Llama Guard 3 8B AWQ is a quantized version of a safety model hosted on Hugging Face. It checks conversations for unsafe content according to a defined policy covering 14 specific categories.
The model evaluates user and agent messages against categories that include violent crimes, non-violent crimes, sex crimes, child exploitation, defamation, specialized advice, privacy, intellectual property, indiscriminate weapons, hate, self-harm, sexual content, elections, and code interpreter abuse. A provided chat template structures the input by labeling roles as User or Agent and framing the task as identifying whether unsafe content appears in the conversation.
It is delivered as a model repository on the Hugging Face platform under the identifier Weni/Llama-Guard-3-8B-AWQ. The repository includes a chat template that enforces alternating user and assistant roles and raises an exception if the conversation structure is invalid.
The page presents the model within the context of open source and open science efforts to advance artificial intelligence.
Llama Guard 3 8B sits in PulseGate's Foundation models & chat category. It focuses on detecting unsafe content across multiple safety categories in user conversations. It is built as an open-source project for AI developers. Llama Guard 3 8B is open source under the Open Source license. The product ships for the web and API.
Weni builds and maintains Llama Guard 3 8B, and the product first shipped in 2024. Key capabilities include Content Safety, moderation, and Quantized Inference.
Latest indexed changes and source events
Weni/Llama-Guard-3-8B-AWQ verified by the PulseGate indexer
