Summary of Generative Reward Models, by Dakota Mahan et al.
Generative Reward Models
by Dakota Mahan, Duy Van Phung, Rafael Rafailov, Chase Blagden, Nathan Lile, Louis Castricato, Jan-Philipp Fränken, Chelsea Finn, Alon Albalak
First submitted to arxiv on: 2 Oct 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed hybrid approach, GenRM, combines Reinforcement Learning from Human Feedback (RLHF) and Reinforcement Learning from AI Feedback (RLAIF) to generate synthetic preference labels that align with human preference judgments. The algorithm trains a Large Language Model (LLM) on self-generated reasoning traces, leading to improved in-distribution accuracy and significantly outperforming RLAIF on out-of-distribution tasks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary GenRM is a new way of training language models using a combination of human feedback and artificial intelligence. This approach helps generate labels that are more like what humans would choose. The results show that GenRM does better than other methods in both cases where the data is similar to what was used to train the model, and when it’s very different. |
Keywords
» Artificial intelligence » Large language model » Reinforcement learning » Reinforcement learning from human feedback » Rlhf
