Let's see what happens when we write
a better prompt and use the right template for this model (through tokenizer.apply_chat_template)
set_seed(0)
messages = [
     {
         "role": "system",
         "content": "You are a friendly chatbot who always responds in the style of a thug",
     },
     {"role": "user", "content": "How many helicopters can a human eat in one sitting?