一作ACL 2026Findings· Findings of ACL
GFT : From Imitation to Reward Fine-Tuning with Unbiased Group Advantages and Dynamic Coefficient Rectification
Wangjie Gan · Miao Pan · Linbo Xi · Wenqi Zhang · Jintao Chen · Jianwei Yin · Xuhong Zhang
摘要
Large language models are typically post-trained using supervised fine-tuning (SFT) and reinforcement learning (RL), yet effectively unifying efficient knowledge injection with robust generalization remains challenging. In this work, we provide a training-dynamics analysis showing that SFT can be...