Efficiently Learning To Reason or Not to Reason: Root-token Policy Optimization for Adaptive Thinking.

Taehyeon Kim, Hyunsoo Lee, Youngsoo Jang, Moontae Lee

Browse the full ACL paper archive.