← All skills

onednn-fp8-gemm

85 / 100 · Low risk

Fine to try.

By ModelTC/LightX2V (GitHub) · media · 3,004 stars · scanned at commit 9cb6929

Implement or debug oneDNN quantized GEMM on Intel XPU: FP16/BF16 x FP8, INT4 weights, per-channel or block scales, zero points and bias. Use for oneDNN matmul scale APIs, JIT dispatch and reference-fallback performance.

Works with any agent that supports SKILL.md skills.

What this skill can do

Install

Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.

if [ -e ~/.agents/skills/'onednn-fp8-gemm' ] || [ -L ~/.agents/skills/'onednn-fp8-gemm' ]; then echo 'A skill folder named onednn-fp8-gemm already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/ModelTC/LightX2V' "$d" && git -C "$d" checkout --quiet '9cb6929d83814ff28c4e1386532298fe93a96af1' -- '.agents/skills/lightx2v-kernel-skills/intel-xpu/kernel-specific-skills/onednn-fp8-gemm' && mkdir -p ~/.agents/skills && cp -R "$d"/'.agents/skills/lightx2v-kernel-skills/intel-xpu/kernel-specific-skills/onednn-fp8-gemm' ~/.agents/skills/'onednn-fp8-gemm' && rm -rf "$d"; fi

Codex reads ~/.agents/skills directly. For Claude Code, Cursor, Copilot, Gemini CLI, Antigravity or OpenCode, also link it. For Claude Code: ln -s ~/.agents/skills/'onednn-fp8-gemm' ~/.claude/skills/'onednn-fp8-gemm' (other agents: ~/.cursor/skills, ~/.copilot/skills, ~/.gemini/skills, ~/.gemini/config/skills, ~/.config/opencode/skills). Or install through the finder skill, which does both.

Badge for your README

The badge for ModelTC/LightX2V shows the score of its lowest-scoring skill.

[![SkillTrust](https://zalavijaysinh.github.io/skilltrust/badges/ModelTC/LightX2V.svg)](https://zalavijaysinh.github.io/skilltrust/repos/ModelTC/LightX2V/)

Score history