TY  - RPRT
TI  - RAIN-Merging: A Gradient-Free Method to Enhance Instruction Following in Large Reasoning Models with Preserved Thinking Format
AU  - Zhehao Huang
AU  - Yuhang Liu
AU  - Baijiong Lin
AU  - Yixin Lou
AU  - Zhengbao He
AU  - Hanling Tian
AU  - Tao Li
AU  - Xiaolin Huang
PY  - 2026
UR  - https://arxiv.org/abs/2602.22538
ID  - 2602.22538
ER  -