[TRTLLM-8579][feat] Support quantized model for nano-v2-vlm #8304
Conversation
* Support for FP8 model. Signed-off-by: Wanli Jiang <35160485+Wanli-Jiang@users.noreply.github.com>
Wanli-Jiang
force-pushed
the
user/williamj/support-nanov2vlm-quant
branch
from
9002587 to
7fc73f7
Compare
Wanli-Jiang
requested review from
brb-nv,
danielafrimi,
jaedeok-nvidia,
yilin-void and
yizhang-nv
|
/bot run |
📝 WalkthroughWalkthroughExpands ModelConfig mutability to include quant_config. Updates RADIOVisionModel to optionally disable quantization by deep-copying and overriding quant_config while preserving kv_cache_quant_algo. Adjusts weight loading key removals to be conditional. Simplifies NanoV2VLM image/video preprocessing by removing tensor-to-PIL conversions and directly invoking the processor. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
actor App as Caller
participant NVLM as NanoV2VLM
participant RVM as RADIOVisionModel
participant VT as VisionTransformer
participant Proc as Processor
App->>NVLM: init(...)
note over NVLM: Create RADIOVisionModel with disable_quantization=true
NVLM->>RVM: __init__(model_config, disable_quantization=true)
activate RVM
RVM->>RVM: model_config' = deepcopy(model_config)
alt quant_config present AND disable_quantization
RVM->>RVM: model_config'.quant_config = QuantConfig()<br/>(preserve kv_cache_quant_algo)
end
RVM->>VT: init(model_config')
deactivate RVM
sequenceDiagram
autonumber
actor User as Caller
participant NVLM as NanoV2VLM
participant Proc as Processor
participant RVM as RADIOVisionModel
User->>NVLM: generate(images/videos, text, ...)
note over NVLM: Directly pass inputs to processor<br/>No Tensor→PIL conversion
NVLM->>Proc: __call__(images or video, text, ...)
Proc-->>NVLM: processed batches (tensors & tokens)
NVLM->>RVM: encode_vision(processed inputs)
RVM-->>NVLM: visual features
NVLM-->>User: outputs
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 (beta)
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 |
📝 WalkthroughWalkthroughModelConfig now permits updating quant_config when frozen. RADIOVisionModel gains a disable_quantization flag, deep-copies and conditionally adjusts model_config, and passes the adjusted config downstream while hardening weight-loading. VLM pipelines bypass tensor-to-PIL conversions, sending images/videos directly to the processor and adjusting video patch/pixel accumulation. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
participant Caller
participant RADIOVisionModel
participant VisionTransformer
participant Base as RADIOVisionModelBase
Caller->>RADIOVisionModel: __init__(model_config, disable_quantization=True)
activate RADIOVisionModel
RADIOVisionModel->>RADIOVisionModel: deepcopy(model_config) -> self.model_config
alt disable_quantization == True
RADIOVisionModel->>RADIOVisionModel: Adjust self.model_config.quant_config (preserve kv_cache_quant_algo)
else disable_quantization == False
RADIOVisionModel->>RADIOVisionModel: Keep quant_config as-is
end
RADIOVisionModel->>VisionTransformer: init(self.model_config)
RADIOVisionModel->>Base: init(self.model_config)
RADIOVisionModel->>RADIOVisionModel: load_weights (safely ignore missing keys)
deactivate RADIOVisionModel
sequenceDiagram
autonumber
participant App as VLM Caller
participant VLM as NanoV2VLM
participant Proc as Processor
participant Vision as RADIOVisionModel
App->>VLM: generate(images/videos, prompts)
Note over VLM: No tensor→PIL conversion
alt Images
VLM->>Proc: process(images)
Proc-->>VLM: pixel_values, num_patches
else Videos
loop each video
VLM->>Proc: process(video)
Proc-->>VLM: pixel_values_i, num_patches_i
VLM->>VLM: accumulate pixel_values, num_patches
end
end
VLM->>Vision: forward(pixel_values, num_patches)
Vision-->>VLM: embeddings
VLM-->>App: outputs
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 (beta)
📜 Recent review detailsConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro 📒 Files selected for processing (3)
🧰 Additional context used📓 Path-based instructions (3)**/*.{h,hpp,hh,hxx,cpp,cxx,cc,cu,cuh,py}📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Files:
**/*.py📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Files:
**/*.{cpp,cxx,cc,h,hpp,hh,hxx,cu,cuh,py}📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Files:
🧬 Code graph analysis (2)tensorrt_llm/_torch/models/modeling_nanov2vlm.py (2)
tensorrt_llm/_torch/models/modeling_radio.py (2)
⏰ 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)
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 |
|
PR_Github #21171 [ run ] triggered by Bot |
|
PR_Github #21171 [ run ] completed with state |
3 participants
Features:
Summary by CodeRabbit
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.