[None][feat] Switch to internal version of MMProjector in Gemma3 #6572
Conversation
📝 WalkthroughWalkthroughThe changes introduce a new local implementation of the Changes
Sequence Diagram(s)sequenceDiagram
participant User
participant Gemma3VLM
participant Gemma3MultiModalProjector
User->>Gemma3VLM: Call load_weights(weights, weight_mapper)
Gemma3VLM->>Gemma3MultiModalProjector: load_weights(weights)
Gemma3MultiModalProjector-->>Gemma3VLM: Weights loaded
User->>Gemma3VLM: Call forward(...)
Gemma3VLM->>Gemma3MultiModalProjector: forward(vision_outputs)
Gemma3MultiModalProjector-->>Gemma3VLM: Projected vision features
Gemma3VLM-->>User: Output tensor
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~15–20 minutes 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 (1)
🚧 Files skipped from review as they are similar to previous changes (1)
⏰ 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)
✨ 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: 2
🔭 Outside diff range comments (1)
tensorrt_llm/_torch/models/modeling_gemma3vl.py (1)
265-277: Remove unused function_load_weights_into_hf_module.This function is no longer used after replacing the HuggingFace multimodal projector with the local implementation.
-def _load_weights_into_hf_module( - model: torch.nn.Module, - weights: dict, - prefix: str, - model_name: str, -) -> None: - filtered_weights = filter_weights(prefix, weights) - missing_keys, _ = model.load_state_dict(filtered_weights) - if len(missing_keys) > 0: - raise KeyError( - f"Missing the following keys for the {model_name} in the checkpoint: " - f"[{', '.join(missing_keys)}].") -
🧹 Nitpick comments (1)
tensorrt_llm/_torch/models/modeling_gemma3vl.py (1)
106-109: Add error handling for missing weights.The weight loading logic is correct, but consider adding error handling for missing keys to provide better debugging information:
def load_weights(self, weights): + required_keys = ["mm_input_projection_weight", "mm_soft_emb_norm.weight"] + missing_keys = [key for key in required_keys if key not in weights] + if missing_keys: + raise KeyError(f"Missing required weights: {missing_keys}") self.mm_input_projection.weight.data.copy_(weights["mm_input_projection_weight"].transpose(0, 1)) self.mm_soft_emb_norm.weight.data.copy_(weights["mm_soft_emb_norm.weight"] + 1.0)
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
tensorrt_llm/_torch/models/modeling_gemma3vl.py(5 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit Inference Engine (CODING_GUIDELINES.md)
**/*.py: The code developed for TensorRT-LLM 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 class in the constructor in Python.
For interfaces that may be used outside a file, prefer docstrings over comments in Python.
Comments in Python should be reserved for code within a function, or interfaces that are local to a file.
Use Google style docstrings for classes and functions in Python, which can be parsed by Sphinx.
Attributes and variables in Python can be documented inline; attribute docstrings will be rendered under the docstring for the class.
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/models/modeling_gemma3vl.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/models/modeling_gemma3vl.py
🧠 Learnings (2)
📓 Common learnings
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.
📚 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/models/modeling_gemma3vl.py
🔇 Additional comments (5)
tensorrt_llm/_torch/models/modeling_gemma3vl.py (5)
19-20: LGTM! Imports align with TRTLLM architecture.The addition of TRTLLM's
LinearandRMSNormmodules is appropriate for implementing the localGemma3MultiModalProjectorclass.
110-123: Forward pass implementation looks good!The reshaping, pooling, normalization, and projection sequence is correctly implemented. Good use of
contiguous()for memory layout optimization andtype_as()for maintaining dtype consistency.
158-161: Correct replacement with local projector implementation.The initialization properly uses the new local
Gemma3MultiModalProjectorclass while maintaining the same initialization pattern withno_init_weights().
196-198: Clean weight loading implementation.The simplified weight loading correctly uses the new
load_weightsmethod of the local projector.
233-233: Good simplification of image feature handling.The removal of redundant reshaping is appropriate since the projector already outputs the correctly shaped tensor. The
contiguous()call ensures proper memory layout for subsequent operations.
brb-nv
changed the title
brb-nv
changed the title
brb-nv
changed the title
brb-nv
force-pushed
the
user/brb/enable-tp-for-gemma3
branch
from
1a42afd to
e314dc8
Compare
|
/bot run |
|
PR_Github #14050 [ run ] triggered by Bot |
|
/bot run |
|
PR_Github #14076 [ run ] triggered by Bot |
|
PR_Github #14050 [ run ] completed with state |
|
PR_Github #14076 [ run ] completed with state |
brb-nv
force-pushed
the
user/brb/enable-tp-for-gemma3
branch
2 times, most recently
from
30880e9 to
7501f2b
Compare
|
/bot run |
|
PR_Github #14173 [ run ] triggered by Bot |
|
PR_Github #14173 [ run ] completed with state |
brb-nv
force-pushed
the
user/brb/enable-tp-for-gemma3
branch
from
7501f2b to
b3e74b1
Compare
Signed-off-by: Balaram Buddharaju <[email protected]>
brb-nv
force-pushed
the
user/brb/enable-tp-for-gemma3
branch
from
b3e74b1 to
38648e9
Compare
|
/bot run |
|
PR_Github #14180 [ run ] triggered by Bot |
|
/bot run |
|
PR_Github #14190 [ run ] triggered by Bot |
|
PR_Github #14180 [ run ] completed with state |
|
PR_Github #14190 [ run ] completed with state |
…DIA#6572) Signed-off-by: Balaram Buddharaju <[email protected]> Signed-off-by: Lanyu Liao <[email protected]>
…DIA#6572) Signed-off-by: Balaram Buddharaju <[email protected]>
3 participants
Description
This MR switches
Gemma3MultimodalProjectorto use TRTLLM components. This makes our implementation less susceptible to dependency changes.Test Coverage
I reran
ai2dandocrbenchmultimodal benchmarks and verified that the scores are the same before and after the change. Also, verified with e2e tests.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
Improvements