modelopt-release-cherry-pick
Skill5.6k starsChanged 3 days agoApache-2.0View on GitHub
Audit merged bug-fix PRs and release NVBugs for missing cherry-pick labels, then cherry-pick labeled PRs into a release branch and open a PR. Use when asked to "cherry-pick PRs for release/X.Y.Z", "pick PRs to release branch", "verify cherry-pick labels", or "cherry-pick labeled PRs".
- Commits and pushes
Trending
More talk lately.about the whole repository
Who's talking about NVIDIA/Model-Optimizer
- @github-trending.bsky.social“📦 NVIDIA / Model-Optimizer ⭐ 3,886 (+22) 🗒 Python A unified library of SOTA model optimization techniques …”Bluesky · Sep 25, 2026 · 5 likes
- onnxsim/onnxsim“Maintaining status of this project”GitHub · Jan 6, 2026 · 5 reactions
- vllm-project/vllm“Add LM head quantization support for ModelOpt”GitHub · May 8, 2026 · 4 reactions
- @chrisshort.net“to optimize inference speed. #devopsish”Bluesky · Sep 27, 2026 · 1 likes
- @dailygithubtrends.bsky.social“今日のGitHubトレンド NVIDIA/Model-Optimizer NVIDIA Model Optimizerは、量子化やプルーニング、蒸留などの手法を用いてモデルを高速化するためのライブラリです。 …”Bluesky · Sep 26, 2026 · 1 likes
The biggest 5 of 54 mentions linking to the repository that our crawl found. Each line is their own headline.
See all 54
Talked about on GitHub and Bluesky (54 times).
On GitHub · 50
- Jan 6, 2026Maintaining status of this project — onnxsim/onnxsim5 reactions · 4 comments
- May 8, 2026Add LM head quantization support for ModelOpt — vllm-project/vllm4 reactions · 7 comments
- Apr 12, 2026Add support for ModelOpt MXFP8 non-gated MoE with ReLU2 — vllm-project/vllm1 reaction · 3 comments
- Apr 2, 2026[Bugfix] Fix NVFP4+MTP crash: force unquantized mtp.fc for Qwen3.5 — vllm-project/vllm1 reaction
- Feb 8, 2026[ModelOpt MXFP8] — sgl-project/sglang1 reaction · 5 comments
45 more on GitHub
- Jan 21, 2026ORT inference fails after upgrading model's opset from 20 to 21 — microsoft/onnxruntime1 reaction · 5 comments
- Nov 15, 2026Watch list: who to follow and where (Edge0 upstream, Qwen4, llama.cpp, MLX, the labs) — toshi38/Edge00 reactions
- Nov 14, 2026Bump nvidia-modelopt from 0.47.0 to 0.48.0 — marek-burza/shrubbery0 reactions
- Nov 11, 2026build(deps-dev): bump nvidia-modelopt from 0.47.0 to 0.48.0 — verl-project/verl0 reactions · 4 comments
- Nov 10, 2026Prepare ModelOpt catalog onboarding: original six skills plus QAD — NVIDIA/skills0 reactions
- Nov 8, 2026build(deps-dev): bump nvidia-modelopt from 0.47.0 to 0.48.0 — ysq111333/verl_jev0 reactions
- Nov 5, 2026python3Packages.nvidia-modelopt: 0.47.0 -> 0.48.0 — NixOS/nixpkgs0 reactions · 1 comment
- Nov 2, 2026claude-code-marketplaces.github.nvidia.model-optimizer: 0.47.0 → 0.48.0 — sudosubin/agents.nix0 reactions
- Nov 2, 2026codex-plugins.github.nvidia.model-optimizer: 0.47.0 → 0.48.0 — sudosubin/agents.nix0 reactions
- Nov 2, 2026claude-code-plugins.github.nvidia.model-optimizer: 0.47.0 → 0.48.0 — sudosubin/agents.nix0 reactions
- Nov 2, 2026build(deps): Update nvidia-modelopt requirement from ~=0.47.0 to ~=0.48.0 — Jonah-May-OSS/wyoming-whisper-trt0 reactions · 2 comments
- Nov 2, 2026codex-marketplaces.github.nvidia.model-optimizer: 0.47.0 → 0.48.0 — sudosubin/agents.nix0 reactions
- Oct 28, 2026claude-code-plugins.github.nvidia.model-optimizer: init at 0.47.0 — sudosubin/agents.nix0 reactions
- Oct 28, 2026claude-code-marketplaces.github.nvidia.model-optimizer: init at 0.47.0 — sudosubin/agents.nix0 reactions
- Oct 28, 2026[ONNX GraphSurgeon] Export fails for a single initializer larger than 2 GiB before external-data saving — NVIDIA/TensorRT0 reactions · 2 comments
- Oct 28, 2026[Bug] make_graph serializes an individual >2 GiB initializer before external-data saving — onnx/onnx0 reactions
- Oct 26, 2026build(deps-dev): bump nvidia-modelopt from 0.44.0 to 0.47.0 — WncFht/verl-opd0 reactions
- Oct 22, 2026chore(deps-dev): Bump nvidia-modelopt from 0.44.0 to 0.47.0 — yiqiu0215/verl-main0 reactions
- Oct 14, 2026Bump nvidia-modelopt from 0.44.0 to 0.47.0 — zery00568/verlxx0 reactions
- Oct 8, 2026Add native packed-cache attention tests — yingguo-trt/Model-Optimizer0 reactions
- Oct 8, 2026Add inline-media CLI HTTP tests — yeyu-nvidia/Model-Optimizer0 reactions
- Oct 8, 2026Add FP8 zero-scale and export tests — raashish1601/Model-Optimizer0 reactions
- Oct 8, 2026Add native CUDA coverage for MXFP4 packing — raashish1601/Model-Optimizer0 reactions
- Oct 8, 2026Add NVFP4 offloaded export tests and collector typing — niko4244/Model-Optimizer0 reactions
- Oct 4, 2026codex-plugins.github.nvidia.model-optimizer: init at 0.47.0 — sudosubin/agents.nix0 reactions
- Oct 4, 2026codex-marketplaces.github.nvidia.model-optimizer: init at 0.47.0 — sudosubin/agents.nix0 reactions
- Oct 2, 2026build(deps-dev): bump nvidia-modelopt from 0.44.0 to 0.47.0 — ydwhhh/gspo0 reactions
- Sep 30, 2026build(deps-dev): bump nvidia-modelopt from 0.44.0 to 0.47.0 — verl-project/verl0 reactions · 1 comment
- Sep 29, 2026Admin AI: Install skills https://github.com/mattpocock/skills https://github.com/ — mehrshadevilb-cell/ArtistYar-Website0 reactions
- Sep 29, 2026Admin AI: Install skills https://github.com/mattpocock/skills https://github.com/ — mehrshadevilb-cell/ArtistYar-Website0 reactions
- Sep 27, 2026🔥 GitHub Trending AI 早报 · 2026-09-27 · Top 6 项目 — happydog-intj/ai-xiaohongshu-daily0 reactions
- Sep 27, 2026🔥 GitHub Trending AI 早报 · 2026-09-27 · Top 6 项目 — Mystery406/ai-xiaohongshu-daily0 reactions
- Sep 27, 2026Github trending [NVIDIA/Model-Optimizer] 一站式模型量化优化加速 — web1992/github-trending0 reactions
- Sep 26, 2026🔥 GitHub Trending AI 早报 · 2026-09-27 · Top 6 项目 — Mystery406/ai-xiaohongshu-daily0 reactions
- Sep 26, 2026🔥 GitHub Trending AI 早报 · 2026-09-27 · Top 6 项目 — happydog-intj/ai-xiaohongshu-daily0 reactions
- Sep 25, 2026Github trending [NVIDIA/Model-Optimizer] 模型推理加速优化库 — web1992/github-trending0 reactions
- Sep 19, 2026Bump nvidia-modelopt from 0.46.0 to 0.46.1 — marek-burza/shrubbery0 reactions
- Sep 17, 2026Add Anima 2B support to Model Optimizer for calibrated PTQ and future NVFP4 QAT/QAD — ruwwww/Model-Optimizer0 reactions
- Sep 16, 2026build(deps-dev): bump nvidia-modelopt from 0.44.0 to 0.46.1 — verl-project/verl0 reactions · 1 comment
- Sep 15, 2026Bump nvidia-modelopt from 0.44.0 to 0.46.1 — HuipengHuang/opd-rl0 reactions
- Sep 9, 2026build(deps-dev): bump nvidia-modelopt from 0.44.0 to 0.46.0 — HuipengHuang/verl-baseline0 reactions
- Sep 9, 2026[None][docs] retarget ModelOpt examples llm_ptq paths to hf_ptq — NVIDIA/TensorRT-LLM0 reactions · 1 comment
- Sep 4, 2026DeepSeek-V4-Pro-0813-NVFP4 — Bibbleq/frontier-model-tracker0 reactions · 1 comment
- Aug 28, 2026[Doc] Broken links in READMEs referencing files removed by legacy TensorRT backend cleanup — NVIDIA/TensorRT-LLM0 reactions · 1 comment
- Aug 19, 2026Fix ModelOpt integration for the >=0.44 list quant_cfg format — huggingface/diffusers0 reactions
On Bluesky · 4
- Sep 25, 2026📦 NVIDIA / Model-Optimizer ⭐ 3,886 (+22) 🗒 Python A unified library of SOTA model optimization techniques like quantization,… — @github-trending.bsky.social5 likes and reposts
- Sep 27, 2026to optimize inference speed. #devopsish — @chrisshort.net1 likes and repost
- Sep 26, 2026今日のGitHubトレンド NVIDIA/Model-Optimizer NVIDIA Model Optimizerは、量子化やプルーニング、蒸留などの手法を用いてモデルを高速化するためのライブラリです。 Hugging… — @dailygithubtrends.bsky.social1 likes and repost
- Sep 25, 20265. NVIDIA Model-Optimizer Library for Deployment LINK 6. World Action Agent for Robot Manipulation via VLMs LINK — @hackobar.bsky.social0 likes and reposts · 1 reply
In this repository33 other files it gives its agents.
- AGENTS.mdAGENTS.md
- tools/debugger/CLAUDE.mdCLAUDE.md
- tools/launcher/CLAUDE.mdCLAUDE.md
- accessing-mlflowSkill
- benchmark-model-kernelsSkill
- commonSkill
- compare-resultsSkill
- day0-releaseSkill
- debugSkill
- deploymentSkill
- evaluationSkill
- launching-evalsSkill
- modelopt-accessing-mlflowSkill
- modelopt-benchmark-model-kernelsSkill
- modelopt-compare-resultsSkill
- modelopt-day0-releaseSkill
- modelopt-debugSkill
- modelopt-deploymentSkill
- modelopt-evaluationSkill
- modelopt-launching-evalsSkill
- modelopt-monitorSkill
- modelopt-ptqSkill
- modelopt-qadSkill
- modelopt-quant-recipe-searchSkill
- modelopt-run-agentxSkill
- modelopt-speculative-decodingSkill
- monitorSkill
- ptqSkill
- qadSkill
- quant-recipe-searchSkill
- release-cherry-pickSkill
- run-agentxSkill
- speculative-decodingSkill
The file
We split this skill into the rules it gives an agent
10 rules. About 0 are common, 5 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 12
Ask the user for the release version (e.g.
- Distinctive: line 18
Before fetching the labeled queue, audit for missing bug fixes:
- Very specific: line 25
Use the NVBugs MCP to search for every NVBug whose Keywords field contains the exact keyword Committed_ModelOpt_<VERSION>.
- Distinctive: line 90
Use these recommendation values and sort the table in this order:
- Very specific: line 96
Use — when no NVBug is known.
- Distinctive: line 104
Do not label unmerged PRs, PRs not based on main, or candidates classified Unclear without explicit approval.
- Distinctive: line 108
Use the GitHub search API to list PRs that have the cherry-pick label but not cherry-pick-done, sorted by merge date ascending:
- Very specific: line 120
Check out release/<VERSION>, creating it from the remote if it doesn't exist locally:
- Distinctive: line 129
All PRs are squash-merged, so each has a single-parent commit.
- Very specific: line 173
Add the cherry-pick-done label to every PR that was successfully cherry-picked:
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Cherry-pick PRs to a Release Branch
Cherry-pick all merged main PRs labeled cherry-pick-<version> (but not cherry-pick-done) into the corresponding release/<version> branch, one by one in merge order.
Step 1 — Identify the target version
Ask the user for the release version (e.g. 0.44.0) and target release candidate (e.g. rc2) if not already provided. Confirm whether this is the final RC.
Set VERSION=<version> for use in subsequent steps.
Step 2 — Audit candidates for missing labels
Before fetching the labeled queue, audit for missing bug fixes:
- Non-final RC: audit release NVBugs and recent merged PRs.
- Final RC: audit only fixes tied to NVBugs with the exact keyword
Committed_ModelOpt_<VERSION>and QA-verification evidence for the fix. Verify that evidence in the NVBug or linked QA results before recommending Needs label; if missing or unclear, classify as Unknown and ask for confirmation. Skip the general recent-PR audit; do not introduce general bug-fix PRs or fixes tied only to NVBugs committed for another release.
Audit release NVBugs
Use the NVBugs MCP to search for every NVBug whose Keywords field contains the exact keyword Committed_ModelOpt_<VERSION>. Follow pagination until no next token is returned.
Fetch each matching NVBug with its comments. Extract every Model-Optimizer PR link or unambiguous PR #<NUM> reference from the comments, not only the latest comment. Keep only PRs that are merged into main, lack cherry-pick-<VERSION>, and are not already included in the release branch. Verify each remaining PR is a bug fix. The NVBug keyword is evidence for review, not by itself proof that every linked PR should be picked.
Audit recent merged PRs
For non-final RCs only: Audit PRs merged into main since the latest release candidate. Exclude PRs that already have cherry-pick-<VERSION> and changes already present on the release branch:
git fetch origin main "release/$VERSION" --tags
RC_TAG=$(git tag --merged "origin/release/$VERSION" \
--list "${VERSION}rc*" --sort=-version:refname | head -1)
test -n "$RC_TAG" || {
echo "No ${VERSION}rc* tag found on origin/release/$VERSION" >&2
exit 1
}
SINCE=$(git for-each-ref --format='%(creatordate:iso-strict)' \
"refs/tags/$RC_TAG")
PATCH_STATUS=$(git cherry "origin/release/$VERSION" origin/main)
SEARCH_RESULTS=$(
gh search prs \
--repo NVIDIA/Model-Optimizer \
--merged \
--base main \
--merged-at ">=$SINCE" \
--limit 1000 \
--json number,title,author,labels,url \
-- "-label:cherry-pick-$VERSION"
)
if test "$(jq 'length' <<<"$SEARCH_RESULTS")" -ge 1000; then
echo "Recent-PR audit reached the 1,000-result limit" >&2
exit 1
fi
jq -r '.[].number' <<<"$SEARCH_RESULTS" \
| while read -r pr; do
sha=$(gh pr view "$pr" --repo NVIDIA/Model-Optimizer \
--json mergeCommit --jq '.mergeCommit.oid')
if grep -q "^+ $sha$" <<<"$PATCH_STATUS"; then
gh pr view "$pr" --repo NVIDIA/Model-Optimizer \
--json number,title,author,labels,url
fi
doneA release is not normally expected to reach the 1,000-PR limit.
Review each PR's title, body, labels, changed files, and linked issue context. Classify it as:
- Yes — repairs incorrect behavior, a regression, crash, compatibility problem, or documentation defect relevant to the release.
- No — feature, refactor, cleanup, dependency refresh, or other change not needed to correct the release.
- Unclear — insufficient evidence or meaningful backport risk; ask the user.
Deduplicate PRs found through both audits.
Report and label
Present the complete audit before changing GitHub labels:
| PR | Title | Author | NVBug(s) | Bug fix? | cherry-pick-<VERSION> present? | Recommendation |
Use these recommendation values and sort the table in this order:
- Needs label — confirmed release-relevant bug fix.
- Unknown — requires user judgment.
- No action needed — not a release-relevant bug fix.
Use — when no NVBug is known. Also list NVBugs with no linked PR. Ask the user to confirm which recommended PRs should receive the missing label. After confirmation, apply it:
for pr in <APPROVED_NUMBERS>; do
gh pr edit "$pr" --repo NVIDIA/Model-Optimizer --add-label "cherry-pick-$VERSION"
doneDo not label unmerged PRs, PRs not based on main, or candidates classified Unclear without explicit approval. Re-run the audit table after edits so it reflects the final label state.
Step 3 — Fetch pending PRs
Use the GitHub search API to list PRs that have the cherry-pick label but not cherry-pick-done, sorted by merge date ascending:
gh api "search/issues?q=repo:NVIDIA/Model-Optimizer+is:pr+is:merged+base:main+label:cherry-pick-<VERSION>+-label:cherry-pick-done&sort=updated&order=asc&per_page=50" \
--jq '.items[] | [.number, .title, .pull_request.merged_at] | @tsv' \
| sort -t$'\t' -k3Present the list to the user before proceeding.
Step 4 — Set up the release branch
Check out release/<VERSION>, creating it from the remote if it doesn't exist locally:
git fetch origin release/<VERSION>
git checkout release/<VERSION>Step 5 — Get merge commit SHAs
All PRs are squash-merged, so each has a single-parent commit. Retrieve the SHA for each PR:
gh pr view <NUM> --repo NVIDIA/Model-Optimizer --json mergeCommit --jq '.mergeCommit.oid'Step 6 — Cherry-pick in merge order
Cherry-pick each commit with -s (DCO sign-off). GPG signing is handled automatically by the repo's git config.
git cherry-pick -s <SHA>On conflict: Tell the user which PR caused the conflict and ask them to fix it, then continue:
git cherry-pick --continueStep 7 — Create a PR to the release branch
Push the cherry-picks to a new branch and open a PR targeting release/<VERSION>. The PR title lists every cherry-picked PR number. The body uses ## Cherry-picked PRs as the only heading with one - #<NUM> bullet per PR — no titles, no links, no extra text.
git checkout -B cherry-picks/release-<VERSION>
git push -u origin cherry-picks/release-<VERSION>
gh pr create \
--title "[Cherry-pick] PRs #<NUM1> #<NUM2> ..." \
--base release/<VERSION> \
--head cherry-picks/release-<VERSION> \
--body "$(cat <<'EOF'
## Cherry-picked PRs
- #<NUM1>
- #<NUM2>
...
EOF
)"Step 8 — Apply cherry-pick-done label
Add the cherry-pick-done label to every PR that was successfully cherry-picked:
for pr in <NUM1> <NUM2> ...; do
gh pr edit $pr --repo NVIDIA/Model-Optimizer --add-label "cherry-pick-done"
doneDoes it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

