[TRTLLM-10076][feat] Serve CLI improvements: renames, new flags, and mm_embedding_serve enhancements

[JunyiXu-nv]

• TRTLLM-10076: Update --tokenizer description for PyTorch backend, add --hf_revision alias for --revision with deprecation warning, support hf_revision key in YAML config, add --enable_attention_dp flag
• TRTLLM-10079: mm_embedding_serve: add --config alias for --extra_encoder_options, expose --hf_revision, --free_gpu_memory_fraction, --tensor_parallel_size
• TRTLLM-10229: Add --config alias for --config_file in disaggregated and disaggregated_mpi_worker commands
• TRTLLM-10078: Improve --server_role help message with role descriptions

Made-with: Cursor

Summary by CodeRabbit

Release Notes

• New Features

  • Added --enable_attention_dp flag to the serve command for distributed attention processing.

• Deprecations

  • Use --hf_revision instead of --revision.
  • Use --config instead of --extra_encoder_options and --config_file.
  • Deprecation warnings will guide migration to new parameter names.

• Improvements

  • Enhanced help documentation for tokenizer and server role configuration.
  • Extended serve_encoder with additional parameters: revision, free_gpu_memory_fraction, and tensor_parallel_size.

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)

• 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 changes add support for a new enable_attention_dp feature flag through CLI argument plumbing in the serve command and related functions. Additionally, deprecation warnings are introduced for legacy options (--revision, --extra_encoder_options, --config_file), and new parameters are added to serve_encoder. A preprocessing step maps hf_revision to revision in llm_args handling.

Changes

Cohort / File(s)	Summary
CLI and Argument Plumbing tensorrt_llm/commands/serve.py	Added --enable_attention_dp CLI option to serve command; extended serve_encoder signature with revision, free_gpu_memory_fraction, and tensor_parallel_size parameters; introduced deprecation warnings for legacy flags (--revision → --hf_revision, --extra_encoder_options → --config, --config_file → --config); enhanced help texts for tokenizer and server role clarification; propagated enable_attention_dp through _serve_llm and launch_server paths (+84/-24).
Argument Preprocessing tensorrt_llm/llmapi/llm_args.py	Added preprocessing step to map hf_revision to revision in llm_args_dict before downstream processing using pop/setdefault (+3/-0).

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

🚥 Pre-merge checks | ✅ 1 | ❌ 2

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 50.00% 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 lists specific JIRA tickets and changes but lacks proper PR title format, missing sections for Description, Test Coverage details, and incomplete PR Checklist verification.	Add a properly formatted PR title following [JIRA][type] format, provide detailed Description and Test Coverage sections, and verify all PR Checklist items are addressed.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately describes the main changes: adding new CLI flags (enable_attention_dp, hf_revision), improving rename/alias support, and enhancing mm_embedding_serve.

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

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment

---

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

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In `@tensorrt_llm/commands/serve.py`:
- Around line 1122-1127: The disaggregated_mpi_worker entrypoint (function
disaggregated_mpi_worker) should mirror the deprecation behavior of
disaggregated by detecting if "--config_file" was passed and emitting a
DeprecationWarning; add a check using sys.argv to see if "--config_file" is
present and call warnings.warn(..., DeprecationWarning, stacklevel=2)
immediately after the disaggregated_mpi_worker docstring/entry log so users see
the same deprecation message as the disaggregated command.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: c1f1f5fa-f1ab-4275-9a8a-9c393cfda74f

📥 Commits

Reviewing files that changed from the base of the PR and between f7255e0 and b398e02.

📒 Files selected for processing (2)

• tensorrt_llm/commands/serve.py
• tensorrt_llm/llmapi/llm_args.py

[JunyiXu-nv]

/bot run

[tensorrt-cicd]

PR_Github #38560 [ run ] triggered by Bot. Commit: 1386e53 Link to invocation

[tensorrt-cicd]

PR_Github #38560 [ run ] completed with state SUCCESS. Commit: 1386e53
/LLM/main/L0_MergeRequest_PR pipeline #29902 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

Link to invocation

[JunyiXu-nv]

/bot run

[tensorrt-cicd]

PR_Github #38585 [ run ] triggered by Bot. Commit: 1386e53 Link to invocation