search

amazon-science / RefChecker

RefChecker provides automatic checking pipeline and benchmark dataset for detecting fine-grained hallucinations generated by Large Language Models.
Apache License 2.0
226 stars 17 forks source link

How to run the demo on a machine with no NVIDIA GPUs? #19

Open saeednj opened 2 months ago

saeednj commented 2 months ago

Hello,

I was trying to use your repo for some fact checking tasks.

I followed the quick start guide, but when I want to use the demo app (demo/main.py), it fails with the assertion that CUDA is not enabled. On may Mac it says AssertionError: Torch not compiled with CUDA enabled, and on my Linux machine it says RuntimeError: Found no NVIDIA driver on your system..

None of my machines have a NVIDIA GPU. Is there a way to run your system on CPU only? Does it work if I change the requirements to Torch-cpu or there are other GPU requirements as well?

Thanks

alexlorenzo commented 6 days ago
  1. Force the device = torch.device('cpu')
  2. Remove all .to(0) that is used for GPU only It should work (at least it work on my Mac)