Accelerate LLM inference using speculative decoding, Medusa multiple heads, and lookahead decoding techniques. Use when optimizing inference speed (1.5-3.6Ã speedup), reducing latency for real-time applications, or deploying models with limited compute. Covers draft models, tree-based attention, Jacobi iteration, parallel token generation, and production deployment strategies.
Install
npx skills add ovachiever/droid-tings --skill speculative-decodingnpx skills add ovachiever/droid-tings --skill speculative-decoding -a claude-code -gnpx skills add ovachiever/droid-tings --skill speculative-decoding -a codex -ggemini skills install https://github.com/ovachiever/droid-tings.git --path skills/speculative-decodinggh skill install ovachiever/droid-tings speculative-decodinggit clone --depth 1 https://github.com/ovachiever/droid-tings /tmp/ovachiever__droid-tings
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/ovachiever__droid-tings/skills/speculative-decoding ~/.claude/skills/speculative-decoding # Claude Code
cp -R /tmp/ovachiever__droid-tings/skills/speculative-decoding ~/.agents/skills/speculative-decoding # Codex, Cursor, Gemini CLI, Copilot, OpenCode