Update app.py

app.py

@@ -23,7 +23,7 @@ rm_model = AutoModelForSequenceClassification.from_pretrained('OpenAssistant/rew
 def generate_text(usertitle, content, temperature, max_length, N=3):
     input_text = f"title: {usertitle}\ncontent: {content}"
     inputs = tokenizer(input_text, return_tensors='pt').to('cuda')
-    attention_mask = torch.ones(inputs.shape, dtype=torch.long)
+    attention_mask = torch.ones(inputs['input_ids'].shape, dtype=torch.long, device='cuda')
     generated_sequences = model.generate(inputs['input_ids'], attention_mask=attention_mask, temperature=temperature, max_length=max_length, pad_token_id=tokenizer.eos_token_id, num_return_sequences=N, do_sample=True)
     decoded_sequences = [tokenizer.decode(g, skip_special_tokens=True) for g in generated_sequences]