ALERT benchmark assesses safety risks in language models to improve their responses.
― 4 min read
Huu Nguyen
RedPajama datasets aim to enhance language model training through transparency and quality data.
― 5 min read
M-ALERT tests language models for safety across five languages.
― 5 min read