(in_features=768, out_features=2, bias=True)81  )82)832026-04-20 11:06:59,803 - INFO - ===== Parameter Summary =====842026-04-20 11:06:59,809 - INFO - Total Parameters:         124,647,170852026-04-20 11:06:59,810 - INFO - Trainable Parameters:     592,130862026-04-20 11:06:59,811 - INFO - Non-trainable Parameters: 124,055,040872026-04-20 11:06:59,813 - INFO - ===== Tokenizer Summary =====882026-04-20 11:06:59,908 - INFO - Vocab size: 50265 | Special tokens: ['<s>', '</s>', '<unk>', '<pad>', '<mask>']892026-04-20 11:06:59,912 - INFO - ===== End of Architecture Log =====902026-04-20 11:06:59,917 - INFO - Data augmentation enabled (rename=0.6, format=0.6)912026-04-20 11:07:10,050 - INFO - === Starting training with MixCode + FFT low-pass consistency ===92