PKU-Alignment

PKU-Alignment is a research team at Peking University that develops the Beaver open-source RLHF framework. It works on alignment research, including constrained value alignment of large language models through Safe RLHF. The team publishes code, datasets and model checkpoints for reproducible research.