cs.AINov 17, 2026

A Full-Ledger Counterfactual Audit of Adaptive Reasoning

Authors: Xiaopeng Chu, Wenzheng Kong, Wei Liu, Wenyi Zhang

Organizations: University of Science and Technology of China Hefei 230027, China

Abstract

Adaptive reasoning must decide whether to answer now or spend more compute. Predicting that the current answer is wrong does not establish whether further reasoning will improve it. We introduce the Full-Ledger Counterfactual Value Audit (FLCVA), which compares an immediate answer with four fixed-length continuations from the same frozen prefix under one answer procedure and accounts for the cost of every signal used by a policy. On GSM8K and MATH-500, a random 256-token continuation recovers a wrong answer in 26.6% of Qwen3-8B cases and 18.9% of DeepSeek-R1-Distill-Qwen-7B cases, while damaging a correct answer in 1.4% and 1.9%. Correctness labels from independently sampled continuations predict a separate continuation's outcome in both primary models. In offline replay, using these labels to allocate compute improves performance on DeepSeek; the Qwen3 gain remains inconclusive. Under full cost accounting, the tested lightweight, myopic controllers mainly recognize current errors and add little continuation-specific value beyond that shortcut. A Llama-8B replication and action/cost ablations examine the robustness of these findings. FLCVA separates measuring local continuation value, predicting it from observable signals, and using those predictions to allocate computation.

Figures & tables

Appendix figures & tables28 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. AERA: Adaptive Evidence Residual Allocation for Efficient Test-Time Reasoning

    Aug 28, 2026Ziming Wang, Ivor Tsang, Hangwei QianAdaptive Reasoning Resource Allocation for Language ModelsAdaptive Inference

  2. Expanding LLM Reasoning

    Oct 4, 2026Rian Atri, Evan LuoLLM Inference EfficiencyInference-Time Scaling

  3. No Judgment Without a Reason: Counterfactual Receipts for Versioned AI Evaluators

    Aug 21, 2026Ye Chen, Weining ZhangLLM EvaluationCounterfactual Evaluation