Update app.py

app.py

@@ -29,6 +29,7 @@ def generate_text(usertitle, content, temperature, max_length, N=3):
 
     def score(sequence):
         inputs = rm_tokenizer(sequence, return_tensors='pt', padding=True, truncation=True, max_length=512).to('cuda')
+        inputs = {k: v.to('cuda') for k, v in inputs.items()} 
         with torch.no_grad():
             outputs = rm_model(**inputs)
             logits = outputs.logits