Is the Llama Guard model only capable of classification?

#16
by adventure2165 - opened

I am planning to introduce the Llama Guard model to protect my Llama model. So, I want to assign some tasks to Llama Guard, such as personal data de-identification. For example, converting [email protected] -> . However, Llama Guard has only been returning "safe" and "unsafe". Can it be fine-tuned to perform other tasks as well?

Sign up or log in to comment