KEMBAR78
[None][infra] Improve the failure message for accuracy test suite by syuoni · Pull Request #7994 · NVIDIA/TensorRT-LLM · GitHub
Skip to content

Conversation

@syuoni
Copy link
Collaborator

@syuoni syuoni commented Sep 25, 2025

Summary by CodeRabbit

  • New Features
    • Structured hypothesis testing report during accuracy evaluations, with clear pass/fail assertions.
  • Refactor
    • Migrated accuracy evaluation to a parameterized workflow, centralizing sample counts and thresholds for consistency.
    • Replaced print statements with structured logging across evaluation and summarization steps for clearer progress visibility.
  • Tests
    • Integration and evaluation flows updated to use the new hypothesis testing parameters; integration mode defaults to 1 sample.
  • Documentation
    • README updated to reference the new method for locating accuracy specification logic.

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.

Signed-off-by: Enwei Zhu <21126786+syuoni@users.noreply.github.com>
@syuoni
Copy link
Collaborator Author

syuoni commented Sep 25, 2025

/bot run

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19941 [ run ] triggered by Bot

@coderabbitai
Copy link
Contributor

coderabbitai bot commented Sep 25, 2025

📝 Walkthrough

Walkthrough

Introduces HypothesisTestingParams dataclass and replaces AccuracyTask.get_num_samples_and_threshold with AccuracyTask.get_hypothesis_testing_params. Updates evaluation/summarization to use the new object’s num_samples and threshold, adds report/assert methods, and switches various print statements to logger.info. Documentation updated to reference the new method.

Changes

Cohort / File(s) Summary
Documentation reference update
tests/integration/defs/accuracy/README.md
Renames referenced API from AccuracyTask.get_num_samples_and_threshold to AccuracyTask.get_hypothesis_testing_params. No other content changes.
Accuracy testing params refactor and logging
tests/integration/defs/accuracy/accuracy_core.py
Adds HypothesisTestingParams dataclass (fields: ref_accuracy, num_samples, alpha, beta, sigma, higher_is_better; computed: theta, threshold) with report and assert_passing. Replaces get_num_samples_and_threshold with get_hypothesis_testing_params. Updates evaluation/summarization to use the new object, including logging reports, and converts print statements to logger.info.

Sequence Diagram(s)

sequenceDiagram
  autonumber
  actor Runner as Test Runner
  participant Task as AccuracyTask
  participant HParams as HypothesisTestingParams
  participant Eval as Evaluator

  Runner->>Task: get_hypothesis_testing_params(acc_specs)
  Task->>HParams: construct with acc_specs
  HParams-->>Task: params (num_samples, threshold, ...)
  Task-->>Runner: HypothesisTestingParams

  Runner->>Eval: evaluate(num_samples = params.num_samples)
  Eval-->>Runner: accuracy
  Runner->>HParams: report(accuracy) / assert_passing(accuracy)
  HParams-->>Runner: report string / assertion outcome
Loading
sequenceDiagram
  autonumber
  participant Runner as Test Runner
  participant Task as AccuracyTask
  participant Eval as Evaluator

  Note over Runner,Eval: Previous flow (removed)
  Runner->>Task: get_num_samples_and_threshold(acc_specs)
  Task-->>Runner: (num_samples, threshold)
  Runner->>Eval: evaluate(num_samples)
  Eval-->>Runner: accuracy
  Runner->>Runner: compare accuracy vs threshold
Loading

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Pre-merge checks and finishing touches

❌ Failed checks (3 warnings)
Check name Status Explanation Resolution
Title Check ⚠️ Warning The pull request title suggests a minor infrastructure change to improve failure messages for the accuracy test suite, but the actual changes introduce a new HypothesisTestingParams dataclass, deprecate and replace get_num_samples_and_threshold with get_hypothesis_testing_params, refactor logging across multiple tasks, and alter the public interface of AccuracyTask. This mismatch means the title does not accurately or concisely capture the primary scope of the change. A concise title should highlight the core refactor and new functionality rather than only mention message improvements. Revise the title to reflect the main changes, for example: “[None][feat] Refactor accuracy testing to use HypothesisTestingParams and enhance logging” so that it clearly summarizes the introduction of the dataclass and interface update.
Description Check ⚠️ Warning The pull request description is just the unfilled template structure with headings for summary, description, and test coverage but contains no actual content under those sections. This leaves critical information about the nature of the changes, the rationale, and relevant tests entirely missing. Without the filled summary and test coverage details, reviewers cannot assess the change impact or verify that appropriate tests exist. Please populate the template by providing a concise summary of the changes, a clear description of what and why, and list relevant tests for new code paths under “Test Coverage.” Ensure each required section contains substantive content rather than placeholder text.
Docstring Coverage ⚠️ Warning Docstring coverage is 7.69% which is insufficient. The required threshold is 80.00%. You can run @coderabbitai generate docstrings to improve docstring coverage.
✨ Finishing touches
  • 📝 Generate Docstrings
🧪 Generate unit tests
  • Create PR with unit tests
  • Post copyable unit tests in a comment

Tip

👮 Agentic pre-merge checks are now available in preview!

Pro plan users can now enable pre-merge checks in their settings to enforce checklists before merging PRs.

  • Built-in checks – Quickly apply ready-made checks to enforce title conventions, require pull request descriptions that follow templates, validate linked issues for compliance, and more.
  • Custom agentic checks – Define your own rules using CodeRabbit’s advanced agentic capabilities to enforce organization-specific policies and workflows. For example, you can instruct CodeRabbit’s agent to verify that API documentation is updated whenever API schema files are modified in a PR. Note: Upto 5 custom checks are currently allowed during the preview period. Pricing for this feature will be announced in a few weeks.

Please see the documentation for more information.

Example:

reviews:
  pre_merge_checks:
    custom_checks:
      - name: "Undocumented Breaking Changes"
        mode: "warning"
        instructions: |
          Pass/fail criteria: All breaking changes to public APIs, CLI flags, environment variables, configuration keys, database schemas, or HTTP/GraphQL endpoints must be documented in the "Breaking Change" section of the PR description and in CHANGELOG.md. Exclude purely internal or private changes (e.g., code not exported from package entry points or explicitly marked as internal).

Please share your feedback with us on this Discord post.


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

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

Copy link
Contributor

@coderabbitai coderabbitai bot left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Actionable comments posted: 1

🧹 Nitpick comments (4)
tests/integration/defs/accuracy/accuracy_core.py (4)

70-80: Add a concise Google‑style docstring to HypothesisTestingParams.

Improves API clarity and aligns with the repo’s Python docstring guidelines.

 @dataclass(slots=True)
 class HypothesisTestingParams:
+    """Hypothesis testing configuration for accuracy checks.
+
+    Attributes:
+        ref_accuracy (float): Reference accuracy for threshold computation.
+            Typically in [0, 100] for percentage-like metrics; may be raw for
+            lower-is-better metrics (e.g., perplexity).
+        num_samples (int): Number of samples to evaluate.
+        alpha (float): Type I error rate in (0, 1). Defaults to 0.05.
+        beta (float): Type II error rate in (0, 1). Defaults to 0.2.
+        sigma (float): Per-sample score standard deviation scaled to the metric
+            range (e.g., 50.0 for Bernoulli accuracy in 0–100 scale).
+        higher_is_better (bool): True if larger metric is better; False otherwise.
+        theta (float): Computed minimum detectable effect (post-init).
+        threshold (float): Computed decision threshold (post-init).
+    """

112-119: Add failure margin to the assertion message for faster debugging.

Including how far the result is from the threshold speeds triage.

     def assert_passing(self, accuracy: float) -> None:
         compare_op = ">=" if self.higher_is_better else "<="
-        err_msg = f"Reference accuracy is {self.ref_accuracy:.3f}, threshold is {self.threshold:.3f}. Expected accuracy {compare_op} threshold, but got {accuracy:.3f}. Please see hypothesis testing report:\n{self.report(accuracy)}"
+        margin = (accuracy - self.threshold) if self.higher_is_better else (self.threshold - accuracy)
+        err_msg = (
+            f"Reference accuracy is {self.ref_accuracy:.3f}, threshold is {self.threshold:.3f}. "
+            f"Expected accuracy {compare_op} threshold (margin: {margin:.3f}), but got {accuracy:.3f}. "
+            f"Please see hypothesis testing report:\n{self.report(accuracy)}"
+        )
         if self.higher_is_better:
             assert accuracy >= self.threshold, err_msg
         else:
             assert accuracy <= self.threshold, err_msg

148-150: Complete the docstring with Returns and Raises sections (Google style).

Aligns with repo docs and clarifies contract.

-    def get_hypothesis_testing_params(self,
-                                      **acc_specs) -> HypothesisTestingParams:
-        """Get hypothesis testing parameters via accuracy specifications.
+    def get_hypothesis_testing_params(self,
+                                      **acc_specs) -> HypothesisTestingParams:
+        """Get hypothesis testing parameters via accuracy specifications.
 
-        Args:
-            acc_specs: Accuracy specifications, currently including:
-                dtype (str): Model data type. Defaults to 'auto'.
-                quant_algo (str): Quantizaion algorithm. Defaults to None.
-                kv_cache_quant_algo (str): KV cache quantizaion algorithm. Defaults to None.
-                spec_dec_algo (str): Speculative decoding algorithm. Defaults to None.
-                extra_acc_spec (str): Extra accuracy specifications. Defaults to None.
-        """
+        Args:
+            acc_specs: Accuracy specifications, currently including:
+                dtype (str): Model data type. Defaults to 'auto'.
+                quant_algo (str): Quantization algorithm. Defaults to None.
+                kv_cache_quant_algo (str): KV cache quantization algorithm. Defaults to None.
+                spec_dec_algo (str): Speculative decoding algorithm. Defaults to None.
+                extra_acc_spec (str): Extra accuracy specification tag. Defaults to None.
+
+        Returns:
+            HypothesisTestingParams: Filled hypothesis testing parameters for this task/model/spec.
+
+        Raises:
+            ValueError: If no matching reference is registered and TRTLLM_ACCURACY_NO_REFERENCE != "1".
+        """

Also applies to: 152-159


204-210: Clarify the integration-test log message.

It does not truly “skip verification”; it uses a permissive threshold. Make the message precise.

         if is_integration_test:
             logger.info(
-                "Running in INTEGRATION_TEST mode: using only 1 sample and skipping accuracy verification"
+                "Running in INTEGRATION_TEST mode: using 1 sample and a permissive threshold (effectively skipping strict accuracy verification)"
             )
             hypothesis_testing_params = HypothesisTestingParams(ref_accuracy=0,
                                                                 num_samples=1)
📜 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 e30d9ac and 1effa0b.

📒 Files selected for processing (2)
  • tests/integration/defs/accuracy/README.md (1 hunks)
  • tests/integration/defs/accuracy/accuracy_core.py (12 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/accuracy_core.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/accuracy_core.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/accuracy_core.py
🧬 Code graph analysis (1)
tests/integration/defs/accuracy/accuracy_core.py (4)
tensorrt_llm/evaluate/lm_eval.py (1)
  • evaluate (385-417)
tensorrt_llm/evaluate/interface.py (1)
  • evaluate (81-110)
tests/integration/defs/conftest.py (2)
  • llm_root (189-190)
  • llm_venv (699-716)
tests/integration/defs/common.py (1)
  • venv_check_call (28-34)
⏰ 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). (1)
  • GitHub Check: Pre-commit Check
🔇 Additional comments (10)
tests/integration/defs/accuracy/accuracy_core.py (9)

19-19: LGTM: new dataclass imports are appropriate.


93-111: LGTM: clear, user‑friendly hypothesis testing report.

This will significantly improve failure diagnostics.


175-182: LGTM: centralizing parameter resolution into HypothesisTestingParams.

Defaulting to task-level ALPHA/BETA/SIGMA/NUM_SAMPLES and HIGHER_IS_BETTER is clean. Combined with post_init validation, bad configs will fail fast with clear errors.


233-235: LGTM: passing num_samples into the evaluator constructor.

Matches how downstream evaluators derive their limits; keeps evaluation volume consistent with hypothesis settings.


242-246: LGTM: improved assertion path with detailed report.

The inline report will be invaluable for CI failures.


499-499: LGTM: switch from prints to structured logger.info.

Improves log consistency and visibility in CI.

Also applies to: 601-601, 620-621, 688-689, 719-719, 727-727


637-647: LGTM: consistent use of HypothesisTestingParams in summarize flow.

num_samples and threshold are consumed uniformly; the pre-log of the report is a nice touch.


688-699: LGTM: MMLU flow now uses unified hypothesis parameters.

Stable and clear.


734-744: LGTM: long-context eval uses unified hypothesis parameters.

Consistent with other flows; batch sizing derived safely.

tests/integration/defs/accuracy/README.md (1)

128-129: LGTM: updated reference to get_hypothesis_testing_params.

Documentation now matches the new API surface.

@tensorrt-cicd
Copy link
Collaborator

PR_Github #19941 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #15013 completed with status: 'SUCCESS'
Pipeline passed with automatic retried tests. Check the rerun report for details.

Signed-off-by: Enwei Zhu <21126786+syuoni@users.noreply.github.com>
@syuoni
Copy link
Collaborator Author

syuoni commented Sep 26, 2025

/bot reuse-pipeline --comment "resolved trivial rebase conflict"

@tensorrt-cicd
Copy link
Collaborator

PR_Github #20013 [ reuse-pipeline ] triggered by Bot

@tensorrt-cicd
Copy link
Collaborator

PR_Github #20013 [ reuse-pipeline ] completed with state SUCCESS
Reusing PR_Github #19941 for commit 119cc56

@syuoni syuoni merged commit d650320 into NVIDIA:main Sep 26, 2025
5 checks passed
@syuoni syuoni deleted the improve-assertion branch September 26, 2025 02:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants