I am trying to install torch 2.5.0 for CPU-only use on Python 3.13. I am facing some issues torchvision as I attempted to install versions from 0.19 to 0.22, but I still got a hard failure. I would like to avoid buildingtorchvision from scratch. As it depends on many dependencies that are, in my case, fixed (e.g., NumPy 2.1.3 and SciPy 1.14.1). Something like:
| torch | vision | python | numpy |
|---|---|---|---|
| 2.5.0 | … | >=3.10 & <=3.13 | … |
| 2.4.0 |
My question: Is there a compatibility matrix for Torch, Vision, and Python versions that is up-to-date (2023 and 2023b) and backed up by the Torch community?
Thanks.