Self-Calibrating Language Models via Test-Time Discriminative Distillation

• URL: http://arxiv.org/abs/2604.09624v1
• Date: Wed, 18 Mar 2026 13:28:50 GMT
• Title: Self-Calibrating Language Models via Test-Time Discriminative Distillation
• Abstract summary: Large language models (LLMs) are systematically overconfident on questions they often answer incorrectly.<n>We introduce $textbfSECL$ ($textbfSE$lf-$textbfC$alibrating $textbfL$anguage Models), a test-time training (TTT) pipeline that exploits this gap as label-free self-supervision.
• Score: 18.46710400838861
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) are systematically overconfident: they routinely express high certainty on questions they often answer incorrectly. Existing calibration methods either require labeled validation data, degrade under distribution shifts, or incur substantial inference costs. Recent work has shown that LLMs already contain a better-calibrated signal than the one they verbalize: the token probability of "True" when the model is asked "Is this answer correct?" ($P(\text{True})$) consistently outperforms their stated confidence, a gap that is theoretically grounded as generative error is lower-bounded by roughly twice the corresponding discriminative error. We introduce $\textbf{SECL}$ ($\textbf{SE}$lf-$\textbf{C}$alibrating $\textbf{L}$anguage Models), a test-time training (TTT) pipeline that exploits this gap as label-free self-supervision, requiring no labeled data or human supervision. SECL adapts only when the input distribution shifts, training on just 6--26% of the question stream at lower cost than the baseline it distills from. Across four small language models from three model families and four diverse domains, SECL reduces Expected Calibration Error (ECE) by 56--78%, outperforming its own supervision signal and matching or outperforming recent inference-time methods. SECL is the first method to apply TTT to calibration; seven ablations covering signal quality, gating strategy, weight accumulation, loss design, domain ordering, hyperparameter sensitivity, and layer selection confirm that each component is crucial and robust across configurations. Code: https://anonymous.4open.science/r/secl-emnlp26-submission-C890

Related papers

• Conformal Cascade: Distribution-Free Accuracy Guarantees for Multi-Tier LLM Inference [6.094431019524038]
  Large language model (LLM) cascades reduce inference cost by routing easy queries to a small model and deferring hard queries to a larger one.<n>We introduce textbfConformal Cascade (CC), a multi-tier inference framework that uses conformal prediction set size as the deferral rule.
  arXiv  Detail & Related papers  (2026-07-27T19:19:58Z)
• Scaling with Confidence: Calibrating Confidence of LLMs for Adaptive Test Time Scaling [2.8793525226737327]
  A novel reinforcement learning algorithm integrates correctness, calibration and dataset-informed reference accuracy rewards together.<n>C3RL enhances calibration without sacrificing accuracy, outperforming the current state-of-the-art method in both performance and calibration metrics.<n>Experiments show that CAS surpasses majority voting on both in-domain and out-of-domain datasets while reducing the inference budget by up to 12.33 times.
  arXiv  Detail & Related papers  (2026-07-02T02:29:33Z)
• CaliDist: Calibrating Large Language Models via Behavioral Robustness to Distraction [51.56484100374058]
  Existing calibration methods for Large Language Models (LLMs) often overlook a critical dimension of trustworthiness: a model's behavioral robustness.<n>We introduce textscCaliDist, a novel post-hoc calibration approach that directly measures and penalizes a model's susceptibility to distraction.<n>textscCaliDist consistently achieves lower Expected Error (ECE) and Brier Score compared with strong baselines.
  arXiv  Detail & Related papers  (2026-06-04T07:27:53Z)
• Causal Label Recovery in Payment Networks [0.0]
  Fraud detection models in payment networks train on chargeback labels that are systematically biased.<n>A companion paper [arXiv05:26.27557] proved that these four impairments impose a minimax lower bound on detection performance.<n>We formalize the observation pipeline as a sequential missing-data problem with three stages and a corruption layer, and construct the Sequential Triply Robust estimator.
  arXiv  Detail & Related papers  (2026-05-28T02:43:34Z)
• The Verification Tax: Fundamental Limits of AI Auditing in the Rare-Error Regime [0.0]
  Most cited calibration result in deep learning -- post-temperature-scaling ECE of 0.012 on CIFAR-100 -- is below the statistical noise floor.<n>We prove this is not a failure of the experiment but a law: the minimax rate for estimating calibration error with model error rate epsilon is Theta((Lepsilon/m)2/3), and no estimator can beat it.
  arXiv  Detail & Related papers  (2026-04-14T16:48:24Z)
• Online Reasoning Calibration: Test-Time Training Enables Generalizable Conformal LLM Reasoning [18.69627681731888]
  Online Reasoning calibration (ORCA) is a framework for calibrating the sampling process that draws upon conformal prediction and test-time training.<n>At risk level $=0.1$, ORCA improves Qwen2.5-32B efficiency on in-distribution tasks with savings up to 47.5% with supervised labels and 40.7% with self-consistency labels.
  arXiv  Detail & Related papers  (2026-04-01T17:21:50Z)
• Cross-Model Disagreement as a Label-Free Correctness Signal [10.66607150500579]
  Cross-model disagreement is a correctness indicator that can be dropped into existing production systems.<n>No generation from the verifying model is required, and no correctness labels are needed.<n>Results establish cross-model disagreement as a practical, training-free approach to label-free correctness estimation.
  arXiv  Detail & Related papers  (2026-03-26T13:46:22Z)
• You only need 4 extra tokens: Synergistic Test-time Adaptation for LLMs [50.54173262572369]
  Large language models (LLMs) are increasingly deployed in specialized domains such as finance, medicine, and agriculture.<n>We study label-free test-time adaptation for language models and present SyTTA, an inference-time framework that adapts models on-the-fly without additional supervision.
  arXiv  Detail & Related papers  (2025-10-11T14:00:39Z)
• CLUE: Non-parametric Verification from Experience via Hidden-State Clustering [64.50919789875233]
  We show that correctness of a solution is encoded as a geometrically separable signature within the trajectory of hidden activations.<n>ClUE consistently outperforms LLM-as-a-judge baselines and matches or exceeds modern confidence-based methods in reranking candidates.
  arXiv  Detail & Related papers  (2025-10-02T02:14:33Z)
• Reinforcement Learning with Verifiable yet Noisy Rewards under Imperfect Verifiers [90.50039419576807]
  Reinforcement Learning with Verifiable Rewards (RLVR) trains policies against automated verifiers to avoid costly human labeling.<n>To reduce vulnerability to verifier hacking, many RLVR systems collapse rewards to binary $0,1$ during training.<n>This choice carries a cost: it introduces textitfalse negatives (rejecting correct answers, FNs) and textitfalse positives (accepting incorrect ones, FPs)
  arXiv  Detail & Related papers  (2025-10-01T13:56:44Z)
• S$^2$R: Teaching LLMs to Self-verify and Self-correct via Reinforcement Learning [51.84977135926156]
  We introduce S$2$R, an efficient framework that enhances LLM reasoning by teaching models to self-verify and self-correct during inference.<n>Our results demonstrate that Qwen2.5-math-7B achieves an accuracy improvement from 51.0% to 81.6%, outperforming models trained on an equivalent amount of long-CoT distilled data.
  arXiv  Detail & Related papers  (2025-02-18T13:40:22Z)
• Label-Retrieval-Augmented Diffusion Models for Learning from Noisy Labels [61.97359362447732]
  Learning from noisy labels is an important and long-standing problem in machine learning for real applications. In this paper, we reformulate the label-noise problem from a generative-model perspective. Our model achieves new state-of-the-art (SOTA) results on all the standard real-world benchmark datasets.
  arXiv  Detail & Related papers  (2023-05-31T03:01:36Z)

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.