Route, Don't Fix: Regime-Dependent Decoding Correction and a Trajectory-Gated Router for Reliable Clinical LLM Answer Selection

• URL: http://arxiv.org/abs/2609.14825v1
• Date: Sun, 13 Sep 2026 22:36:37 GMT
• Title: Route, Don't Fix: Regime-Dependent Decoding Correction and a Trajectory-Gated Router for Reliable Clinical LLM Answer Selection
• Abstract summary: Large language models (LLMs) are often deemed unsafe for clinical question answering because of their tendency to hallucinate.<n>Inference-time correction uses the model's internal logit signals, but a fixed transformation need not suit every question.<n>We introduce ALTAS, which reads terminal entropy and late-layer linearity from one forward pass to choose per question between greedy decoding and late-layer trajectory correction.
• Score: 1.4358829531570645
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large language models (LLMs) are often deemed unsafe for clinical question answering because of their tendency to hallucinate. Retrieval augmentation, fine-tuning, and external verifiers require new infrastructure that clinical governance must approve and may add latency or extra model calls. Inference-time correction uses the model's internal logit signals, but a fixed transformation need not suit every question. A corrector that improves accuracy by about ten percentage points on a truthfulness stress test yields negligible gains on clinical multiple-choice benchmarks, where instruction tuning concentrates output probability on one answer and leaves low terminal entropy. We introduce ALTAS, which reads terminal entropy and late-layer linearity ($R^2$) from one forward pass to choose per question between greedy decoding and late-layer trajectory correction. No classifier, probe, or head is trained; the router operates on candidate-answer logits and adds 6.5% latency overhead. Applied to every question, the correction improves TruthfulQA over greedy at 3B and 8B by 11.4 and 10.0 percentage points, respectively ($p<10^{-10}$). Gated per question, ALTAS retains gains of 8.3 to 9.5 percentage points while keeping MedQA, PubMedQA, and MedHallu within a one-percentage-point do-no-harm band, with no statistically significant differences from greedy. The method passes verification sweeps over frozen thresholds, the scoring rule, and the domain label.

Related papers

• Which Medical Questions Deserve Rationales? Perturbation-Sensitive Selection for Robust QA [12.797519888387628]
  We propose root-mean-square Robustness-based Sample Prioritization (RMS-RSP) to perturb hidden states at rationale tokens.<n> RMS-RSP improves robust accuracy and semantic consistency by 1.91 and 2.85 points on average, with the same direction on all five datasets.
  arXiv  Detail & Related papers  (2026-09-09T03:58:59Z)
• SS-ESOAP: Self-Scaled Adaptive Preconditioning for Physics-Informed Learning [70.71795707610961]
  method augments SOAP-style preconditioning with a scalar secant-energy correction adapted to Kronecker geometry.<n>Method attains the lowest final residual on six of eight PDE benchmarks.<n>Three-seed $L2$ and $H1$ errors on four representative PDEs support the link between lower residuals and improved solution accuracy.
  arXiv  Detail & Related papers  (2026-08-29T21:34:47Z)
• AERA: Adaptive Evidence Residual Allocation for Efficient Test-Time Reasoning [22.964476507912327]
  Test-time scaling improves language-model reasoning by generating additional candidate solutions, but allocating the same inference budget to every problem is computationally wasteful.<n>We show that checkpoint-level correctness evolves non-monotonically, and observable evidence may strengthen before an answer collapses or weaken before it recovers.<n>We introduce Adaptive Evidence Residual Allocation (AERA), a sequential controller that learns whether additional computation is likely to recover a better answer from checkpoint-observable evidence.
  arXiv  Detail & Related papers  (2026-08-28T06:21:07Z)
• SVR: Self-Verifying Refinement via Joint Verdict-Confidence Reinforcement Learning for Adaptive Test-Time Compute [62.3458279176813]
  Self-Verifying Refinement is an oracle-free multi-turn reinforcement learning framework.<n>It learns to use self-verification as a compute-control policy.<n>It achieves a macro-average accuracy of 0.563 with only 2.99 inference turns on average.
  arXiv  Detail & Related papers  (2026-07-30T16:20:58Z)
• Know Before You Fetch: Calibrated Retrieval-Budget Allocation for Retrieval-Augmented Generation [10.60293484295824]
  We formulate adaptive RAG as calibrated retrieval-budget allocation.<n>We calibrate sequence log-probability and prefix-logit uncertainty signals into probabilities of correctness.<n>For sequence log-probability, ECE drops from 0.275 to 0.062 on TriviaQA, 0.643 to 0.009 on NQ, and 0.711 to 0.031 on MS MARCO.
  arXiv  Detail & Related papers  (2026-06-29T08:36:33Z)
• What Makes a Medical Checker Trainable? Diagnosing Signal Collapse and Reward Hacking in Checker-Guided RAG for Biomedical QA [6.163604976615509]
  Medical RAG needs evidence-grounded claims, so plugging a claim-level NLI checker into retrieval-augmented RL is intuitive.<n>We compare four NLI checker back-ends as process rewards inside a GRPO-trained medical RAG agent.
  arXiv  Detail & Related papers  (2026-05-25T16:06:06Z)
• Self-Verified Distillation: Your Language Model Is Secretly Its Own Synthetic Data Pipeline [56.53954182896384]
  We propose Self-Verified Distillation, a simple post-training refinement algorithm for large language models.<n>Self-Verified Distillation generates candidate solutions to unlabeled seed questions.<n>It filters them using prompt-based self-verification, and trains on the resulting self-curated dataset.<n>We find that sampling more candidate generations and using a larger verification budget during training data construction produces higher-quality self-curated data.
  arXiv  Detail & Related papers  (2026-05-20T17:26:10Z)
• Correction and Corruption: A Two-Rate View of Error Flow in LLM Protocols [51.56484100374058]
  We propose a paired-outcome measurement interface for auditing a single protocol step on exact-match tasks.<n>For each instance, the interface records a baseline correctness bit and a post-step correctness bit.<n>These rates predict accuracy changes and define a reusable empirical interface testable across seeds, mixtures, and pipelines.
  arXiv  Detail & Related papers  (2026-04-20T13:25:40Z)
• Learning from Emptiness: De-biasing Listwise Rerankers with Content-Agnostic Probability Calibration [76.08899010904652]
  CapCal is a training-free framework that mechanically decouples positional bias from ranking decisions.<n>It achieves superior performance among training-free methods while preserving single-pass efficiency.
  arXiv  Detail & Related papers  (2026-04-11T10:47:22Z)
• Sandwich Reasoning: An Answer-Reasoning-Answer Approach for Low-Latency Query Correction [14.620900910188382]
  Chain-of-Thought (CoT) reasoning improves accuracy but incurs prohibitive latency for real-time query correction.<n>We propose Sandwich Reasoning (SandwichR), a novel approach that explicitly aligns a fast initial answer with post-hoc reasoning.<n>We show that SandwichR achieves SOTA accuracy comparable to standard CoT while delivering a 40-70% latency reduction.
  arXiv  Detail & Related papers  (2026-01-07T07:52:30Z)
• TARG: Training-Free Adaptive Retrieval Gating for Efficient RAG [46.122203287541005]
  Training-free Adaptive Retrieval Gating (TARG) is a single-shot policy that decides when to retrieve using only a short, no-latency draft from the base model.<n>On NQ-Open, TriviaQA, and PopQA, TARG consistently shifts the accuracy-efficiency frontier.
  arXiv  Detail & Related papers  (2025-11-12T23:09:52Z)
• A Locally Executable AI System for Improving Preoperative Patient Communication: A Multi-Domain Clinical Evaluation [1.9205944025326396]
  LENOHA is a safety-first, local-first system that routes inputs with a high-precision sentence-transformer constraints.<n>It returns verbatim answers from a clinician-curated FAQ for clinical queries.<n>Energy logging shows that the non-generative clinical path consumes 1.0 mWh per input versus 168 mWh per small-talk reply.
  arXiv  Detail & Related papers  (2025-10-02T04:53:11Z)

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.