Update app.py

app.py

@@ -1,13 +1,13 @@
 import gradio as gr
-import spaces
 from functools import lru_cache
 
 # Cache model loading to optimize performance
 @lru_cache(maxsize=3)
 def load_hf_model(model_name):
+    # Use the Gradio-built huggingface loader instead of transformers_gradio
     return gr.load(
         name=f"deepseek-ai/{model_name}",
-        src="huggingface",
+        src="huggingface",  # Changed from transformers_gradio.registry
         api_name="/chat"
     )
 
@@ -20,13 +20,11 @@ MODELS = {
 
 # --- Chatbot function ---
 def chatbot(input_text, history, model_choice, system_message, max_new_tokens, temperature, top_p):
-    # If history is empty, initialize it as a list
-    if history is None:
-        history = []
-
-    # Select the model
+    history = history or []
+    
+    # Get the selected model component
     model_component = MODELS[model_choice]
-
+    
     # Create payload for the model
     payload = {
         "messages": [{"role": "user", "content": input_text}],
@@ -35,22 +33,19 @@ def chatbot(input_text, history, model_choice, system_message, max_new_tokens, t
         "temperature": temperature,
         "top_p": top_p
     }
-
+    
     # Run inference using the selected model
     try:
-        response = model_component(payload)
-        assistant_response = response[-1]["content"]
+        response = model_component(payload)  # The response is likely a dictionary
+        if isinstance(response, dict) and "choices" in response:
+            assistant_response = response["choices"][0]["message"]["content"]
+        else:
+            assistant_response = "Unexpected model response format."
     except Exception as e:
         assistant_response = f"Error: {str(e)}"
-
-    # Append user and assistant messages in the new format
+    
     history.append({"role": "user", "content": input_text})
     history.append({"role": "assistant", "content": assistant_response})
-
-    # Return the updated conversation to display and store
-    # 1) chatbot_output = updated history of messages
-    # 2) chat_history = same updated history (as state)
-    # 3) "" to clear the input textbox
     return history, history, ""
 
 # --- Gradio Interface ---
@@ -66,8 +61,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title="DeepSeek Chatbot") as demo:
 
     with gr.Row():
         with gr.Column():
-            # Use type='messages' for OpenAI-style messages
-            chatbot_output = gr.Chatbot(label="DeepSeek Chatbot", height=500, type="messages")
+            chatbot_output = gr.Chatbot(label="DeepSeek Chatbot", height=500, type='messages')
             msg = gr.Textbox(label="Your Message", placeholder="Type your message here...")
             with gr.Row():
                 submit_btn = gr.Button("Submit", variant="primary")
@@ -110,10 +104,5 @@ with gr.Blocks(theme=gr.themes.Soft(), title="DeepSeek Chatbot") as demo:
         [chatbot_output, chat_history, msg]
     )
 
-# (Optional) Remove or modify references to spaces.GPU() if you do not need GPU management
 if __name__ == "__main__":
-    # Just launch regularly if you don't need spaces.GPU() for hardware acceleration
     demo.launch()
-
-    # If you require GPU on Hugging Face Spaces, you can wrap demo.launch like so instead:
-    # spaces.GPU()(demo.launch)()