Update app.py

app.py

@@ -20,21 +20,17 @@ def respond(
         if val[1]:
             messages.append({"role": "assistant", "content": val[1]})
 
+    # Add the user's current message to the conversation context
     messages.append({"role": "user", "content": message})
 
-    response = ""
+    # The pipeline expects a list of texts, so we will generate a response
+    input_text = "\n".join([msg["content"] for msg in messages if msg["role"] == "user"])
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+    # Generate the response
+    response = pipe(input_text, max_length=max_tokens, temperature=temperature, top_p=top_p)
 
-        response += token
-        yield response
+    # Return the generated response
+    yield response[0]['generated_text']
 
 
 """