dzungpham/graphcodebert-code-classification

12026-04-15 16:22:36,752 - INFO - train_pipeline - Logging to ./output_checkpoints/graphcodebert-focal/training.log22026-04-15 16:22:36,755 - INFO - train_pipeline - Loading model & tokenizer for 'microsoft/graphcodebert-base'32026-04-15 16:22:39,187 - INFO - train_pipeline - Model placed on cuda42026-04-15 16:22:39,191 - INFO - train_pipeline - Base model weights frozen – only classifier head will be trained.52026-04-15 16:22:39,192 - INFO - train_pipeline - ===== Model Architecture =====62026-04-15 16:22:39,196 - INFO - train_pipeline - 7RobertaForSequenceClassification(8  (roberta): RobertaModel(9    (embeddings): RobertaEmbeddings(10      (word_embeddings): Embedding(50265, 768, padding_idx=1)11      (position_embeddings): Embedding(514, 768, padding_idx=1)12      (token_type_embeddings): Embedding(1, 768)13      (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)14      (dropout): Dropout(p=0.2, inplace=False)15    )16    (encoder): RobertaEncoder(17      (layer): ModuleList(18        (0-11): 12 x RobertaLayer(19          (attention): RobertaAttention(20            (self): RobertaSdpaSelfAttention(21              (query): Linear(in_features=768, out_features=768, bias=True)22              (key): Linear(in_features=768, out_features=768, bias=True)23              (value): Linear(in_features=768, out_features=768, bias=True)24              (dropout): Dropout(p=0.2, inplace=False)25            )26            (output): RobertaSelfOutput(27              (dense): Linear(in_features=768, out_features=768, bias=True)28              (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)29              (dropout): Dropout(p=0.2, inplace=False)30            )31          )32          (intermediate): RobertaIntermediate(33            (dense): Linear(in_features=768, out_features=3072, bias=True)34            (intermediate_act_fn): GELUActivation()35          )36          (output): RobertaOutput(37            (dense): Linear(in_features=3072, out_features=768, bias=True)38            (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True)39            (dropout): Dropout(p=0.2, inplace=False)40          )41        )42      )43    )44  )45  (classifier): RobertaClassificationHead(46    (dense): Linear(in_features=768, out_features=768, bias=True)47    (dropout): Dropout(p=0.2, inplace=False)48    (out_proj): Linear(in_features=768, out_features=2, bias=True)49  )50)512026-04-15 16:22:39,199 - INFO - train_pipeline - ===== Parameter Summary =====522026-04-15 16:22:39,200 - INFO - train_pipeline - Total Parameters:         124,647,170532026-04-15 16:22:39,201 - INFO - train_pipeline - Trainable Parameters:     592,130542026-04-15 16:22:39,204 - INFO - train_pipeline - Non-trainable Parameters: 124,055,040552026-04-15 16:22:39,205 - INFO - train_pipeline - ===== Tokenizer Summary =====562026-04-15 16:22:39,227 - INFO - train_pipeline - Vocab size: 50265 | Special tokens: ['<s>', '</s>', '<unk>', '<pad>', '<mask>']572026-04-15 16:22:39,230 - INFO - train_pipeline - ===== End of Architecture Log =====582026-04-15 16:23:26,328 - INFO - train_pipeline - === Starting training ===59