[TRTLLM-6357][test] Add accuracy tests for Qwen3 #6177
Conversation
📝 WalkthroughWalkthroughThis update introduces support and integration tests for the Qwen3 8B model, including accuracy references, disaggregated serving, and single GPU scenarios. It adds new test classes, parameterized test cases, and updates test lists to cover various configurations, including quantized (FP8) variants. Model path resolution logic is refactored for maintainability. Changes
Sequence Diagram(s)sequenceDiagram
participant Tester
participant LLMServer (Disaggregated)
participant GSM8K Evaluation
Tester->>LLMServer (Disaggregated): Launch with Qwen3-8B model (various configs)
LLMServer (Disaggregated)->>Tester: Ready
Tester->>LLMServer (Disaggregated): Submit GSM8K evaluation task
LLMServer (Disaggregated)->>GSM8K Evaluation: Run evaluation
GSM8K Evaluation-->>LLMServer (Disaggregated): Return accuracy
LLMServer (Disaggregated)-->>Tester: Report results
Estimated code review effort🎯 2 (Simple) | ⏱️ ~8 minutes Suggested reviewers
Note ⚡️ Unit Test Generation is now available in beta!Learn more here, or try it out under "Finishing Touches" below. 📜 Recent review detailsConfiguration used: .coderabbit.yaml 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
✨ Finishing Touches🧪 Generate unit tests
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. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments. CodeRabbit Commands (Invoked using PR comments)
Other keywords and placeholders
Documentation and Community
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (1)
tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py (1)
236-236: Fix line length violation.The line exceeds the 120-character limit. Consider breaking it into multiple lines for better readability.
- " The capital of Germany is Berlin. \n\n2. What is the largest planet in our solar system? The largest planet in our", + " The capital of Germany is Berlin. \n\n2. What is the largest planet in our solar system? " + "The largest planet in our",
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (4)
tests/integration/defs/accuracy/references/gsm8k.yaml(1 hunks)tests/integration/defs/accuracy/test_disaggregated_serving.py(1 hunks)tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py(2 hunks)tests/integration/test_lists/test-db/l0_dgx_h100.yml(1 hunks)
🧰 Additional context used
🧬 Code Graph Analysis (2)
tests/integration/defs/accuracy/test_disaggregated_serving.py (2)
tests/integration/defs/accuracy/accuracy_core.py (4)
LlmapiAccuracyTestHarness(726-737)GSM8K(286-299)evaluate(146-199)evaluate(678-688)tests/integration/defs/conftest.py (1)
llm_models_root(77-83)
tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py (3)
tests/unittest/_torch/test_trtllm_sampler.py (1)
model_path(21-22)tests/unittest/llmapi/apps/_test_openai_chat_structural_tag.py (1)
model_name(17-18)tests/integration/defs/conftest.py (1)
llm_models_root(77-83)
🪛 Ruff (0.12.2)
tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py
236-236: Line too long (126 > 120)
(E501)
🔇 Additional comments (5)
tests/integration/defs/accuracy/references/gsm8k.yaml (1)
71-75: Verify identical accuracy values for baseline and FP8 quantization.The baseline accuracy (87.1114) and FP8 quantized accuracy (87.1114) are identical. While this is possible, it's unusual for quantization to preserve accuracy exactly. Please confirm these values are correct and not placeholder data.
tests/integration/test_lists/test-db/l0_dgx_h100.yml (1)
39-40: LGTM! Test configuration is consistent.The new test entries for TestQwen3_8B follow the correct format and are properly configured for the pre_merge stage with appropriate GPU requirements.
tests/integration/defs/accuracy/test_disaggregated_serving.py (1)
426-458: LGTM! Test implementation follows established patterns.The new TestQwen3_8B class is well-implemented:
- Proper inheritance from LlmapiAccuracyTestHarness
- Correct parametrization for overlap_scheduler testing
- Uses GSM8K evaluation task consistent with the accuracy reference data
- Appropriate timeout setting for disaggregated serving tests
tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py (2)
29-43: LGTM! Good refactoring to centralize model path management.The introduction of the MODEL_PATHS dictionary is a clean improvement that makes path resolution more maintainable and easier to extend.
227-241: LGTM! Test implementation is consistent with existing patterns.The new test function follows the established pattern for disaggregated single GPU testing with appropriate parameterization and expected output validation.
reasonsolo
force-pushed
the
trtllm6357_qwen3acctest
branch
from
65457d9 to
714f8e8
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (2)
tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py (2)
29-43: Consider using exact key matching instead of substring matching.The current implementation uses substring matching (
if name in model_name) which could lead to incorrect matches if model names contain substrings of other model names.Consider this more robust approach:
def model_path(model_name): llm_models_root = os.environ["LLM_MODELS_ROOT"] - for name, path in MODEL_PATHS.items(): - if name in model_name: - return os.path.join(llm_models_root, path) + if model_name in MODEL_PATHS: + return os.path.join(llm_models_root, MODEL_PATHS[model_name]) raise ValueError(f"Unknown model: {model_name}")Alternatively, if substring matching is intentional, consider using
model_name.endswith(name)for more predictable behavior.
236-236: Fix line length to comply with style guidelines.The line exceeds the 120-character limit. Consider breaking it:
- " The capital of Germany is Berlin. \n\n2. What is the largest planet in our solar system? The largest planet in our", + " The capital of Germany is Berlin. \n\n2. What is the largest planet in our solar system? " + "The largest planet in our",
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (5)
tests/integration/defs/accuracy/references/gsm8k.yaml(1 hunks)tests/integration/defs/accuracy/test_disaggregated_serving.py(1 hunks)tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py(2 hunks)tests/integration/test_lists/test-db/l0_dgx_h100.yml(1 hunks)tests/integration/test_lists/test-db/l0_h100.yml(1 hunks)
🚧 Files skipped from review as they are similar to previous changes (3)
- tests/integration/test_lists/test-db/l0_dgx_h100.yml
- tests/integration/defs/accuracy/references/gsm8k.yaml
- tests/integration/defs/accuracy/test_disaggregated_serving.py
🧰 Additional context used
🧬 Code Graph Analysis (1)
tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py (3)
tests/unittest/_torch/test_trtllm_sampler.py (1)
model_path(21-22)tests/unittest/llmapi/apps/_test_openai_chat_structural_tag.py (1)
model_name(17-18)tests/integration/defs/conftest.py (1)
llm_models_root(77-83)
🪛 Ruff (0.12.2)
tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py
236-236: Line too long (126 > 120)
(E501)
⏰ 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 (2)
tests/integration/test_lists/test-db/l0_h100.yml (1)
62-65: LGTM! Test entries properly added for Qwen3-8B-FP8.The four new test entries follow the established naming convention and parameter combination pattern consistent with other model tests in this configuration file.
tests/integration/defs/disaggregated/test_disaggregated_single_gpu.py (1)
227-241: Well-implemented test function for Qwen3-8B-FP8 model.The test follows the established pattern and is properly parameterized. The test is now correctly registered in the test list configuration, addressing the previous concern about registration.
|
/bot run |
|
PR_Github #12402 [ run ] triggered by Bot |
|
PR_Github #12402 [ run ] completed with state |
reasonsolo
force-pushed
the
trtllm6357_qwen3acctest
branch
from
e49d5b9 to
ddf761f
Compare
|
/bot run |
|
PR_Github #12434 [ run ] triggered by Bot |
Signed-off-by: Lizhi Zhou <[email protected]>
|
/bot run |
|
PR_Github #13596 [ run ] triggered by Bot |
|
/bot run |
|
PR_Github #13649 [ run ] triggered by Bot |
|
PR_Github #13596 [ run ] completed with state |
|
PR_Github #13649 [ run ] completed with state |
|
/bot run --only-multi-gpu-test --disable-fail-fast |
|
PR_Github #13706 [ run ] triggered by Bot |
|
PR_Github #13706 [ run ] completed with state |
|
/bot run --reuse-test |
|
PR_Github #13719 [ run ] triggered by Bot |
Signed-off-by: Lizhi Zhou <[email protected]>
|
/bot run --reuse-test |
|
PR_Github #13752 [ run ] triggered by Bot |
|
PR_Github #13719 [ run ] completed with state |
|
PR_Github #13752 [ run ] completed with state |
Signed-off-by: Lizhi Zhou <[email protected]> Signed-off-by: Lanyu Liao <[email protected]>
Signed-off-by: Lizhi Zhou <[email protected]>
Description
Add accuracy tests for Qwen3 disaggregated serving.
Test Coverage
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 thestage-listparameter 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.kill
killKill all running builds associated with pull request.
skip
skip --comment COMMENTSkip 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-pipelineReuse 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.
Summary by CodeRabbit
New Features
Refactor
Chores