Set NVFP4 MSE preset effective bits to 4.5

[realAsma]

What does this PR do?

Type of change: Bug fix

Set effective_bits: 4.5 on the NVFP4 W4A4 weight-MSE FP8-scale-sweep preset. This records the NVFP4 storage cost—including its FP8 scale overhead—at the preset level for AutoQuant consumers.

Add a focused regression test to ensure the discovered preset continues to expose the correct effective-bit cost.

Usage

No new API or flag is introduced. Recipes importing nvfp4_w4a4_weight_mse_fp8_sweep automatically receive the corrected metadata.

Testing

• pre-commit run --files modelopt_recipes/configs/ptq/presets/model/nvfp4_w4a4_weight_mse_fp8_sweep.yaml tests/unit/recipe/test_presets.py
• pytest_pwd -q tests/unit/recipe/test_presets.py — 7 passed
• git diff --check 913f5e224cb2a9a7754bd2f17edd50395aad7bf7..5be3e216e57822394f2f6db842f759489c72e2fb

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?: ✅
• 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?: ✅
• Did you update Changelog?: N/A — this focused preset metadata correction is not changelog-worthy
• Did you get Claude approval on this PR?: N/A — this is a small metadata correction with a focused regression test

Additional Information

None.

Summary by CodeRabbit

• New Features

  • Updated the NVFP4 W4A4 MSE FP8-scale preset to report an effective precision of 4.5 bits.

• Bug Fixes

  • Corrected the preset’s storage-cost metadata for more accurate configuration reporting.

• Tests

  • Added regression coverage to verify the updated effective-bits value.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 98b83b5c-d922-4bd4-aadd-ae080c1c5bdb

📥 Commits

Reviewing files that changed from the base of the PR and between 913f5e2 and 5be3e21.

📒 Files selected for processing (2)

• modelopt_recipes/configs/ptq/presets/model/nvfp4_w4a4_weight_mse_fp8_sweep.yaml
• tests/unit/recipe/test_presets.py

Included review availability: Your plan provides up to 12 included reviews per hour; 11 remain after this review.

---

📝 Walkthrough

Walkthrough

The NVFP4 W4A4 MSE FP8-scale preset now declares effective_bits: 4.5. A unit test verifies the configured storage cost.

Changes

NVFP4 storage cost

Layer / File(s)	Summary
Preset storage cost and regression coverage modelopt_recipes/configs/ptq/presets/model/nvfp4_w4a4_weight_mse_fp8_sweep.yaml, tests/unit/recipe/test_presets.py	The preset adds effective_bits: 4.5. The unit test asserts this value.

Estimated code review effort: 1 (Trivial) | ~5 minutes

Merge Risk: ⚪ Minimal · up to 5be3e

This change corrects the NVFP4 preset’s effective-bit metadata and adds regression coverage; no actionable merge-blocking risk remains beyond normal checks and review.

Suggested reviewers: shengliangxu

🚥 Pre-merge checks | ✅ 5 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 3 functions across 1 files. (1 skipped: 1 unsupported.)	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly and concisely describes the main change: setting the NVFP4 MSE preset effective bits to 4.5.
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.
Security Anti-Patterns	✅ Passed	The PR changes only one YAML metadata field and one regression assertion; it adds no modelopt/examples Python security patterns, dependencies, or # nosec bypass.

✨ Finishing Touches 💡 1

📝 Generate docstrings 💡

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch asma/qwen3.8_fp4

---

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

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 78.37%. Comparing base (913f5e2) to head (5be3e21).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2223      +/-   ##
==========================================
- Coverage   78.98%   78.37%   -0.61%     
==========================================
  Files         522      522              
  Lines       60599    60599              
==========================================
- Hits        47862    47497     -365     
- Misses      12737    13102     +365     

Flag	Coverage Δ
examples-diffusers	20.70% <ø> (+<0.01%)	⬆️
examples-gpt-oss	13.23% <ø> (+<0.01%)	⬆️
examples-hf_ptq	21.47% <ø> (-0.04%)	⬇️
examples-llm_distill	13.30% <ø> (+<0.01%)	⬆️
examples-llm_eval	17.06% <ø> (+<0.01%)	⬆️
examples-llm_qat	17.54% <ø> (+<0.01%)	⬆️
examples-llm_sparsity	15.88% <ø> (+<0.01%)	⬆️
examples-megatron_bridge	25.67% <ø> (-0.16%)	⬇️
examples-specdec_bench	12.97% <ø> (+<0.01%)	⬆️
examples-speculative_decoding	17.48% <ø> (-0.07%)	⬇️
examples-torch_onnx	21.79% <ø> (+<0.01%)	⬆️
examples-torch_trt	15.03% <ø> (-0.01%)	⬇️
gpu	58.50% <ø> (-0.72%)	⬇️
unit	55.60% <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.