Failed to install vision based on python 3.13t(free-threaded) on Windows OS

[jameszhouyi]

🐛 Describe the bug

Reproduce steps:
[Windows 11 OS]
conda create -n nogil2 --override-channels -c conda-forge python-freethreading
conda activate nogil2
pip install torch torchvision torchaudio --pre --index-url https://download.pytorch.org/whl/nightly/cu128

ERROR: Cannot install torchvision==0.22.0.dev20250226+cu128, torchvision==0.22.0.dev20250227+cu128, torchvision==0.22.0.dev20250228+cu128, torchvision==0.22.0.dev20250301+cu128, torchvision==0.22.0.dev20250302+cu128, torchvision==0.22.0.dev20250303+cu128, torchvision==0.22.0.dev20250304+cu128, torchvision==0.22.0.dev20250306+cu128, torchvision==0.22.0.dev20250307+cu128, torchvision==0.22.0.dev20250308+cu128, torchvision==0.22.0.dev20250309+cu128, torchvision==0.22.0.dev20250310+cu128, torchvision==0.22.0.dev20250311+cu128 and torchvision==0.22.0.dev20250312+cu128 because these package versions have conflicting dependencies.

The conflict is caused by:
torchvision 0.22.0.dev20250312+cu128 depends on numpy
torchvision 0.22.0.dev20250311+cu128 depends on numpy
torchvision 0.22.0.dev20250310+cu128 depends on numpy
torchvision 0.22.0.dev20250309+cu128 depends on numpy
torchvision 0.22.0.dev20250308+cu128 depends on numpy
torchvision 0.22.0.dev20250307+cu128 depends on numpy
torchvision 0.22.0.dev20250306+cu128 depends on numpy
torchvision 0.22.0.dev20250304+cu128 depends on numpy
torchvision 0.22.0.dev20250303+cu128 depends on numpy
torchvision 0.22.0.dev20250302+cu128 depends on numpy
torchvision 0.22.0.dev20250301+cu128 depends on numpy
torchvision 0.22.0.dev20250228+cu128 depends on numpy
torchvision 0.22.0.dev20250227+cu128 depends on numpy
torchvision 0.22.0.dev20250226+cu128 depends on numpy

To fix this you could try to:

loosen the range of package versions you've specified
remove package versions to allow pip to attempt to solve the dependency conflict
ERROR: ResolutionImpossible: for help visit https://pip.pypa.io/en/latest/topics/dependency-resolution/#dealing-with-dependency-conflicts

Versions

Versions
Collecting environment information...
PyTorch version: N/A
Is debug build: N/A
CUDA used to build PyTorch: N/A
ROCM used to build PyTorch: N/A

OS: Microsoft Windows Server 2022 Datacenter Evaluation (10.0.20348 64-bit)
GCC version: Could not collect
Clang version: Could not collect
CMake version: Could not collect
Libc version: N/A

Python version: 3.13.2 experimental free-threading build | packaged by conda-forge | (main, Feb 17 2025, 13:52:36) [MSC v.1942 64 bit (AMD64)] (64-bit runtime)
Python platform: Windows-2022Server-10.0.20348-SP0
Is CUDA available: N/A
CUDA runtime version: Could not collect
CUDA_MODULE_LOADING set to: N/A
GPU models and configuration: Could not collect
Nvidia driver version: Could not collect
cuDNN version: Could not collect
HIP runtime version: N/A
MIOpen runtime version: N/A
Is XNNPACK available: N/A

CPU:
Name: Intel(R) Xeon(R) Gold 6240 CPU @ 2.60GHz
Manufacturer: GenuineIntel
Family: 179
Architecture: 9
ProcessorType: 3
DeviceID: CPU0
CurrentClockSpeed: 2594
MaxClockSpeed: 2594
L2CacheSize: 18432
L2CacheSpeed: None
Revision: 21767
Name: Intel(R) Xeon(R) Gold 6240 CPU @ 2.60GHz
Manufacturer: GenuineIntel
Family: 179
Architecture: 9
ProcessorType: 3
DeviceID: CPU1
CurrentClockSpeed: 2594
MaxClockSpeed: 2594
L2CacheSize: 18432
L2CacheSpeed: None
Revision: 21767

Versions of relevant libraries:
[pip3] No relevant packages
[conda] No relevant packages

[NicolasHug]

Hi @jameszhouyi , torchvision doesn't support 3.13t yet

[jameszhouyi]

hi @NicolasHug Thanks for your response.