[https://nvbugs/6293015][fix] Add a delegating `@property def vocab_size_padded(self) -> int: return… by tensorrt-cicd · Pull Request #15219 · NVIDIA/TensorRT-LLM

tensorrt-cicd · 2026-06-10T11:14:50Z

Summary

Root cause: KimiK25ForConditionalGeneration (multimodal wrapper around self.llm = DeepseekV3ForCausalLM) does not expose a vocab_size_padded property, unlike every other VLM wrapper.
Fix: Add a delegating @property def vocab_size_padded(self) -> int: return self.llm.vocab_size_padded to KimiK25ForConditionalGeneration, matching the pattern in modeling_qwen2vl.py, modeling_qwen3vl.py, etc.
Automated fix generated by repair-bot

Test plan

Verify fix on the same GPU type as the original failure
Check for regressions in related tests

Links

Bug: https://nvbugs/6293015

Summary by CodeRabbit

New Features
- Exposed padded vocabulary size property, enabling external systems to query vocabulary padding information.

coderabbitai · 2026-06-10T11:18:51Z

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: 753162c0-a0fb-4d79-bcc6-b5006c5ca4ff

📥 Commits

Reviewing files that changed from the base of the PR and between 9635f7d and 51a7963.

📒 Files selected for processing (1)

tensorrt_llm/_torch/models/modeling_kimi_k25.py

📝 Walkthrough

Walkthrough

KimiK25ForConditionalGeneration now exposes a vocab_size_padded property that forwards to the underlying llm.vocab_size_padded, allowing external code to query the padded vocabulary size through the VLM wrapper.

Changes

VLM wrapper vocabulary size property

Layer / File(s)	Summary
VLM wrapper vocab size property `tensorrt_llm/_torch/models/modeling_kimi_k25.py`	A `vocab_size_padded` property is added to KimiK25ForConditionalGeneration to expose the padded vocabulary size from the underlying llm model.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

🚥 Pre-merge checks | ✅ 3 | ❌ 2

❌ Failed checks (1 warning, 1 inconclusive)

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.
Title check	❓ Inconclusive	Title is partially related to the changeset; it refers to adding vocab_size_padded property but is incomplete/truncated mid-word.	Complete the title by finishing the truncated property return statement for clarity.

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description check	✅ Passed	PR description covers root cause, fix, and test plan but lacks the PR Checklist completion marker required by the template.
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.}

zhaoyangwang-nvidia · 2026-06-11T01:48:25Z

Hi @NVIDIA/trt-llm-torch-models-devs please help to review this PR, only a small fix for a NV bug.

zhaoyangwang-nvidia · 2026-06-11T01:49:15Z

/bot run

tensorrt-cicd · 2026-06-11T01:56:26Z

PR_Github #53426 [ run ] triggered by Bot. Commit: 51a7963 Link to invocation

tensorrt-cicd · 2026-06-11T03:37:55Z

PR_Github #53426 [ run ] completed with state FAILURE. Commit: 51a7963
/LLM/main/L0_MergeRequest_PR pipeline #42596 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

zhaoyangwang-nvidia · 2026-06-11T05:40:15Z

/bot run

zhaoyangwang-nvidia · 2026-06-11T07:09:33Z

/bot run

tensorrt-cicd · 2026-06-11T07:15:42Z

PR_Github #53511 [ run ] triggered by Bot. Commit: 51a7963 Link to invocation

tensorrt-cicd · 2026-06-11T08:34:23Z

PR_Github #53511 [ run ] completed with state SUCCESS. Commit: 51a7963
/LLM/main/L0_MergeRequest_PR pipeline #42669 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

xxi-nv · 2026-06-11T08:51:23Z

/bot run

tensorrt-cicd · 2026-06-11T08:57:02Z

PR_Github #53535 [ run ] triggered by Bot. Commit: 51a7963 Link to invocation

tensorrt-cicd · 2026-06-11T10:15:20Z

PR_Github #53535 [ run ] completed with state SUCCESS. Commit: 51a7963
/LLM/main/L0_MergeRequest_PR pipeline #42687 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

xxi-nv · 2026-06-11T10:26:43Z

/bot run

tensorrt-cicd · 2026-06-11T10:33:17Z

PR_Github #53554 [ run ] triggered by Bot. Commit: f0b921c Link to invocation

tensorrt-cicd · 2026-06-11T13:08:41Z

PR_Github #53554 [ run ] completed with state SUCCESS. Commit: f0b921c
/LLM/main/L0_MergeRequest_PR pipeline #42704 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

xxi-nv · 2026-06-11T13:16:53Z

/bot run

xxi-nv · 2026-06-11T13:16:54Z

/bot run

tensorrt-cicd · 2026-06-11T13:22:43Z

PR_Github #53579 [ run ] triggered by Bot. Commit: f0b921c Link to invocation

tensorrt-cicd · 2026-06-11T13:23:13Z

PR_Github #53580 [ run ] triggered by Bot. Commit: f0b921c Link to invocation

tensorrt-cicd · 2026-06-11T13:23:19Z

PR_Github #53579 [ run ] completed with state ABORTED. Commit: f0b921c

Link to invocation

tensorrt-cicd · 2026-06-11T16:18:41Z

PR_Github #53580 [ run ] completed with state SUCCESS. Commit: f0b921c
/LLM/main/L0_MergeRequest_PR pipeline #42727 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

…ditionalGeneration The Kimi-K2.5 multimodal wrapper KimiK25ForConditionalGeneration holds the text backbone as self.llm but does not delegate vocab_size_padded to it. When guided_decoding_backend is set (e.g. xgrammar), py_executor_creator.py reads model_engine.model.vocab_size_padded and raises AttributeError. Add a delegating property matching the pattern used by every other VLM wrapper (qwen2vl, qwen3vl, gemma3vl, llava_next, mistral, phi4mm, etc.). Signed-off-by: tensorrt-cicd <90828364+tensorrt-cicd@users.noreply.github.com>

xxi-nv · 2026-06-12T01:45:46Z

/bot run

tensorrt-cicd · 2026-06-12T01:52:38Z

PR_Github #53748 [ run ] triggered by Bot. Commit: 8129b65 Link to invocation

tensorrt-cicd · 2026-06-12T04:53:39Z

PR_Github #53748 [ run ] completed with state FAILURE. Commit: 8129b65
/LLM/main/L0_MergeRequest_PR pipeline #42872 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

zhaoyangwang-nvidia · 2026-06-12T05:39:42Z

/bot run

tensorrt-cicd · 2026-06-12T05:45:13Z

PR_Github #53801 [ run ] triggered by Bot. Commit: 8129b65 Link to invocation

tensorrt-cicd · 2026-06-12T08:50:45Z

PR_Github #53801 [ run ] completed with state SUCCESS. Commit: 8129b65
/LLM/main/L0_MergeRequest_PR pipeline #42919 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

xxi-nv · 2026-06-12T09:10:05Z

/bot run

tensorrt-cicd · 2026-06-12T09:16:00Z

PR_Github #53862 [ run ] triggered by Bot. Commit: 8129b65 Link to invocation

tensorrt-cicd · 2026-06-12T10:28:16Z

PR_Github #53862 [ run ] completed with state SUCCESS. Commit: 8129b65
/LLM/main/L0_MergeRequest_PR pipeline #42969 completed with status: 'SUCCESS'

CI Report

Link to invocation

The rc18->rc19 rebase kept our rc18-based modeling_kimi_k25.py wholesale, which missed rc19's own evolution of the (shared-lineage) file: - NVIDIA#13864 (EPD): get_prompt_token_ids() -> build_disagg_prefill_multimodal_inputs() returning DisaggPrefillMultimodalInputs. rc19's inputs/registry.py now REQUIRES this method for multimodal disagg; our old file only had the removed get_prompt_token_ids -> disaggregated multimodal Kimi would fail registry validation. - NVIDIA#14741: removed DI's lazy slow-tokenizer shim as redundant. - NVIDIA#15219: native vocab_size_padded delegating property. Re-base our file on rc19's modeling_kimi_k25.py and re-apply only the DI deltas that rc19 lacks: - tokenizer denylist discard ("kimi_k25") for correct gapped special-token ids (kept even though the slow-tokenizer shim was dropped; harmless no-op if upstream resolves the denylist entry); - root-logger filter silencing the per-encode "Calling super().encode" spam; - vision-tower MLP row chunking (TLLM_KIMI_VISION_MLP_CHUNK) to bound peak activation on large images; - keep MoonViT3d LayerNorm params in float32 (TRT-LLM LayerNorm.forward upcasts activations -> bf16 weights crash F.layer_norm); - data.image placeholder deficit top-up (drop the now-removed _ensure_k25_slow_tokenizer call); - spec-decode delegations rc19 lacks (spec_worker, spec_config, set_guided_decoder; rc19 already forwards the rest); - preserve original_input_ids across fuse_input_embeds for the Eagle3 drafter. Net +110 lines over rc19 base. Aggregated and disaggregated multimodal paths both supported now. Needs a vision smoke before prod. Signed-off-by: Bo <bo@deepinfra.com>

tensorrt-cicd requested a review from a team as a code owner June 10, 2026 11:14

tensorrt-cicd requested a review from syuoni June 10, 2026 11:14

tensorrt-cicd assigned zhaoyangwang-nvidia Jun 10, 2026

github-actions Bot assigned tensorrt-cicd Jun 10, 2026

xxi-nv force-pushed the repair-bot-bug6293015 branch from 51a7963 to f0b921c Compare June 11, 2026 10:26

2ez4bz approved these changes Jun 11, 2026

View reviewed changes

xxi-nv force-pushed the repair-bot-bug6293015 branch from f0b921c to 8129b65 Compare June 12, 2026 01:45

sunnyqgg merged commit ec47baa into NVIDIA:main Jun 13, 2026
7 checks passed

Uh oh!

Conversation

tensorrt-cicd commented Jun 10, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test plan

Links

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented Jun 10, 2026

Walkthrough

Changes

Estimated code review effort

❌ Failed checks (1 warning, 1 inconclusive)

Uh oh!

zhaoyangwang-nvidia commented Jun 11, 2026

Uh oh!

zhaoyangwang-nvidia commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

zhaoyangwang-nvidia commented Jun 11, 2026

Uh oh!

zhaoyangwang-nvidia commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

xxi-nv commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

xxi-nv commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

xxi-nv commented Jun 11, 2026

Uh oh!

xxi-nv commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

tensorrt-cicd commented Jun 11, 2026

Uh oh!

xxi-nv commented Jun 12, 2026

Uh oh!

tensorrt-cicd commented Jun 12, 2026

Uh oh!

tensorrt-cicd commented Jun 12, 2026

Uh oh!

zhaoyangwang-nvidia commented Jun 12, 2026

Uh oh!

tensorrt-cicd commented Jun 12, 2026

Uh oh!

tensorrt-cicd commented Jun 12, 2026

Uh oh!

xxi-nv commented Jun 12, 2026

Uh oh!

tensorrt-cicd commented Jun 12, 2026

Uh oh!

tensorrt-cicd commented Jun 12, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

tensorrt-cicd commented Jun 10, 2026 •

edited by coderabbitai Bot

Loading