Update app.py

app.py

@@ -32,7 +32,7 @@ def load_model_and_tokenizer():
         "meta-llama/Llama-2-7b-hf",
         use_auth_token=HUGGINGFACE_API_TOKEN,
         torch_dtype=torch.float16,
-        device_map="auto"
+        device_map="auto",  # This requires the `accelerate` package
     )
 
     return tokenizer, model