Online Reasoning Calibration: Test-Time Training Enables Generalizable Conformal LLM Reasoning

• URL: http://arxiv.org/abs/2604.01170v1
• Date: Wed, 01 Apr 2026 17:21:50 GMT
• Title: Online Reasoning Calibration: Test-Time Training Enables Generalizable Conformal LLM Reasoning
• Abstract summary: 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.
• Score: 18.69627681731888
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: While test-time scaling has enabled large language models to solve highly difficult tasks, state-of-the-art results come at exorbitant compute costs. These inefficiencies can be attributed to the miscalibration of post-trained language models, and the lack of calibration in popular sampling techniques. Here, we present Online Reasoning Calibration (ORCA), a framework for calibrating the sampling process that draws upon conformal prediction and test-time training. Specifically, we introduce a meta-learning procedure that updates the calibration module for each input. This allows us to provide valid confidence estimates under distributional shift, e.g. in thought patterns that occur across different stages of reasoning, or in prompt distributions between model development and deployment. ORCA not only provides theoretical guarantees on conformal risks, but also empirically shows higher efficiency and generalization across different reasoning tasks. At risk level $δ=0.1$, 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. Under zero-shot out-of-domain settings, it improves MATH-500 savings from 24.8% of the static calibration baseline to 67.0% while maintaining a low empirical error rate, and the same trend holds across model families and downstream benchmarks. Our code is publicly available at https://github.com/wzekai99/ORCA.

Related papers

• Calibration-Aware Uncertainty Cascades for Efficient Heterogeneous Model Collaboration [18.045450334322545]
  Heterogeneous model collaboration seeks to exploit the complementary strengths of different models to balance predictive performance and inference cost.<n>Existing approaches typically rely on trained routers, which tie routing decisions to a fixed task and model pool, or on raw-confidence cascades, whose thresholds lack consistent reliability semantics across heterogeneous models.<n>We propose-Aware Uncertainty Cascades (CAUC), a simple post-hoc framework that independently calibrates each model's confidence and selects deployment policies using validation data.
  arXiv  Detail & Related papers  (2026-09-10T12:14:54Z)
• 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)
• MARGIN: Runtime Confidence Calibration for Multi-Agent Foundation Model Coordination [0.0]
  We present MARGIN (Multi-Agent Grading via Incremental Normalisation), an online calibration method that learns per-agent, per-band calibration factors from the task stream itself.<n>Across 18 foundation models, 8 benchmarks, and over 44,000 observations, MARGIN achieves 3-6x lower calibration error than the best design-time baseline under distribution shift.
  arXiv  Detail & Related papers  (2026-05-21T18:25:05Z)
• Self-Calibrating Language Models via Test-Time Discriminative Distillation [18.46710400838861]
  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.
  arXiv  Detail & Related papers  (2026-03-18T13:28:50Z)
• ODAR: Principled Adaptive Routing for LLM Reasoning via Active Inference [60.958331943869126]
  ODAR-Expert is an adaptive routing framework that optimize the accuracy-efficiency trade-off via principled resource allocation.<n>We show strong and consistent gains, including 98.2% accuracy on MATH and 54.8% on Humanity's Last Exam.
  arXiv  Detail & Related papers  (2026-02-27T05:22:01Z)
• Correctness-Optimized Residual Activation Lens (CORAL): Transferrable and Calibration-Aware Inference-Time Steering [3.7758197704962835]
  We introduce CORAL, a regularized-time steering method that captures correctness signals from model internal activations using weight-decay probes.<n>CORAL consistently improves accuracy by 10% and expected calibration error (ECE) by 50% on average.<n>Our results support the hypothesis that distributed information in model internals can be extracted using regularized probes when individual neurons are insufficient.
  arXiv  Detail & Related papers  (2026-02-05T18:55:56Z)
• Classifier Calibration at Scale: An Empirical Study of Model-Agnostic Post-Hoc Methods [0.0]
  We study model-agnostic post-hoc calibration methods to improve probabilistic predictions in supervised binary classification.<n>We benchmark 21 widely used classifiers, including linear models, SVMs, tree ensembles (CatBoost, XGBoost, LightGBM)<n>We find that commonly used calibration procedures, most notably Platt scaling and isotonic regression, can systematically degrade proper scoring performance.
  arXiv  Detail & Related papers  (2026-01-19T18:23:36Z)
• CarBoN: Calibrated Best-of-N Sampling Improves Test-time Reasoning [62.56541355300587]
  We introduce a general test-time calibration framework that adaptively modifies the model toward high-reward reasoning paths.<n>Within this framework, we propose CarBoN, a two-phase method that first explores the solution space and then learns a calibration of the logits.<n>Experiments on MATH-500 and AIME-2024 show that CarBoN improves efficiency, with up to $4times$ fewer rollouts to reach the same accuracy.
  arXiv  Detail & Related papers  (2025-10-17T14:04:37Z)
• Sample Margin-Aware Recalibration of Temperature Scaling [20.87493013833571]
  Recent advances in deep learning have significantly improved predictive accuracy.<n>Modern neural networks remain systematically overconfident, posing risks for deployment in safety-critical scenarios.<n>We propose a lightweight, data-efficient recalibration method that precisely scales logits based on the margin between the top two logits.
  arXiv  Detail & Related papers  (2025-06-30T03:35:05Z)
• Uncertainty Aware Learning for Language Model Alignment [97.36361196793929]
  We propose uncertainty-aware learning (UAL) to improve the model alignment of different task scenarios. We implement UAL in a simple fashion -- adaptively setting the label smoothing value of training according to the uncertainty of individual samples. Experiments on widely used benchmarks demonstrate that our UAL significantly and consistently outperforms standard supervised fine-tuning.
  arXiv  Detail & Related papers  (2024-06-07T11:37:45Z)
• Uncertainty-aware Sampling for Long-tailed Semi-supervised Learning [89.98353600316285]
  We introduce uncertainty into the modeling process for pseudo-label sampling, taking into account that the model performance on the tailed classes varies over different training stages. This approach allows the model to perceive the uncertainty of pseudo-labels at different training stages, thereby adaptively adjusting the selection thresholds for different classes. Compared to other methods such as the baseline method FixMatch, UDTS achieves an increase in accuracy of at least approximately 5.26%, 1.75%, 9.96%, and 1.28% on the natural scene image datasets.
  arXiv  Detail & Related papers  (2024-01-09T08:59:39Z)
• Proximity-Informed Calibration for Deep Neural Networks [49.330703634912915]
  ProCal is a plug-and-play algorithm with a theoretical guarantee to adjust sample confidence based on proximity. We show that ProCal is effective in addressing proximity bias and improving calibration on balanced, long-tail, and distribution-shift settings.
  arXiv  Detail & Related papers  (2023-06-07T16:40:51Z)
• AdaFocal: Calibration-aware Adaptive Focal Loss [8.998525155518836]
  Training with focal loss leads to better calibration than cross-entropy. We propose a calibration-aware adaptive focal loss called AdaFocal.
  arXiv  Detail & Related papers  (2022-11-21T20:19:24Z)
• Leveraging Unlabeled Data to Predict Out-of-Distribution Performance [63.740181251997306]
  Real-world machine learning deployments are characterized by mismatches between the source (training) and target (test) distributions. In this work, we investigate methods for predicting the target domain accuracy using only labeled source data and unlabeled target data. We propose Average Thresholded Confidence (ATC), a practical method that learns a threshold on the model's confidence, predicting accuracy as the fraction of unlabeled examples.
  arXiv  Detail & Related papers  (2022-01-11T23:01:12Z)
• Newer is not always better: Rethinking transferability metrics, their peculiarities, stability and performance [5.650647159993238]
  Fine-tuning of large pre-trained image and language models on small customized datasets has become increasingly popular. We show that the statistical problems with covariance estimation drive the poor performance of H-score. We propose a correction and recommend measuring correlation performance against relative accuracy in such settings.
  arXiv  Detail & Related papers  (2021-10-13T17:24:12Z)
• Test-time Batch Statistics Calibration for Covariate Shift [66.7044675981449]
  We propose to adapt the deep models to the novel environment during inference. We present a general formulation $alpha$-BN to calibrate the batch statistics. We also present a novel loss function to form a unified test time adaptation framework Core.
  arXiv  Detail & Related papers  (2021-10-06T08:45:03Z)

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.