[TRTLLM-6854][feat] Enable guided decoding with disagg serving #6704
Conversation
📝 WalkthroughWalkthroughInternal methods in the Changes
Sequence Diagram(s)sequenceDiagram
participant TestCase as Test (test_guided_decoding)
participant LLM as DuckLLM (with tokenizer)
participant Server as Context/Generation Server
participant GuidedDecoder
TestCase->>LLM: Prepare request with guided decoding params
LLM->>Server: Send request (includes response_format for guided decoding)
Server->>GuidedDecoder: Process request
GuidedDecoder->>GuidedDecoder: _require_matcher_init / _require_matcher_advance
GuidedDecoder->>GuidedDecoder: Initialize or advance matcher as needed
GuidedDecoder-->>Server: Return decoded output
Server-->>LLM: Response
LLM-->>TestCase: Output for evaluation
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Suggested labels
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 (7)
🚧 Files skipped from review as they are similar to previous changes (5)
🧰 Additional context used📓 Path-based instructions (2)**/*.py📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
Files:
**/*.{cpp,h,hpp,cc,cxx,cu,py}📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
Files:
🧠 Learnings (3)📚 Learning: in tensorrt-llm testing, it's common to have both cli flow tests (test_cli_flow.py) and pytorch api ...Applied to files:
📚 Learning: in tensorrt-llm, test files (files under tests/ directories) do not require nvidia copyright headers...Applied to files:
📚 Learning: in tensorrt-llm's multimodal processing pipeline, shared tensor recovery using `from_shared_tensor()...Applied to files:
⏰ 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)
🔇 Additional comments (10)
✨ 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. 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: 1
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
tensorrt_llm/_torch/pyexecutor/guided_decoder.py(3 hunks)tensorrt_llm/_torch/pyexecutor/llm_request.py(1 hunks)tests/integration/defs/accuracy/test_disaggregated_serving.py(6 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
**/*.py: Python code should conform to Python 3.8+.
Indent Python code with 4 spaces. Do not use tabs.
Always maintain the namespace when importing in Python, even if only one class or function from a module is used.
Python filenames should use snake_case (e.g., some_file.py).
Python classes should use PascalCase (e.g., class SomeClass).
Python functions and methods should use snake_case (e.g., def my_awesome_function():).
Python local variables should use snake_case. Prefix k for variable names that start with a number (e.g., k_99th_percentile).
Python global variables should use upper snake_case and prefix G (e.g., G_MY_GLOBAL).
Python constants should use upper snake_case (e.g., MY_CONSTANT).
Avoid shadowing variables declared in an outer scope in Python.
Initialize all externally visible members of a Python class in the constructor.
For interfaces that may be used outside a Python file, prefer docstrings over comments.
Comments in Python should be reserved for code within a function, or interfaces that are local to a file.
Use Google style docstrings for Python classes and functions, which can be parsed by Sphinx.
Attributes and variables in Python can be documented inline; attribute docstrings will be rendered under the class docstring.
Avoid using reflection in Python when functionality can be easily achieved without it.
When using try-except blocks in Python, limit the except to the smallest set of errors possible.
When using try-except blocks to handle multiple possible variable types in Python, keep the body of the try as small as possible, using the else block to implement the logic.
Files:
tensorrt_llm/_torch/pyexecutor/llm_request.pytensorrt_llm/_torch/pyexecutor/guided_decoder.pytests/integration/defs/accuracy/test_disaggregated_serving.py
**/*.{cpp,h,hpp,cc,cxx,cu,py}
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
All TensorRT-LLM Open Source Software code should contain an NVIDIA copyright header that includes the current year. This includes .cpp, .h, .cu, .py, and any other source files which are compiled or interpreted.
Files:
tensorrt_llm/_torch/pyexecutor/llm_request.pytensorrt_llm/_torch/pyexecutor/guided_decoder.pytests/integration/defs/accuracy/test_disaggregated_serving.py
🧠 Learnings (4)
📚 Learning: in tensorrt-llm, test files (files under tests/ directories) do not require nvidia copyright headers...
Learnt from: galagam
PR: NVIDIA/TensorRT-LLM#6487
File: tests/unittest/_torch/auto_deploy/unit/singlegpu/test_ad_trtllm_bench.py:1-12
Timestamp: 2025-08-06T13:58:07.506Z
Learning: In TensorRT-LLM, test files (files under tests/ directories) do not require NVIDIA copyright headers, unlike production source code files. Test files typically start directly with imports, docstrings, or code.
Applied to files:
tensorrt_llm/_torch/pyexecutor/llm_request.pytests/integration/defs/accuracy/test_disaggregated_serving.py
📚 Learning: in tensorrt-llm's multimodal processing pipeline, shared tensor recovery using `from_shared_tensor()...
Learnt from: yechank-nvidia
PR: NVIDIA/TensorRT-LLM#6254
File: tensorrt_llm/_torch/pyexecutor/model_engine.py:1201-1204
Timestamp: 2025-07-22T09:22:14.726Z
Learning: In TensorRT-LLM's multimodal processing pipeline, shared tensor recovery using `from_shared_tensor()` is only needed during the context phase. Generation requests reuse the already-recovered tensor data and only need to call `strip_for_generation()` to remove unnecessary multimodal data while preserving the recovered tensors. This avoids redundant tensor recovery operations during generation.
Applied to files:
tensorrt_llm/_torch/pyexecutor/llm_request.pytests/integration/defs/accuracy/test_disaggregated_serving.py
📚 Learning: in tensorrt-llm testing, it's common to have both cli flow tests (test_cli_flow.py) and pytorch api ...
Learnt from: moraxu
PR: NVIDIA/TensorRT-LLM#6303
File: tests/integration/test_lists/qa/examples_test_list.txt:494-494
Timestamp: 2025-07-28T17:06:08.621Z
Learning: In TensorRT-LLM testing, it's common to have both CLI flow tests (test_cli_flow.py) and PyTorch API tests (test_llm_api_pytorch.py) for the same model. These serve different purposes: CLI flow tests validate the traditional command-line workflow, while PyTorch API tests validate the newer LLM API backend. Both are legitimate and should coexist.
Applied to files:
tensorrt_llm/_torch/pyexecutor/llm_request.pytests/integration/defs/accuracy/test_disaggregated_serving.py
📚 Learning: in tensorrt-llm, examples directory can have different dependency versions than the root requirement...
Learnt from: yibinl-nvidia
PR: NVIDIA/TensorRT-LLM#6506
File: examples/models/core/mixtral/requirements.txt:3-3
Timestamp: 2025-08-01T15:14:45.673Z
Learning: In TensorRT-LLM, examples directory can have different dependency versions than the root requirements.txt file. Version conflicts between root and examples dependencies are acceptable because examples are designed to be standalone and self-contained.
Applied to files:
tensorrt_llm/_torch/pyexecutor/llm_request.py
🔇 Additional comments (7)
tensorrt_llm/_torch/pyexecutor/llm_request.py (1)
332-335: Documentation improvements look good!The added comments clearly explain the purpose of
py_seq_slotandpy_target_seq_slotattributes, which will help developers understand the request sequence slot management and draft request relationships.tests/integration/defs/accuracy/test_disaggregated_serving.py (3)
48-48: LGTM! Tokenizer field addition is appropriate.Adding the tokenizer field to the DuckLLM namedtuple is necessary for guided decoding support and aligns with the test requirements.
215-218: LGTM! Tokenizer loading is properly implemented.The tokenizer is correctly loaded and included in the DuckLLM instance, making it available for tests that require tokenization capabilities.
419-452: LGTM! Well-structured guided decoding test.The test properly configures both context and generation servers with the xgrammar guided decoding backend and uses the appropriate JsonModeEval harness for validation. The memory requirement decorator ensures the test runs only when sufficient resources are available.
tensorrt_llm/_torch/pyexecutor/guided_decoder.py (3)
74-84: LGTM! Method rename and disaggregated serving logic are well-implemented.The rename from
_is_matcher_initto_require_matcher_initbetter reflects the method's intent. The added logic correctly handles the disaggregated generation case where the matcher needs initialization at the first generation step whencontext_phase_paramsis present.
86-92: LGTM! Cleaner logic for matcher advancement check.The rename to
_require_matcher_advanceis consistent with the naming pattern, and the simplified logic clearly identifies when matcher advancement is needed.
109-119: LGTM! Improved control flow and readability.The refactored control flow is cleaner and more maintainable:
- Computing both flags upfront improves clarity
- Early continue reduces nesting
- Separate handling of initialization and advancement is more logical
- The code properly handles cases where both operations may be needed
syuoni
requested review from
QiJune,
Shixiaowei02,
Tabrizian,
sunnyqgg and
yweng0828
|
/bot run --disable-fail-fast |
There was a problem hiding this comment.
Actionable comments posted: 1
♻️ Duplicate comments (1)
tests/integration/defs/accuracy/test_disaggregated_serving.py (1)
167-191: Add error handling for JSON parsing.The
json.loads(schema)call could raise aJSONDecodeErrorif the schema is malformed. Consider adding error handling as previously suggested.
🧹 Nitpick comments (1)
tests/integration/test_lists/qa/llm_function_full.txt (1)
451-454: Consider adding 4-GPU variants for the new guided-decoding combos
We already exercisetest_guided_decoding_4gpus[...]for the plain guided-decoding path (Lines 449-450). For parity and to avoid accidental blind-spots, please add 4-GPU counterparts for the newly introduced*_with_eagle3and*_with_ngramcases.+accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_eagle3_4gpus[xgrammar] +accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_eagle3_4gpus[llguidance] +accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_ngram_4gpus[xgrammar] +accuracy/test_llm_api_pytorch.py::TestLlama3_1_8BInstruct::test_guided_decoding_with_ngram_4gpus[llguidance]
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (6)
docs/source/torch/features/feature_combination_matrix.md(1 hunks)tensorrt_llm/_torch/pyexecutor/guided_decoder.py(4 hunks)tensorrt_llm/_torch/pyexecutor/py_executor.py(3 hunks)tests/integration/defs/accuracy/test_disaggregated_serving.py(6 hunks)tests/integration/test_lists/qa/llm_function_full.txt(2 hunks)tests/integration/test_lists/test-db/l0_dgx_h100.yml(1 hunks)
✅ Files skipped from review due to trivial changes (2)
- tests/integration/test_lists/test-db/l0_dgx_h100.yml
- docs/source/torch/features/feature_combination_matrix.md
🚧 Files skipped from review as they are similar to previous changes (1)
- tensorrt_llm/_torch/pyexecutor/guided_decoder.py
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
**/*.py: Python code should conform to Python 3.8+.
Indent Python code with 4 spaces. Do not use tabs.
Always maintain the namespace when importing in Python, even if only one class or function from a module is used.
Python filenames should use snake_case (e.g., some_file.py).
Python classes should use PascalCase (e.g., class SomeClass).
Python functions and methods should use snake_case (e.g., def my_awesome_function():).
Python local variables should use snake_case. Prefix k for variable names that start with a number (e.g., k_99th_percentile).
Python global variables should use upper snake_case and prefix G (e.g., G_MY_GLOBAL).
Python constants should use upper snake_case (e.g., MY_CONSTANT).
Avoid shadowing variables declared in an outer scope in Python.
Initialize all externally visible members of a Python class in the constructor.
For interfaces that may be used outside a Python file, prefer docstrings over comments.
Comments in Python should be reserved for code within a function, or interfaces that are local to a file.
Use Google style docstrings for Python classes and functions, which can be parsed by Sphinx.
Attributes and variables in Python can be documented inline; attribute docstrings will be rendered under the class docstring.
Avoid using reflection in Python when functionality can be easily achieved without it.
When using try-except blocks in Python, limit the except to the smallest set of errors possible.
When using try-except blocks to handle multiple possible variable types in Python, keep the body of the try as small as possible, using the else block to implement the logic.
Files:
tensorrt_llm/_torch/pyexecutor/py_executor.pytests/integration/defs/accuracy/test_disaggregated_serving.py
**/*.{cpp,h,hpp,cc,cxx,cu,py}
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
All TensorRT-LLM Open Source Software code should contain an NVIDIA copyright header that includes the current year. This includes .cpp, .h, .cu, .py, and any other source files which are compiled or interpreted.
Files:
tensorrt_llm/_torch/pyexecutor/py_executor.pytests/integration/defs/accuracy/test_disaggregated_serving.py
🧠 Learnings (4)
📚 Learning: in tensorrt-llm testing, it's common to have both cli flow tests (test_cli_flow.py) and pytorch api ...
Learnt from: moraxu
PR: NVIDIA/TensorRT-LLM#6303
File: tests/integration/test_lists/qa/examples_test_list.txt:494-494
Timestamp: 2025-07-28T17:06:08.621Z
Learning: In TensorRT-LLM testing, it's common to have both CLI flow tests (test_cli_flow.py) and PyTorch API tests (test_llm_api_pytorch.py) for the same model. These serve different purposes: CLI flow tests validate the traditional command-line workflow, while PyTorch API tests validate the newer LLM API backend. Both are legitimate and should coexist.
Applied to files:
tests/integration/test_lists/qa/llm_function_full.txttests/integration/defs/accuracy/test_disaggregated_serving.py
📚 Learning: in tensorrt-llm, test files (files under tests/ directories) do not require nvidia copyright headers...
Learnt from: galagam
PR: NVIDIA/TensorRT-LLM#6487
File: tests/unittest/_torch/auto_deploy/unit/singlegpu/test_ad_trtllm_bench.py:1-12
Timestamp: 2025-08-06T13:58:07.506Z
Learning: In TensorRT-LLM, test files (files under tests/ directories) do not require NVIDIA copyright headers, unlike production source code files. Test files typically start directly with imports, docstrings, or code.
Applied to files:
tests/integration/test_lists/qa/llm_function_full.txttests/integration/defs/accuracy/test_disaggregated_serving.py
📚 Learning: in tensorrt_llm/executor/worker.py, the lora adapter cache optimization logic that checks `is_adapte...
Learnt from: amitz-nv
PR: NVIDIA/TensorRT-LLM#5616
File: tensorrt_llm/executor/worker.py:375-384
Timestamp: 2025-07-17T09:01:27.402Z
Learning: In tensorrt_llm/executor/worker.py, the LoRA adapter cache optimization logic that checks `is_adapter_in_cpu_cache()` and conditionally passes None for weights/config has a known race condition issue that cannot be solved with simple error handling or verification checks. This is a known limitation that requires a more comprehensive solution.
Applied to files:
tensorrt_llm/_torch/pyexecutor/py_executor.py
📚 Learning: in tensorrt-llm's multimodal processing pipeline, shared tensor recovery using `from_shared_tensor()...
Learnt from: yechank-nvidia
PR: NVIDIA/TensorRT-LLM#6254
File: tensorrt_llm/_torch/pyexecutor/model_engine.py:1201-1204
Timestamp: 2025-07-22T09:22:14.726Z
Learning: In TensorRT-LLM's multimodal processing pipeline, shared tensor recovery using `from_shared_tensor()` is only needed during the context phase. Generation requests reuse the already-recovered tensor data and only need to call `strip_for_generation()` to remove unnecessary multimodal data while preserving the recovered tensors. This avoids redundant tensor recovery operations during generation.
Applied to files:
tensorrt_llm/_torch/pyexecutor/py_executor.pytests/integration/defs/accuracy/test_disaggregated_serving.py
🧬 Code Graph Analysis (1)
tensorrt_llm/_torch/pyexecutor/py_executor.py (1)
tensorrt_llm/_torch/pyexecutor/guided_decoder.py (1)
init_disagg_gen_requests(251-264)
⏰ 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 (7)
tensorrt_llm/_torch/pyexecutor/py_executor.py (3)
748-750: LGTM!The conditional initialization of guided decoder for disaggregated generation requests is correctly placed after resource preparation and before guided decoder execution.
937-940: LGTM!Consistent implementation with the pipeline parallel executor loop. The initialization is properly placed in the execution flow.
1065-1067: LGTM!The guided decoder initialization is consistently implemented across all executor loop variants, ensuring feature parity.
tests/integration/defs/accuracy/test_disaggregated_serving.py (4)
7-7: LGTM! Imports and interface updates are appropriate.The addition of
jsonmodule,load_hf_tokenizer, andJsonModeEvalimports, along with thetokenizerfield inDuckLLM, properly supports the new guided decoding functionality.Also applies to: 23-23, 28-29, 48-48
215-218: LGTM! Tokenizer properly loaded and integrated.The tokenizer is efficiently loaded once and correctly passed to the DuckLLM instance.
419-453: LGTM! Well-structured guided decoding test.The test properly configures both context and generation servers with guided decoding backends and validates the functionality using
JsonModeEval. The parametrization with both "xgrammar" and "llguidance" backends ensures comprehensive coverage.
455-506: LGTM! Comprehensive test for guided decoding with Eagle3.The test effectively validates the compatibility of guided decoding with Eagle3 speculative decoding. The configuration properly combines both features with appropriate parameters (max_draft_len=3, eagle3_one_model=False) and validates the functionality using
JsonModeEval.
|
PR_Github #14480 [ run ] triggered by Bot |
Signed-off-by: Enwei Zhu <[email protected]>
Signed-off-by: Enwei Zhu <[email protected]>
Signed-off-by: Enwei Zhu <[email protected]>
Signed-off-by: Enwei Zhu <[email protected]>
syuoni
force-pushed
the
guided-with-disagg
branch
from
07f60ab to
754865b
Compare
|
/bot run --disable-fail-fast |
|
PR_Github #14491 [ run ] triggered by Bot |
|
PR_Github #14480 [ run ] completed with state |
|
PR_Github #14491 [ run ] completed with state |
|
/bot run |
|
PR_Github #14519 [ run ] triggered by Bot |
|
PR_Github #14519 [ run ] completed with state |
…A#6704) Signed-off-by: Enwei Zhu <[email protected]>
4 participants
Summary by CodeRabbit
New Features
Documentation
Bug Fixes
Description
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.
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.