Skip to content

Bnb/multi gpu#317

Open
bnb32 wants to merge 11 commits intomainfrom
bnb/multi_gpu
Open

Bnb/multi gpu#317
bnb32 wants to merge 11 commits intomainfrom
bnb/multi_gpu

Conversation

@bnb32
Copy link
Copy Markdown
Collaborator

@bnb32 bnb32 commented Apr 29, 2026

  • Moved to MirroredStrategy for multi gpu training.
  • Added loading optmizations for eager batch handling - speeds up loading during training by 2x

bnb32 added 10 commits May 3, 2026 07:22
- Introduced caching mechanisms for eager array views in Sup3rX, including as_array and feature selection.
- Added methods to clear cache and retrieve feature indices and selectors.
- Enhanced the loaded property to utilize caching for performance.
- Updated the shape property to cache results for efficiency.
- Refined data handling in Sup3rDataset for better member access.
- Simplified batch queue and sampler classes by removing unnecessary conversions.
- Added tests for rasterizer caching and dual sampler functionality to ensure correctness.
@bnb32 bnb32 force-pushed the bnb/multi_gpu branch from 00b5035 to 5e3b1ab Compare May 3, 2026 13:22
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant