Right, Wrong, or Unfinished: A Competing Risks Framework for Language-Model Reasoning under Token Budgets

arXiv:2611.05860 · stat.AP, stat.ME · Submitted 2026-11-04 · Read on arXiv

stat.AP, stat.ME

Submitted: 2026-11-04

Updated: 2026-11-04

Code: https://github.com/huggingface/Math-Verify

Project page: https://chiragnagpal.github.io/papers/llm_leng

Terminology

Sources

Related papers