TY  - RPRT
TI  - TPMM-DPO: Trajectory-aware Preference-guided Model Merging for Iterative Direct Preference Optimization
AU  - Lingling Fu
AU  - Yongfu Xu
PY  - 2026
UR  - https://arxiv.org/abs/2605.23398
ID  - 2605.23398
ER  -