NVIDIA NemoGuard JailbreakDetect NIM
About the Microservice
The microservice classifies jailbreak attempts from a user-supplied prompt.
The microservice uses the Ardennes model. The model is a random forest that is trained by NVIDIA on the Snowflake Arctic Embed M Long model. The model is the best-known, publicly-available model for detecting LLM jailbreak attempts.
License
Related Information
-
Improved Large Language Model Jailbreak Detection via Pretrained Embeddings, arXiv:2412.01547.
-
Snowflake/snowflak-arctic-embed-m-long on Hugging Face Hub.