From d704b5e889740903e31aa2d7cab238fd6f8c36aa Mon Sep 17 00:00:00 2001 From: Zhenhua Wang <4936589+zhenhuaw-me@users.noreply.github.com> Date: Mon, 9 Mar 2026 18:20:35 +0800 Subject: [PATCH 1/4] [None][chore] Remove visual_gen benchmark test from YAML (#12027) Signed-off-by: Zhenhua Wang <4936589+zhenhuaw-me@users.noreply.github.com> --- tests/integration/test_lists/test-db/l0_dgx_b200.yml | 1 - 1 file changed, 1 deletion(-) diff --git a/tests/integration/test_lists/test-db/l0_dgx_b200.yml b/tests/integration/test_lists/test-db/l0_dgx_b200.yml index 409a3109199..094408ae1b4 100644 --- a/tests/integration/test_lists/test-db/l0_dgx_b200.yml +++ b/tests/integration/test_lists/test-db/l0_dgx_b200.yml @@ -271,7 +271,6 @@ l0_dgx_b200: - examples/test_visual_gen.py::test_vbench_dimension_score_wan - examples/test_visual_gen.py::test_vbench_dimension_score_wan22_a14b_fp8 - examples/test_visual_gen.py::test_vbench_dimension_score_wan22_a14b_nvfp4 - - visual_gen/test_visual_gen_benchmark.py # ------------- AutoDeploy Backend Stages --------------- - condition: ranges: From a176d8347893cf2242abe7cc7617a50f3d7294bd Mon Sep 17 00:00:00 2001 From: tcherckez-nvidia <127761168+tcherckez-nvidia@users.noreply.github.com> Date: Mon, 9 Mar 2026 14:19:36 +0200 Subject: [PATCH 2/4] [None][fix] Fix the model list as it had a dup model (#12029) Signed-off-by: Tal Cherckez <127761168+tcherckez-nvidia@users.noreply.github.com> --- examples/auto_deploy/model_registry/models.yaml | 2 -- 1 file changed, 2 deletions(-) diff --git a/examples/auto_deploy/model_registry/models.yaml b/examples/auto_deploy/model_registry/models.yaml index 6f6d630d16a..01887699b99 100644 --- a/examples/auto_deploy/model_registry/models.yaml +++ b/examples/auto_deploy/model_registry/models.yaml @@ -181,8 +181,6 @@ models: yaml_extra: ['dashboard_default.yaml', 'world_size_8.yaml', 'simple_shard_only.yaml'] - name: deepseek-ai/DeepSeek-R1 yaml_extra: ['dashboard_default.yaml', 'world_size_8.yaml', 'num_hidden_layers_5.yaml'] -- name: deepseek-ai/DeepSeek-V3 - yaml_extra: ['dashboard_default.yaml', 'world_size_8.yaml', 'num_hidden_layers_5.yaml'] - name: deepseek-ai/DeepSeek-Coder-V2-Instruct yaml_extra: ['dashboard_default.yaml', 'world_size_8.yaml'] - name: Qwen/Qwen3-VL-8B-Instruct From 34a915377f6bf86bc87927d8eaca83e8c7110e68 Mon Sep 17 00:00:00 2001 From: Yiyun Lu <55233584+luyiyun1021@users.noreply.github.com> Date: Mon, 9 Mar 2026 20:31:17 +0800 Subject: [PATCH 3/4] [https://nvbugs/5863806][fix] Fix Python string truthiness bug in FMHA cubin selection (#11909) Signed-off-by: Yiyun Lu <55233584+luyiyun1021@users.noreply.github.com> Co-authored-by: Jie Li <76780849+jieli-matrix@users.noreply.github.com> --- cpp/kernels/fmha_v2/setup.py | 21 ++++++++++++++------- tests/integration/test_lists/waives.txt | 3 --- 2 files changed, 14 insertions(+), 10 deletions(-) diff --git a/cpp/kernels/fmha_v2/setup.py b/cpp/kernels/fmha_v2/setup.py index 88cba8f793f..01dd06c7058 100644 --- a/cpp/kernels/fmha_v2/setup.py +++ b/cpp/kernels/fmha_v2/setup.py @@ -3333,11 +3333,14 @@ def use_cubin_header(sm, head_size, dtype, output_dtype=None, - enable_skip_softmax=False): + enable_skip_softmax=False, + attention_mask_type=None): if enable_skip_softmax: return False if 'e4m3' in dtype and output_dtype in ['bf16', 'fp16']: return False + if attention_mask_type == AttentionMaskType.BIDIRECTIONAL_SLIDING_WINDOW: + return False return (sm == 90 and head_size == 128) or (sm == 89 and 'e4m3' in dtype) @@ -3349,9 +3352,11 @@ def get_cubin_header(kernel_traits, specs_names): cubin_lens_dict = {} launchers_dict = {} for kspec, fname, lname, kname in specs_names: + mask_type = AttentionMaskType.BIDIRECTIONAL_SLIDING_WINDOW \ + if '_bidirectional_sliding_window' in kname else None if generate_cu_trtllm and not use_cubin_header( kspec.sm, kspec.head_size, kspec.dtype, kspec.output_dtype, - kspec.enable_skip_softmax): + kspec.enable_skip_softmax, mask_type): continue name = fname.replace('.', '_') data = 'extern unsigned char cubin_{name}_cubin[];'.format(name=name) @@ -3487,7 +3492,8 @@ def get_cubin_header(kernel_traits, specs_names): return_softmax_stats_flag = pythonBoolean2cpp[sm != '90' or ( sm == '90' and '_softmax' in kname)] - enable_skip_softmax_flag = pythonBoolean2cpp['_skipSoftmax' in kname] + enable_skip_softmax = '_skipSoftmax' in kname + enable_skip_softmax_flag = pythonBoolean2cpp[enable_skip_softmax] # meta_unroll_step meta_unroll_step = unroll_step if ('_nl' in kname @@ -3516,7 +3522,8 @@ def get_cubin_header(kernel_traits, specs_names): def get_lname_from_kname(kname: str) -> str: if use_cubin_header(int(sm), int(head_size), prec.lower(), output_prec.lower(), - enable_skip_softmax_flag): + enable_skip_softmax, + attention_mask_type): return 'nullptr' lname = kname.replace('_kernel', '') mask_types = [ @@ -3537,9 +3544,9 @@ def get_lname_from_kname(kname: str) -> str: {cubin_name}_len, \"{kname}\", {smem}, {threads}, {meta_unroll_step}, {attention_mask_type_value}, \ {attention_input_layout_value}, {is_il}, {is_flash_atten}, {is_warp_specialization}, {is_fp32_accu}, \ {is_alibi_supported}, {is_tiled}, {has_softcapping_scale}, {return_softmax_stats_flag}, {enable_skip_softmax_flag}, {lname}}}\ -'''.format(**locals()) if use_cubin_header(int(sm), int(head_size), - prec.lower(), output_prec.lower(), - enable_skip_softmax_flag) else '''\ +'''.format(**locals()) if use_cubin_header( + int(sm), int(head_size), prec.lower(), output_prec.lower(), + enable_skip_softmax, attention_mask_type) else '''\ {{ DATA_TYPE_{prec}, DATA_TYPE_{output_prec}, {seq_len}, {q_step}, {kv_step}, {head_size}, {head_size_v}, \ {sage_block_sizes[0]}, {sage_block_sizes[1]}, {sage_block_sizes[2]}, kSM_{sm}, nullptr, \ 0, \"{kname}\", {smem}, {threads}, {meta_unroll_step}, {attention_mask_type_value}, \ diff --git a/tests/integration/test_lists/waives.txt b/tests/integration/test_lists/waives.txt index 791866c9e68..b87997f01a2 100644 --- a/tests/integration/test_lists/waives.txt +++ b/tests/integration/test_lists/waives.txt @@ -218,8 +218,6 @@ cpp/test_e2e.py::test_model[-mamba-86] SKIP (https://nvbugs/5781665) unittest/llmapi/test_llm_multi_gpu_pytorch.py::test_tinyllama_logits_processor_tp2pp2 SKIP (https://nvbugs/5781731) accuracy/test_disaggregated_serving.py::TestLlama3_1_8BInstruct::test_eagle3[eagle3_one_model=False-overlap_scheduler=False] SKIP (https://nvbugs/5807902) unittest/_torch/ray_orchestrator/multi_gpu/test_multi_instance.py::test_multi_instance[tp2_2instances] SKIP (https://nvbugs/5784566) -accuracy/test_llm_api_pytorch.py::TestLlama3_2_1B::test_fp8_prequantized SKIP (https://nvbugs/5785465) -accuracy/test_llm_api_pytorch.py::TestMinistral8BInstruct::test_fp8 SKIP (https://nvbugs/5785485) accuracy/test_llm_api_pytorch.py::TestLlama4MaverickInstruct::test_fp8_chunked_prefill[tp8ep8-cuda_graph=False] SKIP (https://nvbugs/5795918) full:RTXPro6000D/accuracy/test_llm_api_pytorch.py::TestDeepSeekV3Lite::test_nvfp4_4gpus[moe_backend=CUTLASS-mtp_nextn=0-ep4-fp8kv=False-attention_dp=False-cuda_graph=False-overlap_scheduler=False-torch_compile=False] SKIP (https://nvbugs/5800672) examples/test_medusa.py::test_llm_medusa_with_qaunt_base_model_1gpu[fp8-use_cpp_session-medusa-vicuna-7b-v1.3-4-heads-float16-bs1] SKIP (https://nvbugs/5802248) @@ -276,7 +274,6 @@ accuracy/test_llm_api_pytorch.py::TestDeepSeekV32::test_fp8_blockscale[disable_s accuracy/test_llm_api_pytorch.py::TestGPTOSS::test_w4_4gpus[v1_kv_cache-tp4-cutlass-fp8] SKIP (https://nvbugs/5651865) accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B::test_w4a16_mxfp4[latency-TRITON] SKIP (https://nvbugs/5864263) accuracy/test_llm_api_pytorch.py::TestGPTOSS::test_w4_1gpu[v1_kv_cache-True-True-triton-auto] SKIP (https://nvbugs/5864187) -accuracy/test_llm_api_pytorch.py::TestQwen3_30B_A3B::test_fp8[latency-torch_compile=False] SKIP (https://nvbugs/5863806) accuracy/test_llm_api_pytorch.py::TestGPTOSS::test_w4_4gpus[v1_kv_cache-dp4-trtllm-auto] SKIP (https://nvbugs/5596343) test_e2e.py::test_trtllm_multimodal_benchmark_serving SKIP (https://nvbugs/5864769) unittest/auto_deploy/multigpu/transformations/library/test_bmm_sharding.py::test_sharding[1-1] SKIP (https://nvbugs/5875203) From 9fe24dbee0bed41bf67c7556c60867233db18c24 Mon Sep 17 00:00:00 2001 From: sunnyqgg <159101675+sunnyqgg@users.noreply.github.com> Date: Mon, 9 Mar 2026 21:00:00 +0800 Subject: [PATCH 4/4] [None][feat] Upgrade xgrammar from 0.1.25 to 0.1.32 (#12016) Signed-off-by: qgai Co-authored-by: Claude Opus 4.6 --- 3rdparty/CMakeLists.txt | 14 ++ 3rdparty/fetch_content.json | 5 +- 3rdparty/patches/xgrammar_constexpr.patch | 19 +++ ATTRIBUTIONS-Python.md | 2 +- requirements.txt | 2 +- .../attribution/data/dependency_metadata.yml | 2 +- .../attribution/data/files_to_dependency.yml | 2 +- security_scanning/poetry.lock | 155 +++++------------- security_scanning/pyproject.toml | 2 +- 9 files changed, 79 insertions(+), 124 deletions(-) create mode 100644 3rdparty/patches/xgrammar_constexpr.patch diff --git a/3rdparty/CMakeLists.txt b/3rdparty/CMakeLists.txt index 7360e13c80d..af2ccf9d693 100644 --- a/3rdparty/CMakeLists.txt +++ b/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 diff --git a/3rdparty/fetch_content.json b/3rdparty/fetch_content.json index f28186b6235..c3d0e1dbcce 100644 --- a/3rdparty/fetch_content.json +++ b/3rdparty/fetch_content.json @@ -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" } ] } diff --git a/3rdparty/patches/xgrammar_constexpr.patch b/3rdparty/patches/xgrammar_constexpr.patch new file mode 100644 index 00000000000..f4e5be99a21 --- /dev/null +++ b/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 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_; diff --git a/ATTRIBUTIONS-Python.md b/ATTRIBUTIONS-Python.md index c21c9063555..b9975173aaa 100644 --- a/ATTRIBUTIONS-Python.md +++ b/ATTRIBUTIONS-Python.md @@ -63471,7 +63471,7 @@ SOFTWARE. - `Homepage`: https://github.com/akshaynagpal/w2n -## xgrammar (0.1.25) +## xgrammar (0.1.32) ### Licenses License: `Apache 2.0` diff --git a/requirements.txt b/requirements.txt index 678d640864e..c41db79b650 100644 --- a/requirements.txt +++ b/requirements.txt @@ -56,7 +56,7 @@ patchelf einops flashinfer-python==0.6.4 opencv-python-headless -xgrammar==0.1.25 +xgrammar==0.1.32 llguidance==0.7.29 jsonschema backoff diff --git a/scripts/attribution/data/dependency_metadata.yml b/scripts/attribution/data/dependency_metadata.yml index 6d0b3380329..456d1446047 100644 --- a/scripts/attribution/data/dependency_metadata.yml +++ b/scripts/attribution/data/dependency_metadata.yml @@ -90,7 +90,7 @@ ucx/1.20: ucxx/16eaa57c8d98c8ef54d666a2d2b11e76cfa565f5: license: 759cb066f14805ef4068f633d9071e1d source: https://github.com/rapidsai/ucxx/tree/16eaa57c8d98c8ef54d666a2d2b11e76cfa565f5 -xgrammar/v0.1.25: +xgrammar/v0.1.32: copyright: 989a9441d689f61fba9f797cc253e51b license: 8e1c96809a7467593130ecc62ae12be9 zeromq/4.3.4-3.el8: diff --git a/scripts/attribution/data/files_to_dependency.yml b/scripts/attribution/data/files_to_dependency.yml index d8243f7689b..70555a26134 100644 --- a/scripts/attribution/data/files_to_dependency.yml +++ b/scripts/attribution/data/files_to_dependency.yml @@ -7507,7 +7507,7 @@ ucxx/16eaa57c8d98c8ef54d666a2d2b11e76cfa565f5: - cb11c17f716ae644b2d74652d3d3232b - e10a9fefe2ef09b9560cc204bd54f728 - f75f9c7cefa54d6626032daa93ef2549 -xgrammar/v0.1.25: +xgrammar/v0.1.32: - 0e2b512f384e122c3b8243ae00256e06 - 1a6e20d89e227a29d674e12e18b5e9e7 - 1acd98aa4050fd0b8cda58d5d4f6ef78 diff --git a/security_scanning/poetry.lock b/security_scanning/poetry.lock index c245778e9f0..6edf13bfc00 100644 --- a/security_scanning/poetry.lock +++ b/security_scanning/poetry.lock @@ -2784,95 +2784,6 @@ numpy = [ [package.extras] dev = ["absl-py", "pyink", "pylint (>=2.6.0)", "pytest", "pytest-xdist"] -[[package]] -name = "mlx" -version = "0.31.0" -description = "A framework for machine learning on Apple silicon." -optional = false -python-versions = ">=3.10" -groups = ["main"] -markers = "platform_system == \"Darwin\" and platform_machine == \"arm64\"" -files = [ - {file = "mlx-0.31.0-cp310-cp310-macosx_14_0_arm64.whl", hash = "sha256:38680838e0dd9a621ed4adc5a9ed8b94aeb6a4798142fbe215b821b8c6b8fc36"}, - {file = "mlx-0.31.0-cp310-cp310-macosx_15_0_arm64.whl", hash = "sha256:7aded590bcf6839307c3acc899e196936991f97b499ddbdd0cd3b228bf10792f"}, - {file = "mlx-0.31.0-cp310-cp310-macosx_26_0_arm64.whl", hash = "sha256:6e3ae83607b798b44cb3e44437095cfd26886fecc15f90f29f9eafd206d4d170"}, - {file = "mlx-0.31.0-cp310-cp310-manylinux_2_35_aarch64.whl", hash = "sha256:b25f785c94eb47d8104604a5de0e7d749b801e7a40073cbf457aa94c372e5593"}, - {file = "mlx-0.31.0-cp310-cp310-manylinux_2_35_x86_64.whl", hash = "sha256:6a4342027e6608ce69807a8f079c750a7c6161f543ebb49e55654edd03c178d6"}, - {file = "mlx-0.31.0-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:285313eaeba425e58cbb3238c2d1a3894e6252d58f243ce56681d5419a568d6c"}, - {file = "mlx-0.31.0-cp311-cp311-macosx_15_0_arm64.whl", hash = "sha256:acf4f04ff33a80784a0f15c492166dc889e65659b41c410ca5a7c2d78bee2a3a"}, - {file = "mlx-0.31.0-cp311-cp311-macosx_26_0_arm64.whl", hash = "sha256:f624571e23a86654496c42a507b4bb42ded0edb91f33161fabafdbf6b81ba024"}, - {file = "mlx-0.31.0-cp311-cp311-manylinux_2_35_aarch64.whl", hash = "sha256:5b5306a0934b15c4e3a1088a10066bdde3966c21b95006c63ecc38ca8e3891e0"}, - {file = "mlx-0.31.0-cp311-cp311-manylinux_2_35_x86_64.whl", hash = "sha256:18078bc67dfb7ed602fca233d00ce93e23d590d9347da5009472455a92831066"}, - {file = "mlx-0.31.0-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:de6c0a3e8aa0e7d1365d46634fdbb3f835c164fbdb6ba8a239e039a4efa07fe2"}, - {file = "mlx-0.31.0-cp312-cp312-macosx_15_0_arm64.whl", hash = "sha256:d6af01b15177da995336a6fd9878e7c5994720a9f1614d8f4d1dbe9293167c30"}, - {file = "mlx-0.31.0-cp312-cp312-macosx_26_0_arm64.whl", hash = "sha256:1ad14ddc3a15818f5bba0de35e88559ed8dcb93ccff2ef879ff604d02d663b25"}, - {file = "mlx-0.31.0-cp312-cp312-manylinux_2_35_aarch64.whl", hash = "sha256:a80754ecf64191f71da1946dc5de6cf903344cc90dd286c589792ee9d3fc62f9"}, - {file = "mlx-0.31.0-cp312-cp312-manylinux_2_35_x86_64.whl", hash = "sha256:363282eb094785f6aba27810ff89331c0f7829c6961f571cd0feaad09d2c809f"}, - {file = "mlx-0.31.0-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:4f33e9aafc6d3ad29e72743dfb786c4ce67397414f0a091469058626381fc1bc"}, - {file = "mlx-0.31.0-cp313-cp313-macosx_15_0_arm64.whl", hash = "sha256:242806b8ad6a4d3ce86cdff513f86520552de7592786712770b2e1ebd178816a"}, - {file = "mlx-0.31.0-cp313-cp313-macosx_26_0_arm64.whl", hash = "sha256:7f0bdbac084017820ce513a12318771a06c7ec10fad159839e27c998bc5dad89"}, - {file = "mlx-0.31.0-cp313-cp313-manylinux_2_35_aarch64.whl", hash = "sha256:8642dda2b23195d9921973749ae9bf764e2c7d70bfc0e60b23b6335e660cc610"}, - {file = "mlx-0.31.0-cp313-cp313-manylinux_2_35_x86_64.whl", hash = "sha256:c6daa671cfa3c194951d742aa09030c5008d9d9657034b2903389fa090b3ba92"}, - {file = "mlx-0.31.0-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:cce3e15cf11c608c9e721502fe56e54f9f48b897e9b80f1204a48643d68710c0"}, - {file = "mlx-0.31.0-cp314-cp314-macosx_15_0_arm64.whl", hash = "sha256:ba330fe40d73b202880bbb5cac62de0b639cf4c44a12853bcadb34a9e3ffe880"}, - {file = "mlx-0.31.0-cp314-cp314-macosx_26_0_arm64.whl", hash = "sha256:d2014d113070846c6cdee980653f561c92a4a663a449f64e70c15bbf74d637e1"}, - {file = "mlx-0.31.0-cp314-cp314-manylinux_2_35_aarch64.whl", hash = "sha256:994fab25ff521621e03001177a8f0f1a7bf8294ff340f89910ec074f9f681ed9"}, - {file = "mlx-0.31.0-cp314-cp314-manylinux_2_35_x86_64.whl", hash = "sha256:c3bb9961f40d098659326b0edb96e2a16adecfaf3c1f2518cad5a0b7e55a3a5d"}, -] - -[package.dependencies] -mlx-metal = {version = "0.31.0", markers = "platform_system == \"Darwin\""} - -[package.extras] -cpu = ["mlx-cpu (==0.31.0) ; platform_system == \"Linux\""] -cuda = ["mlx-cuda-12 (==0.31.0) ; platform_system == \"Linux\""] -cuda12 = ["mlx-cuda-12 (==0.31.0) ; platform_system == \"Linux\""] -cuda13 = ["mlx-cuda-13 (==0.31.0) ; platform_system == \"Linux\""] -dev = ["numpy (>=2)", "pre-commit", "psutil", "torch (>=2.9)", "typing_extensions"] - -[[package]] -name = "mlx-lm" -version = "0.29.1" -description = "LLMs with MLX and the Hugging Face Hub" -optional = false -python-versions = ">=3.8" -groups = ["main"] -markers = "platform_system == \"Darwin\" and platform_machine == \"arm64\"" -files = [ - {file = "mlx_lm-0.29.1-py3-none-any.whl", hash = "sha256:440941b3054c2a2216e97615de584cc90fa1ea874782e20699b9895721fad8dc"}, - {file = "mlx_lm-0.29.1.tar.gz", hash = "sha256:b99180d8f33d33a077b814e550bfb2d8a59ae003d668fd1f4b3fff62a381d34b"}, -] - -[package.dependencies] -jinja2 = "*" -mlx = {version = ">=0.29.2", markers = "platform_system == \"Darwin\""} -numpy = "*" -protobuf = "*" -pyyaml = "*" -sentencepiece = "*" -transformers = ">=4.39.3" - -[package.extras] -cpu = ["mlx[cpu] (>=0.29.2)"] -cuda = ["mlx[cuda] (>=0.29.2)"] -evaluate = ["lm-eval", "tqdm"] -test = ["datasets", "lm-eval"] -train = ["datasets", "tqdm"] - -[[package]] -name = "mlx-metal" -version = "0.31.0" -description = "A framework for machine learning on Apple silicon." -optional = false -python-versions = ">=3.10" -groups = ["main"] -markers = "platform_system == \"Darwin\" and platform_machine == \"arm64\"" -files = [ - {file = "mlx_metal-0.31.0-py3-none-macosx_14_0_arm64.whl", hash = "sha256:1c572a6e3634a63060c103b0c38ac309e2d217be15519e3d8f0d6b452bb015f5"}, - {file = "mlx_metal-0.31.0-py3-none-macosx_15_0_arm64.whl", hash = "sha256:554dc7cb29e0ea5fb6941df42f11a1de385b095848e6183c7a99d7c1f1a11f5d"}, - {file = "mlx_metal-0.31.0-py3-none-macosx_26_0_arm64.whl", hash = "sha256:7fd412f55ddf9f1d90c2cd86ce281d19e8eb93d093c6dbd784a49f8bd7d0a22c"}, -] - [[package]] name = "mpi4py" version = "4.1.1" @@ -6635,41 +6546,50 @@ standard = ["colorama (>=0.4) ; sys_platform == \"win32\"", "httptools (>=0.6.3) [[package]] name = "xgrammar" -version = "0.1.25" +version = "0.1.32" description = "Efficient, Flexible and Portable Structured Generation" optional = false python-versions = "<4,>=3.8" groups = ["main"] files = [ - {file = "xgrammar-0.1.25-cp310-cp310-macosx_10_14_x86_64.whl", hash = "sha256:d12d1078ee2b5c1531610489b433b77694a7786210ceb2c0c1c1eb058e9053c7"}, - {file = "xgrammar-0.1.25-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:c2e940541b7cddf3ef55a70f20d4c872af7f0d900bc0ed36f434bf7212e2e729"}, - {file = "xgrammar-0.1.25-cp310-cp310-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:2063e1c72f0c00f47ac8ce7ce0fcbff6fa77f79012e063369683844e2570c266"}, - {file = "xgrammar-0.1.25-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:9785eafa251c996ebaa441f3b8a6c037538930104e265a64a013da0e6fd2ad86"}, - {file = "xgrammar-0.1.25-cp310-cp310-win_amd64.whl", hash = "sha256:42ecefd020038b3919a473fe5b9bb9d8d809717b8689a736b81617dec4acc59b"}, - {file = "xgrammar-0.1.25-cp311-cp311-macosx_10_14_x86_64.whl", hash = "sha256:27d7ac4be05cf9aa258c109a8647092ae47cb1e28df7d27caced6ab44b72b799"}, - {file = "xgrammar-0.1.25-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:151c1636188bc8c5cdf318cefc5ba23221c9c8cc07cb392317fb3f7635428150"}, - {file = "xgrammar-0.1.25-cp311-cp311-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:35fc135650aa204bf84db7fe9c0c0f480b6b11419fe47d89f4bd21602ac33be9"}, - {file = "xgrammar-0.1.25-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:fc19d6d7e8e51b6c9a266e949ac7fb3d2992447efeec7df32cca109149afac18"}, - {file = "xgrammar-0.1.25-cp311-cp311-win_amd64.whl", hash = "sha256:8fcb24f5a7acd5876165c50bd51ce4bf8e6ff897344a5086be92d1fe6695f7fe"}, - {file = "xgrammar-0.1.25-cp312-cp312-macosx_10_14_x86_64.whl", hash = "sha256:96500d7578c46e8551253b9211b02e02f54e147bc290479a64717d80dcf4f7e3"}, - {file = "xgrammar-0.1.25-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:73ba9031e359447af53ce89dfb0775e7b9f4b358d513bcc28a6b4deace661dd5"}, - {file = "xgrammar-0.1.25-cp312-cp312-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:c519518ebc65f75053123baaf23776a21bda58f64101a64c2fc4aa467c9cd480"}, - {file = "xgrammar-0.1.25-cp312-cp312-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:47fdbfc6007df47de2142613220292023e88e4a570546b39591f053e4d9ec33f"}, - {file = "xgrammar-0.1.25-cp312-cp312-win_amd64.whl", hash = "sha256:c9b3defb6b45272e896da401f43b513f5ac12104ec3101bbe4d3a7d02bcf4a27"}, - {file = "xgrammar-0.1.25-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:2d80d4bfe65d1a3334536c804b6471f32e6759f1972c9abe0ae49d5e21462c0b"}, - {file = "xgrammar-0.1.25-cp313-cp313-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:7a1a6a638167d704a22a0c9670e2176104c38e38c351286a07a77143e22f9053"}, - {file = "xgrammar-0.1.25-cp313-cp313-win_amd64.whl", hash = "sha256:ffadeba0b704667a7eb6202d409533e9d1e80af15a10add107684e0cde45b8e4"}, - {file = "xgrammar-0.1.25-cp39-cp39-macosx_10_14_x86_64.whl", hash = "sha256:a62dea5d73147a254e71e07ceae4a48c0f5a294cce2fa3e028159f48da19a39d"}, - {file = "xgrammar-0.1.25-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:2b309807ad837c1cbe2f833830b665a15309e11692b53795363c59041c65074f"}, - {file = "xgrammar-0.1.25-cp39-cp39-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:073829d8a53ef482e6b51469316f6e505962460bb576ae4d4a606769c4c37678"}, - {file = "xgrammar-0.1.25-cp39-cp39-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:241529d6104d97738b3e29c573bffa6d0fa89a8d0299b2c685358cc13858205c"}, - {file = "xgrammar-0.1.25-cp39-cp39-win_amd64.whl", hash = "sha256:f5d46e1749d9324684d2462e428bc63652096addc1e2c21db2ae66ca88e76a1c"}, - {file = "xgrammar-0.1.25.tar.gz", hash = "sha256:70ce16b27e8082f20808ed759b0733304316facc421656f0f30cfce514b5b77a"}, + {file = "xgrammar-0.1.32-cp310-cp310-macosx_10_14_x86_64.whl", hash = "sha256:1d179511a1258b410d1660ed00032da24256072b44d6743f5bd3f49743d90801"}, + {file = "xgrammar-0.1.32-cp310-cp310-macosx_11_0_arm64.whl", hash = "sha256:b33d2e8f02ca31b93f3cbfea0ed1df7e706a2bea52d7706793884f92a8c1523a"}, + {file = "xgrammar-0.1.32-cp310-cp310-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f7d4f0f2041cdc97bacc096b4c2d1c841e91794b6e7a54e7a4853fc0907956dc"}, + {file = "xgrammar-0.1.32-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:d4f3d18da40cad18e87881925f82d0aed1418cc2fb05886d516e3c9d548db57f"}, + {file = "xgrammar-0.1.32-cp310-cp310-win_amd64.whl", hash = "sha256:e7baf71bba03a5e734df435b6378da4406b1c15f5511ab4f5d4af9f72775c756"}, + {file = "xgrammar-0.1.32-cp311-cp311-macosx_10_14_x86_64.whl", hash = "sha256:51b41c47785aa198d19f8d056b394f75b4421deab88c415568f9c588b1f7e238"}, + {file = "xgrammar-0.1.32-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:d7030192cb1d8579699f1f72fd14d31347a402611aab98a2da6a04c3de07e917"}, + {file = "xgrammar-0.1.32-cp311-cp311-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:a332c0364f665b410a6cfc2ada155c3a6ede430e385ac431015e31735a64fec3"}, + {file = "xgrammar-0.1.32-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:5b8ad132d0fcf3a51dc054ecb0dc9808566b302122de6edaac7b4aca460adbec"}, + {file = "xgrammar-0.1.32-cp311-cp311-win_amd64.whl", hash = "sha256:b8b1ca6d3f3c2842660458660e494aaf0a6745f1b07ae74e4c2230ab4ff70c11"}, + {file = "xgrammar-0.1.32-cp312-cp312-macosx_10_14_x86_64.whl", hash = "sha256:9b78d32265f096e5567ab52c72b681855cf473481a48a1e7e6d97d414ba30b82"}, + {file = "xgrammar-0.1.32-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:23eacaf826c3aeebca0d91fc271417d9d96e157af2bacf6f14277297af7917ef"}, + {file = "xgrammar-0.1.32-cp312-cp312-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:f9a637d4e0c541149e0d409c24f4ec79cd74d87508ee6a17a7e64a9b9c0cf56f"}, + {file = "xgrammar-0.1.32-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f96c7a4fcbd68e18b13cb3b6ed5d24b5326b256933f476bdaf2cc8e609c228db"}, + {file = "xgrammar-0.1.32-cp312-cp312-win_amd64.whl", hash = "sha256:ba6e08c385cce53eda8e9b3bbfba63f100ba3dcb76fa0692a65921a36b20ad0a"}, + {file = "xgrammar-0.1.32-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:4addb8f5d5699e7df7fca6d299a91b3ef1ad799811c0ab7050d6f96d754c9c21"}, + {file = "xgrammar-0.1.32-cp313-cp313-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:028f8d6a105d06549faee0afbebfaada90aa1941c081dcc88f3d5ef373dad934"}, + {file = "xgrammar-0.1.32-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:0c0150c50eb3a56a35d6f0c0af0bce0f113ec5f84f7918bfd46b49e25ecf7fb5"}, + {file = "xgrammar-0.1.32-cp313-cp313-win_amd64.whl", hash = "sha256:e1072d764705c8e87df6136ce3419f96ab3fd423d85f58c2d81c13a647b78894"}, + {file = "xgrammar-0.1.32-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:4e6015ad2b941a292562f68b9a2ee1ddae8e28df840dc39232dcc7007fc6f606"}, + {file = "xgrammar-0.1.32-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:8e8da3e7fc194e098b760bacb2b60ad2227cac70d7be5d2e4f7025b1c360c43d"}, + {file = "xgrammar-0.1.32-cp314-cp314-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6588cfd9754f2c46846276a2e8284a46582a74886d7aaea02cf6ce63ccc397ce"}, + {file = "xgrammar-0.1.32-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:7f740ba83b69abb423167a5d5b13a9fcde89747220e191f6a004fae4a834311f"}, + {file = "xgrammar-0.1.32-cp314-cp314-win_amd64.whl", hash = "sha256:9c0769c3468bd67495c28a03dc5ce3948d83cddaf0a59c6d992b12fc683a1c3e"}, + {file = "xgrammar-0.1.32-cp314-cp314t-macosx_10_15_x86_64.whl", hash = "sha256:da8339b38e96d105868c14b2cb2df4b7c83d7a49f8539c74fd7470d61043e5b1"}, + {file = "xgrammar-0.1.32-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:b938a9096bccc06c30abb5304b2b39c272a924ca002e19421cce5e6ee9670f4f"}, + {file = "xgrammar-0.1.32-cp314-cp314t-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:fe2ee94080d77b84e38cb6643b75a6ca29cf814a3e5d5da8e1176eae4034d662"}, + {file = "xgrammar-0.1.32-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:70ddbf7216e1e7ec96134a2474a6b84d2b14439a6f6379e079b7c557131be41d"}, + {file = "xgrammar-0.1.32-cp314-cp314t-win_amd64.whl", hash = "sha256:4f68e591a6e9e121d5f03821ab2c44a7af092dc8bf7c9cde1a776871c6bd4dc5"}, + {file = "xgrammar-0.1.32-cp39-cp39-macosx_10_14_x86_64.whl", hash = "sha256:88f3a00623b197c6b76dea3c0c7f326702209c1066eb9993b4723929a106c8ba"}, + {file = "xgrammar-0.1.32-cp39-cp39-macosx_11_0_arm64.whl", hash = "sha256:0ddf7ca871c7ec8c9ce8c34b8547344fce70ef3faa33b6e29f6826d098bc7714"}, + {file = "xgrammar-0.1.32-cp39-cp39-manylinux_2_26_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6925258c8e188fbddcf2962c6a0819f71cf664263d2346545feadee92994685e"}, + {file = "xgrammar-0.1.32-cp39-cp39-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:2691be7077e2eae34cdefce2ef3f88e2f018158ccff0eb4b94bb8725a98e2f27"}, + {file = "xgrammar-0.1.32-cp39-cp39-win_amd64.whl", hash = "sha256:82b5c3d6b3a6fe01a1957d311473808aee90e946d38bd942c7161f1c4aedce11"}, + {file = "xgrammar-0.1.32.tar.gz", hash = "sha256:5d424d52779ca2d3ccaf72f2289d6519efe308e933d0d3fc3c292c780825bb12"}, ] [package.dependencies] -mlx-lm = {version = "*", markers = "platform_system == \"Darwin\" and platform_machine == \"arm64\""} -ninja = "*" numpy = "*" pydantic = "*" torch = ">=1.10.0" @@ -6678,6 +6598,7 @@ triton = {version = "*", markers = "platform_system == \"Linux\" and platform_ma typing-extensions = ">=4.9.0" [package.extras] +metal = ["mlx-lm ; platform_system == \"Darwin\" and platform_machine == \"arm64\""] test = ["huggingface-hub[cli]", "protobuf", "pytest", "sentencepiece", "tiktoken", "transformers (<4.50.0) ; platform_system == \"Darwin\""] [[package]] @@ -6996,4 +6917,4 @@ type = ["pytest-mypy"] [metadata] lock-version = "2.1" python-versions = ">=3.10,<3.13" -content-hash = "2a21a66f0b0512caff14b587ff5a19a8d42d3ae9b5624172c4e68f8d07044c48" +content-hash = "593920d3d4ce64e19ce8ffc9e66e2f4e0337a23b36ae8a85ed99c9c87b427877" diff --git a/security_scanning/pyproject.toml b/security_scanning/pyproject.toml index ccb19badfc0..84c9d57426d 100644 --- a/security_scanning/pyproject.toml +++ b/security_scanning/pyproject.toml @@ -57,7 +57,7 @@ dependencies = [ "patchelf (>=0.17.2.4,<0.18.0.0)", "einops (>=0.8.2,<0.9.0)", "flashinfer-python (==0.6.4)", - "xgrammar (==0.1.25)", + "xgrammar (==0.1.32)", "llguidance (==0.7.29)", "jsonschema (>=4.26.0,<5.0.0)", "backoff (>=2.2.1,<3.0.0)",