Optimizes transformer attention with Flash Attention for 2-4x speedup and 10-20x memory reduction. Use when training/running transformers with long sequences (>512 tokens), encountering GPU memory issues with attention, or need faster inference. Supports PyTorch native SDPA, flash-attn library, H100 FP8, and sliding window attention.
安装
npx skills add davila7/claude-code-templates --skill optimizing-attention-flashnpx skills add davila7/claude-code-templates --skill optimizing-attention-flash -a claude-code -gnpx skills add davila7/claude-code-templates --skill optimizing-attention-flash -a codex -ggemini skills install https://github.com/davila7/claude-code-templates.git --path cli-tool/components/skills/ai-research/optimization-flash-attentiongh skill install davila7/claude-code-templates optimizing-attention-flashgit clone --depth 1 https://github.com/davila7/claude-code-templates /tmp/davila7__claude-code-templates
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/davila7__claude-code-templates/cli-tool/components/skills/ai-research/optimization-flash-attention ~/.claude/skills/optimizing-attention-flash # Claude Code
cp -R /tmp/davila7__claude-code-templates/cli-tool/components/skills/ai-research/optimization-flash-attention ~/.agents/skills/optimizing-attention-flash # Codex, Cursor, Gemini CLI, Copilot, OpenCode