TY  - RPRT
TI  - GFT: From Imitation to Reward Fine-Tuning with Unbiased Group Advantages and Dynamic Coefficient Rectification
AU  - Wangjie Gan
AU  - Miao Pan
AU  - Linbo Xi
AU  - Wenqi Zhang
AU  - Jintao Chen
AU  - Jianwei Yin
AU  - Xuhong Zhang
PY  - 2026
UR  - https://arxiv.org/abs/2604.14258
ID  - 2604.14258
ER  -