| 2025-11-29 21:57:53 - train - INFO - Logging to: logs/codet5/train_20251129_215753.log |
| 2025-11-29 21:57:53 - train - INFO - Monitor progress: tail -f logs/codet5/train_20251129_215753.log |
| 2025-11-29 21:57:53 - train - INFO - ============================================================ |
| 2025-11-29 21:57:53 - train - INFO - CodeT5+ Training |
| 2025-11-29 21:57:53 - train - INFO - ============================================================ |
| 2025-11-29 21:57:53 - train - INFO - Using CUDA device: 0 |
| 2025-11-29 21:57:53 - train - INFO - GPU: NVIDIA GeForce RTX 5090 |
| 2025-11-29 21:57:53 - train - INFO - Configuration: |
| 2025-11-29 21:57:53 - train - INFO - model: Salesforce/codet5p-220m |
| 2025-11-29 21:57:53 - train - INFO - data: datasets/java |
| 2025-11-29 21:57:53 - train - INFO - output: model/checkpoints/run1-java-codet5 |
| 2025-11-29 21:57:53 - train - INFO - batch_size: 10 |
| 2025-11-29 21:57:53 - train - INFO - gradient_accumulation_steps: 4 |
| 2025-11-29 21:57:53 - train - INFO - effective_batch_size: 40 |
| 2025-11-29 21:57:53 - train - INFO - learning_rate: 5e-05 |
| 2025-11-29 21:57:53 - train - INFO - epochs: 5 |
| 2025-11-29 21:57:53 - train - INFO - max_source_len: 1024 |
| 2025-11-29 21:57:53 - train - INFO - max_target_len: 32 |
| 2025-11-29 21:57:53 - train - INFO - fp16: True |
| 2025-11-29 21:57:53 - train - INFO - seed: 42 |
| 2025-11-29 21:57:53 - train - INFO - Loading tokenizer and model... |
| 2025-11-29 21:58:04 - train - INFO - Model loaded: Salesforce/codet5p-220m |
| 2025-11-29 21:58:04 - train - INFO - Loading and preprocessing dataset... |
| 2025-11-29 21:58:06 - train - INFO - Train examples: 275962 |
| 2025-11-29 21:58:06 - train - INFO - Validation examples: 34495 |
| 2025-11-29 22:08:37 - train - INFO - Dataset preprocessing completed |
| 2025-11-29 22:08:37 - train - INFO - Starting training... |
| 2025-11-29 22:08:37 - train - INFO - Total training steps: 34495 |
| 2025-11-29 22:08:37 - train - INFO - No checkpoint found for auto-resume, starting from scratch |
| 2025-11-30 08:45:17 - train - INFO - Training completed in 38843.71 seconds (10.79 hours) |
| 2025-11-30 08:45:17 - train - INFO - Saving model to model/checkpoints/run1-java-codet5 |
| 2025-11-30 08:45:18 - train - INFO - Model and tokenizer saved successfully |
| 2025-11-30 08:45:18 - train - INFO - ============================================================ |
| 2025-11-30 08:45:18 - train - INFO - Training Summary |
| 2025-11-30 08:45:18 - train - INFO - ============================================================ |
| 2025-11-30 08:45:18 - train - INFO - Total time: 10.79 hours |
| 2025-11-30 08:45:18 - train - INFO - Output directory: model/checkpoints/run1-java-codet5 |
| 2025-11-30 08:45:18 - train - INFO - Training log: model/checkpoints/run1-java-codet5/training_log.csv |
|
|