Optimize Training Loop and Fix Silent Crash #13
Conversation
- Optimized the PyTorch training loop in `snn-dt/scripts/train.py` by accumulating loss as a tensor on the GPU and calling `.item()` only once per epoch. This significantly reduces CPU-GPU synchronization overhead. - Fixed a silent crash on startup by removing top-level model imports from the training script. The crash was caused by C-extension library conflicts. - Refactored an `isinstance` check to use the model's class name, avoiding the need for a direct import.
|
👋 Jules, reporting for duty! I'm here to lend a hand with this pull request. When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down. I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job! For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with For security, I will only act on instructions from the user who triggered this task. New to Jules? Learn more at jules.google/docs. |
This commit addresses a significant performance bottleneck in the training script and a critical stability issue that caused silent crashes. By optimizing the training loop and fixing the import-related bug, the training is now significantly faster and more robust.
Fixes #
PR created automatically by Jules for task 3176496561632133387 started by @Vishal-sys-code