Update app.py

app.py

@@ -4,9 +4,10 @@ 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"
     )
 
@@ -17,57 +18,56 @@ MODELS = {
     "DeepSeek-R1-Zero": load_hf_model("DeepSeek-R1-Zero")
 }
 
-def parse_response(response):
-    """Handle different response formats from various models"""
-    if isinstance(response, list):
-        if len(response) > 0:
-            # Handle list of messages format
-            return response[0].get('generated_text', 
-                   response[0].get('content', str(response[0])))
-    elif isinstance(response, dict):
-        # Handle OpenAI-style format
-        if 'choices' in response:
-            return response['choices'][0]['message']['content']
-        # Handle standard text generation format
-        elif 'generated_text' in response:
-            return response['generated_text']
-    return f"Unsupported response format: {type(response)}"
-
+# --- Chatbot function ---
 def chatbot(input_text, history, model_choice, system_message, max_new_tokens, temperature, top_p):
     history = history or []
     
+    # Get the selected model component
     model_component = MODELS[model_choice]
     
-    # Construct messages with optional system message
-    messages = []
-    if system_message.strip():
-        messages.append({"role": "system", "content": system_message})
-    messages.append({"role": "user", "content": input_text})
-    
+    # Create payload for the model
     payload = {
-        "messages": messages,
+        "messages": [{"role": "user", "content": input_text}],
+        "system": system_message,
         "max_tokens": max_new_tokens,
         "temperature": temperature,
         "top_p": top_p
     }
     
+    # Run inference using the selected model
     try:
-        response = model_component(payload)
-        assistant_response = parse_response(response)
+        response = model_component(payload)  # The response is likely a dictionary
+        if isinstance(response, dict) and "choices" in response:
+            # Assuming the response structure is similar to OpenAI's API
+            assistant_response = response["choices"][0]["message"]["content"]
+        elif isinstance(response, dict) and "generated_text" in response:
+            # If the response is in a different format, adjust accordingly
+            assistant_response = response["generated_text"]
+        else:
+            assistant_response = "Unexpected model response format."
     except Exception as e:
         assistant_response = f"Error: {str(e)}"
     
+    # Append user and assistant messages to history
     history.append((input_text, assistant_response))
+    
     return history, history, ""
 
-# Interface setup remains the same
+# --- Gradio Interface ---
 with gr.Blocks(theme=gr.themes.Soft(), title="DeepSeek Chatbot") as demo:
-    gr.Markdown("""# DeepSeek Chatbot""")
+    gr.Markdown(
+        """
+        # DeepSeek Chatbot
+        Created by [ruslanmv.com](https://ruslanmv.com/)
+        This is a demo of different DeepSeek models. Select a model, type your message, and click "Submit".
+        You can also adjust optional parameters like system message, max new tokens, temperature, and top-p.
+        """
+    )
 
     with gr.Row():
         with gr.Column():
-            chatbot_output = gr.Chatbot(height=500)
-            msg = gr.Textbox(placeholder="Type your message...")
+            chatbot_output = gr.Chatbot(label="DeepSeek Chatbot", height=500)
+            msg = gr.Textbox(label="Your Message", placeholder="Type your message here...")
             with gr.Row():
                 submit_btn = gr.Button("Submit", variant="primary")
                 clear_btn = gr.ClearButton([msg, chatbot_output])
@@ -76,19 +76,28 @@ with gr.Blocks(theme=gr.themes.Soft(), title="DeepSeek Chatbot") as demo:
         with gr.Accordion("Options", open=True):
             model_choice = gr.Radio(
                 choices=list(MODELS.keys()),
+                label="Choose a Model",
                 value="DeepSeek-R1"
             )
             with gr.Accordion("Optional Parameters", open=False):
                 system_message = gr.Textbox(
-                    value="You are a helpful AI assistant",
-                    lines=2
+                    label="System Message",
+                    value="You are a friendly Chatbot created by ruslanmv.com",
+                    lines=2,
+                )
+                max_new_tokens = gr.Slider(
+                    minimum=1, maximum=4000, value=200, label="Max New Tokens"
+                )
+                temperature = gr.Slider(
+                    minimum=0.10, maximum=4.00, value=0.70, label="Temperature"
+                )
+                top_p = gr.Slider(
+                    minimum=0.10, maximum=1.00, value=0.90, label="Top-p (nucleus sampling)"
                 )
-                max_new_tokens = gr.Slider(1, 4000, 200)
-                temperature = gr.Slider(0.1, 4.0, 0.7)
-                top_p = gr.Slider(0.1, 1.0, 0.9)
 
     chat_history = gr.State([])
 
+    # Event handling
     submit_btn.click(
         chatbot,
         [msg, chat_history, model_choice, system_message, max_new_tokens, temperature, top_p],