Update app.py

app.py

@@ -46,9 +46,10 @@ Assistant: girlfriend;mother;father;friend
 
 # setup
 torch.set_grad_enabled(False)
+device = "cpu"
 model_name = "TheBloke/OpenHermes-2.5-Mistral-7B-GPTQ"
 token = os.environ['hf_token']
-pipe = pipeline("text-generation", model=model_name, device="cuda")
+pipe = pipeline("text-generation", model=model_name, device=device)
 pipe.tokenizer.chat_template = chatml_template # TheBloke says this is the right template for this model
 generate_kwargs = {'max_new_tokens': 20}