TY  - RPRT
TI  - Sparse Reward Subsystem in Large Language Models
AU  - Guowei Xu
AU  - Mert Yuksekgonul
AU  - James Zou
PY  - 2026
UR  - https://arxiv.org/abs/2602.00986
ID  - 2602.00986
ER  -