bugfixes, update response streaming, thought

src/app.py

@@ -17,10 +17,12 @@ try:
     import spaces
     @spaces.GPU(duration=60)
     def run_respond(*args, **kwargs):
-        return respond(*args, **kwargs)
+        for token in respond(*args, **kwargs):
+            yield token
 except ImportError:
     def run_respond(*args, **kwargs):
-        return respond(*args, **kwargs)
+        for token in respond(*args, **kwargs):
+            yield token
 
 def respond(
     message,
@@ -102,7 +104,7 @@ def respond(
 
 # Create Gradio interface
 demo = gr.ChatInterface(
-    respond,
+    run_respond,
     additional_inputs=[
         gr.Dropdown(
             choices=["Qwen3-14B", "Qwen3-8B", "Qwen3-0.6B"],