Update app.py

app.py

@@ -62,8 +62,8 @@ def calculate_eou(chat_ctx, session):
 def respond(
     message,
     history: list[tuple[str, str]],
-    max_tokens=256,
-    temperature=0.7,
+    max_tokens=1024,
+    temperature=0.6,
     top_p=0.95,
 ):
     messages = [{"role": "system", "content": os.environ.get("CHARACTER_DESC", "You are a helpful assistant.")}]
@@ -81,22 +81,23 @@ def respond(
         yield "[Wait... Keep typing...]"
         return
 
-    # Generate response incrementally and yield each token
-    accumulated_response = ""
-    # Corrected the chat completions method call
-    for chunk in qwen_client.chat.completions.create(
+    # Generate raw response without any processing
+    full_response = ""
+    stream = qwen_client.chat.completions.create(
         messages=messages,
         max_tokens=max_tokens,
         stream=True,
         temperature=temperature,
         top_p=top_p,
-    ):
-        token = chunk.choices[0].delta.content or ""
-        accumulated_response += token
-        yield accumulated_response  # Yield accumulated response for live updates
+    )
 
-    print(f"Final response: {accumulated_response}")
+    for chunk in stream:
+        token = chunk.choices[0].delta.content or ""
+        full_response += token
+        yield full_response  # Send raw unmodified response to Gradio
 
+    # This will match both console and Gradio output
+    print(f"Final response: {full_response}")
 # Create Gradio interface
 demo = gr.ChatInterface(
     respond,