.png){kind=icon}
.png){kind=icon}
Return
Refinement-Guided Critique Learning: A Framework for Training Critique Models
.png){kind=icon}
PRE
AI
DOI：10.1016/j.inffus.2025.104002.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• We propose Refinement-Guided Critique Learning, a framework for training critique models. • We design a new critic reward that address inaccurate annotation of critique preferences and poor consistency. • We evaluate the RGCL framework across the five tasks, where it achieves superior performance in terms of critique quality and refinement outcomes.
Journal
IF:
15.5
Papers:
4.1K
Citations:
2.7W
View more organizations