Update app.py

app.py

@@ -128,11 +128,10 @@ def render_latex(text):
     return text
 
 def respond(message, history, system_message, max_tokens, temperature, top_p):
-    """Non-streaming response for stability"""
+    """Streaming response with yield"""
     client = InferenceClient(model="Qwen/Qwen2.5-Math-7B-Instruct")
     
     messages = [{"role": "system", "content": system_message}]
-    # Iterate over history dicts and add user/assistant pairs
     for msg in history:
         if msg["role"] == "user":
             messages.append({"role": "user", "content": msg["content"]})
@@ -140,17 +139,21 @@ def respond(message, history, system_message, max_tokens, temperature, top_p):
             messages.append({"role": "assistant", "content": msg["content"]})
     messages.append({"role": "user", "content": message})
     
+    response = ""
     try:
-        completion = client.chat_completion(
+        for chunk in client.chat_completion(
             messages,
             max_tokens=max_tokens,
             temperature=temperature,
             top_p=top_p,
-        )
-        response = completion.choices[0].message.content
-        return render_latex(response)
+            stream=True  # Enable streaming
+        ):
+            if chunk.choices[0].delta.content:
+                response += chunk.choices[0].delta.content
+                yield render_latex(response)  # Yield progressively
+                
     except Exception as e:
-        return f"❌ Error: {str(e)[:100]}... Try a simpler problem."
+        yield f"❌ Error: {str(e)[:100]}..."
 
 def get_random_sample():
     """Get a random sample problem - loads datasets if needed"""
@@ -199,12 +202,13 @@ with gr.Blocks(title="🧮 Mathetics AI") as demo:
     )
     
     def chat_response(message, history):
-        """Updated to use dict-based history for type='messages'."""
-        bot_response = respond(message, history, create_math_system_message(), 1024, 0.3, 0.85)
-        # Append as dicts, not tuples
-        history.append({"role": "user", "content": message})
-        history.append({"role": "assistant", "content": bot_response})
-        return history, ""
+    """Streaming chat response"""
+    history.append({"role": "user", "content": message})
+    history.append({"role": "assistant", "content": ""})
+    
+    for partial_response in respond(message, history[:-1], create_math_system_message(), 1024, 0.3, 0.85):
+        history[-1]["content"] = partial_response
+        yield history, ""
     
     def clear_chat():
         """Clear the chat history and textbox."""