conftrace_

comparative policy optimization

1 paper

Explore in graph

Also known as

CPO

Co-occurring keywords · papers mentioning both

role-playing dialogue (1) role playing (1) reinforcement learning fine-tuning (1) comparative evaluation (1) dialogue generation (1) policy optimization (1) reward modeling (1)

Papers

CPO: Addressing Reward Ambiguity in Role-playing Dialogue via Comparative Policy Optimization EMNLP 2025