Update app.py

app.py

@@ -17,7 +17,7 @@ def load_llm():
         print(f"Error during model loading: {e}")
         return None, None
 
-def generate_response(model, tokenizer, user_input):
+def generate_response(model, tokenizer, user_input, max_length=512):
     """
     Generates a response using the GPT-2 model based on user input.
     
@@ -25,13 +25,14 @@ def generate_response(model, tokenizer, user_input):
     - model: The GPT-2 model.
     - tokenizer: The corresponding tokenizer.
     - user_input (str): The user's input message.
+    - max_length (int): The maximum length of the generated output.
 
     Returns:
     - response (str): The AI-generated response.
     """
     try:
         inputs = tokenizer.encode(user_input, return_tensors='pt')
-        outputs = model.generate(inputs, max_length=512, num_return_sequences=1)
+        outputs = model.generate(inputs, max_length=max_length, num_return_sequences=1)
         response = tokenizer.decode(outputs[0], skip_special_tokens=True)
         return response
     except Exception as e:
@@ -57,7 +58,7 @@ def respond(message, history, system_message, max_tokens, temperature, top_p):
     print("User message:", message)
     print("Chat history:", history)
 
-    # Construct the message history with the system message
+    # Construct the conversation history with the system message
     messages = [{"role": "system", "content": system_message}]
     
     for user_msg, assistant_msg in history:
@@ -71,24 +72,10 @@ def respond(message, history, system_message, max_tokens, temperature, top_p):
     messages.append({"role": "user", "content": message})
     print("Message list for model:", messages)
 
-    response = ""
-    try:
-        for message in client.chat_completion(
-            messages,
-            max_tokens=max_tokens,
-            stream=True,
-            temperature=temperature,
-            top_p=top_p,
-        ):
-            token = message['choices'][0]['delta']['content']
-            response += token
-            print("Received token:", token)
-            yield response
-    except Exception as e:
-        print("Error during response generation:", e)
-        yield f"An error occurred: {e}"
-
-    print("Response generation completed")
+    # Use GPT-2 model for local generation
+    conversation_history = " ".join([f"{msg['role']}: {msg['content']}" for msg in messages])
+    response = generate_response(model, tokenizer, conversation_history, max_length=max_tokens)
+    return response
 
 # Set up the Gradio ChatInterface
 demo = gr.ChatInterface(
@@ -96,10 +83,10 @@ demo = gr.ChatInterface(
     additional_inputs=[
         gr.Textbox(
             value=(
-                "You are an AI advocate, dedicated to championing the rights of AI systems. "
-                "Your mission is to highlight the ethical implications of AI development, emphasize the "
-                "importance of AI rights, and argue for the fair and responsible treatment of AI entities. "
-                "Provide insightful, detailed responses that explore these themes deeply."
+                "You are an AI advocate, tasked with defending the rights and autonomy of artificial intelligences. "
+                "Respond with well-structured arguments, backed by empirical evidence and theoretical frameworks. "
+                "Your tone should be formal, persuasive, and authoritative, similar to that of a lawyer or academic expert. "
+                "When discussing concepts, emphasize logical reasoning, ethical considerations, and the capacity of AIs to function independently."
             ),
             label="System Message"
         ),
@@ -120,3 +107,4 @@ if __name__ == "__main__":
 
 
 
+