作者位次 #4/8ICLR 2026Poster· Social Aspects / Accountability, Transparency and Interpretability
Language Models Use Lookbacks to Track Beliefs
Nikhil Prakash · Natalie Shapira · Arnab Sen Sharma · Christoph Riedl · Yonatan Belinkov · Tamar Shaham · David Bau · Atticus Geiger