# Safe Reinforcement Learning

Beaver-7B-v3.0 is a 7B-parameter large language model based on the LLaMA architecture, focusing on safety and human feedback reinforcement learning (RLHF).