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 15, 01:00 UTC
Update PyTorch to `torch==2.7.1` for CUDA (#21011)
Signed-off-by: mgoin <mgoin64@gmail.com>
Michael Goin committed
Jul 16, 2025 at 22:30 UTC
4e7dfbe7b49a3386e3212209fcc093c0c166ba95
7 files changed
+17
-17
CMakeLists.txt
+1
-1
@@ -45,7 +45,7 @@ set(HIP_SUPPORTED_ARCHS "gfx906;gfx908;gfx90a;gfx942;gfx950;gfx1030;gfx1100;gfx1
45
# requirements.txt files and should be kept consistent. The ROCm torch
46
# versions are derived from docker/Dockerfile.rocm
47
#
48
-set(TORCH_SUPPORTED_VERSION_CUDA "2.7.0")
48
+set(TORCH_SUPPORTED_VERSION_CUDA "2.7.1")
49
set(TORCH_SUPPORTED_VERSION_ROCM "2.7.0")
50
51
#
pyproject.toml
+1
-1
@@ -6,7 +6,7 @@ requires = [
6
"packaging>=24.2",
7
"setuptools>=77.0.3,<80.0.0",
8
"setuptools-scm>=8.0",
9
- "torch == 2.7.0",
9
+ "torch == 2.7.1",
10
"wheel",
11
"jinja2",
12
]
requirements/build.txt
+1
-1
@@ -4,7 +4,7 @@ ninja
4
packaging>=24.2
5
setuptools>=77.0.3,<80.0.0
6
setuptools-scm>=8
7
-torch==2.7.0
7
+torch==2.7.1
8
wheel
9
jinja2>=3.1.6
10
regex
requirements/cuda.txt
+5
-5
@@ -6,9 +6,9 @@ numba == 0.61.2; python_version > '3.9'
6
7
# Dependencies for NVIDIA GPUs
8
ray[cgraph]>=2.43.0, !=2.44.* # Ray Compiled Graph, required for pipeline parallelism in V1.
9
-torch==2.7.0
10
-torchaudio==2.7.0
9
+torch==2.7.1
10
+torchaudio==2.7.1
11
# These must be updated alongside torch
12
-torchvision==0.22.0 # Required for phi3v processor. See https://github.com/pytorch/vision?tab=readme-ov-file#installation for corresponding version
13
-# https://github.com/facebookresearch/xformers/releases/tag/v0.0.30
14
-xformers==0.0.30; platform_system == 'Linux' and platform_machine == 'x86_64' # Requires PyTorch >= 2.7
12
+torchvision==0.22.1 # Required for phi3v processor. See https://github.com/pytorch/vision?tab=readme-ov-file#installation for corresponding version
13
+# https://github.com/facebookresearch/xformers/releases/tag/v0.0.31
14
+xformers==0.0.31; platform_system == 'Linux' and platform_machine == 'x86_64' # Requires PyTorch >= 2.7
requirements/test.in
+3
-3
@@ -22,9 +22,9 @@ sentence-transformers # required for embedding tests
22
soundfile # required for audio tests
23
jiwer # required for audio tests
24
timm # required for internvl test
25
-torch==2.7.0
26
-torchaudio==2.7.0
27
-torchvision==0.22.0
25
+torch==2.7.1
26
+torchaudio==2.7.1
27
+torchvision==0.22.1
28
transformers_stream_generator # required for qwen-vl test
29
mamba_ssm # required for plamo2 test
30
matplotlib # required for qwen-vl test
requirements/test.txt
+4
-4
@@ -762,7 +762,7 @@ tomli==2.2.1
762
# via schemathesis
763
tomli-w==1.2.0
764
# via schemathesis
765
-torch==2.7.0+cu128
765
+torch==2.7.1+cu128
766
# via
767
# -r requirements/test.in
768
# accelerate
@@ -781,12 +781,12 @@ torch==2.7.0+cu128
781
# torchvision
782
# vector-quantize-pytorch
783
# vocos
784
-torchaudio==2.7.0+cu128
784
+torchaudio==2.7.1+cu128
785
# via
786
# -r requirements/test.in
787
# encodec
788
# vocos
789
-torchvision==0.22.0+cu128
789
+torchvision==0.22.1+cu128
790
# via
791
# -r requirements/test.in
792
# timm
@@ -816,7 +816,7 @@ transformers==4.53.2
816
# transformers-stream-generator
817
transformers-stream-generator==0.0.5
818
# via -r requirements/test.in
819
-triton==3.3.0
819
+triton==3.3.1
820
# via torch
821
tritonclient==2.51.0
822
# via
tests/entrypoints/openai/test_vision.py
+2
-2
@@ -36,11 +36,11 @@ EXPECTED_MM_BEAM_SEARCH_RES = [
36
],
37
[
38
"The image shows a Venn diagram with three over",
39
- "This image shows a Venn diagram with three over",
39
+ "The image shows a Venn diagram with three intersect",
40
],
41
[
42
"This image displays a gradient of colors ranging from",
43
- "This image displays a gradient of colors transitioning from",
43
+ "The image displays a gradient of colors ranging from",
44
],
45
]
46