Fix token and temperature inputs

app.py

@@ -31,7 +31,7 @@ def show_model_change_info(model_name):
     return model_name
 
 
-async def respond(message, history, model_name="command-a-03-2025"):
+async def respond(message, history, model_name="command-a-03-2025", temperature=0.7, max_tokens=None):
     """Generate streaming response using Cohere API"""
     
     # Convert Gradio history format to API format
@@ -75,7 +75,9 @@ async def respond(message, history, model_name="command-a-03-2025"):
             user_message=message,
             conversation_history=conversation_history,
             api_key=api_key,
-            model_name=model_name
+            model_name=model_name,
+            temperature=temperature,
+            max_tokens=max_tokens
         ):
             partial_message += chunk
             yield partial_message
@@ -84,14 +86,33 @@ async def respond(message, history, model_name="command-a-03-2025"):
 
 with gr.Blocks() as demo:
     gr.Markdown("## Modular Chatbot")
+    
+    # State components to track current values
+    temperature_state = gr.State(value=0.7)
+    max_tokens_state = gr.State(value=None)
+    model_state = gr.State(value=COHERE_MODELS[0])
 
     with gr.Row():
         with gr.Column(scale=2):
+             # Define wrapper function after all components are created
+            async def chat_wrapper(message, history, model_val, temp_val, tokens_val):
+                # Use the state values directly
+                current_model = model_val if model_val else COHERE_MODELS[0]
+                current_temp = temp_val if temp_val is not None else 0.7
+                current_max_tokens = tokens_val
+                
+                # Stream the response
+                async for chunk in respond(message, history, current_model, current_temp, current_max_tokens):
+                    yield chunk
+
+            # Create chat interface using the wrapper with additional inputs
             chat_interface = gr.ChatInterface(
-                fn=respond,
+                fn=chat_wrapper,
                 type="messages",
-                save_history=True
+                save_history=True,
+                additional_inputs=[model_state, temperature_state, max_tokens_state]
             )
+
             with gr.Accordion("Chat Settings", elem_id="chat_settings_group"):
                 with gr.Row():
                     with gr.Column(scale=3):
@@ -126,6 +147,15 @@ with gr.Blocks() as demo:
                         outputs=[model]
                     )
                     
+                    # Update state when model changes
+                    model.change(
+                        fn=lambda x: x,
+                        inputs=[model],
+                        outputs=[model_state]
+                    )
+                    
+                    
+                    
                     with gr.Column(scale=1):
                         temperature = gr.Slider(
                             label="Temperature",
@@ -140,12 +170,30 @@ with gr.Blocks() as demo:
                         )
                         max_tokens = gr.Textbox(
                             label="Max Tokens",
-                            info="Higher values allow longer responses.",
+                            info="Higher values allow longer responses. Leave empty for default.",
                             value="8192",
                             elem_id="max_tokens_input",
                             interactive=True,
-                            show_label=False,
+                            show_label=True,
+                        )
+
+                        # Update state when temperature changes
+                        temperature.change(
+                            fn=lambda x: x,
+                            inputs=[temperature],
+                            outputs=[temperature_state]
+                        )
+                        
+                        # Update state when max_tokens changes
+                        max_tokens.change(
+                            fn=lambda x: int(x) if x and str(x).strip() else None,
+                            inputs=[max_tokens],
+                            outputs=[max_tokens_state]
                         )
+            
+           
+            
+            
         
 if __name__ == "__main__":
     demo.launch()