A training pipeline that scores a model's own responses for semantic, factual, and safety uncertainty, builds preference pairs from those scores, and trains in three difficulty stages improves reported alignment scores across four benchmarks.
Direct Large Language Model Alignment Through Self-Rewarding Contrastive Prompt Distillation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
Aligning large language models (LLMs) with human expectations without human-annotated preference data is an important problem. In this paper, we propose a method to evaluate the response preference by using the output probabilities of response pairs under contrastive prompt pairs, which could achieve better performance on LLaMA2-7B and LLaMA2-13B compared to RLAIF. Based on this, we propose an automatic alignment method, Direct Large Model Alignment (DLMA). First, we use contrastive prompt pairs to automatically generate preference data. Then, we continue to evaluate the generated preference data using contrastive prompt pairs and calculate a self-rewarding score. Finally, we use the DPO algorithm to effectively align LLMs by combining this self-rewarding score. In the experimental stage, our DLMA method could surpass the \texttt{RLHF} method without relying on human-annotated preference data.
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
An Uncertainty-Driven Adaptive Self-Alignment Framework for Large Language Models
A training pipeline that scores a model's own responses for semantic, factual, and safety uncertainty, builds preference pairs from those scores, and trains in three difficulty stages improves reported alignment scores across four benchmarks.