[TOC]
- Title: Reward Engineering for Generating Semi-Structured Explanation
- Author: Jiuzhou Han et. al.
- Publish Year: EACL2024
- Review Date: Thu, Jun 20, 2024
- url: https://github.com/Jiuzhouh/Reward-Engineering-for-Generating-SEG
Summary of paper
Motivation
Contribution
- the objective is to equip moderately-sized LMs with the ability to not only provide answers but also generate structured explanations
Some key terms
Intro
the author talked about some background on Cui et al. incorporate a generative pre-training mechanism over synthetic graphs by aligning inputs pairs of text-graph to improve the model’s capability in generating semi-structured explanation.
- the author first stated that they utilise SFT as teh de-facto solution. We then turn our focus to RLHF as a mechanism to further align the explanations with ground-truth on top of SFT.
Related work
the author mention in details about the benchmark they use – ExplaGraph and COPA-SSE
Results
Experiments
the author first talked about the dataset size and said that “The detailed descriptions of all evaluation metrics are provided in Appendix A.”
Human evaluation
