Anchored Alignment for Self-Explanations Enhancement
Keywords: LLM, Self-Explanation, Alignment, Preference Pairs, DPO, SFT, RLAIF, Self-Alignment, Self-instruction
TL;DR: In this work, we introduce a methodology for alignment tailored to enhance the ability of large language models (LLMs) to articulate their reasoning—self-explanation—in the absence of annotated rationale explanations.
Abstract: In this work, we introduce a methodology for alignment designed to enhance the ability of large language models (LLMs) to articulate their reasoning—\textit{self-explanation}—even in the absence of annotated rationale explanations. Our alignment methodology comprises three key components: explanation quality assessment, self-instruction dataset generation, and model alignment. Additionally, we present a novel technique called \textit{Alignment with Anchor Preference Pairs}, which improves the selection of preference pairs by categorizing model outputs into three groups: consistently correct, consistently incorrect, and variable. By applying tailored strategies to each category, we enhance the effectiveness of Direct Preference Optimization (DPO). Our experimental results demonstrate that this approach significantly improves explanation quality while maintaining accuracy compared to other fine-tuning strategies.
Primary Area: interpretability and explainable AI
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 12006
Loading