Think Again or Think Longer? Selective Verification for Budget-Aware Reasoning

• URL: http://arxiv.org/abs/2606.19808v1
• Date: Thu, 18 Jun 2026 05:25:43 GMT
• Title: Think Again or Think Longer? Selective Verification for Budget-Aware Reasoning
• Abstract summary: sevra is a serving-layer controller that decides whether to preserve a frozen solver's initial answer or invoke active verification.<n>On mathfive, selective verification reaches 76.3% accuracy, compared with 75.5% for always verifying.<n>On CommonsenseQA, always-on verification hurts, while Self-Consistency@5 improves accuracy about five times the realized token cost.
• Score: 15.231928045776293
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Test-time reasoning is increasingly used as a serving-time control knob, but extra reasoning is not uniformly valuable: it can repair failed attempts, waste compute on already-correct answers, or introduce harmful answer changes. We study this as a deployment allocation problem rather than a new-verifier problem. We introduce \sevra, Selective Verification for Reasoning Allocation, a serving-layer controller that decides whether to preserve a frozen solver's initial answer or invoke active verification. Using a frozen Qwen3-4B solver, we log intervention outcomes and train recoverability-aware gates from serving-visible attempt state. On \mathfive, selective verification reaches 76.3\% accuracy, compared with 75.5\% for always verifying, while reducing post-generation tokens by 26.8\% and harmful flips from 2.2\% to 1.0\%. However, an 8,192-token initial solve reaches 76.0\% accuracy with 28\% fewer total model tokens, showing that selective recovery is useful but not the best tested cost frontier. In frozen transfer to \gsm, the selective policy verifies only 3.0\% of examples, improves accuracy from 93.4\% to 94.5\%, and reduces verification tokens by 91.2\% relative to always verifying; again, a longer initial solve matches its accuracy with fewer realized tokens. On CommonsenseQA, always-on verification hurts, while Self-Consistency@5 improves accuracy at about five times the realized token cost. The resulting deployment rule is: tune the initial budget first, then use selective recovery when explicit checks, bounded retries, auditability, or regression-risk control matter.

Related papers

• What Eviction Destroys: A Restore-Counterfactual Audit of Forgetting in Agent Memory [1.6558929815405155]
  We introduce the restore counterfactual, a per-question paired intervention that reinstates the question's gold evidence in the read-time context.<n>We evaluate FIFO, random, redundancy-aware, and LLM-importance eviction on LongMemEval-S at three budgets and under two retrieval regimes.
  arXiv  Detail & Related papers  (2026-09-08T05:48:10Z)
• Same Agent, Different Answers: A Repeat-Aware Audit of Corpus-Induced Answer Churn in Retrieval-Augmented QA [0.4351760066527804]
  A retrieval-augmented QA system can return different answers after an index expansion even when its requested model identifier, prompt, retrieval policy, evidence depth, rendering, and exposed generation controls are held fixed.<n>We call the hidden phenomenon accuracy-blind answer churn and introduce the emphSnapshot Compatibility Audit, which estimates excess answer churn by subtracting same-snapshot repeat disagreement from cross-snapshot disagreement.
  arXiv  Detail & Related papers  (2026-08-24T06:40:09Z)
• No Judgment Without a Reason: Counterfactual Receipts for Versioned AI Evaluators [10.23869027501654]
  Evaluators often produce correct labels via flawed reasoning, a critical failure for agentic systems gating actions, routing reviews, or supplying training feedback.<n>We formalize evaluator reasoning accountability via three core sources: grounds, norms, and authority. Varying these sources yields an eight-cell counterfactual judgment cube to characterize judgment updates.<n>We show reason-aware evaluation must decouple prediction and certification, reporting transformation consistency alongside standard accuracy for trustworthy evaluator auditing.
  arXiv  Detail & Related papers  (2026-08-21T10:01:58Z)
• Calibrated e-CUSUM Decoding for Quantized Reasoning Models: Why Token Log-Probability Is the Wrong Observable for Decoding Monitors [0.6999740786886536]
  Low-bit quantization makes small reasoning models inexpensive to deploy but can degrade their chains of thought.<n>We show that the centered token log-probability increment $log p(w_t)+H_t$ is the wrong observable for this purpose.<n>We introduce a training-free decoding controller that combines a degeneration-aware alarm score fusing token uncertainty with explicit repetition.
  arXiv  Detail & Related papers  (2026-07-13T09:34:36Z)
• When Does Learning to Stop Help? A Cost-Aware Study of Early Exits in Reasoning Models [4.8190992438931035]
  LearnStop is a hidden-state-free logistic stopper over prefix-observable features.<n>A trajectory decomposition predicts the regime: learning pays where answers oscillate and correctness evidence is spread.<n>Cost accounting sharpens the picture further -- the same policy that saves 32% of tokens under KV-cache forking costs 121% extra under black-box repeated prefilling.
  arXiv  Detail & Related papers  (2026-06-29T19:33:42Z)
• Re-feeding Is Not Replaying: Measuring Replay Noise in Counterfactual Token-Credit Estimation [0.0]
  Per-token counterfactual credit estimation asks which token in a language-model rollout caused the final answer to be right or wrong.<n>We measure what that assumption costs on a stock inference engine, with a three-pass design.<n>Across six configurations and three models, re-feeding changes the credit estimate at rates 14-28 percentage points above the replica floor.
  arXiv  Detail & Related papers  (2026-06-14T06:09:16Z)
• PACE: Anytime-Valid Acceptance Tests for Self-Evolving Agents [0.0]
  Self-evolving agents improve by repeatedly proposing changes to their own prompts, skills, ormodifications.<n>On Qwen2.5 agents (0.5B-3B) self-evolving at the prompt level on GSM8K, SVAMP, and ARC-Challenge, greedy acceptance commits 30-42% false and 10-33% harmful edits.<n> PACE commits the real one and essentially nothing else, matching greedy's held-out accuracy at sharply lower variance and about 18% lower evaluation cost.
  arXiv  Detail & Related papers  (2026-06-06T11:12:11Z)
• Right Makes Might: Aligning Verified Hidden States Empowers RL Reasoning [55.264863369127774]
  Current methods reduce each correct rollout to a single reward bit, ignoring the geometric structure shared among their hidden states.<n>We propose Hidden-Align, an auxiliary loss function that aligns the last-layer hidden states of correct rollouts at the anchor token during RL training, with zero overhead in both training and inference.<n>On eight mathematical reasoning benchmarks, Hidden-Align improves average pass@1 over the DAPO baseline by 3.8, 6.2, and 5.4 percentage points on Qwen3-1.7B, 4B, and 14B respectively, with consistent pass@k gains across all three scales.
  arXiv  Detail & Related papers  (2026-06-02T06:51:15Z)
• MAIGO: Mitigating Lost-in-Conversation with History-Cleaned On-Policy Self-Distillation [51.32912774464992]
  Large language models often solve tasks from a fully specified prompt but degrade when the same requirements unfold over multiple turns.<n>We propose MAIGO, an on-policy self-distillation method that reduces this contamination using history-cleaned references from the model's own policy.<n>For middle turns, MAIGO removes prior assistant replies while preserving the user-visible sharded prefix; for answer turns, it distills from paired full-view references conditioned on the completed user-side dialogue.
  arXiv  Detail & Related papers  (2026-05-26T15:38:46Z)
• Guarded Repair for Harm-Aware Post-hoc Replacement of LLM Mathematical Reasoning [0.0]
  GuardedRepair is a best-of-N repair framework that diagnoses cached reasoning traces, selectively triggers repair, and accepts answer-changing candidates.<n>We show that GuardedRepair substantially improves the fixed/broken tradeoff, while also revealing that replacement risk is reduced rather than eliminated.
  arXiv  Detail & Related papers  (2026-05-23T14:51:13Z)
• Pause and Reflect: Conformal Aggregation for Chain-of-Thought Reasoning [8.024041325202612]
  Chain-of-thought (CoT) reasoning with self-consistency improves performance by aggregating multiple sampled reasoning paths.<n>We introduce a conformal procedure for CoT reasoning that directly addresses aggregation uncertainty.<n>Our approach replaces majority voting with weighted score aggregation over reasoning paths and calibrates an abstention rule using conformal risk control.
  arXiv  Detail & Related papers  (2026-05-13T20:33:59Z)
• Rethinking Dense Sequential Chains: Reasoning Language Models Can Extract Answers from Sparse, Order-Shuffling Chain-of-Thoughts [51.84894623128418]
  Modern reasoning language models generate dense, sequential chain-of-thought traces implicitly assuming that every token contributes and that steps must be consumed in order.<n>We challenge both assumptions through a systematic intervention pipeline--removal, masking, shuffling, and noise injection--applied to model-generated reasoning chains.<n>We find that answer extraction operates on a sparse, order-insensitive, and structurally robust informational substrate.
  arXiv  Detail & Related papers  (2026-05-08T06:15:50Z)
• Correct Answers from Sound Reasoning: Verifiable Process Supervision for Language Models [94.68358825189738]
  We propose a post-training framework for verifiable domains that jointly optimize prediction accuracy and reasoning quality.<n>We evaluate VPS on chess, a controlled testbed where reasoning steps can be deterministically verified against engine signals.<n>VPS preserves accuracy while significantly improving reasoning quality, reducing win-rate error by up to 30% and restoring consistency to near saturation.
  arXiv  Detail & Related papers  (2026-04-03T15:19:46Z)
• Self-Verification Dilemma: Experience-Driven Suppression of Overused Checking in LLM Reasoning [27.979418392479033]
  Large Reasoning Models (LRMs) achieve strong performance by generating long reasoning traces with reflection.<n>We find that a substantial fraction of reflective steps consist of self-verification (recheck) that repeatedly confirm intermediate results.<n>This reveals a mismatch between how often self-verification is activated and how often it is actually useful.<n>We propose a novel, experience-driven test-time framework that reduces the overused verification.
  arXiv  Detail & Related papers  (2026-02-03T12:58:23Z)
• Re-FORC: Adaptive Reward Prediction for Efficient Chain-of-Thought Reasoning [85.76121000710522]
  We propose Re-FORC, an adaptive reward prediction method.<n>It enables prediction of the expected future rewards as a function of the number of future thinking tokens.
  arXiv  Detail & Related papers  (2025-11-03T23:47:49Z)
• VeriThinker: Learning to Verify Makes Reasoning Model Efficient [52.74493506816969]
  Large Reasoning Models excel at complex tasks using Chain-of-Thought (CoT) reasoning.<n>Their tendency to overthinking leads to unnecessarily lengthy reasoning chains.<n>We introduce VeriThinker, a novel approach for CoT compression.
  arXiv  Detail & Related papers  (2025-05-23T14:17:56Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.