[https://nvbugs/5808603][fix] Add bias support to WeightOnlyQuantLinearMethod

[stnie]

Summary by CodeRabbit

• Bug Fixes

  • Fixed bias handling in linear layer operations to ensure bias is correctly applied to outputs across all execution paths.

• Tests

  • Enhanced test coverage for quantized linear operations with parameterized bias configuration validation.

Description

WeightOnlyQuantLinearMethods did not apply bias, which is for example used in Qwen2.5 models

This PR:

• Modified WeightOnlyQuantLinearMethod to include bias in the output calculation.
• Updated test_weight_only_quant_linear to parameterize bias and validate its effect on output.

Test Coverage

• Updated test_weight_only_quant_linear to parameterize bias and validate its effect on output.

PR Checklist

Please review the following before submitting your PR:

• PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.

• PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.

• Test cases are provided for new code paths (see test instructions)

• Any new dependencies have been scanned for license and vulnerabilities

• CODEOWNERS updated if ownership changes

• Documentation updated as needed

• Update tava architecture diagram if there is a significant design change in PR.

• The reviewers assigned automatically/manually are appropriate for the PR.

• Please check this after reviewing the above items as appropriate for this PR.

GitHub Bot Help

To see a list of available CI bot commands, please comment /bot help.

[stnie]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #39446 [ run ] triggered by Bot. Commit: 5763719 Link to invocation

[tensorrt-cicd]

PR_Github #39446 [ run ] completed with state SUCCESS. Commit: 5763719
/LLM/main/L0_MergeRequest_PR pipeline #30673 completed with status: 'SUCCESS'
Pipeline passed with automatic retried tests. Check the rerun report for details.

CI Report

Link to invocation

[coderabbitai]

📝 Walkthrough

Walkthrough

Adds bias support to the Linear module's apply method by conditionally adding a bias tensor to the output. Extends the weight-only quantized linear test with parameterized bias cases, including bias tensor creation, loading, and reference output adjustment.

Changes

Cohort / File(s)	Summary
Linear Module Bias Support tensorrt_llm/_torch/modules/linear.py	Adds three lines to conditionally apply bias in the Linear.apply path: checks if bias tensor is provided and adds it to the computed output.
Quantized Linear Test Extension tests/unittest/_torch/thop/parallel/test_weight_only_quant_linear.py	Introduces bias as a pytest parameterized test parameter (True/False). Creates bias tensor when enabled, passes it to Linear constructor and load_weights payload, and adjusts reference output computation to include bias term conditionally.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

🚥 Pre-merge checks | ✅ 2 | ❌ 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%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically describes the main change: adding bias support to WeightOnlyQuantLinearMethod, which aligns perfectly with the code changes in the PR.
Description check	✅ Passed	The PR description includes a clear problem statement, explains the changes made, and lists test coverage, following the required template structure.

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

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

📝 Coding Plan

• Generate coding plan for human review comments

---

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

Tip

You can disable the changed files summary in the walkthrough.

Disable the reviews.changed_files_summary setting to disable the changed files summary in the walkthrough.