nvidia/Llama-3.1-Nemotron-Safety-Guard-8B-v3

The nvidia/Llama-3.1-Nemotron-Safety-Guard-8B-v3 is an 8 billion parameter multilingual content safety model developed by NVIDIA, based on Meta's Llama-3.1-8B-Instruct. It is LoRa-tuned to classify user prompts and LLM responses as safe or unsafe, providing specific violation categories for unsafe content. This model supports 9 core languages and demonstrates strong zero-shot generalization across over 20 languages, making it suitable for moderating human-LLM interactions.