Update app.py

app.py

@@ -98,13 +98,14 @@ def predict_chat(message: str, history: list):
                 prompt_input += f"Assistant: {msg['content']}\n"
         prompt_input += "Assistant:"
 
+        # FIXED: Use max_tokens instead of max_new_tokens for ctransformers
         for token in model.generate(
             prompt_input,
-            max_new_tokens=MAX_NEW_TOKENS,
+            max_tokens=MAX_NEW_TOKENS,  # Changed from max_new_tokens
             temperature=TEMPERATURE,
             top_k=TOP_K,
             top_p=TOP_P,
-            do_sample=DO_SAMPLE,
+            sample=DO_SAMPLE,  # Changed from do_sample
             repetition_penalty=1.1,
             stop=["User:", "\nUser", "\n#", "\n##", "<|endoftext|>"] 
         ):
@@ -168,4 +169,4 @@ if __name__ == "__main__":
 
     demo.chatbot.value = initial_messages_for_value 
 
-    demo.launch()
+    demo.launch()