#ML Agents, what can I use for AMD?
1 messages · Page 1 of 1 (latest)
Pytorch doesn't have cuda as a requirement, it can run on AMD cards and even CPU.
And regardless, you don't need to install anything extra. Just the right pytorch version.
Oh, fair enough. They mentioned installing it on the install guide so thought I needed it. Thanks a lot!
If you install it via pip, you probably need to provide the right args or url for a version that supports amd gpus(which I assume you have) if you want it to run on the gpu.
So what would be the simplest way for me to get this going? Ideally I want to just get up and running, quite a lot to pick up atm 😅
Follow the installation guides of ml agents package for starters. You can worry about it using the gpu or not when you have it working.
Sounds good. Miniconda being a bit of a struggle to get working but should be goochie
Miniconda is just for virtual environment, to avoid dirtying your PC env. You could probably skip it or use venv.
Okey, skipping that for now then. But you mentioned pip and extra steps? Would it be easier with libTorch or source instead?