Confidence-Calibrated Inference Expansion for Evaluator-Guided Test-Time Reasoning

Key points

• We formulate this choice as a compute-aware three-action controller, scoring terminal actions by predicted answer success and expansion by a rollout-based continuation-success surrogate, with compute cost priced explicitly.1
• A local action-gap analysis characterizes when score errors can change a decision, while an expected regret bound accounts for errors along the realized expansion path.1
• Guided by this analysis, we learn and calibrate action-specific confidence estimates, with local uncertainty penalties for ambiguous decisions.1
• The results frame evaluator-guided reasoning as resource allocation under uncertainty, rather than a uniformly stronger reasoner.1

Sources (1)

• [1]Confidence-Calibrated Inference Expansion for Evaluator-Guided Test-Time Reasoning
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 14, 07:25 PM

  Evaluator-guided reasoning presents a local control problem at every step: trust the current answer, return a fallback, or spend more inference.

  We formulate this choice as a compute-aware three-action controller, scoring terminal actions by predicted answer success and expansion by a rollout-based continuation-success surrogate, with compute cost priced explicitly.

Extractive summary: sentences quoted from the sources.