Optimizing Attention Flash

Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Supports PyTorch native SDPA, flash-attn library, H100 FP8, and sliding window attention.

Orchestra Research Updated 14.5k repo stars
AI Research Skills Demo

File contents

Orchestra-Research/AI-Research-SKILLs commit 773a52944b

Frequently asked questions

npx get-skill add orchestra-research/optimizing-attention-flash