[TRTLLM-15076][test] Remove Bielik-11B-v2.2-Instruct an… by xinhe-nv · Pull Request #17345 · NVIDIA/TensorRT-LLM

xinhe-nv · 2026-08-06T06:51:41Z

…d kanana-1.5-2.1b-instruct-2505 tests

Remove all test cases, test list entries, model path registrations, accuracy references, and waives for these two models.

Dev Engineer Review

Removed Bielik-11B-v2.2-Instruct test, performance, model path, and multi-LoRA references.
Removed Kanana accuracy references, test entries, and waiver entry.
Configuration and test-list changes match the removal scope.
No production-code or API changes are included.

QA Engineer Review

Removed TestKanana_Instruct::test_auto_dtype and its MMLU and GSM8K accuracy coverage.
Removed the Kanana test from qa/llm_function_core.txt and waives.txt.
Removed test_bielik_11b_v2_2_instruct_multi_lora.
Removed the Bielik performance entry from test-db/l0_perf.yml.
Removed the Bielik model path registration.
The Kanana test was covered in the QA test list.
The Bielik multi-LoRA test was not listed in the modified integration test lists.
Verdict: sufficient.

Description

The background is https://docs.google.com/spreadsheets/d/17qLRr5oQESqn1rT9LrIUnpjvhH8wcC59273YEPT9k44/edit?gid=177162358#gid=177162358

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.

…d kanana-1.5-2.1b-instruct-2505 tests Remove all test cases, test list entries, model path registrations, accuracy references, and waives for these two models. Signed-off-by: Xin He (SW-GPU) <200704525+xinhe-nv@users.noreply.github.com>

coderabbitai · 2026-08-06T06:52:48Z

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

@coderabbitai resume to resume automatic reviews.
@coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

▶️ Resume reviews
🔍 Trigger review

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: 22109856-a550-43fc-92ce-b8bb4aed42e5

📥 Commits

Reviewing files that changed from the base of the PR and between ea9ff4b and 90d030e.

📒 Files selected for processing (1)

tests/integration/test_lists/waives.txt

🚧 Files skipped from review as they are similar to previous changes (1)

tests/integration/test_lists/waives.txt

Walkthrough

The PR removes Kanana accuracy coverage and Bielik model-path, performance, and multi-LoRA test coverage from integration and unit test definitions.

Changes

Model test coverage removal

Layer / File(s)	Summary
Kanana accuracy coverage removal `tests/integration/defs/accuracy/references/*.yaml`, `tests/integration/defs/accuracy/test_llm_api_pytorch.py`, `tests/integration/test_lists/qa/llm_function_core.txt`	Removes Kanana GSM8K and MMLU references, the PyTorch accuracy test, and its test-list entry.
Bielik test coverage removal `tests/integration/defs/perf/_model_paths.py`, `tests/integration/test_lists/test-db/l0_perf.yml`, `tests/unittest/llmapi/test_llm_pytorch.py`	Removes the Bielik model path, L40S performance configuration, and multi-LoRA unit test.

Estimated code review effort: 2 (Simple) | ~10 minutes

Possibly related PRs

NVIDIA/TensorRT-LLM#17351: Both PRs remove obsolete model-specific accuracy, integration, performance, and test-list coverage.
NVIDIA/TensorRT-LLM#17442: Both PRs remove obsolete test waiver entries.
NVIDIA/TensorRT-LLM#17444: Both PRs remove obsolete test waiver entries.

Suggested reviewers: qijune, chenfeiz0326, ruodil

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title identifies the ticket, test scope, and removal of the Bielik model, which is a major change in the pull request.
Description check	✅ Passed	The description explains the removal scope and provides background, but the Test Coverage section is empty and most checklist items are unconfirmed.
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.}

xinhe-nv · 2026-08-06T06:53:54Z

/bot run

tensorrt-cicd · 2026-08-06T07:00:27Z

PR_Github #64254 [ run ] triggered by Bot. Commit: d9bc9b6 Link to invocation

BowenFu · 2026-08-06T07:37:21Z

            task.evaluate(llm)


-class TestKanana_Instruct(LlmapiAccuracyTestHarness):


TestKanana_Instruct is the only coverage for an in-flight product fix. #14707 (open) fixes the very bug this test is waived under — nvbugs/6209806 — by relaxing transformers 5.x LlamaConfig.validate_architecture in _torch/pyexecutor/config_utils.py and tokenizer/tokenizer.py so models with an explicit head_dim load. It removes this exact waives.txt line to re-enable this test, and ships no replacement test of its own.

So if this lands first, that fix merges with zero regression coverage for the load path it changes.

Fine either way once it's a decision rather than a side effect — if kanana is genuinely retired, say so on #14707 so it gets a focused config_utils unit test instead; if it isn't, keep this class and let #14707 do the unwaive. The waives.txt overlap itself is not a problem (three-way merge is clean).

Everything else here checks out: no dangling references left for either model, check_test_list.py --validate passes, the bielik_11b_v2.2_instruct alias is removed from _model_paths.py too (a real retirement, not just a coverage drop), and no import in test_llm_pytorch.py is orphaned.

@BowenFu should we keep this model or not?

tensorrt-cicd · 2026-08-06T09:50:07Z

PR_Github #64254 [ run ] completed with state FAILURE. Commit: d9bc9b6
/LLM/main/L0_MergeRequest_PR pipeline #52159 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

xinhe-nv · 2026-08-06T12:34:53Z

/bot run --disable-fail-fast

tensorrt-cicd · 2026-08-06T12:40:47Z

PR_Github #64338 [ run ] triggered by Bot. Commit: 7300624 Link to invocation

tensorrt-cicd · 2026-08-06T15:03:24Z

PR_Github #64338 [ run ] completed with state ABORTED. Commit: 7300624

Link to invocation

xinhe-nv · 2026-08-07T03:12:59Z

/bot run --disable-fail-fast

tensorrt-cicd · 2026-08-07T03:19:34Z

PR_Github #64475 [ run ] triggered by Bot. Commit: 7300624 Link to invocation

tensorrt-cicd · 2026-08-07T05:45:56Z

PR_Github #64475 [ run ] completed with state SUCCESS. Commit: 7300624
/LLM/main/L0_MergeRequest_PR pipeline #52349 completed with status: 'SUCCESS'

CI Report

Link to invocation

zhaoyangwang-nvidia

Cleanup looks complete — no leftover references, no unused imports, CI green. One question below.

zhaoyangwang-nvidia · 2026-08-07T05:59:59Z

-
-    @pytest.mark.skip_device_not_contain(["H20", "H100"])
-    def test_auto_dtype(self):
-        "RCCA: https://nvbugspro.nvidia.com/bug/5310520"


This test was the guard for RCCA nvbugs/5310520 (attention-DP + CUDA-graph padding), and grep shows no other reference to that bug in the repo. Is that RCCA covered by an equivalent config elsewhere, or should the enable_attention_dp + CudaGraphConfig(enable_padding=True) combo be moved onto a model that is staying?

@zhaoyangwang-nvidia this bug is about a variant llama model, but llama family is going to be retired in TensorRT-LLM, should we keep this case? if we only want to cover enable_attention_dp + CudaGraphConfig(enable_padding=True), we can use another model to cover, such as gpt-oss and qwen3

after checking the bug again, only llama-related models hit the same TP-mode inconsistency under attention DP. Qwen3 was explicitly reported as not having the issue in the same scenario. According to AI suggestion, we can de-prioritizing or ignoring this case if its only coverage value is for the retiring LLaMA-variant path.

coderabbitai · 2026-08-10T03:58:37Z

Note

GitHub couldn't provide a complete incremental comparison for this pull request, so CodeRabbit is performing a full review instead. This review may take a little longer.

xinhe-nv requested review from a team as code owners August 6, 2026 06:51

xinhe-nv requested review from BowenFu, JunyiXu-nv, QiJune, crazydemo, jieli-matrix, tburt-nv, yuanjingx87 and zhaoyangwang-nvidia August 6, 2026 06:51

github-actions Bot assigned xinhe-nv Aug 6, 2026

Merge branch 'main' into prune

d5753fc

xinhe-nv enabled auto-merge (squash) August 6, 2026 06:52

Merge branch 'main' into prune

d9bc9b6

BowenFu reviewed Aug 6, 2026

View reviewed changes

coderabbitai Bot mentioned this pull request Aug 6, 2026

[TRTLLM-15078][test] Remove all DeepSeek-R1-Distill-* tests #17351

Open

1 task

Merge branch 'main' into prune

7300624

zhaoyangwang-nvidia approved these changes Aug 7, 2026

View reviewed changes

Merge branch 'main' into prune

ea9ff4b

Merge branch 'main' into prune

90d030e

		task.evaluate(llm)


		class TestKanana_Instruct(LlmapiAccuracyTestHarness):

Uh oh!

Conversation

xinhe-nv commented Aug 6, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Dev Engineer Review

QA Engineer Review

Description

Test Coverage

PR Checklist

GitHub Bot Help

Uh oh!

coderabbitai Bot commented Aug 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Reviews paused

Walkthrough

Changes

Uh oh!

xinhe-nv commented Aug 6, 2026

Uh oh!

tensorrt-cicd commented Aug 6, 2026

Uh oh!

BowenFu Aug 6, 2026

Choose a reason for hiding this comment

Uh oh!

xinhe-nv Aug 6, 2026

Choose a reason for hiding this comment

Uh oh!

tensorrt-cicd commented Aug 6, 2026

Uh oh!

xinhe-nv commented Aug 6, 2026

Uh oh!

tensorrt-cicd commented Aug 6, 2026

Uh oh!

tensorrt-cicd commented Aug 6, 2026

Uh oh!

xinhe-nv commented Aug 7, 2026

Uh oh!

tensorrt-cicd commented Aug 7, 2026

Uh oh!

tensorrt-cicd commented Aug 7, 2026

Uh oh!

zhaoyangwang-nvidia left a comment

Choose a reason for hiding this comment

Uh oh!

zhaoyangwang-nvidia Aug 7, 2026

Choose a reason for hiding this comment

Uh oh!

xinhe-nv Aug 10, 2026

Choose a reason for hiding this comment

Uh oh!

xinhe-nv Aug 10, 2026

Choose a reason for hiding this comment

Uh oh!

coderabbitai Bot commented Aug 10, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

xinhe-nv commented Aug 6, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented Aug 6, 2026 •

edited

Loading