Update space

app.py

@@ -15,26 +15,29 @@ def respond(
     temperature,
     top_p,
 ):
-    # Format the conversation history into a prompt
-    prompt = f"{system_message}\n\n"
+    messages = [{"role": "system", "content": system_message}]
+
     for val in history:
         if val[0]:
-            prompt += f"User: {val[0]}\n"
+            messages.append({"role": "user", "content": val[0]})
         if val[1]:
-            prompt += f"Assistant: {val[1]}\n"
-    prompt += f"User: {message}\nAssistant:"
+            messages.append({"role": "assistant", "content": val[1]})
+
+    messages.append({"role": "user", "content": message})
 
     response = ""
     
-    # Use text generation instead of chat completion
-    for message in client.text_generation(
-        prompt,
-        max_new_tokens=max_tokens,
+    # Use chat completion instead of text generation
+    for message in client.chat_completion(
+        messages,
+        max_tokens=max_tokens,
         stream=True,
         temperature=temperature,
         top_p=top_p,
     ):
-        response += message
+        token = message.choices[0].delta.content
+        if token:
+            response += token
         yield response