huggingface-accelerate

Run PyTorch training across GPUs with minimal changes.

Reference it in any coding agent with:

@skills nousresearch/huggingface-accelerate