awesome-RLHF

Why we included this project

For anyone trying to understand how language models get aligned to human preferences, this curated index of RLHF research is a strong starting point. It organizes papers by year, from the early foundational work through 2026, so you can follow how methods like PPO and DPO developed without hunting through conference proceedings. The maintainers also collect codebases, datasets, blogs, and books, which helps when you want to move from reading about a technique to actually running it. The overview section explains the core idea of RLHF in plain language, so even someone new to reinforcement learning can get oriented. Researchers, grad students, and engineers scoping out the alignment space will find it saves real time.