Revise supported formats to reflect new quantization types

[dsikka]

Updated the supported formats section to include new precisions and types for quantization.

[github-actions]

👋 Hi! Thank you for contributing to llm-compressor. Please add the ready label when the PR is ready for review.

Note: This is required to complete the testing suite, please only add the label once the PR is code complete and local testing has been performed.

[coderabbitai]

Important

Review skipped

Auto incremental reviews are disabled on this repository.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 7a13ba46-0e90-42a2-82bd-1357ebdba5ae

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

Walkthrough

Updates the README documentation to reflect expanded quantization capabilities by renaming the "Supported Formats" section to "Supported Precisions and Types," introducing new precision variants (W4AFP8, NVFP4, MXFP4, MXFP8), adding mixed-precision combinations, explicitly documenting attention and KV cache quantization, and extending the "Supported Algorithms" list with rotation-based methods.

Changes

Cohort / File(s)	Summary
README Documentation README.md	Updated quantization capability documentation with renamed section ("Supported Precisions and Types"), new precision names and microscale variants (W4AFP8, NVFP4, MXFP4, MXFP8), mixed-precision entries (MXFP4A16, NVFP4A16), explicit attention and KV-cache quantization callouts, and rotation-based algorithms (SpinQuant, QuIP).

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Possibly related PRs

• Updates to docs; move mxfp8 examples #2673: Updates README quantization documentation with overlapping precision names and quantization types (MXFP4/MXFP8, NVFP4, attention/KV-cache quantization).

Suggested labels

enhancement, nvfp4, fp8, w4a16

Suggested reviewers

• brian-dellabetta
• kylesayrs

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately reflects the main change in the pull request, which updates the README documentation to reflect new quantization types and expanded precision names.
Description check	✅ Passed	The description relates to the changeset by mentioning the update to the supported formats section with new precisions and types, though it contains placeholder text indicating incomplete documentation.
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.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch dsikka-update-precision

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[gemini-code-assist]

Code Review

This pull request updates the README.md to reflect expanded support for various quantization precisions, types, and algorithms, including microscale formats and rotation-based methods. Review feedback suggests correcting the naming convention for 4-bit weight/8-bit activation quantization and specifying the supported formats for attention and KV cache quantization to improve clarity and consistency.

[brian-dellabetta]

nice!