[https://nvbugs/6442074][fix] Rename MTPEagleDynamicTreeWorker.forward to _forward_impl

[sunnyqgg]

Summary

• Fix an import-time TypeError on current main: MTPEagleDynamicTreeWorker (merged in [None][feat] Support Nemotron dynamic-tree MTP decoding #15582) overrides SpecWorkerBase.forward, which is rejected by the __init_subclass__ guard added in [https://nvbugs/6442074][fix] Make one-model spec-dec attn-metadata save/restore exception-safe #16382 (https://nvbugs/6442074). Since speculative/utils.py imports mtp_dynamic_tree at module level, any import of the PyTorch backend fails.
• [None][feat] Support Nemotron dynamic-tree MTP decoding #15582's CI ran before the guard landed, so this semantic conflict only surfaced after merge.

Changes

• tensorrt_llm/_torch/speculative/mtp_dynamic_tree.py: rename the forward override to _forward_impl per the guard's contract (1 line). SpecWorkerBase.forward now wraps it with the guaranteed spec-dec attn-metadata cleanup, same as Eagle3OneModelDynamicTreeWorker.

Test plan

• Reproduced the failure on main tip: python -c "import tensorrt_llm._torch.speculative.utils" raises the guard TypeError
• Verified the same import passes the class-definition guard with this fix (A/B on the same tree)

Dev Engineer Review

• Renamed MTPEagleDynamicTreeWorker.forward to _forward_impl, satisfying the SpecWorkerBase.__init_subclass__ guard.
• Preserved the method body, parameters, and NVTX label (mtp_dyn.forward).
• SpecWorkerBase.forward remains the public entry point, preserving speculative-decoding attention-metadata cleanup.
• No configuration or test-list changes.

QA Engineer Review

No test changes.

[sunnyqgg]

/bot run --disable-fail-fast

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 6799847d-da73-4c1f-a9b6-7df36d9ec35d

📥 Commits

Reviewing files that changed from the base of the PR and between 9cd6a49 and 132f121.

📒 Files selected for processing (1)

• tensorrt_llm/_torch/speculative/mtp_dynamic_tree.py

---

Walkthrough

MTPEagleDynamicTreeWorker.forward(...) is renamed to _forward_impl(...) in the MTP dynamic tree worker. The NVTX label and method body remain unchanged.

Changes

MTP dynamic tree worker

Layer / File(s)	Summary
Rename worker entry method tensorrt_llm/_torch/speculative/mtp_dynamic_tree.py	Renames the method declaration from forward to _forward_impl while preserving its decorator and implementation.

Estimated code review effort: 1 (Trivial) | ~2 minutes

Suggested reviewers: asfiyab-nvidia

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title follows the required ticket/type format and clearly states the rename.
Description check	✅ Passed	The description includes Summary, Changes, and Test plan sections and is mostly complete.
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.

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

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

[tensorrt-cicd]

PR_Github #60962 [ run ] triggered by Bot. Commit: 132f121 Link to invocation

[tensorrt-cicd]

PR_Github #60962 [ run ] completed with state FAILURE. Commit: 132f121
/LLM/main/L0_MergeRequest_PR pipeline #49224 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

[Hudayday]

/bot run --disable-fail-fast

[tensorrt-cicd]

PR_Github #60978 [ run ] triggered by Bot. Commit: 132f121 Link to invocation

[tensorrt-cicd]

PR_Github #60978 [ run ] completed with state SUCCESS. Commit: 132f121
/LLM/main/L0_MergeRequest_PR pipeline #49240 completed with status: 'SUCCESS'

CI Report

Link to invocation