Reasons to Reject? Aligning Language Models with Judgments
This work addresses the alignment of LLMs for safer and more effective AI interactions, offering a novel approach that could enhance model behavior in real-world applications, though it builds incrementally on existing alignment methods.
The paper tackles the problem of aligning large language models (LLMs) using natural language feedback (judgments) instead of scalar rewards, proposing Contrastive Unlikelihood Training (CUT) to improve fine-grained content detection and correction. Results show that CUT with LLaMA2-13b outperforms a 175B model and beats baselines by 50.84 points on AlpacaEval, and iterative alignment boosts performance from 81.09 to 91.68 points.
As humans, we consistently interact with our peers and receive feedback in the form of natural language. This language feedback allows us to maintain appropriate behavior, and rectify potential errors. The question arises naturally: can we use language feedback to align large language models (LLMs)? In contrast to previous research that aligns LLMs with scalar rewards, we present the first systematic exploration of alignment through the lens of language feedback (i.e., judgment). We start with an in-depth investigation of potential methods that can be adapted for aligning LLMs with judgments, revealing that these methods cannot fully capitalize on judgments. To facilitate more effective utilization of judgments, we propose a novel framework, Contrastive Unlikelihood Training (CUT), that allows for fine-grained inappropriate content detection and correction based on judgments. Our results show that, with merely 1317 off-the-shelf judgment data, CUT (LLaMA2-13b) can beat the 175B DaVinci003 and surpass the best baseline by 50.84 points on AlpacaEval. CUT (LLaMA2-chat-13b) can also align LLMs in an iterative fashion using up-to-date model-specific judgments, improving performance from 81.09 to 91.68 points on AlpacaEval. Further analysis suggests that judgments hold greater potential than rewards in LLM alignment.