Giters

LINCellularNeuroscience / VAME

Variational Animal Motion Embedding - A tool for time series embedding and clustering

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Compatibility with GeForce 30xx series GPUs

ssterrett opened this issue · comments

commented

I'm a new user setting up VAME on my desktop (OS: Linux Ubuntu 18.04 GPU: nvidia GeForce RTX 3070). Initially, when I tried to run the demo I received an error when running import vame:
GeForce RTX 3070 with CUDA capability sm_86 is not compatible with the current PyTorch installation
I resolved this issue by updating to Cuda 11 and Pytorch 1.8 nightly in my VAME environment by running the following command
conda install pytorch torchvision cudatoolkit=11 -c pytorch-nightly
I was then able to successfully run the full demo script. This is a closed issue, but I wanted to add as documentation for any other 30xx series users.