Update app.py

app.py

@@ -163,6 +163,9 @@ def respond(
     client_llama = InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct")
     generate_kwargs = dict( max_new_tokens=2000, do_sample=True, stream=True, details=True, return_full_text=False )
 
+    for qas in history:
+        messages.append({"role": "user", "content": f"{str(msg[0])}"})
+    
     messages.append({"role": "user", "content": f'[SYSTEM]You are a helpful assistant with access to the following functions: \n {str(functions_metadata)}\n\nTo use these functions respond with:\n<functioncall> {{ "name": "function_name", "arguments": {{ "arg_1": "value_1", "arg_1": "value_1", ... }} }}  </functioncall>  [USER] {message} {vqa}'})
 
     response = client.chat_completion( messages, max_tokens=150)