TY  - RPRT
TI  - TAB-PO: Preference Optimization with a Token-Level Adaptive Barrier for Token-Critical Structured Generation
AU  - Samah Fodeh
AU  - Linhai Ma
AU  - Ganesh Puthiaraju
AU  - Srivani Talakokkul
AU  - Afshan Khan
AU  - Sreeraj Ramachandran
AU  - Elyas Irankhah
AU  - Aimee Kendall Roundtree
PY  - 2026
UR  - https://arxiv.org/abs/2603.00025
ID  - 2603.00025
ER  -