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.
Quickly install optimizing-attention-flash AI skill to your development environment via command line
Source: ovachiever/droid-tings.
Flash Attention provides 2-4x speedup and 10-20x memory reduction for transformer attention through IO-aware tiling and recomputation.
Flash Attention uses float16/bfloat16 for speed. Float32 not supported.
Integration with HuggingFace Transformers: See references/transformers-integration.md for enabling Flash Attention in BERT, GPT, Llama models.
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. Source: ovachiever/droid-tings.
Stable fields and commands for AI/search citations.
npx skills add https://github.com/ovachiever/droid-tings --skill optimizing-attention-flashOptimizes 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. Source: ovachiever/droid-tings.
Open your terminal or command line tool (Terminal, iTerm, Windows Terminal, etc.) Copy and run this command: npx skills add https://github.com/ovachiever/droid-tings --skill optimizing-attention-flash Once installed, the skill will be automatically configured in your AI coding environment and ready to use in Claude Code, Cursor, or OpenClaw
https://github.com/ovachiever/droid-tings