Home » Authors » Jingwei Ni, Dmitry Abulkhanov, Vladislav Smirnov, Markus Leippold, Elliott Ash, Timothy Baldwin, Artem Shelmanov, Mrinmaya Sachan

Jingwei Ni, Dmitry Abulkhanov, Vladislav Smirnov, Markus Leippold, Elliott Ash, Timothy Baldwin, Artem Shelmanov, Mrinmaya Sachan

Public reference bibliography and cataloged PDF materials authored or published by Jingwei Ni, Dmitry Abulkhanov, Vladislav Smirnov, Markus Leippold, Elliott Ash, Timothy Baldwin, Artem Shelmanov, Mrinmaya Sachan.

Found 1 document(s) Page 1 of 1

Same Beliefs, Different Policies: Learning Reusable Representations for Early Stopping in LLM Reasoning

Year: 2026 • Source: arxiv.org • Listed PDF

Reasoning LLMs often continue reasoning after reaching a correct answer, wasting computation and sometimes replacing it with an incorrect one. Deciding when to stop is a metareasoning problem: the model must weigh what f...