Check pre commit

.github/CODEOWNERS

@@ -59,8 +59,10 @@
 /tensorrt_llm/_torch/pyexecutor @NVIDIA/trt-llm-torch-runtime-devs
 ## TensorRT-LLM Pytorch backend - AutoDeploy flow
 /tensorrt_llm/_torch/auto_deploy @NVIDIA/trt-llm-torch-autodeploy-devs
-/examples/auto_deploy @NVIDIA/trt-llm-torch-autodeploy-devs @NVIDIA/trt-llm-doc-owners
-/tests/unittest/_torch/auto_deploy @NVIDIA/trt-llm-torch-autodeploy-devs
+/examples/auto_deploy @NVIDIA/trt-llm-torch-autodeploy-devs
+/docs/source/features/auto_deploy @NVIDIA/trt-llm-torch-autodeploy-devs @NVIDIA/trt-llm-doc-owners
+/tests/unittest/auto_deploy @NVIDIA/trt-llm-torch-autodeploy-devs
+/tests/integration/defs/accuracy/test_llm_api_autodeploy.py @NVIDIA/trt-llm-torch-autodeploy-devs @NVIDIA/trt-llm-qa-function
 
 ## TensorRT-LLM Pytorch - Speculative Decoding
 /tensorrt_llm/_torch/speculative @NVIDIA/trt-llm-torch-spec-decoding

.github/workflows/blossom-ci.yml

@@ -101,10 +101,12 @@ jobs:
         "dpitman-nvda",
         "DylanChen-NV",
         "ebarilanM",
+        "ekou24",
         "elvischenv",
         "EmmaQiaoCh",
         "eopXD",
         "esha-nvidia",
+        "etz-lmn",
         "evezhier",
         "faradawn",
         "farazkh80",
@@ -140,6 +142,7 @@ jobs:
         "JadoTu",
         "jaedeok-nvidia",
         "jdemouth-nvidia",
+        "janbernloehr",
         "JennyLiu-nv",
         "jershi425",
         "jgangani",
@@ -227,6 +230,7 @@ jobs:
         "nvzhihanj",
         "nvzhou",
         "nzmora-nvidia",
+        "o-stoner",
         "omera-nv",
         "pamelap-nvidia",
         "pcastonguay",
@@ -317,6 +321,7 @@ jobs:
         "xinhe-nv",
         "xmchen1987",
         "xrq-phys",
+        "xuantengh",
         "xuanzic",
         "xueweilnvidia",
         "xupinjie",
@@ -351,6 +356,7 @@ jobs:
         "zerollzeng",
         "zhanga5",
         "zhangcl",
+        "zhaoyangwang-nvidia",
         "ZhanruiSunCh",
         "zhengd-nv",
         "zhenhuaw-me",

.github/workflows/model-registry-check.yml

@@ -0,0 +1,40 @@
+# SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+name: Model Registry Check
+
+on:
+  pull_request:
+    types: [opened, edited, synchronize, reopened]
+    paths:
+      - examples/auto_deploy/model_registry/models.yaml
+
+jobs:
+  validate-model-registry:
+    name: Validate AutoDeploy Model Registry
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v6
+
+      - uses: actions/setup-python@v6
+        with:
+          python-version: "3.12"
+          cache: "pip"
+
+      - name: Install validator dependency
+        run: python3 -m pip install PyYAML
+
+      - name: Validate model registry
+        run: python3 scripts/check_model_registry.py

.github/workflows/precommit-check.yml

@@ -38,6 +38,21 @@ jobs:
           python-version: '3.12'
           cache: 'pip'
 
+      - name: Get changed files
+        id: changed-files
+        if: github.event_name == 'pull_request'
+        env:
+          GH_TOKEN: ${{ github.token }}
+        run: |
+          files=$(gh api "repos/${{ github.repository }}/pulls/${{ github.event.pull_request.number }}/files" \
+            --paginate --jq '.[].filename' | paste -sd ' ' -)
+          echo "files=$files" >> $GITHUB_OUTPUT
+
       - name: Run pre-commit checks
         run: |
-          python3 -u scripts/release_check.py
+          if [ "${{ github.event_name }}" = "pull_request" ]; then
+            echo "${{ steps.changed-files.outputs.files }}" | tr ' ' '\n' | sed '/^$/d' > changed_files.txt
+            python3 -u scripts/release_check.py --files-from changed_files.txt
+          else
+            python3 -u scripts/release_check.py
+          fi

.pre-commit-config.yaml

@@ -1368,7 +1368,7 @@ common-files: &common_files |
     )$
 
 # Global exclude pattern for vendored third-party code
-exclude: '^triton_kernels/'
+exclude: '(^triton_kernels/|trtllmGenKernels/fmha/cubin/kernelMetaInfo\.h$|cubin\.cpp$|cubin\.h$)'
 
 default_install_hook_types: [pre-commit, commit-msg]
 repos:
@@ -1390,14 +1390,10 @@ repos:
     rev: v6.0.0
     hooks:
     -   id: check-added-large-files
-        exclude: |
-            (?x)^(.*cubin.cpp | .*cubin.h)$
     -   id: check-merge-conflict
     -   id: check-symlinks
     -   id: detect-private-key
     -   id: end-of-file-fixer
-        exclude: |
-            (?x)^(.*cubin.cpp | .*cubin.h)$
     -   id: check-yaml
         args: [--allow-multiple-documents, --unsafe]
         exclude: ".*/gitlab/.*.yml"
@@ -1429,8 +1425,6 @@ repos:
     hooks:
     -   id: clang-format
         types_or: [c++, c, cuda]
-        exclude: |
-            (?x)^(.*cubin.cpp$ | .*_cubin.h)$
 -   repo: https://github.com/cheshirekow/cmake-format-precommit
     rev: v0.6.10
     hooks:
@@ -1442,7 +1436,7 @@ repos:
         additional_dependencies:
         - tomli
         # add ignore words list
-        args: ["-L", "Mor,ans,thirdparty,subtiles,PARD,pard", "--skip", "ATTRIBUTIONS-*.md,*.svg", "--skip", "security_scanning/*"]
+        args: ["-L", "Mor,ans,thirdparty,subtiles,PARD,pard,therefrom", "--skip", "ATTRIBUTIONS-*.md,*.svg", "--skip", "security_scanning/*", "--skip", "tensorrt_llm/_torch/visual_gen/jit_kernels/*"]
         exclude: 'scripts/attribution/data/cas/.*$'
 -   repo: https://github.com/astral-sh/ruff-pre-commit
     rev: v0.9.4
@@ -1459,6 +1453,14 @@ repos:
         files: ".*/auto_deploy/.*"
 -   repo: local
     hooks:
+    -   id: model-registry-check
+        name: Validate AutoDeploy model registry
+        entry: python scripts/check_model_registry.py
+        language: python
+        additional_dependencies:
+        - PyYAML
+        files: ^examples/auto_deploy/model_registry/models\.yaml$
+        pass_filenames: false
     -   id: test lists format
         name: Check for tabs and multiple spaces in test_lists txt files
         entry: ./scripts/format_test_list.py

3rdparty/CMakeLists.txt

@@ -30,6 +30,7 @@ foreach(DEP_IDX RANGE ${DEP_COUNT_MINUS_ONE})
   string(JSON DEP_SOURCE_SUBDIR          ERROR_VARIABLE _err GET "${DEP_OBJECT}" "source_subdir")
   string(JSON DEP_GIT_SUBMODULES_RECURSE ERROR_VARIABLE _err GET "${DEP_OBJECT}" "git_submodules_recurse")
   string(JSON DEP_USE_URL                ERROR_VARIABLE _err GET "${DEP_OBJECT}" "use_url")
+  string(JSON DEP_PATCH_FILE            ERROR_VARIABLE _err GET "${DEP_OBJECT}" "patch_file")
   # cmake-format: on
 
   # Build FetchContent_Declare arguments
@@ -53,6 +54,19 @@ foreach(DEP_IDX RANGE ${DEP_COUNT_MINUS_ONE})
     list(APPEND FETCH_ARGS SOURCE_SUBDIR "${DEP_SOURCE_SUBDIR}")
   endif()
 
+  if(DEP_PATCH_FILE AND NOT DEP_PATCH_FILE STREQUAL "")
+    list(
+      APPEND
+      FETCH_ARGS
+      PATCH_COMMAND
+      patch
+      -p1
+      --forward
+      --batch
+      -i
+      "${CMAKE_CURRENT_SOURCE_DIR}/${DEP_PATCH_FILE}")
+  endif()
+
   FetchContent_Declare(${FETCH_ARGS})
 
   # Special handling: Export deep_ep commit to global property

3rdparty/fetch_content.json

@@ -11,7 +11,7 @@
     {
       "name": "cutlass",
       "git_repository": "https://github.com/NVIDIA/cutlass",
-      "git_tag": "v4.3.0",
+      "git_tag": "v4.4.1",
       "git_shallow": true,
       "source_subdir": "dont-add-this-project-with-add-subdirectory"
     },
@@ -93,9 +93,10 @@
     {
       "name": "xgrammar",
       "git_repository": "https://github.com/mlc-ai/xgrammar",
-      "git_tag": "v0.1.25",
+      "git_tag": "v0.1.32",
       "git_shallow": true,
-      "source_subdir": "dont-add-this-project-with-add-subdirectory"
+      "source_subdir": "dont-add-this-project-with-add-subdirectory",
+      "patch_file": "patches/xgrammar_constexpr.patch"
     }
   ]
 }

3rdparty/patches/xgrammar_constexpr.patch

@@ -0,0 +1,19 @@
+--- a/cpp/grammar_functor.cc
++++ b/cpp/grammar_functor.cc
+@@ -1750,11 +1750,11 @@
+   void Apply(Grammar* grammar);
+   static std::optional<uint64_t> HashSequence(const Grammar& grammar, int32_t sequence_id);
+
+-  static const int16_t kNotEndStateFlag = -0x100;
+-  static const int16_t kEndStateFlag = -0x200;
+-  static const int16_t kSelfRecursionFlag = -0x300;
+-  static const int16_t kSimpleCycleFlag = -0x400;
+-  static const int16_t kUnKnownFlag = -0x500;
++  static constexpr int16_t kNotEndStateFlag = -0x100;
++  static constexpr int16_t kEndStateFlag = -0x200;
++  static constexpr int16_t kSelfRecursionFlag = -0x300;
++  static constexpr int16_t kSimpleCycleFlag = -0x400;
++  static constexpr int16_t kUnKnownFlag = -0x500;
+
+  private:
+   Grammar* grammar_;

AGENTS.md

@@ -10,11 +10,10 @@ Python and C++ codebase supporting TensorRT engine-based and PyTorch-based execu
 **CRITICAL (YOU MUST):**
 - Read and follow `CODING_GUIDELINES.md` for ALL code changes (C++ and Python)
 - NVIDIA copyright header on ALL new files (update year on modified files)
-- `git commit -s` (DCO sign-off required). Never attribute AI tools in sign-off line. Do not add
-  co-authors to the git commit message unless explicitly instructed to do so by the user.
+- `git commit -s` (DCO sign-off required). Never attribute AI tools in sign-off line. Always rely on `git` to do the sign off instead of directly adding sign off in commit message.
+- Do not add co-authors to the git commit message unless explicitly instructed to do so by the user.
 - `pre-commit` hooks run on commit — if files are modified by hooks, re-stage and commit again
 - PR title format: `[JIRA/NVBUG/None][type] description` (e.g., `[TRTLLM-5516][perf] optimize cuda graph padding`)
-- Python imports: `from package.subpackage import module` (never `from module import Class`)
 - Set `LLM_MODELS_ROOT` env var when running tests that need model weights
 
 ## Common Commands
@@ -101,7 +100,6 @@ HuggingFace Model → LLM API → Executor (PyTorch/AutoDeploy/TensorRT)
 - **Integration tests need GPUs + models** — always set `LLM_MODELS_ROOT` and ensure GPU access. Unit tests don't.
 - **Copyright year** — update to current year when modifying existing files; add full header to new files.
 - **Avoid broad exception handling** — catch specific exceptions, not bare `except:` (see `CODING_GUIDELINES.md`).
-- **Python import style is enforced** — `from package.subpackage import module`, never `from module import Class`. Pre-commit will not catch this.
 - **One concern per PR** — avoid scope creep. If a PR touches unrelated areas, split it.
 - **User-facing configuration classes** - when editing or defining any user-facing configuration classes (particularly `LlmArgs` or any class used in its fields), you **MUST** follow the Pydantic guidelines in `CODING_GUIDELINES.md`.
 

ATTRIBUTIONS-Python.md

@@ -5261,7 +5261,7 @@ For more information, please refer to <http://unlicense.org>
   - `Tracker`: https://github.com/tox-dev/py-filelock/issues
 
 
-## flashinfer-python (0.6.4)
+## flashinfer-python (0.6.6)
 
 ### Licenses
 License: `Apache-2.0`
@@ -63471,7 +63471,7 @@ SOFTWARE.
   - `Homepage`: https://github.com/akshaynagpal/w2n
 
 
-## xgrammar (0.1.25)
+## xgrammar (0.1.32)
 
 ### Licenses
 License: `Apache 2.0`