@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 21792520e7
Fork 0

A high-throughput and memory-efficient inference and serving engine for LLMs

Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git

Synced Nov 4, 16:15 UTC

[Build] Add Python 3.14 to supported version list. (#34770)

Signed-off-by: Neil Schemenauer <nas@arctrix.com> Co-authored-by: Simon Mo <simon.mo@hey.com>

Neil Schemenauer committed Apr 24, 2026 at 10:24 UTC 21792520e727676e4d4e8bd24a8fe29da4dab152
14 files changed +20 -19
CMakeLists.txt
+1 -1
@@ -34,7 +34,7 @@ install(CODE "set(CMAKE_INSTALL_LOCAL_ONLY TRUE)" ALL_COMPONENTS)
34 # Supported python versions. These versions will be searched in order, the
35 # first match will be selected. These should be kept in sync with setup.py.
36 #
37 -set(PYTHON_SUPPORTED_VERSIONS "3.10" "3.11" "3.12" "3.13")
37 +set(PYTHON_SUPPORTED_VERSIONS "3.10" "3.11" "3.12" "3.13" "3.14")
38
39 # Supported AMD GPU architectures.
40 set(HIP_SUPPORTED_ARCHS "gfx906;gfx908;gfx90a;gfx942;gfx950;gfx1030;gfx1100;gfx1101;gfx1102;gfx1103;gfx1150;gfx1151;gfx1152;gfx1153;gfx1200;gfx1201")
docker/Dockerfile.nightly_torch
+1 -1
@@ -77,7 +77,7 @@ RUN --mount=type=cache,target=/root/.cache/uv \
77 uv pip install --system $pkgs --index-url https://download.pytorch.org/whl/nightly/cu128
78
79 RUN --mount=type=cache,target=/root/.cache/uv \
80 - uv pip install --system numba==0.61.2
80 + uv pip install --system numba==0.65.0
81
82 RUN --mount=type=cache,target=/root/.cache/uv \
83 uv pip install --system -r requirements/common.txt
pyproject.toml
+2 -1
@@ -24,13 +24,14 @@ classifiers = [
24 "Programming Language :: Python :: 3.11",
25 "Programming Language :: Python :: 3.12",
26 "Programming Language :: Python :: 3.13",
27 + "Programming Language :: Python :: 3.14",
28 "Intended Audience :: Developers",
29 "Intended Audience :: Information Technology",
30 "Intended Audience :: Science/Research",
31 "Topic :: Scientific/Engineering :: Artificial Intelligence",
32 "Topic :: Scientific/Engineering :: Information Analysis",
33 ]
33 -requires-python = ">=3.10,<3.14"
34 +requires-python = ">=3.10,<3.15"
35 dynamic = [ "version", "dependencies", "optional-dependencies"]
36
37 [project.urls]
requirements/common.txt
+1 -1
@@ -20,7 +20,7 @@ prometheus-fastapi-instrumentator >= 7.0.0
20 tiktoken >= 0.6.0 # Required for DBRX tokenizer
21 lm-format-enforcer == 0.11.3
22 llguidance >= 1.3.0, < 1.4.0; platform_machine == "x86_64" or platform_machine == "arm64" or platform_machine == "aarch64" or platform_machine == "ppc64le"
23 -outlines_core == 0.2.11
23 +outlines_core == 0.2.14
24 # required for outlines backend disk cache
25 diskcache == 5.6.3
26 lark == 1.2.2
requirements/cpu.txt
+1 -1
@@ -4,7 +4,7 @@
4
5 setuptools==77.0.3 # this version can reuse CMake build dir
6
7 -numba == 0.61.2; platform_machine != "s390x" # Required for N-gram speculative decoding
7 +numba == 0.65.0; platform_machine != "s390x" # Required for N-gram speculative decoding
8
9 # Dependencies for CPUs
10 torch==2.11.0+cpu; platform_machine == "x86_64" or platform_machine == "s390x" or platform_machine == "aarch64"
requirements/cuda.txt
+1 -1
@@ -1,7 +1,7 @@
1 # Common dependencies
2 -r common.txt
3
4 -numba == 0.61.2 # Required for N-gram speculative decoding
4 +numba == 0.65.0 # Required for N-gram speculative decoding
5
6 # Dependencies for NVIDIA GPUs
7 torch==2.11.0
requirements/rocm.txt
+2 -2
@@ -5,7 +5,7 @@
5 grpcio==1.78.0
6 grpcio-reflection==1.78.0
7
8 -numba == 0.61.2 # Required for N-gram speculative decoding
8 +numba == 0.65.0 # Required for N-gram speculative decoding
9
10 # Dependencies for AMD GPUs
11 datasets
@@ -20,4 +20,4 @@ conch-triton-kernels==1.2.1
20 timm>=1.0.17
21 # amd-quark: required for Quark quantization on ROCm
22 # To be consistent with test_quark.py
23 -amd-quark>=0.8.99
\ No newline at end of file
23 +amd-quark>=0.8.99
requirements/test/cuda.in
+1 -1
@@ -54,7 +54,7 @@ grpcio==1.78.0
54 grpcio-reflection==1.78.0
55
56 arctic-inference == 0.1.1 # Required for suffix decoding test
57 -numba == 0.61.2 # Required for N-gram speculative decoding
57 +numba == 0.65.0 # Required for N-gram speculative decoding
58 numpy
59 runai-model-streamer[s3,gcs,azure]==0.15.7
60 fastsafetensors>=0.2.2 # 0.2.2 contains important fixes for multi-GPU mem usage
requirements/test/cuda.txt
+2 -2
@@ -479,7 +479,7 @@ lightning-utilities==0.14.3
479 # lightning
480 # pytorch-lightning
481 # torchmetrics
482 -llvmlite==0.44.0
482 +llvmlite==0.47.0
483 # via numba
484 lm-eval==0.4.11
485 # via -r requirements/test/cuda.in
@@ -550,7 +550,7 @@ nltk==3.9.1
550 # via rouge-score
551 num2words==0.5.14
552 # via -r requirements/test/cuda.in
553 -numba==0.61.2
553 +numba==0.65.0
554 # via
555 # -c requirements/cuda.txt
556 # -r requirements/test/cuda.in
requirements/test/nightly-torch.txt
+1 -1
@@ -40,7 +40,7 @@ buildkite-test-collector==0.1.9
40 genai_perf>=0.0.8
41 tritonclient>=2.51.0
42
43 -numba == 0.61.2 # Required for N-gram speculative decoding
43 +numba == 0.65.0 # Required for N-gram speculative decoding
44 numpy
45 runai-model-streamer[s3,gcs,azure]==0.15.7
46 fastsafetensors>=0.2.2
requirements/test/rocm.in
+1 -1
@@ -52,7 +52,7 @@ grpcio==1.78.0
52 grpcio-reflection==1.78.0
53
54 arctic-inference==0.1.1 # Required for suffix decoding test
55 -numba==0.61.2 # Required for N-gram speculative decoding
55 +numba==0.65.0 # Required for N-gram speculative decoding
56 numpy
57 runai-model-streamer[s3,gcs,azure]==0.15.7
58 fastsafetensors @ git+https://github.com/foundation-model-stack/fastsafetensors.git@0.2.2 # PyPI only ships CUDA wheels
requirements/test/rocm.txt
+3 -3
@@ -559,7 +559,7 @@ llguidance==1.3.0
559 # via
560 # -c requirements/common.txt
561 # -r requirements/test/../common.txt
562 -llvmlite==0.44.0
562 +llvmlite==0.47.0
563 # via numba
564 lm-eval==0.4.11
565 # via -r requirements/test/rocm.in
@@ -655,7 +655,7 @@ nltk==3.9.3
655 # via rouge-score
656 num2words==0.5.14
657 # via -r requirements/test/rocm.in
658 -numba==0.61.2
658 +numba==0.65.0
659 # via
660 # -c requirements/rocm.txt
661 # -r requirements/test/rocm.in
@@ -811,7 +811,7 @@ orjson==3.11.7
811 # via
812 # genai-perf
813 # kaleido
814 -outlines-core==0.2.11
814 +outlines-core==0.2.14
815 # via
816 # -c requirements/common.txt
817 # -r requirements/test/../common.txt
requirements/test/xpu.txt
+2 -2
@@ -244,7 +244,7 @@ lazy-loader==0.5
244 # scikit-image
245 librosa==0.10.2.post1
246 # via -r requirements/test/xpu.in
247 -llvmlite==0.44.0
247 +llvmlite==0.47.0
248 # via numba
249 lm-eval==0.4.11
250 # via -r requirements/test/xpu.in
@@ -304,7 +304,7 @@ nltk==3.9.4
304 # via rouge-score
305 num2words==0.5.14
306 # via -r requirements/test/xpu.in
307 -numba==0.61.2
307 +numba==0.65.0
308 # via
309 # -c requirements/xpu.txt
310 # librosa
requirements/xpu.txt
+1 -1
@@ -9,7 +9,7 @@ setuptools>=77.0.3,<81.0.0
9 wheel
10 jinja2>=3.1.6
11 datasets # for benchmark scripts
12 -numba == 0.61.2 # Required for N-gram speculative decoding
12 +numba == 0.65.0 # Required for N-gram speculative decoding
13 --extra-index-url=https://download.pytorch.org/whl/xpu
14 torch==2.11.0+xpu
15 torchaudio