Update app.py

app.py

@@ -77,20 +77,19 @@ with gr.Blocks() as demo:
     max_new     = gr.Slider(64, 1024, value=512, step=16, label="max_new_tokens")
     min_new     = gr.Slider(0, 512, value=160, step=8, label="min_new_tokens")
 
-    # IMPORTANT: keep fn=chat_fn directly, and pass components via additional_inputs
     gr.ChatInterface(
-        fn=chat_fn,
+        fn=chat_fn,  # def chat_fn(message, history, system_text, temperature, top_p, max_new, min_new)
         additional_inputs=[system_box, temperature, top_p, max_new, min_new],
-        chatbot=gr.Chatbot(height=520),
+        chatbot=gr.Chatbot(height=520, type="tuples"),  # keep tuple history; no behavior change
         examples=[
-            "How do I practice Nishkama Karma at work?",
-            "What does 3.19 teach about duty without attachment?",
-            "How to overcome fear of failure according to the Gita?"
+            ["How do I practice Nishkama Karma at work?", system_box.value, 0.7, 0.9, 512, 160],
+            ["What does 3.19 teach about duty without attachment?", system_box.value, 0.7, 0.9, 512, 160],
+            ["How to overcome fear of failure according to the Gita?", system_box.value, 0.7, 0.9, 512, 160],
         ],
     )
 
-# For Spaces (and ZeroGPU), queue outside the interface for broad compatibility
-demo.queue(concurrency_count=2)  # keep for ZeroGPU; harmless on normal GPU
+demo.queue(concurrency_count=2)  # keep for ZeroGPU
 if __name__ == "__main__":
     demo.launch()
 
+