Casper, 2023 — Open Problems and Fundamental Limitations of Reinforcement Learning from Human Feedback

DOI: <a href="https://doi.org/10.48550/arXiv.2307.15217" target="_blank" rel="noopener noreferrer">https://doi.org/10.48550/arXiv.2307.15217
URL: <a href="https://arxiv.org/abs/2307.15217" target="_blank" rel="noopener noreferrer">https://arxiv.org/abs/2307.15217

Covers preference-learning or reward-modeling methods used in modern alignment pipelines.

Publication links

Covers preference-learning or reward-modeling methods used in modern alignment pipelines.