Skip to content
NeurIPS 2026 planner

Poster · Sydney

Stabilizing Off-policy LLM Optimization with Prefix Importance Ratio

When
Thu 10 Dec 2026, 10:00–13:00 AEDT
Where
Hall 1-4, ICC Sydney
Exhibition Centre, Level 1
Session
Sydney Poster Session 5
Authors
Shiye Lei, Zhihao Cheng, Dacheng Tao
Organisations
ByteDance, University of Sydney, Nanyang Technological University
QRS classification
Reinforcement learning, Language models

Open in scheduleOfficial page