[None][chore] Remove unnecessary buffer to save memory during refit

[shuyixiong]

Summary by CodeRabbit

• Refactor
  • Optimized weight tensor handling in FP8 quantization module to enhance loading performance and reliability of quantized models.

Description

Test Coverage

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)

• If PR introduces API changes, an appropriate PR label is added - either api-compatible or api-breaking. For api-breaking, include BREAKING in the PR title.

• 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.

[coderabbitai]

📝 Walkthrough

Walkthrough

The change modifies weight parameter application in DeepSeek FP8 quantization's resmoothing path. Resmoothed weight tensors are now updated in-place via .data.copy_() rather than full parameter replacement, while scaling factors remain on the full replacement path.

Changes

FP8 Weight Resmoothing Strategy

Layer / File(s)	Summary
Weight resmoothing parameter update strategy tensorrt_llm/_torch/modules/fused_moe/quantization.py	Resmoothed w3_w1_weight and w2_weight tensors are now updated in-place via .data.copy_() into existing parameter buffers, while corresponding scaling factors (w3_w1_weight_scaling_factor, w2_weight_scaling_factor) continue to use replace_parameter_and_save_metadata(...).

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 minutes

Suggested reviewers

• xinhe-nv

🚥 Pre-merge checks | ✅ 2 | ❌ 3

❌ Failed checks (3 warnings)

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.
Description check	⚠️ Warning	The PR description is incomplete. All content sections (Description, Test Coverage) are empty, containing only template comments without actual details about the changes, rationale, or test coverage.	Fill in the Description section explaining why the buffer is unnecessary and how removing it saves memory. Add Test Coverage section listing relevant tests that validate the changes.
Title check	⚠️ Warning	The title mentions 'Remove unnecessary buffer to save memory' but the actual change modifies weight tensor parameter replacement methodology in the resmoothing path, not buffer removal.	Update the title to accurately reflect the main change, such as 'Use in-place copy for weight updates in DeepSeekFP8 resmoothing path' or similar.

✅ Passed checks (2 passed)

Check name	Status	Explanation
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

---

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

[shuyixiong]

/bot run --stage-list "DGX_B200-4_GPUs-PyTorch-Ray-1"

[tensorrt-cicd]

PR_Github #49336 [ run ] triggered by Bot. Commit: 8dddab6 Link to invocation

[tensorrt-cicd]

PR_Github #49336 [ run ] completed with state SUCCESS. Commit: 8dddab6
/LLM/main/L0_MergeRequest_PR pipeline #38993 (Partly Tested) completed with status: 'SUCCESS'

CI Report

Link to invocation

[shuyixiong]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #49387 [ run ] triggered by Bot. Commit: 8dddab6 Link to invocation

[tensorrt-cicd]

PR_Github #49387 [ run ] completed with state SUCCESS. Commit: 8dddab6
/LLM/main/L0_MergeRequest_PR pipeline #39037 completed with status: 'FAILURE'

CI Report

⚠️ Action Required:

• Please check the failed tests and fix your PR
• If you cannot view the failures, ask the CI triggerer to share details
• Once fixed, request an NVIDIA team member to trigger CI again

CI Agent Failure Analysis

Link to invocation

[shuyixiong]

/bot run --disable-fail-fast

[shuyixiong]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #49570 [ run ] triggered by Bot. Commit: 8dddab6 Link to invocation

[tensorrt-cicd]

PR_Github #49570 [ run ] completed with state SUCCESS. Commit: 8dddab6
/LLM/main/L0_MergeRequest_PR pipeline #39197 completed with status: 'SUCCESS'

CI Report

Link to invocation