| 2026-04-23 14:52:26,578 - INFO - ===== Training Configuration ===== |
| 2026-04-23 14:52:26,580 - INFO - model_name : microsoft/unixcoder-base |
| 2026-04-23 14:52:26,582 - INFO - output_dir : output_checkpoints/unixcoder-base/ |
| 2026-04-23 14:52:26,583 - INFO - num_epochs : 0.5 |
| 2026-04-23 14:52:26,584 - INFO - max_steps : 50 |
| 2026-04-23 14:52:26,585 - INFO - batch_size : 32 |
| 2026-04-23 14:52:26,587 - INFO - learning_rate : 1e-06 |
| 2026-04-23 14:52:26,588 - INFO - max_length : 512 |
| 2026-04-23 14:52:26,589 - INFO - num_labels : 2 |
| 2026-04-23 14:52:26,591 - INFO - use_wandb : True |
| 2026-04-23 14:52:26,592 - INFO - freeze_base : True |
| 2026-04-23 14:52:26,593 - INFO - loss_type : r-drop |
| 2026-04-23 14:52:26,594 - INFO - focal_alpha : 1.0 |
| 2026-04-23 14:52:26,595 - INFO - focal_gamma : 2.0 |
| 2026-04-23 14:52:26,596 - INFO - r_drop_alpha : 6.0 |
| 2026-04-23 14:52:26,597 - INFO - infonce_temperature : 0.07 |
| 2026-04-23 14:52:26,598 - INFO - infonce_weight : 0.5 |
| 2026-04-23 14:52:26,599 - INFO - seed : 42 |
| 2026-04-23 14:52:26,600 - INFO - resume_from_checkpoint : None |
| 2026-04-23 14:52:26,601 - INFO - label_smoothing : 0.3 |
| 2026-04-23 14:52:26,602 - INFO - adversarial_epsilon : 0.5 |
| 2026-04-23 14:52:26,604 - INFO - use_swa : False |
| 2026-04-23 14:52:26,606 - INFO - swa_start_epoch : 0 |
| 2026-04-23 14:52:26,606 - INFO - swa_lr : 1e-05 |
| 2026-04-23 14:52:26,607 - INFO - data_augmentation : True |
| 2026-04-23 14:52:26,609 - INFO - aug_rename_prob : 0.6 |
| 2026-04-23 14:52:26,610 - INFO - aug_format_prob : 0.6 |
| 2026-04-23 14:52:26,612 - INFO - mixup_alpha : 1.0 |
| 2026-04-23 14:52:26,613 - INFO - low_pass_keep_ratio : 0.5 |
| 2026-04-23 14:52:26,614 - INFO - freq_consistency_weight : 0.2 |
| 2026-04-23 14:52:26,615 - INFO - hidden_dropout_prob : 0.3 |
| 2026-04-23 14:52:26,616 - INFO - attention_probs_dropout_prob : 0.3 |
| 2026-04-23 14:52:26,618 - INFO - classifier_dropout : 0.3 |
| 2026-04-23 14:52:26,619 - INFO - ================================= |
| 2026-04-23 14:52:27,631 - INFO - Model placed on cuda |
| 2026-04-23 14:52:27,635 - INFO - ===== Model Architecture ===== |
| 2026-04-23 14:52:27,638 - INFO - |
| RobertaForSequenceClassification( |
| (roberta): RobertaModel( |
| (embeddings): RobertaEmbeddings( |
| (word_embeddings): Embedding(51416, 768, padding_idx=1) |
| (position_embeddings): Embedding(1026, 768, padding_idx=1) |
| (token_type_embeddings): Embedding(10, 768) |
| (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True) |
| (dropout): Dropout(p=0.3, inplace=False) |
| ) |
| (encoder): RobertaEncoder( |
| (layer): ModuleList( |
| (0-11): 12 x RobertaLayer( |
| (attention): RobertaAttention( |
| (self): RobertaSdpaSelfAttention( |
| (query): Linear(in_features=768, out_features=768, bias=True) |
| (key): Linear(in_features=768, out_features=768, bias=True) |
| (value): Linear(in_features=768, out_features=768, bias=True) |
| (dropout): Dropout(p=0.3, inplace=False) |
| ) |
| (output): RobertaSelfOutput( |
| (dense): Linear(in_features=768, out_features=768, bias=True) |
| (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True) |
| (dropout): Dropout(p=0.3, inplace=False) |
| ) |
| ) |
| (intermediate): RobertaIntermediate( |
| (dense): Linear(in_features=768, out_features=3072, bias=True) |
| (intermediate_act_fn): GELUActivation() |
| ) |
| (output): RobertaOutput( |
| (dense): Linear(in_features=3072, out_features=768, bias=True) |
| (LayerNorm): LayerNorm((768,), eps=1e-05, elementwise_affine=True) |
| (dropout): Dropout(p=0.3, inplace=False) |
| ) |
| ) |
| ) |
| ) |
| ) |
| (classifier): RobertaClassificationHead( |
| (dense): Linear(in_features=768, out_features=768, bias=True) |
| (dropout): Dropout(p=0.3, inplace=False) |
| (out_proj): Linear(in_features=768, out_features=2, bias=True) |
| ) |
| ) |
| 2026-04-23 14:52:27,640 - INFO - ===== Parameter Summary ===== |
| 2026-04-23 14:52:27,642 - INFO - Total Parameters: 125,931,266 |
| 2026-04-23 14:52:27,643 - INFO - Trainable Parameters: 592,130 |
| 2026-04-23 14:52:27,646 - INFO - Non-trainable Parameters: 125,339,136 |
| 2026-04-23 14:52:27,647 - INFO - ===== Tokenizer Summary ===== |
| 2026-04-23 14:52:27,663 - INFO - Vocab size: 51416 | Special tokens: ['<s>', '</s>', '<unk>', '<pad>', '<mask>'] |
| 2026-04-23 14:52:27,665 - INFO - ===== End of Architecture Log ===== |
| 2026-04-23 14:52:27,666 - INFO - Data augmentation enabled (rename=0.6, format=0.6) |
| 2026-04-23 14:52:28,797 - INFO - === Starting training with MixCode + FFT low-pass consistency === |
| 2026-04-23 14:54:25,677 - INFO - Training complete! |
| 2026-04-23 14:54:26,437 - INFO - Final model saved to output_checkpoints/unixcoder-base/final_model |
|
|
