[None][fix] piecewise cudagraph node name for gemma-3-1b-it

[amukkara]

@coderabbitai summary

Description

1. For gemma-3-1b-it, input_ids node name in fx graph is l_kwargs_input_ids_. For other models, the node name is l_input_ids_. This PR pattern matches against both names when estimating input_num_tokens.
2. Fixes initialization of input_num_tokens to avoid "Undefined variable" error when pattern match in 1 fails.
3. Common to all models: torch_inductor and piecewise_cuda_graph are incompatible according to our docs. This PR keeps the default value TorchCompileConfig.enable_inductor=False in example script to avoid conflict with piecewise_cuda_graph.

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

Changes remove a configuration parameter from an example file and refine backend compilation logic to initialize a variable, expand placeholder name matching to include an additional pattern, and add early termination after capturing input token counts.

Changes

Cohort / File(s)	Summary
Example Configuration examples/llm-api/quickstart_advanced.py	Removed enable_inductor parameter from TorchCompileConfig initialization call within setup_llm function.
Backend Compilation Logic tensorrt_llm/_torch/compilation/backend.py	Initialize self.input_num_tokens to None at method start. Expanded placeholder name check to match both "l_input_ids_" and "l_kwargs_input_ids_" patterns. Added early break after capturing the first matching placeholder to prevent redundant iterations.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~8 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
Description check	✅ Passed	The PR description is mostly complete with three clear points explaining what was changed and why, along with a filled PR checklist.
Title check	✅ Passed	The title is partially related to the changeset, focusing on a specific fix for piecewise cudagraph node names (related to the backend.py changes), but doesn't address the removal of enable_inductor from quickstart_advanced.py, which is also a significant change.

_{✏️ 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

📝 Coding Plan

• Generate coding plan for human review comments

---

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

Tip

CodeRabbit can enforce grammar and style rules using `languagetool`.

Configure the reviews.tools.languagetool setting to enable/disable rules and categories. Refer to the LanguageTool Community to learn more.