npu-torchair-infer

75 / 100 · Review before installing

Check before installing. Some of what it can do is expected: a media skill usually calls an AI service over the internet.

By ascend-ai-coding/awesome-ascend-skills (GitHub) · media · 184 stars · scanned at commit 75ec89c

Migrate any HuggingFace model to Ascend NPU torchair graph mode (torch.compile) and benchmark it for accuracy and performance against NPU eager and CPU eager. Use when running, compiling, or benchmarking HF models (vision, text, image-text encoders such as SigLIP2, DINOv3, ViT, CLIP, Qwen-VL, SAM) on Ascend 910B/CANN with torch_npu and torchair; when a torch.compile graph-mode run on NPU fails (Dynamo TorchRuntimeError, unsupported op, interpolate/contiguous errors); or when comparing torchair …

Works with any agent that supports SKILL.md skills.

What this skill can do

Runs commands on your computer

The author does not mention this.

Found in 1 place.

skills/inference/npu-torchair-infer/SKILL.md, line 39source $ASCEND_HOME/ascend-toolkit/set_env.sh # CANN runtime first

Downloads and installs other software

The author does not mention this.

Found in 1 place.

skills/inference/npu-torchair-infer/references/environment.md, line 49- torchair ships with torch_npu — never `pip install torchair` separately.

Connects to the internet

The author does not mention this. Expected for a media skill.

Found in 2 places.

skills/inference/npu-torchair-infer/scripts/run_benchmark.sh, line 49curl -fsSL --retry 3 --connect-timeout 15 --max-time 120 -o "${dest}/${f}" "${base}/${f}" 2>/dev/null || true
skills/inference/npu-torchair-infer/scripts/run_benchmark.sh, line 52if curl -fSL --retry 3 --connect-timeout 15 --max-time "${WEIGHT_MAX_TIME}" \

Install

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

if [ -e ~/.agents/skills/'npu-torchair-infer' ] || [ -L ~/.agents/skills/'npu-torchair-infer' ]; then echo 'A skill folder named npu-torchair-infer already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/ascend-ai-coding/awesome-ascend-skills' "$d" && git -C "$d" checkout --quiet '75ec89ce9d401be8ac95980e1c3f42f8d4315057' -- 'skills/inference/npu-torchair-infer' && mkdir -p ~/.agents/skills && cp -R "$d"/'skills/inference/npu-torchair-infer' ~/.agents/skills/'npu-torchair-infer' && 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/'npu-torchair-infer' ~/.claude/skills/'npu-torchair-infer' (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 ascend-ai-coding/awesome-ascend-skills shows the score of its lowest-scoring skill.

SkillTrust badge for ascend-ai-coding/awesome-ascend-skills

[![SkillTrust](https://zalavijaysinh.github.io/skilltrust/badges/ascend-ai-coding/awesome-ascend-skills.svg)](https://zalavijaysinh.github.io/skilltrust/repos/ascend-ai-coding/awesome-ascend-skills/)

Score history

Date	Version	Score
2026-11-08	`75ec89c`	100