# llama2 70b chat was instruct-tuned in a way where the model itself refuses some requests. try other models or different prompts