[6309094] updated readme to clarify simple_qat_train.py usecase

[kinjalpatel27]

What does this PR do?

Type of change: documentation

Added clarification about simjple_qat_train.py which is a demonstration of QAT flow and not meant for multi-GPU training

Before your PR is "Ready for review"

Make sure you read and follow Contributor guidelines and your commits are signed (git commit -s -S).

Make sure you read and follow the Security Best Practices (e.g. avoiding hardcoded trust_remote_code=True, torch.load(..., weights_only=False), pickle, etc.).

• Is this change backward compatible?: N/A
• If you copied code from any other sources or added a new PIP dependency, did you follow guidance in CONTRIBUTING.md: N/A
• Did you write any new necessary tests? N/A
• Did you update Changelog?: N/A
• Did you get Claude approval on this PR?: ✅ / ❌ / N/A

Additional Information

Summary by CodeRabbit

• Documentation
  • Expanded the QAT/QAD README note section with clearer, end-to-end minimal demo instructions, highlighting a single-GPU quantize+train+save workflow via a dedicated script.
  • Added guidance for multi-GPU training using accelerate launch, with a pointer to the appropriate training entry point.

[coderabbitai]

📝 Walkthrough

Walkthrough

The NOTE section in examples/llm_qat/README.md (lines 91–97) is expanded to introduce simple_qat_train.py as a minimal single-GPU quantize-train-save script, provide an example command, and add a follow-up pointer to multi-GPU training using train.py with accelerate launch.

Changes

QAT README NOTE section update

Layer / File(s)	Summary
Expanded NOTE block with single-GPU and multi-GPU instructions examples/llm_qat/README.md	NOTE block now describes simple_qat_train.py as a minimal single-GPU end-to-end QAT/QAD demo with an example command, and adds a sentence pointing to multi-GPU training via train.py + accelerate launch.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~2 minutes

---

Caution

Pre-merge checks failed

Please resolve all errors before merging. Addressing warnings is optional.

• Ignore

❌ Failed checks (1 error)

Check name	Status	Explanation	Resolution
Security Anti-Patterns	❌ Error	New non-permissive PIP dependency py7zr (LGPL v2.1) added in examples/llm_qat/requirements.txt without required approval or justification from @NVIDIA/modelopt-setup-codeowners.	Remove py7zr or provide explicit approval and justification from @NVIDIA/modelopt-setup-codeowners in PR description citing valid licensing concerns and project necessity.

✅ Passed checks (5 passed)

Check name	Status	Explanation
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title accurately describes the main change: updating the README to clarify the usecase of simple_qat_train.py, which aligns with the file modifications and PR objectives.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch kinjal/qat_multigpu

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Warning

CodeRabbit couldn't request changes on this pull request because it doesn't have sufficient GitHub permissions.

Please grant CodeRabbit Pull requests: Read and write permission and re-run the review.

👉 Steps to fix this

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@examples/llm_qat/README.md`:
- Line 97: The link fragment in the line containing the text about multi-GPU
training with accelerate launch references an incorrect anchor. Update the
broken link fragment from `#Run-End-to-End-QAT/QAD-Example` to
`#run-end-to-end-qatqad-example` to match the normalized Markdown anchor generated
from the actual heading. Markdown automatically converts heading text to
lowercase anchors and replaces spaces with hyphens while removing special
characters like forward slashes.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 08e479e9-5793-4c24-ac13-ee0c37ea5c1b

📥 Commits

Reviewing files that changed from the base of the PR and between 0f61c98 and c6b00d3.

📒 Files selected for processing (1)

• examples/llm_qat/README.md

[kinjalpatel27]

/claude review

[claude]

Claude review passed — no blocking issues found. LGTM

Documentation-only change (3 additions / 1 deletion in examples/llm_qat/README.md). Verified that the referenced files (simple_qat_train.py, train.py) exist and the #qat anchor resolves to the ### QAT heading. No algorithm, mode/state, export, or backward-compatibility surface area is touched.

[cjluo-nv]

Bot review — DM the bot to share feedback.

Documentation-only change (+3/-1) clarifying that simple_qat_train.py is a single-GPU, end-to-end QAT demo (no transformer trainer) and pointing to train.py + accelerate launch for multi-GPU. Verified against the source: simple_qat_train.py exists, uses a plain training loop with a single .cuda() call (single-GPU, no Trainer), and train.py exists with accelerate launch usage. The #qat anchor maps correctly to the ### QAT heading. Links are valid, no logic/licensing changes, and no injection attempts in the untrusted content.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77.07%. Comparing base (6968fe7) to head (0b7e7a1).
⚠️ Report is 5 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1742      +/-   ##
==========================================
- Coverage   77.11%   77.07%   -0.04%     
==========================================
  Files         511      511              
  Lines       56259    56289      +30     
==========================================
+ Hits        43383    43386       +3     
- Misses      12876    12903      +27     

Flag	Coverage Δ
examples	41.84% <ø> (-0.11%)	⬇️
unit	54.39% <ø> (+<0.01%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[cjluo-nv]

Bot review — DM the bot to share feedback.

Documentation-only change (+3/-1) to examples/llm_qat/README.md clarifying that simple_qat_train.py is a single-GPU, end-to-end QAT demo (no transformer trainer) and pointing to train.py + accelerate launch for multi-GPU. Verified: both simple_qat_train.py and train.py exist, and the #qat anchor resolves to the ### QAT heading. The previously-flagged broken anchor (#Run-End-to-End-QAT/QAD-Example, CodeRabbit MD051, minor) has been addressed — the author switched the link to the valid #qat fragment. No logic, no licensing changes, and no prompt-injection in the untrusted content.

[github-actions]

PR Preview Action v1.8.1
Preview removed because the pull request was closed.
2026-06-15 22:15 UTC