🛰️ Daily AI Frontier
‹ back to 2026-09-10

Stable Answers, Unfinished Reasoning: Why Self-Consensus Is Not a Safe Early-Exit Signal

Research Efficiency & Systems

Ranking

Overall 81
Content 100
Popularity 37

Observed public metrics from 1 member.

Representative image for Stable Answers, Unfinished Reasoning: Why Self-Consensus Is Not a Safe Early-Exit Signal

Merged summary

TL;DR - Self-consensus—ending inference when repeated probes agree on an answer—is not a reliable early-exit signal for reasoning models. It can reduce token use, but often stops unfinished reasoning before the model corrects a provisional answer.

  • A preregistered evaluation of 3,520 consensus rules across two models and three benchmarks found that none met all predefined safety and token-saving criteria.
  • The failure reproduced on a held-out split and two unseen models, while the boundary-confidence method DEER passed all three acceptance gates.
  • At 32% token savings, roughly one in nine early exits selected an answer the trajectory later abandoned, usually cutting off a subsequent correction.
  • Larger agreement windows reduced but did not eliminate premature stops: the rate plateaued near 7% while token savings fell to 8%.

Sources (1)

Stable Answers, Unfinished Reasoning: Why Self-Consensus Is Not a Safe Early-Exit Signal

arXiv cs.CL Yunxiang Mo, Donghao Zhao, Hejia Geng 2026-09-09 arXiv:2609.09989
Public signals Semantic Scholar citations 0 · Semantic Scholar influential citations 0
Providers: Hugging Face · N/A OpenAlex · N/A Publisher · N/A Semantic Scholar · Citations 0 · Influential citations 0 X · N/A Fetched 2026-09-12 14:13:50.932394 UTC

TL;DR - Self-consensus—ending inference when repeated probes agree on an answer—is not a reliable early-exit signal for reasoning models. It can reduce token use, but often stops unfinished reasoning before the model corrects a provisional answer.

  • A preregistered evaluation of 3,520 consensus rules across two models and three benchmarks found that none met all predefined safety and token-saving criteria.
  • The failure reproduced on a held-out split and two unseen models, while the boundary-confidence method DEER passed all three acceptance gates.
  • At 32% token savings, roughly one in nine early exits selected an answer the trajectory later abandoned, usually cutting off a subsequent correction.
  • Larger agreement windows reduced but did not eliminate premature stops: the rate plateaued near 7% while token savings fell to 8%.
item →