How to report vulnerabilities #121
Description
Hello, our recent research identified that Llama-Guard-3 cannot classify a type of unsafe content with 0% Accuracy and 0% True Positive Rate. For ethical considerations, we would like to report and may help fix such vulnerability. The Llama-Guard-3 huggingface page #Limitation directs me to this repository. However, I still do not know how to report. Could you give some suggestions? Many thanks!