[AutoDeploy] HF factory improvements

[lucaslie]

This PR introduces several improvements to our HF model factory:

• Change name to AutoModelForCausalLM to reflect there are separate factories for different auto model types
• Better loading reusing HF utilities
• Ability to correctly customize tokenizer from native HF if desired

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 [--disable-fail-fast --skip-test --stage-list "A10-1, xxx" --gpu-type "A30, H100_PCIe" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-[Post-Merge]-1, xxx"]

Launch build/test pipelines. All previously running jobs will be killed.

--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 "A10-1, xxx" (OPTIONAL) : Only run the specified test stages. Examples: "A10-1, 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.

--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. Will also run 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-[Post-Merge]-1, xxx" (OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-[Post-Merge]-1, xxx".

kill

kill

Kill all running builds associated with pull request.

skip

skip --comment COMMENT

Skip 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-pipeline

Reuse 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.

[Copilot]

Pull Request Overview

This PR improves the HF model factory by renaming the model factory string, refining weight loading, and enhancing tokenizer customization. Key changes include updating method signatures to use a device parameter instead of keyword arguments, renaming the factory to "AutoModelForCausalLM", and patching configuration updates to recursively apply model_kwargs.

Reviewed Changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
tensorrt_llm/_torch/auto_deploy/transformations/transform.py	Updates the load_or_random_init call to pass device instead of kwargs.
tensorrt_llm/_torch/auto_deploy/shim/interface.py	Updates model_factory default from "hf" to "AutoModelForCausalLM".
tensorrt_llm/_torch/auto_deploy/shim/demollm.py	Passes None instead of model to create_input_processor; behavior should be verified.
tensorrt_llm/_torch/auto_deploy/models/hf.py	Adds a recursive config update and adjusts checkpoint loading to use the device parameter.
tensorrt_llm/_torch/auto_deploy/models/factory.py	Modifies load_or_random_init and _load_random_init signatures to require a device.
examples/auto_deploy/simple_config.py	Switches model_factory type to a Literal for clarity.
examples/auto_deploy/README.md	Reflects the updated model factory naming in the supported models table.
examples/auto_deploy/.vscode/settings.json & launch.json	Updates testing configuration to align with the new factory naming and environment paths.

Comments suppressed due to low confidence (2)

tensorrt_llm/_torch/auto_deploy/shim/demollm.py:375

• Passing None in place of a model instance to create_input_processor may lead to unintended behavior if the processor expects a valid model. Please verify that this change is intentional and that create_input_processor can handle a None value.

self.input_processor = create_input_processor(None, self.tokenizer)

tensorrt_llm/_torch/auto_deploy/models/factory.py:87

• Ensure that the updated signature using the device parameter is propagated to all calling contexts and that the corresponding documentation is updated for clarity.

def load_or_random_init(self, model: nn.Module, device: DeviceLikeType):

[lucaslie]

/bot run

[lucaslie]

/bot run

[tensorrt-cicd]

PR_Github #5413 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #5415 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #5413 [ run ] completed with state ABORTED

[tensorrt-cicd]

PR_Github #5415 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #3952 completed with status: 'FAILURE'

[lucaslie]

/bot run --disable-fail-fast --extra-stage "DGX_H100-4_GPUs-PyTorch-[Post-Merge]"

[tensorrt-cicd]

PR_Github #5522 [ run ] triggered by Bot

[lucaslie]

/bot run

[tensorrt-cicd]

PR_Github #5535 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #5522 [ run ] completed with state ABORTED

[lucaslie]

/bot run --disable-fail-fast --extra-stage "DGX_H100-4_GPUs-PyTorch-[Post-Merge]"

[tensorrt-cicd]

PR_Github #5542 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #5535 [ run ] completed with state ABORTED
/LLM/main/L0_MergeRequest_PR pipeline #4036 completed with status: 'FAILURE'

[tensorrt-cicd]

PR_Github #5542 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #4043 completed with status: 'FAILURE'

[lucaslie]

/bot run

[tensorrt-cicd]

PR_Github #5760 [ run ] triggered by Bot

[tensorrt-cicd]

PR_Github #5760 [ run ] completed with state SUCCESS
/LLM/main/L0_MergeRequest_PR pipeline #4214 completed with status: 'SUCCESS'