Move Fuse RMSNorm to new Inf Optimizer

[h-guo18]

Summary by CodeRabbit

• New Features

  • Introduced a modular, staged graph transformation and export pipeline for PyTorch models, including dynamic YAML configuration, deep merging, and advanced CLI argument handling.
  • Added backend-specific RMSNorm, quantized MoE (FP8/NVFP4), and Torch attention custom operators, with corresponding graph pattern fusion transforms.
  • Integrated a flexible patch system for export compatibility (e.g., SDPA, ModuleList, linear ops, meta device, transformers).
  • Added a comprehensive inference optimizer for efficient model deployment.

• Enhancements

  • Improved configuration validation, error checking, and support for nested YAML files.
  • Expanded attention and MoE quantization support, including new test coverage.
  • Optimized CUDA graph capture with memory pool reuse and improved logging.

• Bug Fixes

  • Fixed parameter deduplication, device info cleanup, and input constraint handling in exported graphs.
  • Corrected sharding, caching, and input preparation logic for distributed and single-GPU scenarios.

• Documentation

  • Updated and expanded documentation for configuration, advanced usage, and expert options.

• Tests

  • Added and refactored extensive unit and integration tests for new transforms, quantization, sharding, custom ops, and export compatibility.

• Chores

  • Refactored and reorganized codebase for modularity, maintainability, and extensibility, including deprecating legacy transformation modules.

Description

Issue #4403 . This PR moves fuse_rmsnorm only.

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.

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 the stage-list parameter 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.md
and the scripts/test_to_stage_mapping.py helper.

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.

[coderabbitai]

Caution

Review failed

The pull request is closed.

Walkthrough

This update introduces a major modularization and refactor of the AutoDeploy graph transformation, export, and configuration system. Key changes include new modular export and transform frameworks, dynamic YAML-based config merging, new backend-specific custom ops, quantization and sharding enhancements, and extensive updates to test infrastructure. Numerous new modules and classes were added, legacy transformation code was deprecated or replaced, and documentation was expanded.

Changes

File(s) / Path(s)	Change Summary
examples/auto_deploy/.vscode/launch.json, README.md, build_and_run_ad.py	Refactored experiment config for dynamic YAML merging, enhanced CLI arg parsing, clarified docs, and improved prompt/model kwarg handling.
tensorrt_llm/_torch/auto_deploy/export/, transform/, utils/_config.py	Introduced modular export and transform frameworks with patch/transform registries, deep YAML config merging, and new patch/transform libraries.
tensorrt_llm/_torch/auto_deploy/llm_args.py, models/hf.py, shim/ad_executor.py	Refactored LLM argument/config structure for stricter validation, dynamic merging, and support for new fields (e.g., max_beam_width).
tensorrt_llm/_torch/auto_deploy/custom_ops/, custom_ops/torch_backend_attention.py, custom_ops/rms_norm.py, custom_ops/torch_moe.py	Added/extended custom ops for Torch/triton/flashinfer backends, including new RMSNorm and quantized MoE implementations, and enhanced attention ops with sinks and sliding window support.
tensorrt_llm/_torch/auto_deploy/transform/library/, transformations/library/	Added new graph transforms for model building, export, RMSNorm fusion, quantized MoE, input constraint cleanup, and more.
tensorrt_llm/_torch/auto_deploy/transformations/ (legacy)	Deprecated/replaced legacy transformation and export utilities; removed or refactored direct graph returns to in-place mutation.
tensorrt_llm/_torch/auto_deploy/models/patches/	Modularized model-specific export patches; encapsulated monkey-patches in patch classes for registry-based management.
tensorrt_llm/_torch/auto_deploy/utils/quantization_utils.py, utils/node_utils.py	Added quantization skip/extract helpers, improved pattern matching and node filtering utilities.
tensorrt_llm/_torch/auto_deploy/transformations/library/sharding.py	Refactored sharding logic: introduced typed config objects, deferred application, and modularized detection/execution of TP/BMM/EP sharding.
tensorrt_llm/_torch/auto_deploy/models/__init__.py, models/factory.py	Reduced wildcard imports, minor signature simplifications.
tensorrt_llm/_torch/auto_deploy/compile/backends/torch_cudagraph.py	CUDA graph batch sizes now sorted descending; introduced memory pool reuse and warm-up logging.
requirements.txt, setup.py	Added omegaconf and enabled YAML support for pydantic-settings; included YAML files in package data.
tensorrt_llm/bench/benchmark/throughput.py	Adjusted backend-specific argument handling for AutoDeployLLM instantiation.
tests/unittest/_torch/auto_deploy/	Extensive test refactor: new reference modules, quantized MoE/attention tests, modular optimizer usage, pattern detection tests, and in-place transform handling.
Miscellaneous (models/patches/*.py, transformations/*.py, etc.)	Minor comment, import, and docstring updates; migration to new modular interfaces.

Sequence Diagram(s)

sequenceDiagram
    participant User
    participant CLI/Script
    participant ConfigLoader
    participant ExportRegistry
    participant TransformRegistry
    participant ModelFactory
    participant GraphModule
    participant CustomOps

    User->>CLI/Script: Launch with CLI args/YAML
    CLI/Script->>ConfigLoader: Parse CLI args, merge YAML configs
    ConfigLoader->>ConfigLoader: Deep merge, validate config
    CLI/Script->>ModelFactory: Create model factory from config
    CLI/Script->>ExportRegistry: Apply export patches (as context managers)
    ExportRegistry->>ModelFactory: Build model (possibly on meta device)
    ModelFactory->>GraphModule: Export model to FX graph
    ExportRegistry->>GraphModule: Deduplicate params, clean up devices
    CLI/Script->>TransformRegistry: Apply graph transforms in stage order
    TransformRegistry->>GraphModule: Apply transforms (e.g., fuse RMSNorm, quantize MoE)
    GraphModule->>CustomOps: Replace patterns with backend-specific ops
    CLI/Script->>GraphModule: Finalize, run inference/benchmark

Loading

Estimated code review effort

🎯 5 (Critical) | ⏱️ ~90+ minutes

This is a critical, high-complexity refactor involving new frameworks, deep config changes, new custom ops, sharding/quantization logic, and extensive test updates across many files.

Possibly related PRs

• [AutoDeploy] merge feat/ad-2025-07-07 #6196: Shares identical changes to launch config, README, experiment config refactor, dynamic YAML merging, CLI argument parsing, and export submodule import; these changes are directly related at the code level.

Suggested labels

Community want to contribute

Suggested reviewers

• shaharmor98
• nv-guomingz
• litaotju

Poem

A rabbit hops through fields of code,
With YAML, ops, and graphs bestowed.
It patches, fuses, quantizes, too—
Modular magic, configs anew!
From sharding fields to custom norms,
This bunny leaps through transform storms.
Review this garden, see it bloom—
🐇✨ Modular AutoDeploy in full costume!

Note

⚡️ Unit Test Generation is now available in beta!

Learn more here, or try it out under "Finishing Touches" below.

✨ Finishing Touches

• 📝 Generate Docstrings

🧪 Generate unit tests

• Create PR with unit tests
• Post copyable unit tests in a comment

---

🪧 Tips

Chat

There are 3 ways to chat with CodeRabbit:

• Review comments: Directly reply to a review comment made by CodeRabbit. Example:
  • I pushed a fix in commit <commit_id>, please review it.
  • Explain this complex logic.
  • Open a follow-up GitHub issue for this discussion.
• Files and specific lines of code (under the "Files changed" tab): Tag @coderabbitai in a new review comment at the desired location with your query. Examples:
  • @coderabbitai explain this code block.
  • @coderabbitai modularize this function.
• PR comments: Tag @coderabbitai in a new PR comment to ask questions about the PR branch. For the best results, please provide a very specific query, as very limited context is provided in this mode. Examples:
  • @coderabbitai gather interesting stats about this repository and render them as a table. Additionally, render a pie chart showing the language distribution in the codebase.
  • @coderabbitai read src/utils.ts and explain its main purpose.
  • @coderabbitai read the files in the src/scheduler package and generate a class diagram using mermaid and a README in the markdown format.
  • @coderabbitai help me debug CodeRabbit configuration file.

Support

Need help? Create a ticket on our support page for assistance with any issues or questions.

Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments.

CodeRabbit Commands (Invoked using PR comments)

• @coderabbitai pause to pause the reviews on a PR.
• @coderabbitai resume to resume the paused reviews.
• @coderabbitai review to trigger an incremental review. This is useful when automatic reviews are disabled for the repository.
• @coderabbitai full review to do a full review from scratch and review all the files again.
• @coderabbitai summary to regenerate the summary of the PR.
• @coderabbitai generate docstrings to generate docstrings for this PR.
• @coderabbitai generate sequence diagram to generate a sequence diagram of the changes in this PR.
• @coderabbitai generate unit tests to generate unit tests for this PR.
• @coderabbitai resolve resolve all the CodeRabbit review comments.
• @coderabbitai configuration to show the current CodeRabbit configuration for the repository.
• @coderabbitai help to get help.

Other keywords and placeholders

• Add @coderabbitai ignore anywhere in the PR description to prevent this PR from being reviewed.
• Add @coderabbitai summary to generate the high-level summary at a specific location in the PR description.
• Add @coderabbitai anywhere in the PR title to generate the title automatically.

Documentation and Community

• Visit our Documentation for detailed information on how to use CodeRabbit.
• Join our Discord Community to get help, request features, and share feedback.
• Follow us on X/Twitter for updates and announcements.