[TRTLLM-7723][feat] sampling using FlashInfer.sampling #8581
Conversation
|
/bot run |
|
PR_Github #22176 [ run ] triggered by Bot. Commit: |
|
PR_Github #22176 [ run ] completed with state |
ixlmar
force-pushed
the
feat/flashinfer-sampling
branch
from
f2af5a0 to
530e64b
Compare
|
/bot run |
|
PR_Github #22257 [ run ] triggered by Bot. Commit: |
|
PR_Github #22257 [ run ] completed with state |
|
/bot run --disable-fail-fast |
1 similar comment
|
/bot run --disable-fail-fast |
|
PR_Github #22294 [ run ] triggered by Bot. Commit: |
|
PR_Github #22294 [ run ] completed with state |
ixlmar
force-pushed
the
feat/flashinfer-sampling
branch
from
530e64b to
6994f9c
Compare
|
/bot run --disable-fail-fast |
|
PR_Github #22907 [ run ] triggered by Bot. Commit: |
|
/bot run --disable-fail-fast |
|
PR_Github #22914 [ run ] triggered by Bot. Commit: |
|
PR_Github #22907 [ run ] completed with state |
|
PR_Github #22914 [ run ] completed with state |
ixlmar
force-pushed
the
feat/flashinfer-sampling
branch
from
6994f9c to
6ec395d
Compare
|
/bot run --disable-fail-fast |
|
PR_Github #23019 [ run ] triggered by Bot. Commit: |
|
PR_Github #23019 [ run ] completed with state |
ixlmar
force-pushed
the
feat/flashinfer-sampling
branch
from
6ec395d to
bb06fbf
Compare
|
/bot run --disable-fail-fast |
|
PR_Github #23396 [ run ] triggered by Bot. Commit: |
|
/bot run --disable-fail-fast |
|
PR_Github #23495 [ run ] triggered by Bot. Commit: |
|
Note: The FlashInfer path currently fails two unit tests (see CI pipeline in #8980). Those failures are to be addressed later. |
|
/bot run --disable-fail-fast |
|
PR_Github #23849 [ run ] triggered by Bot. Commit: |
|
PR_Github #23829 [ run ] completed with state |
|
PR_Github #23849 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #23870 [ run ] triggered by Bot. Commit: |
|
PR_Github #23870 [ run ] completed with state |
|
/bot run |
|
PR_Github #24010 [ run ] triggered by Bot. Commit: |
|
PR_Github #24010 [ run ] completed with state |
|
/bot run |
|
PR_Github #24025 [ run ] triggered by Bot. Commit: |
|
PR_Github #24025 [ run ] completed with state |
|
/bot run |
|
PR_Github #24136 [ run ] triggered by Bot. Commit: |
|
PR_Github #24136 [ run ] completed with state |
Signed-off-by: ixlmar <[email protected]>
|
/bot reuse-pipeline |
|
PR_Github #24174 [ reuse-pipeline ] triggered by Bot. Commit: |
|
PR_Github #24174 [ reuse-pipeline ] completed with state |
Signed-off-by: ixlmar <[email protected]>
4 participants
Description
This implements batched sampling using the FlashInfer sampling functions (TRTLLM-7723).
FlashInfer is used if possible, unless
_disable_flash_infer_sampling=True. The latter is to remain the default at least until there is better test coverage.When using FlashInfer, requests with different parameter values (e.g. top-p requests with different thresholds) can be batched together, which is expected to reduce latency for mixed batches (TRTLLM-8375). Also, temperature and SoftMax are applied only once (TRTLLM-8376).
Test Coverage
Coverage for the default functionality is unchanged (because
_disable_flash_infer_sampling=Trueis the default). Test coverage will be improved in a separate PR.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 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.
Summary by CodeRabbit
New Features
Refactor