[https://nvbugs/5575913][fix] Use separate thresholds for 120b/20b gptoss

[dongfengy]

…ptoss

Summary by CodeRabbit

• Updates
  • Model identifier updates: GPT-OSS names now include explicit size variants (20B and 120B options)
  • New model available: GPT-OSS/20B-MXFP4 with multiple quantization configurations
  • Both variants maintain strong performance metrics (85.0-90.3 accuracy)
  • Enhanced test suite with updated model references

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

• 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

/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...

Provide a user friendly way for developers to interact with a Jenkins server.

Run /bot [-h|--help] to print this help message.

See details below for each supported subcommand.

run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]

Launch build/test pipelines. All previously running jobs will be killed.

--reuse-test (optional)pipeline-id (OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.

--disable-reuse-test (OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.

--disable-fail-fast (OPTIONAL) : Disable fail fast on build/tests/infra failures.

--skip-test (OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.

--stage-list "A10-PyTorch-1, xxx" (OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.

--gpu-type "A30, H100_PCIe" (OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.

--test-backend "pytorch, cpp" (OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.

--only-multi-gpu-test (OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.

--disable-multi-gpu-test (OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.

--add-multi-gpu-test (OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.

--post-merge (OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.

--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" (OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".

--detailed-log (OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.

--debug (OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in the stage-list parameter to access the appropriate container environment. Note: Does NOT update GitHub check status.

For guidance on mapping tests to stage names, see docs/source/reference/ci-overview.md
and the scripts/test_to_stage_mapping.py helper.

kill

kill

Kill all running builds associated with pull request.

skip

skip --comment COMMENT

Skip testing for latest commit on pull request. --comment "Reason for skipping build/test" is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

reuse-pipeline

reuse-pipeline

Reuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.

[dongfengy]

/bot run

[coderabbitai]

📝 Walkthrough

Walkthrough

The changes update model identifiers in accuracy test references and corresponding test configurations. The generic identifier GPT-OSS/MXFP4 is renamed to GPT-OSS/120B-MXFP4, and a new model variant GPT-OSS/20B-MXFP4 is added to the accuracy reference file. Integration tests are updated to use the new explicit model names.

Changes

Cohort / File(s)	Summary
Model Reference Configuration tests/integration/defs/accuracy/references/gsm8k.yaml	Renamed GPT-OSS/MXFP4 to GPT-OSS/120B-MXFP4 (accuracy 90.3), added new GPT-OSS/20B-MXFP4 entry with quantization variants (accuracy 85.0)
Integration Test Updates tests/integration/defs/accuracy/test_llm_api_pytorch.py	Updated model name references in test functions from GPT-OSS/MXFP4 to GPT-OSS/120B-MXFP4 or GPT-OSS/20B-MXFP4 across test_w4_1gpu, test_dummy_load_format, test_w4_4gpus, test_w4_2gpus, and test_w4_chunked_prefill

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~5 minutes

• The changes are straightforward and homogeneous—simple model identifier substitutions with no logic modifications
• Verify that the new model names (GPT-OSS/120B-MXFP4 and GPT-OSS/20B-MXFP4) are properly registered in the system and test environment

Pre-merge checks and finishing touches

❌ Failed checks (2 warnings)

Check name	Status	Explanation	Resolution
Description Check	⚠️ Warning	The PR description is largely incomplete. While the template structure is present, the critical sections are empty or minimal: the "Description" section (to explain the issue and solution) is not filled out, the "Test Coverage" section is not filled out, and only the final checkbox in the PR Checklist is marked without any substantive content addressing the other checklist items. The only content provided is a partial note beginning with "…ptoss" and the "@coderabbitai summary" marker, which does not constitute meaningful description content that would help reviewers understand the purpose, justification, and testing strategy for these changes.	The author should fill out the Description section to explain why separate model variants are needed and what problem this solves. The Test Coverage section should explicitly list which tests (such as test_w4_1gpu, test_w4_4gpus, test_w4_2gpus, test_dummy_load_format, and test_w4_chunked_prefill mentioned in the changes) validate these updates. The PR Checklist items should be reviewed and documented to confirm the changes follow coding guidelines, include proper test coverage, and have been reviewed by appropriate team members.
Docstring Coverage	⚠️ Warning	Docstring coverage is 0.00% which is insufficient. The required threshold is 80.00%.	You can run @coderabbitai generate docstrings to improve docstring coverage.

✅ Passed checks (1 passed)

Check name	Status	Explanation
Title Check	✅ Passed	The PR title "[https://nvbugs/5575913][fix] Use separate thresholds for 120b/20b gptoss" directly corresponds to the changes in the pull request. The changes rename the model identifier from a generic "GPT-OSS/MXFP4" to separate variants ("GPT-OSS/120B-MXFP4" and "GPT-OSS/20B-MXFP4") with different accuracy thresholds (90.3 for 120B and 85.0 for 20B), and update corresponding test references. The title clearly conveys that the main change is differentiating model configurations by model size, and includes the proper NVBugs reference and [fix] type designation per template requirements.

✨ Finishing touches

• 📝 Generate docstrings

🧪 Generate unit tests (beta)

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

---

📜 Recent review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between a6d20f6 and aa1f1d7.

📒 Files selected for processing (2)

• tests/integration/defs/accuracy/references/gsm8k.yaml (2 hunks)
• tests/integration/defs/accuracy/test_llm_api_pytorch.py (5 hunks)

🧰 Additional context used

📓 Path-based instructions (3)

**/*.{h,hpp,hh,hxx,cpp,cxx,cc,cu,cuh,py}

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

Use only spaces, no tabs; indent with 4 spaces.

Files:

• tests/integration/defs/accuracy/test_llm_api_pytorch.py

**/*.py

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

**/*.py: Python code must target Python 3.8+.
Indent Python code with 4 spaces; do not use tabs.
Maintain module namespace when importing; prefer 'from package.subpackage import foo' then 'foo.SomeClass()' instead of importing the class directly.
Python filenames should be snake_case (e.g., some_file.py).
Python classes use PascalCase names.
Functions and methods use snake_case names.
Local variables use snake_case; prefix 'k' for variables that start with a number (e.g., k_99th_percentile).
Global variables use upper SNAKE_CASE prefixed with 'G' (e.g., G_MY_GLOBAL).
Constants use upper SNAKE_CASE (e.g., MY_CONSTANT).
Avoid shadowing variables from an outer scope.
Initialize all externally visible members of a class in the constructor.
Prefer docstrings for interfaces that may be used outside a file; comments for in-function or file-local interfaces.
Use Google-style docstrings for classes and functions (Sphinx-parsable).
Document attributes and variables inline so they render under the class/function docstring.
Avoid reflection when a simpler, explicit approach suffices (e.g., avoid dict(**locals()) patterns).
In try/except, catch the most specific exceptions possible.
For duck-typing try/except, keep the try body minimal and use else for the main logic.

Files:

• tests/integration/defs/accuracy/test_llm_api_pytorch.py

**/*.{cpp,cxx,cc,h,hpp,hh,hxx,cu,cuh,py}

📄 CodeRabbit inference engine (CODING_GUIDELINES.md)

Prepend the NVIDIA Apache-2.0 copyright header with current year to the top of all source files (e.g., .cpp, .h, .cu, .py).

Files:

• tests/integration/defs/accuracy/test_llm_api_pytorch.py

🧬 Code graph analysis (1)

tests/integration/defs/accuracy/test_llm_api_pytorch.py (2)

tensorrt_llm/llmapi/llm_args.py (1)

• model_name (1490-1491)

tensorrt_llm/llmapi/llm_utils.py (1)

• model_name (55-59)

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (2)

• GitHub Check: Check PR Checklist Resolution
• GitHub Check: Pre-commit Check

🔇 Additional comments (2)

tests/integration/defs/accuracy/references/gsm8k.yaml (1)

215-243: LGTM! Clean separation of model variants.

The changes correctly split the generic GPT-OSS/MXFP4 identifier into size-specific variants (120B-MXFP4 and 20B-MXFP4) with appropriate accuracy thresholds. The accuracy difference (90.3 vs 85.0) is reasonable given the model size difference.

tests/integration/defs/accuracy/test_llm_api_pytorch.py (1)

3453-3453: LGTM! Model identifiers correctly updated.

All test methods now use the appropriate size-specific model identifiers:

• Tests using gpt-oss-20b model path correctly reference GPT-OSS/20B-MXFP4
• Tests using gpt-oss-120b model path correctly reference GPT-OSS/120B-MXFP4

The changes are consistent with the updated accuracy reference definitions in gsm8k.yaml.

Also applies to: 3464-3464, 3511-3511, 3598-3598, 3625-3625

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[tensorrt-cicd]

PR_Github #22511 [ run ] triggered by Bot. Commit: aa1f1d7

[tensorrt-cicd]

PR_Github #22511 [ run ] completed with state SUCCESS. Commit: aa1f1d7
/LLM/main/L0_MergeRequest_PR pipeline #16968 completed with status: 'SUCCESS'
Pipeline passed with automatic retried tests. Check the rerun report for details.