[https://nvbugs/5536131][fix] Fix illegal access issue when scale is not provided in Llama3/4. #7960
Conversation
|
/bot run --disable-fail-fast |
📝 WalkthroughWalkthroughAdds guarded checks around FP8/NVFP4 scaling and fusion paths in Llama model forward passes to avoid missing input_scale attributes; mirrors logic for post-MLP paths. Introduces a new FP8 TP2/PP2 integration test and registers it in the DGX B200 L0 test list. One typo (elf vs self) added. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
participant M as LlamaModel
participant A as next_attn
participant Q as qkv_proj
participant F as FusionOp
Note over M: Forward pass (attention/MLP post-processing)
M->>A: Check existence of next_attn
alt next_attn exists
M->>Q: Check quant mode in {FP8, NVFP4} and hasattr(Q, 'input_scale')
alt scale available
M->>Q: Read input_scale
M->>F: Use scaled post-processing path
else no scale
M->>F: Use RESIDUAL_RMS_NORM path
end
else no next_attn
M->>F: Use RESIDUAL_RMS_NORM path
end
sequenceDiagram
autonumber
participant ML as LlamaModel (min_latency)
participant AR as AllReduce
participant A as next_attn
participant Q as qkv_proj
Note over ML: Post-attention all-reduce (FP8/NVFP4)
ML->>AR: AllReduce activations
ML->>A: Verify next_attn present
ML->>Q: Check mode {FP8, NVFP4} and hasattr(Q, 'input_scale')
alt input_scale present
ML->>Q: Apply scaling using input_scale
else missing input_scale
ML->>ML: Skip scaling branch
end
Note over ML: One condition contains a typo (`elf` vs `self`)
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Pre-merge checks and finishing touches❌ Failed checks (2 warnings)
✅ Passed checks (1 passed)
✨ 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 |
![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
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
tensorrt_llm/_torch/models/modeling_llama_min_latency.py (1)
821-834: Typo bug:elfis undefined; useselfThis will raise NameError and break the FP8 post-allreduce path.
Apply this diff:
- if use_fp8_allreduce and self.next_attn is not None \ - and hasattr(elf.next_attn.qkv_proj, 'input_scale'): + if use_fp8_allreduce and self.next_attn is not None \ + and hasattr(self.next_attn.qkv_proj, 'input_scale'):
🧹 Nitpick comments (2)
tests/integration/defs/accuracy/test_llm_api_pytorch.py (2)
646-669: Align skip marker with existing FP8 test (run on Hopper+)The TP4 FP8 test for this model uses skip_pre_hopper, but this new TP2PP2 FP8 test uses skip_pre_blackwell. Unless there’s a known Hopper limitation for TP2PP2 FP8 here, align with the TP4 test to keep coverage consistent across Hopper and Blackwell.
Apply this diff:
- @pytest.mark.skip_less_device(4) - @skip_pre_blackwell + @pytest.mark.skip_less_device(4) + @skip_pre_hopper def test_fp8_tp2pp2(self):
646-669: Optional: match max_seq_len to TP4 FP8 test for consistencyThe TP4 FP8 test sets max_seq_len=8192; mirroring it here helps keep resource profiles comparable across configs.
Example change (optional):
with LLM(model_path, tensor_parallel_size=2, pipeline_parallel_size=2, + max_seq_len=8192, max_batch_size=32, kv_cache_config=kv_cache_config) as llm:
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (4)
tensorrt_llm/_torch/models/modeling_llama.py(2 hunks)tensorrt_llm/_torch/models/modeling_llama_min_latency.py(2 hunks)tests/integration/defs/accuracy/test_llm_api_pytorch.py(1 hunks)tests/integration/test_lists/test-db/l0_dgx_b200.yml(1 hunks)
🧰 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:
tests/integration/defs/accuracy/test_llm_api_pytorch.pytensorrt_llm/_torch/models/modeling_llama_min_latency.pytensorrt_llm/_torch/models/modeling_llama.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:
tests/integration/defs/accuracy/test_llm_api_pytorch.pytensorrt_llm/_torch/models/modeling_llama_min_latency.pytensorrt_llm/_torch/models/modeling_llama.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:
tests/integration/defs/accuracy/test_llm_api_pytorch.pytensorrt_llm/_torch/models/modeling_llama_min_latency.pytensorrt_llm/_torch/models/modeling_llama.py
🧠 Learnings (4)
📚 Learning: 2025-07-28T17:06:08.621Z
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/defs/accuracy/test_llm_api_pytorch.pytests/integration/test_lists/test-db/l0_dgx_b200.yml
📚 Learning: 2025-08-14T06:36:40.701Z
Learnt from: timlee0212
PR: NVIDIA/TensorRT-LLM#6886
File: tensorrt_llm/_torch/models/modeling_deepseekv3.py:0-0
Timestamp: 2025-08-14T06:36:40.701Z
Learning: In DeepSeek V3 model (tensorrt_llm/_torch/models/modeling_deepseekv3.py), the disagreement between AllReduce.__init__ guard and _compute_mlp_tp_size logic for MNNVL usage is expected by design. The AllReduce component and MLP TP-size computation intentionally use different criteria for MNNVL availability decisions.
Applied to files:
tensorrt_llm/_torch/models/modeling_llama_min_latency.py
📚 Learning: 2025-08-26T09:49:04.956Z
Learnt from: pengbowang-nv
PR: NVIDIA/TensorRT-LLM#7192
File: tests/integration/test_lists/test-db/l0_dgx_b200.yml:56-72
Timestamp: 2025-08-26T09:49:04.956Z
Learning: In TensorRT-LLM test configuration files, the test scheduling system handles wildcard matching with special rules that prevent duplicate test execution even when the same tests appear in multiple yaml files with overlapping GPU wildcards (e.g., "*b200*" and "*gb200*").
Applied to files:
tests/integration/test_lists/test-db/l0_dgx_b200.yml
📚 Learning: 2025-09-09T09:40:45.658Z
Learnt from: fredricz-20070104
PR: NVIDIA/TensorRT-LLM#7645
File: tests/integration/test_lists/qa/llm_function_core.txt:648-648
Timestamp: 2025-09-09T09:40:45.658Z
Learning: In TensorRT-LLM test lists, it's common and intentional for the same test to appear in multiple test list files when they serve different purposes (e.g., llm_function_core.txt for comprehensive core functionality testing and llm_function_core_sanity.txt for quick sanity checks). This duplication allows tests to be run in different testing contexts.
Applied to files:
tests/integration/test_lists/test-db/l0_dgx_b200.yml
🧬 Code graph analysis (2)
tests/integration/defs/accuracy/test_llm_api_pytorch.py (5)
tests/integration/defs/conftest.py (1)
llm_models_root(77-91)tensorrt_llm/llmapi/llm_args.py (1)
KvCacheConfig(976-1110)tensorrt_llm/quantization/mode.py (1)
QuantAlgo(23-47)tensorrt_llm/sampling_params.py (1)
SamplingParams(126-512)tests/integration/defs/accuracy/accuracy_core.py (5)
MMLU(276-290)evaluate(147-206)evaluate(712-722)GSM8K(293-308)GPQADiamond(311-324)
tensorrt_llm/_torch/models/modeling_llama.py (2)
tensorrt_llm/functional.py (1)
AllReduceFusionOp(3888-3897)cpp/tensorrt_llm/kernels/customAllReduceKernels.h (1)
AllReduceFusionOp(69-171)
🪛 Ruff (0.13.1)
tensorrt_llm/_torch/models/modeling_llama_min_latency.py
822-822: Undefined name elf
(F821)
⏰ 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 (4)
tests/integration/test_lists/test-db/l0_dgx_b200.yml (1)
54-54: LGTM: new TP2PP2 FP8 test correctly registered in 4-GPU B200 pre-merge blockThis aligns with the added test and target hardware.
tensorrt_llm/_torch/models/modeling_llama.py (2)
564-571: Good guard to avoid missing input_scale and reset fusion opThis prevents illegal access when scale isn’t present and safely falls back to RESIDUAL_RMS_NORM.
780-787: Mirror guard looks correct for LlamaDecoderLayerConsistent hasattr check and fallback op avoid attribute errors and maintain correctness.
tensorrt_llm/_torch/models/modeling_llama_min_latency.py (1)
821-834: Search for remainingelf.typosThe previous rg returned no output — cannot confirm absence of other occurrences. Re-run with a literal search to be safe: rg -n --type=py -C1 -F 'elf.'
|
PR_Github #19799 [ run ] triggered by Bot |
|
PR_Github #19799 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #19856 [ run ] triggered by Bot |
|
PR_Github #19856 [ run ] completed with state |
|
/bot run |
|
/bot run --disable-fail-fast |
|
PR_Github #20028 [ run ] triggered by Bot |
|
PR_Github #20028 [ run ] completed with state |
|
PR_Github #20210 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #20237 [ run ] triggered by Bot |
|
PR_Github #20237 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #20247 [ run ] triggered by Bot |
|
PR_Github #20247 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #20269 [ run ] triggered by Bot |
|
PR_Github #20269 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #20693 [ run ] triggered by Bot |
|
PR_Github #20693 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #20709 [ run ] triggered by Bot |
|
PR_Github #20709 [ run ] completed with state |
|
/bot run |
|
PR_Github #20720 [ run ] triggered by Bot |
|
PR_Github #20720 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
PR_Github #20748 [ run ] triggered by Bot |
|
PR_Github #20748 [ run ] completed with state |
…not provided in Llama3/4. (NVIDIA#7960) Signed-off-by: Yukun He <[email protected]>
…not provided in Llama3/4. (NVIDIA#7960) Signed-off-by: Yukun He <[email protected]> Signed-off-by: Mike Iovine <[email protected]>
…not provided in Llama3/4. (NVIDIA#7960) Signed-off-by: Yukun He <[email protected]>
…not provided in Llama3/4. (NVIDIA#7960) Signed-off-by: Yukun He <[email protected]>
…not provided in Llama3/4. (NVIDIA#7960) Signed-off-by: Yukun He <[email protected]> Signed-off-by: Mike Iovine <[email protected]>
When the scale is not provided, the fusion pattern needs to be reset to RESIDUAL_RMS_NORM.
Summary by CodeRabbit
Bug Fixes
Tests
Chores
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.