[None][feat] Refactor scaffolding streaming feature and fix openai wo…#8622
Conversation
WeiHaocheng
force-pushed
the
feat/main_scaffolding_refactor_streaming_fix_openai
branch
2 times, most recently
from
dfdd847 to
14f2a7c
Compare
|
/bot run |
|
PR_Github #22435 [ run ] triggered by Bot. Commit: |
📝 WalkthroughWalkthroughThis PR refactors the scaffolding result and streaming architecture. It replaces a single-output model with a multi-output list-based system using a new ScaffoldingOutput dataclass, eliminates the central streaming_event in favor of streaming_output_flag and streaming_output_list, removes AsyncGeneration examples, and updates all callers to use result.outputs[0].text for result access. Changes
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Rationale: The diff spans 18 files with mixed complexity. Core architectural changes (result.py, task.py, worker.py, scaffolding_llm.py, controller.py) introduce new data structures (ScaffoldingOutput), replace event-based streaming with list-based streaming, and refactor inter-dependent logic paths. These require understanding both old and new flows and validating correctness across multiple modules. Examples and tests apply a consistent result-access pattern migration across 6 example files and 2 test files, which is more straightforward but still requires verification. Additionally, public API changes (task fields, result structure, worker methods) necessitate careful review of backward compatibility implications. Pre-merge checks and finishing touches❌ Failed checks (2 warnings)
✅ Passed checks (1 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
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. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 13
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (6)
examples/scaffolding/token_budget_majority_vote.py (2)
1-1: Missing NVIDIA Apache-2.0 copyright header.All Python source files must include the NVIDIA Apache-2.0 copyright header with the current year at the top of the file.
As per coding guidelines
16-17: Fix typo in parameter name.The parameter name
sumple_num_per_turnshould besample_num_per_turn. This typo appears throughout the file (lines 17, 21, 27, 34, 71, 97).Apply this diff to fix the typo in the constructor:
- def __init__(self, generation_controller: Controller, token_budget: int, - sumple_num_per_turn: int): + def __init__(self, generation_controller: Controller, token_budget: int, + sample_num_per_turn: int): super().__init__() self.generation_controller = generation_controller self.token_budget = token_budget - self.sumple_num_per_turn = sumple_num_per_turn + self.sample_num_per_turn = sample_num_per_turnSimilar changes are needed at lines 27, 34, 71, and 97.
examples/scaffolding/contrib/Dynasor/scaffolding_dynasor_run.py (2)
1-1: Add the required NVIDIA Apache-2.0 copyright header.This Python source file is missing the NVIDIA Apache-2.0 copyright header with the current year (2025), which is required by the coding guidelines.
Add the following header at the top of the file:
+# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved. +# SPDX-License-Identifier: Apache-2.0 +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + import argparseAs per coding guidelines
80-82: Add bounds checking forresult.outputslist access.The non-streaming path also accesses
result.outputs[0]without verifying that theoutputslist is non-empty, risking anIndexError.Apply this diff:
results = llm.generate(prompts) for result in results: - print(result.outputs[0].text) + if result.outputs: + print(result.outputs[0].text)tensorrt_llm/scaffolding/contrib/Dynasor/dynasor_controller.py (1)
1-1: Add the required NVIDIA Apache-2.0 copyright header.The coding guidelines require an Apache-2.0 copyright header with the current year (2025) at the top of all Python source files.
Add this header at the top of the file:
+# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved. +# SPDX-License-Identifier: Apache-2.0 +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License. + from enum import EnumAs per coding guidelines.
tensorrt_llm/scaffolding/task.py (1)
1-1: Add NVIDIA Apache-2.0 header (2025).Header is required for this Python file too.
Use the same header snippet as in result.py.
🧹 Nitpick comments (2)
examples/scaffolding/token_budget_majority_vote.py (1)
107-108: API change looks correct, but consider adding defensive checks.The updated call to
result.outputs[0].textcorrectly uses the new multi-output API structure. However, consider adding defensive checks:
- Verify that
result.outputsis not empty before accessing index 0- Handle the case where
extract_answer_from_boxedreturnsNone(when no boxed answer is found)Example refactor for defensive coding:
- extracted_answer = extract_answer_from_boxed(result.outputs[0].text) - print(f'extracted_answer={extracted_answer}') + if not result.outputs: + print('No outputs generated') + else: + extracted_answer = extract_answer_from_boxed(result.outputs[0].text) + if extracted_answer is None: + print('No boxed answer found in output') + else: + print(f'extracted_answer={extracted_answer}')tensorrt_llm/scaffolding/task.py (1)
86-87: Minor: comment typo.“don’t not use TokenLogprobs” → “avoid using TokenLogprobs for general support.”
Apply:
- # TODO: don't not use TokenLogprobs for general support + # TODO: Avoid using TokenLogprobs for general support
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (18)
examples/scaffolding/contrib/AsyncGeneration/README.md(0 hunks)examples/scaffolding/contrib/AsyncGeneration/stream_generation_run.py(0 hunks)examples/scaffolding/contrib/Dynasor/scaffolding_dynasor_run.py(1 hunks)examples/scaffolding/contrib/mcp/mcptest.py(4 hunks)examples/scaffolding/run_basic_generation.py(1 hunks)examples/scaffolding/token_budget_majority_vote.py(1 hunks)tensorrt_llm/scaffolding/__init__.py(1 hunks)tensorrt_llm/scaffolding/contrib/AsyncGeneration/README.md(0 hunks)tensorrt_llm/scaffolding/contrib/AsyncGeneration/__init__.py(0 hunks)tensorrt_llm/scaffolding/contrib/Dynasor/dynasor_controller.py(1 hunks)tensorrt_llm/scaffolding/contrib/mcp/chat_handler.py(1 hunks)tensorrt_llm/scaffolding/controller.py(1 hunks)tensorrt_llm/scaffolding/result.py(1 hunks)tensorrt_llm/scaffolding/scaffolding_llm.py(2 hunks)tensorrt_llm/scaffolding/task.py(5 hunks)tensorrt_llm/scaffolding/worker.py(3 hunks)tests/unittest/scaffolding/test_bench.py(1 hunks)tests/unittest/scaffolding/test_scaffolding.py(4 hunks)
💤 Files with no reviewable changes (4)
- examples/scaffolding/contrib/AsyncGeneration/README.md
- tensorrt_llm/scaffolding/contrib/AsyncGeneration/README.md
- tensorrt_llm/scaffolding/contrib/AsyncGeneration/init.py
- examples/scaffolding/contrib/AsyncGeneration/stream_generation_run.py
🧰 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:
tensorrt_llm/scaffolding/contrib/mcp/chat_handler.pytensorrt_llm/scaffolding/scaffolding_llm.pyexamples/scaffolding/contrib/mcp/mcptest.pytests/unittest/scaffolding/test_scaffolding.pytensorrt_llm/scaffolding/contrib/Dynasor/dynasor_controller.pytensorrt_llm/scaffolding/worker.pytensorrt_llm/scaffolding/result.pytests/unittest/scaffolding/test_bench.pytensorrt_llm/scaffolding/__init__.pyexamples/scaffolding/token_budget_majority_vote.pytensorrt_llm/scaffolding/controller.pyexamples/scaffolding/run_basic_generation.pyexamples/scaffolding/contrib/Dynasor/scaffolding_dynasor_run.pytensorrt_llm/scaffolding/task.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:
tensorrt_llm/scaffolding/contrib/mcp/chat_handler.pytensorrt_llm/scaffolding/scaffolding_llm.pyexamples/scaffolding/contrib/mcp/mcptest.pytests/unittest/scaffolding/test_scaffolding.pytensorrt_llm/scaffolding/contrib/Dynasor/dynasor_controller.pytensorrt_llm/scaffolding/worker.pytensorrt_llm/scaffolding/result.pytests/unittest/scaffolding/test_bench.pytensorrt_llm/scaffolding/__init__.pyexamples/scaffolding/token_budget_majority_vote.pytensorrt_llm/scaffolding/controller.pyexamples/scaffolding/run_basic_generation.pyexamples/scaffolding/contrib/Dynasor/scaffolding_dynasor_run.pytensorrt_llm/scaffolding/task.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:
tensorrt_llm/scaffolding/contrib/mcp/chat_handler.pytensorrt_llm/scaffolding/scaffolding_llm.pyexamples/scaffolding/contrib/mcp/mcptest.pytests/unittest/scaffolding/test_scaffolding.pytensorrt_llm/scaffolding/contrib/Dynasor/dynasor_controller.pytensorrt_llm/scaffolding/worker.pytensorrt_llm/scaffolding/result.pytests/unittest/scaffolding/test_bench.pytensorrt_llm/scaffolding/__init__.pyexamples/scaffolding/token_budget_majority_vote.pytensorrt_llm/scaffolding/controller.pyexamples/scaffolding/run_basic_generation.pyexamples/scaffolding/contrib/Dynasor/scaffolding_dynasor_run.pytensorrt_llm/scaffolding/task.py
🧬 Code graph analysis (9)
tensorrt_llm/scaffolding/contrib/mcp/chat_handler.py (1)
tensorrt_llm/scaffolding/worker.py (1)
add_param_if_not_none(47-51)
tensorrt_llm/scaffolding/scaffolding_llm.py (2)
examples/scaffolding/contrib/Dynasor/scaffolding_dynasor_run.py (1)
task(67-75)tensorrt_llm/scaffolding/result.py (3)
result(45-49)set_output_streaming(30-31)ScaffoldingResult(12-67)
examples/scaffolding/contrib/mcp/mcptest.py (2)
tensorrt_llm/scaffolding/contrib/mcp/chat_handler.py (1)
chat_handler(36-52)tensorrt_llm/scaffolding/result.py (1)
result(45-49)
tests/unittest/scaffolding/test_scaffolding.py (1)
tensorrt_llm/scaffolding/result.py (1)
result(45-49)
tensorrt_llm/scaffolding/worker.py (5)
tensorrt_llm/llmapi/llm.py (1)
generate_async(332-497)tensorrt_llm/executor/executor.py (1)
GenerationExecutor(77-578)tensorrt_llm/executor/result.py (1)
GenerationResult(656-830)tensorrt_llm/scaffolding/result.py (3)
result(45-49)ScaffoldingOutput(7-9)_aresult_step(37-43)tensorrt_llm/scaffolding/task.py (3)
GenerationTask(46-99)StreamGenerationTask(103-125)StreamGenerationTask(136-150)
examples/scaffolding/token_budget_majority_vote.py (2)
tensorrt_llm/scaffolding/math_utils.py (1)
extract_answer_from_boxed(14-53)tensorrt_llm/scaffolding/result.py (1)
result(45-49)
examples/scaffolding/run_basic_generation.py (2)
tensorrt_llm/scaffolding/result.py (1)
result(45-49)tensorrt_llm/scaffolding/scaffolding_llm.py (1)
generate_async(173-193)
examples/scaffolding/contrib/Dynasor/scaffolding_dynasor_run.py (2)
tensorrt_llm/scaffolding/result.py (1)
result(45-49)tensorrt_llm/scaffolding/scaffolding_llm.py (1)
generate_async(173-193)
tensorrt_llm/scaffolding/task.py (2)
tensorrt_llm/scaffolding/result.py (2)
result(45-49)ScaffoldingOutput(7-9)tensorrt_llm/scaffolding/contrib/AsyncGeneration/stream_generation.py (1)
StreamGenerationTask(10-33)
🪛 Ruff (0.14.1)
examples/scaffolding/contrib/mcp/mcptest.py
31-31: Redefinition of unused MCPController from line 7
Remove definition: MCPController
(F811)
31-31: Redefinition of unused MCPWorker from line 8
Remove definition: MCPWorker
(F811)
tensorrt_llm/scaffolding/result.py
34-34: Undefined name TaskCollection
(F821)
45-45: Unused method argument: timeout
(ARG002)
tensorrt_llm/scaffolding/task.py
136-136: Redefinition of unused StreamGenerationTask from line 103
(F811)
⏰ 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 (12)
tensorrt_llm/scaffolding/contrib/Dynasor/dynasor_controller.py (2)
108-108: Code is correct; no changes needed.The
GenerationTaskclass maintains backward compatibility by keeping theoutput_strattribute, which is populated fromresult.outputs[0].textvia thefill_task_with_result()method (seetensorrt_llm/scaffolding/worker.py:215). The dynasor_controller code reads fromoutput_strafter tasks have been filled, so the attribute access is valid. The internalGenerationTaskAPI layer is properly bridged to the external output structure—no update required.
129-132: The review comment is based on incorrect assumptions about the codebase.The original comment references "tasks[0].result assignment," but this attribute does not exist in the
GenerationTaskorTaskclasses. The codebase uses onlyoutput_str, which the current code at line 140 correctly assigns to:tasks[0].output_str = current_prompt.The control flow refactoring is sound:
- Certitude path: appends answer to
current_prompt, breaks, then assigns at line 140- Max-tokens path: exits loop normally, then assigns at line 140
- Both paths converge on the same output mechanism (
output_str)The suffix selection logic (line 129) correctly chooses
answer_suffix_with_markerwhen</think>is not yet present to close the think tag.Likely an incorrect or invalid review comment.
tensorrt_llm/scaffolding/contrib/mcp/chat_handler.py (1)
22-24: Worker-level parameter fallbacks removed—verify intended behavior.The parameter selection now only considers task-level values for
max_tokens,temperature, andtop_p, removing worker-level fallbacks. If a task attribute is None, the parameter will not be set at all. Ensure this behavior is intentional and that downstream OpenAI API calls handle missing optional parameters correctly.examples/scaffolding/contrib/mcp/mcptest.py (1)
64-64: LGTM! Result access updated to new structure.The change to
result.outputs[0].textcorrectly aligns with the new multi-output result structure introduced in this PR.tests/unittest/scaffolding/test_bench.py (1)
59-59: LGTM! Test assertion updated to new outputs structure.The change correctly accesses the first output's text via
results[0].outputs[0].text, aligning with the new multi-output result model.tensorrt_llm/scaffolding/controller.py (1)
70-70: LGTM! Attribute renamed to match new streaming architecture.The rename from
task.streamingtotask.streaming_output_flagaligns with the broader streaming refactoring. The logic remains unchanged.examples/scaffolding/run_basic_generation.py (1)
57-63: LGTM! Async streaming simplified with new outputs structure.The refactored loop removes the inner iteration and directly accesses
result.outputs[0], making the code cleaner and consistent with the new multi-output result model.tests/unittest/scaffolding/test_scaffolding.py (1)
52-53: LGTM! All test assertions updated to new outputs structure.All test assertions now correctly access the output text via
result.outputs[0].text, consistently reflecting the new multi-output result model across unbatched, batched, async, and majority vote scenarios.Also applies to: 65-66, 77-78, 89-90
tensorrt_llm/scaffolding/scaffolding_llm.py (2)
84-87: LGTM! Streaming refactored to use output list.The streaming logic now correctly iterates over
task.streaming_output_listand emits outputs viaset_output_streaming, replacing the previous event-based mechanism. Clearing the list after emission prevents duplicate outputs.
174-174: LGTM! ScaffoldingResult initialization simplified.The removal of the
streaming_eventargument aligns with the streaming architecture refactoring where results now use a queue-based approach instead of event-based synchronization.tensorrt_llm/scaffolding/task.py (1)
81-89: Result fields and create_scaffolding_output look consistent.Mapping to ScaffoldingOutput aligns with the new outputs[0] API.
Also applies to: 98-100
tensorrt_llm/scaffolding/worker.py (1)
214-219: Field mapping confirmed in GenerationResult outputs API.The GenerationResult class exposes the accessed fields correctly. Verification confirms:
outputs[0].text,outputs[0].token_ids, andoutputs[0].logprobsare consistently accessed throughout the codebasecontext_logitsis present at the GenerationResult level as expectedThe field mapping is sound and aligns with existing API usage patterns.
|
PR_Github #22435 [ run ] completed with state |
WeiHaocheng
force-pushed
the
feat/main_scaffolding_refactor_streaming_fix_openai
branch
from
14f2a7c to
fbda2be
Compare
|
/bot run |
|
PR_Github #22514 [ run ] triggered by Bot. Commit: |
|
PR_Github #22514 [ run ] completed with state |
WeiHaocheng
force-pushed
the
feat/main_scaffolding_refactor_streaming_fix_openai
branch
from
fbda2be to
24576d4
Compare
|
/bot run --reuse-pipeline |
|
PR_Github #22858 Bot args parsing error: usage: /bot [-h] |
|
/bot --help |
GitHub Bot Help
Provide a user friendly way for developers to interact with a Jenkins server. Run See details below for each supported subcommand. Details
Launch build/test pipelines. All previously running jobs will be killed.
kill
Kill all running builds associated with pull request. skip
Skip testing for latest commit on pull request. 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. |
WeiHaocheng
force-pushed
the
feat/main_scaffolding_refactor_streaming_fix_openai
branch
from
24576d4 to
9194149
Compare
|
/bot reuse-pipeline |
|
PR_Github #22960 [ reuse-pipeline ] triggered by Bot. Commit: |
|
PR_Github #22960 [ reuse-pipeline ] completed with state |
…rker Signed-off-by: Fred Wei <20514172+WeiHaocheng@users.noreply.github.com>
WeiHaocheng
force-pushed
the
feat/main_scaffolding_refactor_streaming_fix_openai
branch
from
9194149 to
09074aa
Compare
|
/bot reuse-pipeline |
1 similar comment
|
/bot reuse-pipeline |
|
PR_Github #23007 [ reuse-pipeline ] triggered by Bot. Commit: |
|
PR_Github #23007 [ reuse-pipeline ] completed with state |
NVIDIA#8622) Signed-off-by: Fred Wei <20514172+WeiHaocheng@users.noreply.github.com> Signed-off-by: FredricZ-2007 <226039983+fredricz-20070104@users.noreply.github.com>
6 participants
…rker
Summary by CodeRabbit
Release Notes
New Features
ScaffoldingOutputstructure for improved result handling with text and token ID tracking.Documentation
Breaking Changes
result.outputs[0].textinstead of previous direct output access methods.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.
Details
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.mdand the
scripts/test_to_stage_mapping.pyhelper.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.