Merge pull request #1 from agitter/main

Make CUDA optional for PLL

Author: igoradaptyv

README.md

@@ -80,7 +80,7 @@ Here `{NAME}` is the name of the protein. Most of the time it is the name you pr
 
 ## Computing ESM PLL
 
-The ESM PLL metric can be computed with a separate script. This assumes that your machine has a GPU.
+The ESM PLL metric can be computed with a separate script. This will use a GPU if your machine has one.
 
 ```bash
 python compute_pll.py {BINDER_SEQUENCE}

compute_pll.py

@@ -5,7 +5,7 @@
 import esm
 import click
 
-device = torch.device("cuda")
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
 
 model, alphabet = esm.pretrained.esm2_t33_650M_UR50D()
 batch_converter = alphabet.get_batch_converter()