[None][feat] Return topk logprobs in torch backend #7756
Conversation
Signed-off-by: Dong Cao <docao@nvidia.com>
…topk_logprobs_torch_backend Signed-off-by: Dong Cao <docao@nvidia.com>
Signed-off-by: Dong Cao <docao@nvidia.com>
svc-trtllm-gh-bot
added
the
Community want to contribute
📝 WalkthroughWalkthroughAdds end-to-end support for top‑k token log probabilities: introduces a logprobs parameter in sampling, plumbs it through workers and request objects, computes per-step top‑k logprobs in the sampler via log_softmax, removes prior constraint (logprobs > 1), and adds a unit test validating structure and ordering. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
participant Client
participant LLMAPI as LLM API
participant Scaff as Scaffolding Worker
participant Exec as Executor Worker
participant Req as LlmRequest
participant Sampler
Client->>LLMAPI: generate(prompt, SamplingParams{logprobs=N})
LLMAPI-->>LLMAPI: validate args (no N==1 restriction)
LLMAPI->>Scaff: task params
Scaff-->>Scaff: build SamplingParams(..., logprobs=N)
Scaff->>Exec: enqueue with SamplingParams
Exec-->>Exec: set py_num_logprobs = SamplingParams.logprobs
Exec->>Req: construct LlmRequest(..., num_logprobs=N)
loop decode steps
Sampler->>Sampler: log_softmax(raw_logits)
alt logprobs requested
Sampler-->>Sampler: topk per step (vals, indices)
Sampler-->>Req: store py_topk_logprobs_vals/indices
end
end
Sampler-->>Client: outputs with per-token top‑k logprobs
Estimated code review effort🎯 4 (Complex) | ⏱️ ~60 minutes ✨ Finishing touches
🧪 Generate unit tests
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.
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. Comment Pre-merge checks❌ Failed checks (2 warnings)
✅ Passed checks (1 passed)
|
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 6
🧹 Nitpick comments (7)
tensorrt_llm/executor/worker.py (1)
1-1: Missing NVIDIA Apache-2.0 header (2025).Add the required copyright/license header per repo convention.
tests/unittest/llmapi/test_llm_pytorch.py (1)
1-1: Missing NVIDIA Apache-2.0 header (2025).Add the required copyright/license header per repo convention.
tensorrt_llm/_torch/pyexecutor/llm_request.py (1)
1-1: Missing NVIDIA Apache-2.0 header (2025).Add the required copyright/license header per repo convention.
tensorrt_llm/scaffolding/worker.py (2)
184-186: Minor: coerce task.num_logprobs to int for SamplingParamsProtect against float/str inputs from upstream task sources.
Apply this diff:
- return_context_logits=task.return_context_logits, - logprobs=task.num_logprobs) + return_context_logits=task.return_context_logits, + logprobs=(int(task.num_logprobs) if task.num_logprobs is not None else None))
1-1: Missing NVIDIA Apache-2.0 header (2025).Add the required copyright/license header per repo convention.
tensorrt_llm/_torch/pyexecutor/sampler.py (2)
590-594: Type: log_probs_host is a boolean nowFix the signature/type to avoid confusion and accidental tensor operations.
Apply this diff:
- def log_probs_host(self, scheduled_requests: ScheduledRequests): + def log_probs_host(self, scheduled_requests: ScheduledRequests) -> bool:And in
_process_requests:- log_probs_host: torch.Tensor | None = None): + log_probs_host: bool = False):
1-1: Missing NVIDIA Apache-2.0 header (2025).Add the required copyright/license header per repo convention.
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (6)
tensorrt_llm/_torch/pyexecutor/llm_request.py(3 hunks)tensorrt_llm/_torch/pyexecutor/sampler.py(7 hunks)tensorrt_llm/executor/worker.py(1 hunks)tensorrt_llm/llmapi/llm.py(0 hunks)tensorrt_llm/scaffolding/worker.py(1 hunks)tests/unittest/llmapi/test_llm_pytorch.py(1 hunks)
💤 Files with no reviewable changes (1)
- tensorrt_llm/llmapi/llm.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/executor/worker.pytests/unittest/llmapi/test_llm_pytorch.pytensorrt_llm/scaffolding/worker.pytensorrt_llm/_torch/pyexecutor/llm_request.pytensorrt_llm/_torch/pyexecutor/sampler.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/executor/worker.pytests/unittest/llmapi/test_llm_pytorch.pytensorrt_llm/scaffolding/worker.pytensorrt_llm/_torch/pyexecutor/llm_request.pytensorrt_llm/_torch/pyexecutor/sampler.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/executor/worker.pytests/unittest/llmapi/test_llm_pytorch.pytensorrt_llm/scaffolding/worker.pytensorrt_llm/_torch/pyexecutor/llm_request.pytensorrt_llm/_torch/pyexecutor/sampler.py
🧠 Learnings (1)
📚 Learning: 2025-08-28T10:25:22.370Z
Learnt from: ixlmar
PR: NVIDIA/TensorRT-LLM#7294
File: tensorrt_llm/_torch/pyexecutor/sampler.py:887-891
Timestamp: 2025-08-28T10:25:22.370Z
Learning: In tensorrt_llm/_torch/pyexecutor/sampler.py, the draft_probs and target_probs tensors have shapes [1, steps] not [steps, vocab_size] as might be expected, making the .squeeze(0) operations appropriate for removing the batch dimension of size 1.
Applied to files:
tensorrt_llm/_torch/pyexecutor/sampler.py
🧬 Code graph analysis (4)
tensorrt_llm/executor/worker.py (1)
tensorrt_llm/scaffolding/task.py (1)
logprobs(99-100)
tests/unittest/llmapi/test_llm_pytorch.py (4)
tensorrt_llm/llmapi/llm.py (1)
generate(238-316)tensorrt_llm/sampling_params.py (1)
SamplingParams(125-486)tensorrt_llm/scaffolding/task.py (1)
logprobs(99-100)tensorrt_llm/executor/result.py (1)
outputs(198-213)
tensorrt_llm/scaffolding/worker.py (2)
tests/unittest/llmapi/test_llm.py (6)
task(480-487)task(527-532)task(1862-1871)task(1965-1978)task(2326-2327)task(2417-2436)tensorrt_llm/scaffolding/task.py (1)
logprobs(99-100)
tensorrt_llm/_torch/pyexecutor/sampler.py (2)
tensorrt_llm/executor/result.py (1)
Logprob(37-40)tensorrt_llm/_torch/pyexecutor/scheduler.py (1)
all_requests(38-39)
|
/bot run |
|
PR_Github #18737 [ run ] triggered by Bot |
|
PR_Github #18737 [ run ] completed with state |
WeiHaocheng
force-pushed
the
docao/support_topk_logprobs_torch_backend
branch
from
fb6f39f to
2aaf2a7
Compare
|
/bot run |
|
PR_Github #18876 [ run ] triggered by Bot |
|
PR_Github #18876 [ run ] completed with state |
WeiHaocheng
removed
the
Community want to contribute
Signed-off-by: Dong Cao <docao@nvidia.com>
Signed-off-by: Dong Cao <docao@nvidia.com>
WeiHaocheng
force-pushed
the
docao/support_topk_logprobs_torch_backend
branch
from
e3d28ab to
e01e84d
Compare
|
/bot run |
|
PR_Github #18924 [ run ] triggered by Bot |
dcaox
requested review from
Shixiaowei02,
brb-nv,
dongjiyingdjy,
galagam,
mikeiovine,
nv-guomingz and
nv-yilinf
dcaox
force-pushed
the
docao/support_topk_logprobs_torch_backend
branch
from
b872425 to
95d23f4
Compare
…gprobs_torch_backend Signed-off-by: Dong Cao <docao@nvidia.com>
|
/bot run --disable-fail-fast |
1 similar comment
|
/bot run --disable-fail-fast |
|
PR_Github #19573 [ run ] triggered by Bot |
|
/bot run --disable-fail-fast |
|
PR_Github #19634 [ run ] triggered by Bot |
|
PR_Github #19573 [ run ] completed with state |
|
PR_Github #19634 [ run ] completed with state |
…gprobs_torch_backend Signed-off-by: Dong Cao <docao@nvidia.com>
…gprobs_torch_backend Signed-off-by: Dong Cao <docao@nvidia.com>
|
/bot reuse-pipeline |
|
PR_Github #19753 [ reuse-pipeline ] triggered by Bot |
|
PR_Github #19753 [ reuse-pipeline ] completed with state |
…)" This reverts commit 2f8dc6f. Signed-off-by: Iman Tabrizian <10105175+tabrizian@users.noreply.github.com>
6 participants
Summary by CodeRabbit
New Features
Tests
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 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.