Llama Guard

Llama Guard 4 12B is an AI safety and content moderation model designed to identify and classify potentially unsafe content in AI applications. It can be used to evaluate user inputs and AI-generated responses as part of responsible AI and content safety workflows.

Features of Llama Guard:

• Llama Guard provides automated AI safety and content moderation capabilities.
• It helps identify potentially harmful or unsafe content in AI interactions.
• It can be used to classify user prompts and AI-generated responses.
• It is designed for integration with generative AI and machine-learning applications.
• It supports safety filtering workflows for applications using large language models.
• Llama Guard is suitable for responsible AI development, evaluation, and content moderation workloads.

Usage instructions:

$ sudo su
$ cd /opt/llama-guard
$ source venv/bin/activate
$ python -c "from transformers import AutoConfig; c=AutoConfig.from_pretrained('meta-llama/Llama-Guard-4-12B'); print('Llama Guard 4 12B')"

Disclaimer: Llama Guard is provided as an AI safety and content moderation model. Performance, supported features, and compatibility may depend on the installed Llama Guard version, Transformers and PyTorch versions, Python version, available GPU/CPU resources, hardware architecture, and system environment.