add cuda

app.py

@@ -133,6 +133,7 @@ def encode_queries(dataset_name, postfix):
     
     encoded_embeds = []
     batch_size = 32
+    model = model.cuda()
     
     for start_idx in tqdm.tqdm(range(0, len(input_texts), batch_size), desc="Encoding queries"):
         batch_input_texts = input_texts[start_idx: start_idx + batch_size]
@@ -147,6 +148,7 @@ def encode_queries(dataset_name, postfix):
                 embeds = F.normalize(embeds, p=2, dim=-1)
                 encoded_embeds.append(embeds.float().cpu().numpy())
 
+    model = model.cpu()
     return np.concatenate(encoded_embeds, axis=0)