dzungpham/graphcodebert-code-classification

12026-04-23 14:52:26,578 - INFO - ===== Training Configuration =====22026-04-23 14:52:26,580 - INFO - model_name           : microsoft/unixcoder-base32026-04-23 14:52:26,582 - INFO - output_dir           : output_checkpoints/unixcoder-base/42026-04-23 14:52:26,583 - INFO - num_epochs           : 0.552026-04-23 14:52:26,584 - INFO - max_steps            : 5062026-04-23 14:52:26,585 - INFO - batch_size           : 3272026-04-23 14:52:26,587 - INFO - learning_rate        : 1e-0682026-04-23 14:52:26,588 - INFO - max_length           : 51292026-04-23 14:52:26,589 - INFO - num_labels           : 2102026-04-23 14:52:26,591 - INFO - use_wandb            : True112026-04-23 14:52:26,592 - INFO - freeze_base          : True122026-04-23 14:52:26,593 - INFO - loss_type            : r-drop132026-04-23 14:52:26,594 - INFO - focal_alpha          : 1.0142026-04-23 14:52:26,595 - INFO - focal_gamma          : 2.0152026-04-23 14:52:26,596 - INFO - r_drop_alpha         : 6.0162026-04-23 14:52:26,597 - INFO - infonce_temperature  : 0.07172026-04-23 14:52:26,598 - INFO - infonce_weight       : 0.5182026-04-23 14:52:26,599 - INFO - seed                 : 42192026-04-23 14:52:26,600 - INFO - resume_from_checkpoint : None202026-04-23 14:52:26,601 - INFO - label_smoothing      : 0.3212026-04-23 14:52:26,602 - INFO - adversarial_epsilon  : 0.5222026-04-23 14:52:26,604 - INFO - use_swa              : False232026-04-23 14:52:26,606 - INFO - swa_start_epoch      : 0242026-04-23 14:52:26,606 - INFO - swa_lr               : 1e-05252026-04-23 14:52:26,607 - INFO - data_augmentation    : True262026-04-23 14:52:26,609 - INFO - aug_rename_prob      : 0.6272026-04-23 14:52:26,610 - INFO - aug_format_prob      : 0.6282026-04-23 14:52:26,612 - INFO - mixup_alpha          : 1.0292026-04-23 14:52:26,613 - INFO - low_pass_keep_ratio  : 0.5302026-04-23 14:52:26,614 - INFO - freq_consistency_weight : 0.2312026-04-23 14:52:26,615 - INFO - hidden_dropout_prob  : 0.3322026-04-23 14:52:26,616 - INFO - attention_probs_dropout_prob : 0.3332026-04-23 14:52:26,618 - INFO - classifier_dropout   : 0.3342026-04-23 14:52:26,619 - INFO - =================================352026-04-23 14:52:27,631 - INFO - Model placed on cuda362026-04-23 14:52:27,635 - INFO - ===== Model Architecture =====372026-04-23 14:52:27,638 - INFO - 38RobertaForSequenceClassification(39  (roberta): RobertaModel(40    (embeddings): RobertaEmbeddings(41      (word_embeddings): Embedding(51416, 768, padding_idx=1)42      (position_embeddings): Embedding(1026, 768, padding_idx=1)43      (token_type_embeddings): Embedding(10, 768)44      (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)45      (dropout): Dropout(p=0.3, inplace=False)46    )47    (encoder): RobertaEncoder(48      (layer): ModuleList(49        (0-11): 12 x RobertaLayer(50          (attention): RobertaAttention(51            (self): RobertaSdpaSelfAttention(52              (query): Linear(in_features=768, out_features=768, bias=True)53              (key): Linear(in_features=768, out_features=768, bias=True)54              (value): Linear(in_features=768, out_features=768, bias=True)55              (dropout): Dropout(p=0.3, inplace=False)56            )57            (output): RobertaSelfOutput(58              (dense): Linear(in_features=768, out_features=768, bias=True)59              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)60              (dropout): Dropout(p=0.3, inplace=False)61            )62          )63          (intermediate): RobertaIntermediate(64            (dense): Linear(in_features=768, out_features=3072, bias=True)65            (intermediate_act_fn): GELUActivation()66          )67          (output): RobertaOutput(68            (dense): Linear(in_features=3072, out_features=768, bias=True)69            (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)70            (dropout): Dropout(p=0.3, inplace=False)71          )72        )73      )74    )75  )76  (classifier): RobertaClassificationHead(77    (dense): Linear(in_features=768, out_features=768, bias=True)78    (dropout): Dropout(p=0.3, inplace=False)79    (out_proj): Linear(in_features=768, out_features=2, bias=True)80  )81)822026-04-23 14:52:27,640 - INFO - ===== Parameter Summary =====832026-04-23 14:52:27,642 - INFO - Total Parameters:         125,931,266842026-04-23 14:52:27,643 - INFO - Trainable Parameters:     592,130852026-04-23 14:52:27,646 - INFO - Non-trainable Parameters: 125,339,136862026-04-23 14:52:27,647 - INFO - ===== Tokenizer Summary =====872026-04-23 14:52:27,663 - INFO - Vocab size: 51416 | Special tokens: ['<s>', '</s>', '<unk>', '<pad>', '<mask>']882026-04-23 14:52:27,665 - INFO - ===== End of Architecture Log =====892026-04-23 14:52:27,666 - INFO - Data augmentation enabled (rename=0.6, format=0.6)902026-04-23 14:52:28,797 - INFO - === Starting training with MixCode + FFT low-pass consistency ===912026-04-23 14:54:25,677 - INFO - Training complete!922026-04-23 14:54:26,437 - INFO - Final model saved to output_checkpoints/unixcoder-base/final_model93