Meta AI has released Llama Guard 3-1B-INT4, a compact and high-performance AI moderation model for human-AI conversations.The model is 7 times smaller than its predecessor, Llama Guard 3-1B, with a size of 440MB.Advanced compression techniques, including pruning and quantization, were used to achieve this reduction in size.Llama Guard 3-1B-INT4 maintains robust safety moderation capabilities while operating efficiently on mobile devices.