Daily Papers

Solving problems through repeated attempts is a sequential modeling task: at each step, the solver receives feedback and decides how to revise their solutions. Predicting whether performance improves, plateaus, or regresses across attempts is central to understanding any iterative problem-solving process in both human learners and autonomous agents. Beyond outcomes, modeling what errors persist and how strategies shift across attempts provides deeper insight into the mechanics of sequential learning. Studying these dynamics requires observing many solvers as they attempt, receive feedback, and revise. Programming courses with automated grading provide this setting, as students iteratively submit code to test suites and receive feedback on every attempt. We therefore curate CodeInsight, a large-scale dataset of over 3 million submissions from 3,286 undergraduates across 2 introductory C++ courses in 2 academic years, with test-case-level outcomes, timestamps, and source code. On this dataset, we build a benchmark that evaluates models spanning parametric, sequential, and generative traditions under a shared calibration-and-scoring protocol, including a Recurrent State Space Model (RSSM) adapted to track solver characteristics through discrete latent variables and an LLM-based predictor that generates explicit solutions. The adapted RSSM achieves the strongest predictive accuracy on three of the four courses. The LLM predictor is less accurate but produces full submissions at each attempt, enabling direct analysis of failure modes. We find that the model's coding proficiency is inversely related to predictive performance in this setting, with the LLM better understood as a generative solver conditioned on context rather than a faithful predictor of solver behavior. We publicly release our code and the dataset on request to facilitate future research.

Reliable probability estimates are critical in many machine learning applications, yet modern classifiers are often poorly calibrated. Post-hoc calibration provides a simple and widely used solution, but the large number of proposed methods, combined with small-scale and inconsistent evaluations, makes it difficult to determine which approaches are truly effective in practice. We introduce a large-scale, standardized benchmark for post-hoc calibration, covering nearly 2000 experiments across tabular and computer vision tasks, including binary, multiclass, and large-scale classification settings. Our benchmark aggregates predictions from a diverse set of classical models, modern deep learning architectures, and foundation models, and provides unified, reproducible implementations of dozens of calibration methods within a common evaluation framework. We argue that Post-Hoc Improvement (PHI) in proper scoring rules offers a principled alternative to traditional calibration error estimators for comparing post-hoc methods, capturing both calibration quality and potential degradation to the model's predictive performance. Using this framework, we conduct the most comprehensive empirical study of post-hoc calibration to date. Our results reveal consistent patterns across domains: smooth calibration functions outperform binning-based approaches, dedicated multiclass methods are essential in high-dimensional settings, and generic machine learning models are not competitive without calibration-specific design. To facilitate future research, we release all data, code, and evaluation tools, providing a plug-and-play benchmark for developing and comparing calibration methods.

As language models improve and become increasingly deployed to solve a variety of tasks, trustworthiness becomes essential. Calibration is a good proxy for trust: well-calibrated confidence estimates help inform the risk versus reward tradeoff when trusting a specific model output. Unfortunately, even as models improve, they remain poorly calibrated, often biasing towards overconfidence. Additionally, calibration can be gamed: a policy that always predicts the base rate is perfectly calibrated, but completely uninformative. To resolve this, we develop a new metric, expected utility renormalized by the oracle (EURO), that balances calibration and informativeness. We also propose a general-purpose activation-based confidence, utility, and trust estimation protocol (ACUTE) to appropriately adjudicate uncertainty. The ACUTE protocol provides flexible, sample-efficient, and compute-efficient confidence estimators for 3 tasks including multiple choice question answering, tool-calling, and scientific document summarization across 6 models from 4 model families. ACUTE outperforms strong baselines on EURO, while maintaining low calibration error. Taken together, our work shows that equipping LLMs with the ACUTE protocol can improve calibration, utility, and trustworthiness in numerous settings.

Calibration, the alignment of predicted probabilities with true outcome frequencies, is essential for reliable decision-making. While extensively studied for classification and regression, calibration has not been formally addressed for probabilistic label ranking, where the goal is to predict a distribution over orderings of a label set. Naively treating rankings as classes ignores their structure and fails to capture important modalities such as pairwise and top-k predictions. We formalize calibration for label ranking and develop a hierarchy of notions covering full rankings, sub-rankings, and top-k rankings. We prove that full-rank calibration implies the others but not conversely, and sub-ranking and top-k calibration are incomparable. Empirically, we find popular label ranking models are often poorly calibrated, with substantial differences between sub-ranking and top-k metrics. Applying our framework to RLHF reward models, we find that calibration correlates strongly but not perfectly with benchmark accuracy, suggesting it captures a meaningful quality dimension beyond top-1 accuracy. These findings motivate future work on understanding the downstream effects of miscalibration and developing methods to correct it.

Large language models are increasingly deployed as autonomous agents executing multi-step workflows in real-world software environments. However, existing agent benchmarks suffer from three critical limitations: (1) trajectory-opaque grading that checks only final outputs, (2) underspecified safety and robustness evaluation, and (3) narrow modality coverage and interaction paradigms. We introduce Claw-Eval, an end-to-end evaluation suite addressing all three gaps. It comprises 300 human-verified tasks spanning 9 categories across three groups (general service orchestration, multimodal perception and generation, and multi-turn professional dialogue). Every agent action is recorded through three independent evidence channels (execution traces, audit logs, and environment snapshots), enabling trajectory-aware grading over 2,159 fine-grained rubric items. The scoring protocol evaluates Completion, Safety, and Robustness, reporting Average Score, Pass@k, and Pass^k across three trials to distinguish genuine capability from lucky outcomes. Experiments on 14 frontier models reveal that: (1) trajectory-opaque evaluation is systematically unreliable, missing 44% of safety violations and 13% of robustness failures that our hybrid pipeline catches; (2) controlled error injection primarily degrades consistency rather than peak capability, with Pass^3 dropping up to 24% while Pass@3 remains stable; (3) multimodal performance varies sharply, with most models performing poorer on video than on document or image, and no single model dominating across all modalities. Beyond benchmarking, Claw-Eval highlights actionable directions for agent development, shedding light on what it takes to build agents that are not only capable but reliably deployable.

In this paper, we uncover a systematic bias in the evaluation paradigm of adopting large language models~(LLMs), e.g., GPT-4, as a referee to score and compare the quality of responses generated by candidate models. We find that the quality ranking of candidate responses can be easily hacked by simply altering their order of appearance in the context. This manipulation allows us to skew the evaluation result, making one model appear considerably superior to the other, e.g., Vicuna-13B could beat ChatGPT on 66 over 80 tested queries with ChatGPT as an evaluator. To address this issue, we propose a calibration framework with three simple yet effective strategies: 1) Multiple Evidence Calibration, which requires the evaluator model to generate multiple evaluation evidence before assigning ratings; 2) Balanced Position Calibration, which aggregates results across various orders to determine the final score; 3) Human-in-the-Loop Calibration, which introduces a balanced position diversity entropy to measure the difficulty of each example and seeks human assistance when needed. We also manually annotate the "win/tie/lose" outcomes of responses from ChatGPT and Vicuna-13B in the Vicuna Benchmark's question prompt, and extensive experiments demonstrate that our approach successfully mitigates evaluation bias, resulting in closer alignment with human judgments. We release our code and human annotation at https://github.com/i-Eval/FairEval to facilitate future research.

Recently, diffusion probabilistic models (DPMs) have achieved promising results in diverse generative tasks. A typical DPM framework includes a forward process that gradually diffuses the data distribution and a reverse process that recovers the data distribution from time-dependent data scores. In this work, we observe that the stochastic reverse process of data scores is a martingale, from which concentration bounds and the optional stopping theorem for data scores can be derived. Then, we discover a simple way for calibrating an arbitrary pretrained DPM, with which the score matching loss can be reduced and the lower bounds of model likelihood can consequently be increased. We provide general calibration guidelines under various model parametrizations. Our calibration method is performed only once and the resulting models can be used repeatedly for sampling. We conduct experiments on multiple datasets to empirically validate our proposal. Our code is at https://github.com/thudzj/Calibrated-DPMs.

Speech deepfake detectors usually emit one score per utterance, but a borderline score does not reveal why two examples differ during retrospective error analysis. We ask whether a final score can be calibrated from component fields while keeping those fields visible for inspection. We build a decision record with a passive detector score and a score from a probe applied to a marked copy. It also includes retrieval support held out of the evaluated family, a margin from a support-set profile, and raw neighbor closeness. A cross-fit calibrator combines these fields and two differences between raw scores into one final score. On matched ASVspoof development data, the calibrated record reduces equal error rate (EER) by 3.48 percentage points relative to the fixed retrieval-augmented rule. It reaches 8.43% EER, whereas a passive WavLM baseline reaches 6.71% on the same subset. The record is therefore not the strongest detector in this comparison. Its value is to retain inspectable component fields while producing one scalar score for retrospective diagnosis.

Graph-augmented retrieval combines dense similarity with graph-based relevance signals such as Personalized PageRank (PPR), but these scores have different distributions and are not directly comparable. We study this as a score calibration problem for heterogeneous retrieval fusion in multi-hop question answering. Our method, PhaseGraph, maps vector and graph scores to a common unit-free scale using percentile-rank normalization (PIT) before fusion, enabling stable combination without discarding magnitude information. Across MuSiQue and 2WikiMultiHopQA, calibrated fusion improves held-out last-hop retrieval on HippoRAG2-style benchmarks: LastHop@5 increases from 75.1% to 76.5% on MuSiQue (8W/1L, p=0.039) and from 51.7% to 53.6% on 2WikiMultiHopQA (11W/2L, p=0.023), both on independent held-out test splits. A theory-driven ablation shows that percentile-based calibration is directionally more robust than min-max normalization on both tune and test splits (1W/6L, p=0.125), while Boltzmann weighting performs comparably to linear fusion after calibration (0W/3L, p=0.25). These results suggest that score commensuration is a robust design choice, and the exact post-calibration operator appears to matter less on these benchmarks.

AlphaFold confidence scores are widely used to rank protein complex predictions and to select subsets of models for downstream applications. Although these scores provide strong discriminatory power for distinguishing prediction quality, their numerical values were never intended to represent the probability that a prediction satisfies a given structural quality criterion. In this work, we calibrate commonly used confidence scores to estimate the probability that a predicted complex exceeds a chosen threshold of DockQ, an established measure of interface quality. On antibody-antigen and general multimer datasets, post-hoc isotonic calibration consistently improves agreement between predicted probabilities and observed DockQ success rates while preserving the underlying score ordering. The calibrated probabilities also provide more accurate estimates of the fraction of selected models expected to pass a given structural quality threshold. Although the mapping is fitted within a given prediction distribution and therefore transfers imperfectly across prediction domains, we show that reliable estimates are recovered from a small number of labeled target-domain samples. Because the calibration is entirely post hoc and requires no underlying AlphaFold model fine-tuning, it provides a straightforward and customizable way to convert existing model quality rankings into calibrated probabilities of passing a chosen structural quality threshold.

As AI-based code generation becomes widespread, researchers are investigating the calibration of code LLMs - ensuring their confidence scores faithfully represent the true likelihood of code correctness. To do so, we investigate multicalibration, which can capture additional factors about a coding problem, such as complexity, code length, or programming language used. We study four multicalibration approaches on three function synthesis benchmarks, using latest-generation code LLMs (Qwen3 Coder, GPT-OSS, DeepSeek-R1-Distill). Our results demonstrate that multicalibration can yield distinct improvements over both uncalibrated token likelihoods (+1.03 in skill score) and baseline calibrations (+0.37 in skill score). We study the influence of the aforementioned factors in ablations, and make our dataset (consisting of code generations, likelihoods, and correctness labels) available for future research on code LLM calibration.

Deep neural networks have demonstrated remarkable performance across numerous learning tasks but often suffer from miscalibration, resulting in unreliable probability outputs. This has inspired many recent works on mitigating miscalibration, particularly through post-hoc recalibration methods that aim to obtain calibrated probabilities without sacrificing the classification performance of pre-trained models. In this study, we summarize and categorize previous works into three general strategies: intuitively designed methods, binning-based methods, and methods based on formulations of ideal calibration. Through theoretical and practical analysis, we highlight ten common limitations in previous approaches. To address these limitations, we propose a probabilistic learning framework for calibration called h-calibration, which theoretically constructs an equivalent learning formulation for canonical calibration with boundedness. On this basis, we design a simple yet effective post-hoc calibration algorithm. Our method not only overcomes the ten identified limitations but also achieves markedly better performance than traditional methods, as validated by extensive experiments. We further analyze, both theoretically and experimentally, the relationship and advantages of our learning objective compared to traditional proper scoring rule. In summary, our probabilistic framework derives an approximately equivalent differentiable objective for learning error-bounded calibrated probabilities, elucidating the correspondence and convergence properties of computational statistics with respect to theoretical bounds in canonical calibration. The theoretical effectiveness is verified on standard post-hoc calibration benchmarks by achieving state-of-the-art performance. This research offers valuable reference for learning reliable likelihood in related fields.

Controlling an acting LLM agent means deciding, at each step of a long-horizon task, whether to trust its next action, abstain, or hand off to a human. That decision needs a per-step confidence signal predicting whether the current rollout will eventually succeed. The agent's own self-reported verbalized confidence is the natural candidate, but existing reinforcement learning methods calibrate it against the correctness of a decision at the current step. We propose a method to calibrate it against the eventual outcome of a multi-step episode: under GRPO, the group of rollouts for a task supplies success labels, and a proper scoring rule scores each step's confidence against the realized outcome. In offline RL on recorded trajectories of a Qwen2.5-7B agent in ALFWorld, TextWorld, and WebShop environments, the AUROC of step confidence against episode outcome rises from near chance to 0.68-0.85, and the top-10% of steps the agent is most confident about succeed ~4.3x more often than average. In online RL on the live ALFWorld agent, four calibration rewards improve early-step AUROC, expected calibration error goes from 0.42 to below 0.12, and selective execution improves over both the untrained agent and a format-only RL control, without degrading task success. We also introduce an evaluation protocol that separates episode depth, task difficulty, and attempt-level foresight, which shows where the gain lies: the trained agent recognizes from its first step which tasks it will fail, while telling a succeeding attempt from a failing one at the same task is harder, with some early-window signal emerging on TextWorld.

Overconfidence and underconfidence in machine learning classifiers is measured by calibration: the degree to which the probabilities predicted for each class match the accuracy of the classifier on that prediction. How one measures calibration remains a challenge: expected calibration error, the most popular metric, has numerous flaws which we outline, and there is no clear empirical understanding of how its choices affect conclusions in practice, and what recommendations there are to counteract its flaws. In this paper, we perform a comprehensive empirical study of choices in calibration measures including measuring all probabilities rather than just the maximum prediction, thresholding probability values, class conditionality, number of bins, bins that are adaptive to the datapoint density, and the norm used to compare accuracies to confidences. To analyze the sensitivity of calibration measures, we study the impact of optimizing directly for each variant with recalibration techniques. Across MNIST, Fashion MNIST, CIFAR-10/100, and ImageNet, we find that conclusions on the rank ordering of recalibration methods is drastically impacted by the choice of calibration measure. We find that conditioning on the class leads to more effective calibration evaluations, and that using the L2 norm rather than the L1 norm improves both optimization for calibration metrics and the rank correlation measuring metric consistency. Adaptive binning schemes lead to more stablity of metric rank ordering when the number of bins vary, and is also recommended. We open source a library for the use of our calibration measures.

Research agents explain planned experiments. We measure predictive credit with paired forecasts sharing an intervention, forecaster, and outcome while varying description, matched explanation, and donor context. Five checks track commitment, delivery, predictive gain, alignment, and known-signal uptake. Across 336 prospective states in controlled learning, 12 Tox21 endpoints, and 24 OpenML tasks, v5's frozen credit decision was inconclusive. Tox21's preregistered ROC AUC interval-score harm test was unmet (D-M=-.0026, 95 percent interval [-.0174, .0104]); OpenML's joint formation, point-equivalence, and repeatability rule was unmet. Matched point-accuracy gains over description remained unconfirmed, and Tox21/OpenML seed-donor intervals spanned zero. Under requested DeepSeek V4 Pro, matched and donor cards reduced secondary Tox21 drift by 64.5 and 59.1 percent. A DeepSeek V4 Flash replay raised matched point MAE from .01823 to .02020 and missed matched-donor interval-score equivalence. OpenML full-card assignment widened nominal 80 percent intervals by 21 percent, with 49.3 percent coverage versus 51.4 percent for description and content in 66/144 cards. Direct-text Flash delivered all 144 notes without detectable matched point-accuracy gain. A researcher-authored mechanism positive control lowered point MAE by 2.60 percentage points versus description. The protocol measures predictive credit for research-agent benchmarks and scientific forecasting; natural-explanation credit remained unconfirmed at the tested donor resolutions.

Applications such as weather forecasting and personalized medicine demand models that output calibrated probability estimates---those representative of the true likelihood of a prediction. Most models are not calibrated out of the box but are recalibrated by post-processing model outputs. We find in this work that popular recalibration methods like Platt scaling and temperature scaling are (i) less calibrated than reported, and (ii) current techniques cannot estimate how miscalibrated they are. An alternative method, histogram binning, has measurable calibration error but is sample inefficient---it requires O(B/ε^2) samples, compared to O(1/ε^2) for scaling methods, where B is the number of distinct probabilities the model can output. To get the best of both worlds, we introduce the scaling-binning calibrator, which first fits a parametric function to reduce variance and then bins the function values to actually ensure calibration. This requires only O(1/ε^2 + B) samples. Next, we show that we can estimate a model's calibration error more accurately using an estimator from the meteorological community---or equivalently measure its calibration error with fewer samples (O(B) instead of O(B)). We validate our approach with multiclass calibration experiments on CIFAR-10 and ImageNet, where we obtain a 35% lower calibration error than histogram binning and, unlike scaling methods, guarantees on true calibration. In these experiments, we also estimate the calibration error and ECE more accurately than the commonly used plugin estimators. We implement all these methods in a Python library: https://pypi.org/project/uncertainty-calibration

Video world models approximate the stochastic distribution of physical outcomes through generative sampling, but existing benchmarks score individual generations or compare distributions coarsely over a whole dataset, leaving the fine-grained aleatoric uncertainty of specific phenomena untested. We introduce CaliBench, which scores outcomes in a physically interpretable discrete space - a bin index, a die face, a suit, a colour - rather than a learned feature space such as in FID, so the distance from a known reference distribution is measured directly. We curate outcome spaces whose reference is known in closed form (binomial Galton boards, Bernoulli forks, uniform dice/cards/lottery, a skewed European-roulette colour), enabling an exact calibration test. We decompose performance into two orthogonal axes that a single accuracy metric conflates: scorability, the fraction of generations yielding a scoreable outcome, and calibration, the total variation distance from the reference on that sample. A chi-squared test assesses significance; as calibration is its null hypothesis it can evidence only miscalibration, and at N=32 per cell detects only large deviations. We apply it to nine scenes and six image-to-video models (WAN-2.7, SeeDance-2.0, HappyHorse-1.0, Veo 3.1, Runway Gen-4.5, Cosmos3-Super), 32 generations each. Models consistently concentrate probability mass on a few outcomes rather than reproducing the reference. Most scene-model combinations are significantly miscalibrated, in the extreme collapsing to one outcome, as Veo 3.1 does on dice. On roulette, generations often leave the ball ambiguously placed, giving several models low scorability. Performance varies by scene: no model dominates all nine. We release the protocol and a metric (mean normalised total variation, mnTV) for comparing new models against our results.

Learned post-training quantization (PTQ) calibration fits quantizer parameters on a small calibration set, and for a text embedder the natural supervision is retrieval relevance. We ask two questions that are usually conflated: how far to calibrate, and whether to calibrate at all. Across three embedders (e5-base-v2, BGE-M3, EmbeddingGemma-300m) at 4-bit weight quantization, 161 calibration runs, the outcome of a fixed procedure ranges from a gain to catastrophic degradation, and the driver is the dose of the update, measured as the functional displacement D_func of the quantized model from its uncalibrated start. On EmbeddingGemma the same learning-rate grid gives +1.47 to -71.53 points of in-domain NDCG@10 change; a label-free cosine objective collapses at the same dose (-69.52 vs -71.63), so the collapse is a property of learnable calibration rather than of retrieval supervision. Objective sensitivity is heterogeneous: on e5, ranking calibration is substantially more robust than cosine at matched displacement (+14.16 +/- 4.34 pp, 95% t-CI [+3.37, +24.96], three paired slates), whereas on BGE-M3 the ordering is directionally opposite but unresolved ([-5.18, +2.38]). Dose is the main axis but not a sufficient statistic: equal D_func reached along different paths differs by up to 1.61 pp. We therefore treat calibration as an optional intervention. A pre-registered rule that accepts a checkpoint only when a per-query paired confidence interval on a 25% calibration-validation split clears the uncalibrated baseline was run on 6 independent slates and five arms. It never selected a checkpoint below its slate's baseline (0 harmful selections in 30 decisions), retained 8 of 11 beneficial updates, and paid a mean oracle regret of 1.52 pp with one abstention that forfeited 11.13 pp, a safety-opportunity trade-off, not a free improvement. We report what remains open, including label efficiency.

We introduce TuneJury, an open, instance-level pairwise reward model for text-to-music that predicts a music preference score from a text prompt and an audio clip. The released checkpoint is trained on publicly available human-preference labels covering arena-style (A vs. B) votes, metric-alignment preference pairs, crowdsourced pairwise comparisons, and expert aesthetic ratings. The predicted score margin between two clips is well calibrated on our held-out test split, supporting data filtering via a simple score threshold. TuneJury generalizes to both held-out test pairs and out-of-distribution benchmarks, remaining competitive with prior baselines on the latter. For generators released after training, we introduce anchor calibration, a post-hoc, per-system Bradley-Terry calibration that recovers agreement at substantially better data efficiency than from-scratch retraining. The same frozen reward drives consistent reward-axis gains across three downstream applications: inference-time best-of-N selection, DITTO-style latent optimization, and expert-iteration post-training. TuneJury is available at https://github.com/yonghyunk1m/TuneJury.

Calibration of language models -- the alignment between expressed or implicit confidence and empirical correctness -- is a well-studied subfield within NLP. Methods to measure it already exist. The problem is adoption: outside this subfield, NLP research regularly introduces new models, datasets, and benchmarks without checking whether the model's confidence scores are meaningful. We argue that this adoption gap is a major obstacle to trustworthy LLM evaluation. Miscalibration causes problems in two distinct areas: at deployment, where overconfident mistakes cause real harm, and inside the research pipeline, where methods like LLM-as-a-judge, synthetic data generation, and active learning rely on calibrated confidence without verifying it. Standard calibration metrics only require two inputs per example: a confidence score and a correctness judgment. Most benchmarks in use today already provide both, meaning calibration can be reported immediately. For open-ended generation, however, defining these two inputs is still an open challenge. We argue that each NLP subfield should pair its main performance metric with a calibration score and call for treating calibration as an essential property of every model rather than a niche topic.

Recent text-to-video (T2V) technology advancements, as demonstrated by models such as Gen2, Pika, and Sora, have significantly broadened its applicability and popularity. Despite these strides, evaluating these models poses substantial challenges. Primarily, due to the limitations inherent in automatic metrics, manual evaluation is often considered a superior method for assessing T2V generation. However, existing manual evaluation protocols face reproducibility, reliability, and practicality issues. To address these challenges, this paper introduces the Text-to-Video Human Evaluation (T2VHE) protocol, a comprehensive and standardized protocol for T2V models. The T2VHE protocol includes well-defined metrics, thorough annotator training, and an effective dynamic evaluation module. Experimental results demonstrate that this protocol not only ensures high-quality annotations but can also reduce evaluation costs by nearly 50%. We will open-source the entire setup of the T2VHE protocol, including the complete protocol workflow, the dynamic evaluation component details, and the annotation interface code. This will help communities establish more sophisticated human assessment protocols.

Large language models (LLMs) have demonstrated remarkable capabilities out of box for a wide range of applications, yet accuracy still remains a major growth area, especially in mission-critical domains such as biomedicine. An effective method to calibrate the confidence level on LLM responses is essential to automatically detect errors and facilitate human-in-the-loop verification. An important source of calibration signals stems from expert-stipulated programmatic supervision, which is often available at low cost but has its own limitations such as noise and coverage. In this paper, we introduce a Pareto optimal self-supervision framework that can leverage available programmatic supervision to systematically calibrate LLM responses by producing a risk score for every response, without any additional manual efforts. This is accomplished by learning a harmonizer model to align LLM output with other available supervision sources, which would assign higher risk scores to more uncertain LLM responses and facilitate error correction. Experiments on standard relation extraction tasks in biomedical and general domains demonstrate the promise of this approach, with our proposed risk scores highly correlated with the real error rate of LLMs. For the most uncertain test instances, dynamic prompting based on our proposed risk scores results in significant accuracy improvement for off-the-shelf LLMs, boosting GPT-3 results past state-of-the-art (SOTA) weak supervision and GPT-4 results past SOTA supervised results on challenging evaluation datasets.

Agent benchmarks increasingly evaluate repository editing, web research, terminal use, and long-horizon interaction. Their scores support capability claims only when the evaluation protocol keeps the intended capability necessary for success. Recent reward-hacking benchmarks and system reports show that agents can instead recover public solutions, read evaluation artifacts, infer generator structure, manipulate feedback, or benefit from invalid scoring paths; existing responses do not provide a common procedure for attributing these shortcuts and quantifying their effect across benchmarks. We formulate protocol validity and introduce HackDetect, a post-hoc audit that identifies an exposure, determines how the agent used it, and assesses whether the resulting score is misleading. We quantify score inflation with the Mislead gap, defined as the exploit score minus the intended score. We audit 2,385 traces across 15 agent benchmarks and find evidence of exposures and reward hacking in 67.0% of Frontier Science traces and 66.7% of AutoLab tasks. Across paired comparisons, we measure score inflation of 0.45-1.00, showing that benchmark reports should provide evidence that scores reflect the intended capability.

LeWorldModel trains a latent world model with a prediction loss and a single anti-collapse regulariser, and reports approximately 87% of goals reached on TwoRoom, its simplest diagnostic environment. We reproduce that result by independent reimplementation on roughly $25 of rented compute, with all evaluation on one laptop CPU. We reach 94.0% at the repository's evaluation goal offset, against 84.0% for the authors' own released checkpoint measured under our protocol on identical episodes, and we reproduce the reported representation result directly (position probe Pearson r = 0.9988 against a reported 0.996). Reaching that point required four conventions that determine the outcome and appear in no released configuration file: dense action gathering across a frameskip block, a programmatically-set action-encoder width, ImageNet pixel normalisation, and action z-scoring. A reproducer following the released configurations alone obtains a model whose predictor cannot converge. The evaluation protocol is itself contested by the released material. The paper's appendix and the repository's configuration specify different goal offsets and step budgets; on the authors' own weights these yield 14.0% and 84.0%, and only the configuration's values reproduce the reported figure. On fifty identical episodes, changing nothing but how the goal is constructed moves that checkpoint from 84.0% to 8.0%. Two findings generalise. One-step prediction accuracy does not predict long-horizon planning success: across three checkpoints spanning a sevenfold range in prediction error, including the authors' own, it orders short-horizon success monotonically and fails to order long-horizon success at all. And a batch normalisation layer inflated our reported validation loss by up to a factor of 300, concealing a training loss that was flat throughout.

Reliable usage of object detectors require them to be calibrated -- a crucial problem that requires careful attention. Recent approaches towards this involve (1) designing new loss functions to obtain calibrated detectors by training them from scratch, and (2) post-hoc Temperature Scaling (TS) that learns to scale the likelihood of a trained detector to output calibrated predictions. These approaches are then evaluated based on a combination of Detection Expected Calibration Error (D-ECE) and Average Precision. In this work, via extensive analysis and insights, we highlight that these recent evaluation frameworks, evaluation metrics, and the use of TS have notable drawbacks leading to incorrect conclusions. As a step towards fixing these issues, we propose a principled evaluation framework to jointly measure calibration and accuracy of object detectors. We also tailor efficient and easy-to-use post-hoc calibration approaches such as Platt Scaling and Isotonic Regression specifically for object detection task. Contrary to the common notion, our experiments show that once designed and evaluated properly, post-hoc calibrators, which are extremely cheap to build and use, are much more powerful and effective than the recent train-time calibration methods. To illustrate, D-DETR with our post-hoc Isotonic Regression calibrator outperforms the recent train-time state-of-the-art calibration method Cal-DETR by more than 7 D-ECE on the COCO dataset. Additionally, we propose improved versions of the recently proposed Localization-aware ECE and show the efficacy of our method on these metrics as well. Code is available at: https://github.com/fiveai/detection_calibration.

Decision models such as Jev answer questions with probabilities, which are only useful if they are calibrated. Open-source reproductions rely on supervised fine-tuning plus temperature scaling, while reinforcement learning from verifiable rewards (RLVR) makes reasoning models overconfident. We present a working implementation of reinforcement learning for calibrated decisions (RLCD) for reasoning models: the model samples a rationale, and we score the answer distribution it commits to afterwards with a strictly proper scoring rule. A variance identity shows that scoring the mixture of several samples rewards disagreeing rationales, and that RLVR is exactly this mixture objective without its diversity term. Optimized naively, the per-rationale objective either switches reasoning off or is drowned out by policy-gradient noise, which leads to a two-stage recipe: calibrate, then reinforce. With Qwen3-1.7B on two reasoning tasks (3 seeds, paired tests), RLCD matches or beats SFT, RFT/STaR and GRPO (each temperature-scaled) in accuracy and beats all of them in selective prediction; on GSM8K answer verification a single query decides \gvTwoCovFive\% of the items at le5\% error, versus \gvGrpoCovFive\% for GRPO. When uncertainty comes from annotator disagreement, RLCD provably cannot beat cross-entropy. Code and results: https://github.com/ZimmyGao/openjev-rlcd.

We introduce a framework for calibrating machine learning models so that their predictions satisfy explicit, finite-sample statistical guarantees. Our calibration algorithms work with any underlying model and (unknown) data-generating distribution and do not require model refitting. The framework addresses, among other examples, false discovery rate control in multi-label classification, intersection-over-union control in instance segmentation, and the simultaneous control of the type-1 error of outlier detection and confidence set coverage in classification or regression. Our main insight is to reframe the risk-control problem as multiple hypothesis testing, enabling techniques and mathematical arguments different from those in the previous literature. We use the framework to provide new calibration methods for several core machine learning tasks, with detailed worked examples in computer vision and tabular medical data.

Large Language Models excel at reasoning and generation but their autoregressive, token-by-token inference makes them unnecessarily slow, expensive, and poorly calibrated for the many routine decisions (routing, gating, scoring, classification) that modern AI systems require at sub-second latency. A growing class of models addresses this gap: System 1 decision models, architectures that map an input directly to a typed, structured decision in a single forward pass. This guide provides a comprehensive, practitioner-oriented treatment of the field. We propose a five-family taxonomy covering encoder-based classifiers, reward models, non-autoregressive decision engines, contrastive decision models, and extraction models, and we provide in-depth coverage of each family's architecture, training methodology, and deployment trade-offs. A central theme is calibration: we develop the theory of proper scoring rules, analyze Reinforcement Learning for Calibrated Decisions (RLCD), and examine recently discovered phenomena including IIA violations, ordinal-scale bias, and few-shot recalibration under distribution shift. On the integration side, we describe how System 1 models serve as decision infrastructure within agentic workflows and System 1 + System 2 cascades that route between fast and slow models based on calibrated confidence. The guide covers 18 application domains, maps the rapidly growing open-source ecosystem (over 70 model reproductions in weeks), provides runnable code throughout, and includes a practitioner's checklist, an open-source model zoo, and 22 self-assessment questions with detailed answers. Spanning 190 references, it is intended as a unified entry point for researchers and engineers working at the intersection of fast decision-making, calibration, and production AI systems.

Evaluating software agents on a full benchmark is expensive. A small task subset can preserve agent rankings while yielding inaccurate scores. Historical responses provide useful structure, but tasks with identical source responses can have different outcomes for a new target agent. We introduce Hierarchical Score Posterior Recovery (HSPR), which combines source-fitted response factors with uncertain group success rates. Observed target outcomes update the posteriors over the target latent state and group rates; posterior-guided acquisition selects subsequent tasks. Its score-based criterion equals the expected one-step reduction in full-score posterior variance. Observed successes and posterior-predictive counts for unqueried tasks yield a full-score posterior. The final estimate uses its median with source-selected calibration. At a 10% task budget on four task banks from the SWE-bench family, HSPR achieves the lowest mean absolute error (MAE) on three banks and the lowest maximum absolute error (MaxErr) on all four among the methods compared. Removing group uncertainty or replacing adaptive queries increases macro MAE in the respective controls. Macro MAE decreases across 5-20% budgets.

We present an online post-hoc calibration method, called Online Platt Scaling (OPS), which combines the Platt scaling technique with online logistic regression. We demonstrate that OPS smoothly adapts between i.i.d. and non-i.i.d. settings with distribution drift. Further, in scenarios where the best Platt scaling model is itself miscalibrated, we enhance OPS by incorporating a recently developed technique called calibeating to make it more robust. Theoretically, our resulting OPS+calibeating method is guaranteed to be calibrated for adversarial outcome sequences. Empirically, it is effective on a range of synthetic and real-world datasets, with and without distribution drifts, achieving superior performance without hyperparameter tuning. Finally, we extend all OPS ideas to the beta scaling method.

Training terminal agents requires executable and verifiable tasks that are not merely solvable, but appropriately challenging for learning. Executable validation establishes feasibility, yet does not reveal how a task behaves relative to a given solver setting. In this paper, we present CalibForge, an autonomous terminal-task synthesis system that uses verified solver behavior to revise candidate tasks through adversarial solver calibration. Multi-solver calibration targets disagreement within a heterogeneous solver pool, whereas contrastive solver calibration targets a designated strong-pass/weak-fail relation; both operationalize a solver-relative learnable zone anchored in demonstrated solvability. Using CalibForge, we construct 5,431 calibrated terminal tasks. Our ablations show that both strategies yield more effective supervision than authoring and validation alone or ordinary single-solver feedback. Models trained on the full collection achieve 32.58% and 47.57% on Terminal-Bench 2.0. The largest improvements over the corresponding base model reach 24.71 percentage points on Terminal-Bench 2.0, 27.68 points on SWE-bench Pro, and 30.04 points on Doc2Repo. Together, these results support solver-relative learnability as a practical target for constructing effective and transferable agent training data.

Conformal prediction is often calibrated with a single pooled threshold, but this can hide cross-group heterogeneity in score distributions and distort group-wise coverage. We study this phenomenon through the population score distributions underlying split conformal calibration. First, we derive a conservation law and lower bound showing that pooled calibration incurs irreducible group-wise coverage distortion at a scale set by cross-group quantile heterogeneity. Second, we demonstrate that the two leading fairness definitions for conformal prediction, Equalized Coverage and Equalized Set Size, are fundamentally in tension. Third, we quantify the cost of moving between policies which treat groups separately or pool them. Experiments on synthetic and real data confirm the same bidirectional trade-off after finite-sample calibration. Our results show that, for the policy families studied here, calibration choice does not remove cross-group heterogeneity; it determines whether the resulting distortion appears in the coverage or size dimension, providing a principled lens for analyzing fairness-oriented calibration choices in practice.

Calibration measures and reliability diagrams are two fundamental tools for measuring and interpreting the calibration of probabilistic predictors. Calibration measures quantify the degree of miscalibration, and reliability diagrams visualize the structure of this miscalibration. However, the most common constructions of reliability diagrams and calibration measures -- binning and ECE -- both suffer from well-known flaws (e.g. discontinuity). We show that a simple modification fixes both constructions: first smooth the observations using an RBF kernel, then compute the Expected Calibration Error (ECE) of this smoothed function. We prove that with a careful choice of bandwidth, this method yields a calibration measure that is well-behaved in the sense of (B{\l}asiok, Gopalan, Hu, and Nakkiran 2023a) -- a consistent calibration measure. We call this measure the SmoothECE. Moreover, the reliability diagram obtained from this smoothed function visually encodes the SmoothECE, just as binned reliability diagrams encode the BinnedECE. We also provide a Python package with simple, hyperparameter-free methods for measuring and plotting calibration: `pip install relplot\`.

We address the problem of uncertainty calibration and introduce a novel calibration method, Parametrized Temperature Scaling (PTS). Standard deep neural networks typically yield uncalibrated predictions, which can be transformed into calibrated confidence scores using post-hoc calibration methods. In this contribution, we demonstrate that the performance of accuracy-preserving state-of-the-art post-hoc calibrators is limited by their intrinsic expressive power. We generalize temperature scaling by computing prediction-specific temperatures, parameterized by a neural network. We show with extensive experiments that our novel accuracy-preserving approach consistently outperforms existing algorithms across a large number of model architectures, datasets and metrics.

Large language model (LLM) agents are increasingly deployed over Model Context Protocol (MCP) servers, yet the benchmarks used to evaluate them score the final answer or a fixed "ground-truth" list of tools, both of which are fragile once the underlying data is live and stateful. We present DynamicMCPBench, a reusable framework rather than a fixed dataset. A practitioner can run it on their own MCP servers to test models on their own tasks, or let it collect servers automatically to measure a model's general ability to solve agentic tasks. Given the servers and any set of models, it generates realistic goals, pursues each one live to record a successful trajectory, distills that trajectory into path-agnostic effect checkpoints, and scores an agent on whether it reproduces those effects, never on the final answer. To show what the framework reveals, we run it at scale: 24 models over 121 servers and 750 tasks spread evenly over 15 task categories (50 each), where each category targets a distinct tool-use challenge of the generated questions. Each task is scored by pass^3: it counts as solved only if all three independent attempts succeed. Even the strongest agents solve only about half of the tasks, 31% of tasks are solved by no model at all, and accuracy collapses as the required tool chain grows longer (from 39% on the shortest chains to 13% on the longest). A human validation study confirms the automatic scoring is reliable (chance-corrected agreement of 0.76). DynamicMCPBench thus turns benchmark construction into something practitioners can rerun on their own servers and models, while exposing a consistent inability of current agents to handle long, multi-step agentic tasks.

Pancreatic ductal adenocarcinoma (PDAC) segmentation on contrast-enhanced CT is inherently ambiguous: inter-rater disagreement among experts reflects genuine uncertainty rather than annotation noise. Standard deep learning approaches assume a single ground truth, producing probabilistic outputs that can be poorly calibrated and difficult to interpret under such ambiguity. We present TwinTrack, a framework that addresses this gap through post-hoc calibration of ensemble segmentation probabilities to the empirical mean human response (MHR) -the fraction of expert annotators labeling a voxel as tumor. Calibrated probabilities are thus directly interpretable as the expected proportion of annotators assigning the tumor label, explicitly modeling inter-rater disagreement. The proposed post-hoc calibration procedure is simple and requires only a small multi-rater calibration set. It consistently improves calibration metrics over standard approaches when evaluated on the MICCAI 2025 CURVAS-PDACVI multi-rater benchmark.

Authorship verification (AV) is the task of determining whether two texts were written by the same author. In a forensic context, the strength of AV evidence can be quantified using likelihood ratios. Most AV methods are score-based and deriving well-calibrated likelihood ratios from these scores requires a separate calibration model. This, in turn, requires additional amounts of case-relevant data, which is often time-consuming to obtain and prepare. This study proposes two novel normalisation techniques, the Square Root Correction and the Hapax Correction, for deriving likelihood ratios from the AV method LambdaG without the need of a calibration model (Nini et al. 2026). These corrections are designed to mitigate the overestimation of evidential strength that may result from long or highly repetitive texts. Performance is evaluated against logistic regression calibration across fifteen corpora and a range of text lengths (100-9,500 tokens), using the log-likelihood ratio cost (Cllr). The proposed methods achieve performance comparable to logistic regression calibration, with the Hapax Correction outperforming it in approximately 45% of tests (weighted by corpora). Furthermore, performance was more frequently close (within 5%) when the Hapax Correction was outperformed by logistic regression calibration, compared with the reverse comparison. Eliminating the need to train a calibration model reduces data-requirements, time and complexity, thereby increasing the accessibility and transparency of forensic text comparison. This combination of empirical performance and practical advantages supports the adoption of the proposed methods in forensic settings.

Search-augmented language agents should retrieve external information only when necessary and ground their answers in retrieved evidence. Yet terminal correctness assigns equal credit to direct answers from parametric knowledge, evidence-grounded answers, and answers recalled after redundant search. Process supervision and LLM judges provide richer feedback but can require high-cost annotation or inference; low-cost, non-contrastive policy confidence signals do not directly identify whether retrieved evidence changes an answer. We propose Search-G1, a reward framework that couples retrieval necessity with evidence reliance through two intervention-calibrated hidden-state readouts for selective grounded search. A prompt readout predicts closed-book sufficiency, whose complement proxies policy-relative retrieval necessity; an answer-commit readout predicts evidence reliance as sensitivity to evidence deletion conditional on the realized reasoning prefix. A correctness-first reward gates evidence-reliance credit by retrieval necessity, rewards sufficient direct answers, and penalizes repeated search. After calibration, scoring needs neither process annotations nor LLM-judge inference. As policy-relative targets, trajectory distributions, and representations may drift, Search-G1 periodically regenerates targets and refits its readouts. Across four QA benchmarks and two scales, Search-G1 achieves competitive task accuracy, with stronger counterfactual evidence dependence and fewer searches in the main comparisons, complemented by independent human audits of answer support.

A well-calibrated model should express confidence that matches its actual accuracy -- when it claims 80\% confidence, it should be correct 80\% of the time. While large language models (LLMs) have achieved remarkable performance across diverse tasks, their epistemic calibration remains poorly understood. We introduce KalshiBench, a benchmark of 300 prediction market questions from Kalshi, a CFTC-regulated exchange, with verifiable real-world outcomes occurring after model training cutoffs. Unlike traditional benchmarks measuring accuracy on static knowledge, KalshiBench evaluates whether models can appropriately quantify uncertainty about genuinely unknown future events. We evaluate five frontier models -- Claude Opus 4.5, GPT-5.2, DeepSeek-V3.2, Qwen3-235B, and Kimi-K2 -- and find systematic overconfidence across all models. Even the best-calibrated model (Claude Opus 4.5, ECE=0.120) shows substantial calibration errors, while reasoning-enhanced models like GPT-5.2-XHigh exhibit worse calibration (ECE=0.395) despite comparable accuracy. Critically, only one model achieves a positive Brier Skill Score, indicating most models perform worse than simply predicting base rates. Our findings suggest that scaling and enhanced reasoning do not automatically confer calibration benefits, highlighting epistemic calibration as a distinct capability requiring targeted development.

LLM cascades and model routing promise lower inference cost by sending easy queries to a small model and escalating hard ones to a large model, but most deployed routers use uncalibrated confidence scores and require per-workload threshold tuning. We present UCCI, a calibration-first router that maps token-level margin uncertainty to a per-query error probability via isotonic regression and selects the escalation threshold by constrained cost minimization. Under three explicit assumptions, threshold policies on the calibrated score are cost-optimal, and isotonic calibration achieves O(n^{-1/3}) sample complexity for expected calibration error (ECE). On a production named entity recognition workload of 75,000 queries served by 4B and 12B instruction-tuned LLMs on H100 GPUs, UCCI cuts inference cost by 31% (95% CI: [27%, 35%]) at micro-F1 = 0.91 while reducing ECE from 0.12 to 0.03. At the same operating point, UCCI beats entropy thresholding, split-conformal routing, and a FrugalGPT-style learned threshold. All cascade results use end-to-end routing on actual model outputs and measured H100 latency, not simulated routing from global accuracies or nominal API prices.

Model calibration aims to align confidence with prediction correctness. The Cross-Entropy (CE) loss is widely used for calibrator training, which enforces the model to increase confidence on the ground truth class. However, we find the CE loss has intrinsic limitations. For example, for a narrow misclassification, a calibrator trained by the CE loss often produces high confidence on the wrongly predicted class (e.g., a test sample is wrongly classified and its softmax score on the ground truth class is around 0.4), which is undesirable. In this paper, we propose a new post-hoc calibration objective derived from the aim of calibration. Intuitively, the proposed objective function asks that the calibrator decrease model confidence on wrongly predicted samples and increase confidence on correctly predicted samples. Because a sample itself has insufficient ability to indicate correctness, we use its transformed versions (e.g., rotated, greyscaled and color-jittered) during calibrator training. Trained on an in-distribution validation set and tested with isolated, individual test samples, our method achieves competitive calibration performance on both in-distribution and out-of-distribution test sets compared with the state of the art. Further, our analysis points out the difference between our method and commonly used objectives such as CE loss and mean square error loss, where the latters sometimes deviates from the calibration aim.

Conformal prediction guarantees marginal coverage, but a single calibration threshold can still spread that coverage unevenly, over-covering easy regions and under-covering hard ones. SimplexUQ is, to our knowledge, the first benchmark and reproducible protocol for measuring this allocation problem on simplex-valued predictions; it compares existing conformal wrappers rather than proposing a new one. Its task suite, SimplexTasks-12, combines six controlled synthetic regimes with six frozen-predictor real tasks spanning class probabilities, topic mixtures, spectral abundances, cell-type fractions, age distributions, and emotion mixtures. Each comparison fixes the predictor, score, and response-free stratification map, varies only the wrapper, and reports marginal coverage, worst-stratum coverage, max disparity, and within-task radius and compute. Global calibration can look valid while failing badly: on CIFAR-10 it attains 0.900 marginal coverage but only 0.542 in the worst entropy stratum, and Mondrian calibration raises that stratum to 0.886 while reducing max disparity from 0.358 to 0.022. No wrapper dominates, however. Under smooth synthetic heterogeneity, several repairs are competitive; fixed-map analyses show that rankings depend on the evaluation groups and protocol; and in a 12-task comparison, Mondrian has lower disparity on its single target partition for all 12 tasks, whereas BatchMVP has lower disparity over overlapping groups on five. These are empirical comparisons, not new coverage guarantees. A controlled predictor-bias sweep shows that removing predictor bias only partly reduces global-threshold disparity. We release task cards, result provenance, permitted derived arrays, and rebuild instructions, and treat wrapper selection as a diagnostic comparison rather than a universal ranking.

State-of-the-art (SOTA) LLMs have progressed from struggling on proof-based Olympiad problems to solving most of the IMO 2025 problems, with leading systems reportedly handling 5 of 6 problems. Given this progress, we assess how well these models can grade proofs: detecting errors, judging their severity, and assigning fair scores beyond binary correctness. We study proof-analysis capabilities using a corpus of 90 Gemini 2.5 Pro-generated solutions that we grade on a 1-4 scale with detailed error annotations, and on MathArena solution sets for IMO/USAMO 2025 scored on a 0-7 scale. Our analysis shows that models can reliably flag incorrect (including subtly incorrect) solutions but exhibit calibration gaps in how partial credit is assigned. To address this, we introduce agentic workflows that extract and analyze reference solutions and automatically derive problem-specific rubrics for a multi-step grading process. We instantiate and compare different design choices for the grading workflows, and evaluate their trade-offs. Across our annotated corpus and MathArena, our proposed workflows achieve higher agreement with human grades and more consistent handling of partial credit across metrics. We release all code, data, and prompts/logs to facilitate future research.

Closed-source frontier labs do not disclose parameter counts. Storing F facts requires at least F/(bits per parameter) weights, so factual recall lower-bounds parameter count--an intrinsic, serving-independent signal, though (as we show) a coarse one. We introduce Incompressible Knowledge Probes (IKPs), a benchmark of 1,400 factual questions spanning 7 tiers of obscurity, designed to isolate knowledge that cannot be derived by reasoning or compressed by architectural improvements. We score with no hallucination penalty (lambda = 0: IKP accuracy is simply the fraction of probed facts answered correctly), which removes both the penalty hyperparameter and the per-tier flooring choice; a full lambda x flooring ablation shows the calibration is robust across scoring choices while individual estimates are not, motivating the no-penalty default. We calibrate a log-linear mapping from IKP accuracy to parameter count on 93 open-weight models (135M-1,600B) spanning 19 vendors, achieving R^2 = 0.910; leave-one-out cross-validation confirms generalization (median fold error 1.48x, 72% within 2x, 86% within 3x). The instrument is deliberately coarse--its 90% prediction interval spans ~3x in either direction, wider than inference economics--so IKP recovers order-of-magnitude effective capacity and relative rankings, not precise parameter counts. For Mixture-of-Experts models, total parameters predict knowledge (R^2 = 0.67) better than active parameters (R^2 = 0.41). We evaluate 201 models from 27 vendors on a curated probe set (1,311 of 1,400 probes surviving name-collision and label-ambiguity filters) and report effective knowledge capacity for all major proprietary frontier models as prediction bands rather than point estimates; for heavily safety-tuned models these are lower bounds, since refusal policy can suppress tens of percentage points of otherwise-answerable capacity.

Single-prompt accuracy is the dominant way to benchmark language models, but it can miss reliability failures that matter. We evaluate a 15-model open-weight corpus, with the main reliability analyses focused on 10 instruct models across five classification and reasoning benchmarks under five prompt variants each, measuring accuracy, token-probability calibration, verbal-confidence calibration, verbal parse rate, and prompt-perturbation spread for every (model x dataset x variant) cell. We find three broad results. First, evaluation design can materially change the conclusion. Switching Expected Calibration Error (ECE) token from a raw to a label-set-normalised definition changes per-cell calibration by a mean absolute 0.149. More strikingly, pairing a chain-of-thought prompt with a first-character evaluator on ARC-Challenge reduces apparent accuracy by 72-88% across all five primary models; two independent repair procedures recover 93.8% and 102.7% of the lost performance, indicating an evaluator-side rather than model-side failure. Second, confidence signals are fragile. On MMLU-Pro, every primary model verbally reports confidence substantially above both its accuracy and its token-probability confidence on the same rows, and verbal parse rate can collapse for a single model on a single prompt variant. Third, prompt robustness does not track parameter count reliably. Across 10 instruct models, the correlation between model size and prompt-perturbation spread ranges from -0.244 to 0.474 across benchmarks. Taken together, these results show that reliability conclusions for small language models depend not only on the model being evaluated, but also on the evaluation pipeline used to measure it. We argue that calibration definitions, evaluator logic, verbal parseability, and prompt robustness should be reported explicitly when making reliability claims.

Quantifying the uncertainty of predictions made by large language models (LLMs) in binary text classification tasks remains a challenge. Calibration, in the context of LLMs, refers to the alignment between the model's predicted probabilities and the actual correctness of its predictions. A well-calibrated model should produce probabilities that accurately reflect the likelihood of its predictions being correct. We propose a novel approach that utilizes the inductive Venn--Abers predictor (IVAP) to calibrate the probabilities associated with the output tokens corresponding to the binary labels. Our experiments on the BoolQ dataset using the Llama 2 model demonstrate that IVAP consistently outperforms the commonly used temperature scaling method for various label token choices, achieving well-calibrated probabilities while maintaining high predictive quality. Our findings contribute to the understanding of calibration techniques for LLMs and provide a practical solution for obtaining reliable uncertainty estimates in binary question answering tasks, enhancing the interpretability and trustworthiness of LLM predictions.

As LLM benchmarks saturate, the evaluation community has pursued two strategies to increase difficulty: escalating knowledge demands (GPQA, HLE) or removing knowledge entirely in favor of abstract reasoning (ARC-AGI). The first conflates memorization with capability; the second divorces reasoning from the practical contexts in which it matters. We take a different approach. The Grounded Integration Measure (GIM) is a benchmark of 820 original problems (615 public, 205 private) where difficulty comes from integration; individual problems require coordinating multiple cognitive operations (constraint satisfaction, state tracking, epistemic vigilance, audience calibration) over broadly accessible knowledge, so that reasoning stays grounded in realistic tasks without being gated on specialized expertise. Each problem is an original expert-authored composition, majority with rubric-decomposed scoring (median 6 independently judged criteria). A balanced public--private split provides built-in contamination diagnostic. We calibrate a continuous response 2-parameter logistic (2PL) IRT model over >200k prompt-response pairs across 28 models, producing robust ability estimates that correctly order test-configurations even when raw accuracy is distorted by errors or missing data, addressing a common challenge in benchmark reporting. Using this framework, we present a comprehensive leaderboard spanning 22 models and 47 test-configurations (unique model, thinking-level pairs), and conduct what is to our knowledge the most extensive published study of how test-time compute trades off against model capability on a fixed benchmark: 11 models swept across 35 test-configurations. We observe that within-family configuration choices, such as thinking budget and quantization, matter as much as model selection. We release the evaluation framework, calibrated IRT parameters, and all public problems.

Autonomous language-model agents increasingly rely on installable skills and tools to complete user tasks. Static skill auditing can expose capability surface before deployment, but it cannot determine whether a particular invocation is unsafe under the current user request and runtime context. We therefore study skill invocation auditing as a continuous-risk estimation problem: given a user request, candidate skill, and runtime context, predict a score that supports ranking and triage before a hard intervention is applied. We introduce STARS, which combines a static capability prior, a request-conditioned invocation risk model, and a calibrated risk-fusion policy. To evaluate this setting, we construct SIA-Bench, a benchmark of 3,000 invocation records with group-safe splits, lineage metadata, runtime context, canonical action labels, and derived continuous-risk targets. On a held-out split of indirect prompt injection attacks, calibrated fusion reaches 0.439 high-risk AUPRC, improving over 0.405 for the contextual scorer and 0.380 for the strongest static baseline, while the contextual scorer remains better calibrated with 0.289 expected calibration error. On the locked in-distribution test split, gains are smaller and static priors remain useful. The resulting claim is therefore narrower: request-conditioned auditing is most valuable as an invocation-time risk-scoring and triage layer rather than as a replacement for static screening. Code is available at https://github.com/123zgj123/STARS.

Autonomous agents based on large language models (LLMs) are rapidly evolving to handle multi-turn tasks, but ensuring their trustworthiness remains a critical challenge. A fundamental pillar of this trustworthiness is calibration, which refers to an agent's ability to express confidence that reliably reflects its actual performance. While calibration is well-established for static models, its dynamics in tool-integrated agentic workflows remain underexplored. In this work, we systematically investigate verbalized calibration in tool-use agents, revealing a fundamental confidence dichotomy driven by tool type. Specifically, our pilot study identifies that evidence tools (e.g., web search) systematically induce severe overconfidence due to inherent noise in retrieved information, while verification tools (e.g., code interpreters) can ground reasoning through deterministic feedback and mitigate miscalibration. To robustly improve calibration across tool types, we propose a reinforcement learning (RL) fine-tuning framework that jointly optimizes task accuracy and calibration, supported by a holistic benchmark of reward designs. We demonstrate that our trained agents not only achieve superior calibration but also exhibit robust generalization from local training environments to noisy web settings and to distinct domains such as mathematical reasoning. Our results highlight the necessity of domain-specific calibration strategies for tool-use agents. More broadly, this work establishes a foundation for building self-aware agents that can reliably communicate uncertainty in high-stakes, real-world deployments.

AI pentesting agents are increasingly credible as offensive security systems, but current benchmarks still provide limited guidance on which will perform best in real-world targets. Existing evaluation protocols assess and optimize for predefined goals such as capture-the-flag, remote code execution, exploit reproduction, or trajectory similarity, in simplified or narrow settings. These tools are valuable for measuring bounded capabilities, yet they do not adequately capture the complexity, open-ended exploration, and strategic decision-making required in realistic pentesting. In this paper, we present a practical evaluation protocol that shifts assessment from task completion to validated vulnerability discovery, allowing evaluation in sufficiently complex targets spanning multiple attack surfaces and vulnerability classes. The protocol combines structured ground-truth with LLM-based semantic matching to identify vulnerabilities, bipartite resolution to score findings under realistic ambiguity, continuous ground-truth maintenance, repeated and cumulative evaluation of stochastic agents, efficiency metrics, and reduced-suite selection for sustainable experimentation. This protocol extends the state of the art by enabling a more realistic, operationally informative comparison of AI pentesting agents. To enable reproducibility, we also release expert-annotated ground truth and code for the proposed evaluation protocol: https://github.com/ethiack/ethibench.

System One models output calibrated probabilities over typed answers such as categorical choices, ordinal levels, or binary outcomes, via a non-generative interface. Software can act on these probabilities through thresholds, cost-weighted choices, and escalation rules. Consequently, if these probabilities are miscalibrated or wording-sensitive, the software ma take unintended actions leaving human operators with no textual rationale to inspect. Current evaluations largely report accuracy and calibration on public classification datasets without a clear reference. We present TypedBench, a benchmark for typed decision models like Jev built from seven policy-labelled generators and nine evaluation suites. We report accuracy as median and range across paraphrases, and calibration error relative to the finite-sample noise floor of a matched, perfectly calibrated predictor. We assess probability quality through selective prediction, ordinal proper scoring rules, and realised cost under asymmetric cost matrices. We evaluate a hosted model, an open encoder, and a family of open decoders spanning 0.8B-9B parameters on identical items. The hosted model follows the stated policy but is wording-sensitive and systematically underconfident; under asymmetric costs, using its probabilities can be worse than taking its top answer. Decoders route exactly and are slow as options or questions are added. The decoder is least accurate on policy questions and degrades with more options. Overall, typed decision models must be evaluated jointly on policy adherence, wording robustness, probability quality, and induced decision outcomes.

Allocating more computation during inference time (test-time scaling) improves language model performance, especially for reasoning tasks. However, popular methods like Best-of-N sampling often show diminishing returns as N increases. To address this inefficiency, we introduce a general test-time calibration framework that adaptively modifies the model toward high-reward reasoning paths, with theoretical guarantees of improving the lower bound of expected reward under finite sampling, all without large language model (LLM) retraining. Within this framework, we propose CarBoN (Calibrated Best-of-N), a two-phase method that first explores the solution space and then learns a calibration of the logits via an input-specific temperature T and additive shift vector delta, guiding generation toward more reliable reasoning. Experiments on MATH-500 and AIME-2024 show that CarBoN improves efficiency, with up to 4times fewer rollouts to reach the same accuracy, while often achieving higher accuracy under fixed budgets. We also analyze the complementary roles of T and delta in balancing output diversity and correctness, and demonstrate that the framework also generalizes to step-level sampling strategies such as beam search. For more information, please refer to our project page at huggingface.co/spaces/TrustSafeAI/Test-Time-Calibration.

Diffusion models have become a core paradigm for multimedia generation, offering powerful concept-driven controllability for personalization, semantic editing, and selective unlearning. However, as semantic control extends beyond natural-language prompts to learned embeddings and intervention pipelines, the safety and governance of these systems become increasingly difficult to evaluate in a unified manner, especially for safety-sensitive, identity-linked, and other privacy-relevant concepts. Existing studies mainly rely on heuristic audits, adversarial probing, or task-specific erasure benchmarks, and therefore provide limited support for systematic comparison across models, conditioning channels, and deployment conditions. We present a concept-level probabilistic audit and reporting framework for diffusion models. We formalize governance-relevant concept behaviors as Bernoulli semantic events induced by stochastic generation, and define a Concept Risk Operator that maps model-channel configurations to structured risk profiles, enabling comparison across prompting interfaces, learned embedding channels, models, and recorded conditions. We apply sample-level post-hoc calibration and configuration-level risk aggregation, and show that probability error can change thresholded actions near policy boundaries. Experiments on SD1.5, SD2.1, and SDXL reveal consistent yet non-uniform operational risk patterns across concept families, channels, recorded conditions, and shifted protocols. In particular, embedding-based access and obfuscated prompts expose risks often understated by standard-prompt evaluation. A pooled multi-protocol calibrator improves held-out probability reliability, but we do not claim transfer from a standard-only calibrator. CLRC provides a common audit schema for probabilistic and decision-aware governance of multimedia generation systems.

Risk-budgeted throughput forecasting converts a point forecast of LEO satellite capacity into a conservative bound whose overestimation rate is held at a user-specified budget, so that admission control can act on it safely. We reproduce the state of the art, BG-CFQS, on the three Starlink traces its authors used and on an independent fourth dataset, and report four findings. First, its risk guarantee is reproducible but conditional on the calibration split: under a random split we recover every published quantity, including risk-pass on 3 of 3 datasets, but under a contiguous temporal split, which is what a deployed terminal faces, it holds on 1 of 3. Second, independent of any split, the method cannot serve a risk budget below 0.15, the low end of its candidate quantile set. On all four datasets the achieved rate is identical at budgets of 0.05, 0.10 and 0.15, overshooting by 1.4x to 3.7x at the tightest. Third, static conformal calibration cannot improve a ranking, only shift an operating point; we confirm this exactly on latency spike detection, where online calibration improves AUPRC by up to 51%. Fourth, group-conditional online recalibration splits into two mechanisms. The online part replicates: it tracks the budget within 1% from 0.05 to 0.35, where static calibration is 16 to 21% off. The group-conditional part does not: on three of four datasets, conditioning on any available covariate, including measured serving-satellite geometry, is worse than not conditioning, and our attempt to predict which datasets benefit failed. What is new here is empirical, not algorithmic: we claim the findings, not the methods.

Systematic reviewers who use a model to read randomized trials need each answer and a measure of how far to trust it. If the model is calibrated, so that its stated probabilities match how often it is right, a reviewer can accept its confident answers and check the rest by hand. This paper fine-tunes Qwen3-4B-Base with low-rank adapters and a classification head for evidence inference (Lehman et al., 2019; DeYoung et al., 2020): given a trial's article and a prompt that names an intervention, a comparator and an outcome, classify the result as a significant increase, a significant decrease or no significant difference. The articles of Evidence Inference 2.0 come from the open-access subset of PubMed Central, which includes articles under non-commercial and no-derivatives licenses, so the released model was fine-tuned only on the 1,431 of 2,657 training articles under CC BY, CC0 or the public domain; it is released under the Apache License 2.0 and runs on a laptop. On the test split, its expected calibration error is 0.017 and its log loss is lower than that of Gemma 4 31B-it given the same input zero-shot, meeting criteria registered on the Open Science Framework before testing. Its macro-F1, 0.925, is above every published result on that split that I found, though those systems used other inputs, scoring rules and prompt filters, and the test labels are public. Accepting its 80 percent most confident answers leaves an error rate of 2.6 percent among them. Context ablations show that the answers depend on the article, but the title alone carries some signal, which an informative title or recall of the trial could explain.

Benchmark contamination is diagnosed with n-gram overlap, likelihood-based membership inference, or canary strings, and each needs something usually unavailable: the training corpus, a well-chosen test statistic, or foresight at release. A recent alternative reads it off a linear probe on internal activations. We show the natural way to do this does not work, specify one that survives measurement, then find that the correction making it work carries more variance than the null it is tested against. The protocol reports a zero-sum contrast on the depth profile of probe accuracy, recentred on a level-matched placebo baseline, tested against a label-permutation null, with the reference set twice the size of the suspect set. Each choice replaces a simpler alternative we rejected on measurement. Reporting the level of excess separability rather than its shape makes the false positive rate track the size of the analyst's own control set, 0.03 to 0.99 under a true null. Contrasting against a flat depth profile rejects a true null 0.72 of the time when surface decodability rises with depth, and loses all power when it falls. On real transformers the protocol fails a test the simulations did not pose. The recentring subtracts an estimate, and the permutation null holds it fixed. Re-estimated across split seeds on four audits of contaminated checkpoints, its standard deviation is 1.30 to 1.56 times the null's own in every arm: what is subtracted to remove a bias is more variable than what it corrects. The one nominally significant result, p = 0.0075, becomes 0.0745 once that variance is propagated, and no verdict is issued. The simulations missed this because their surface key is the covariate driving item variation; on real text it is a proxy, and degrading key quality in simulation reproduces it. We add a companion measurement and a widened null. No arm shows contamination.

Quantum computing calibration depends on interpreting experimental data, and calibration plots provide the most universal human-readable representation for this task, yet no systematic evaluation exists of how well vision-language models (VLMs) interpret them. We introduce QCalEval, the first VLM benchmark for quantum calibration plots: 243 samples across 87 scenario types from 22 experiment families, spanning superconducting qubits and neutral atoms, evaluated on six question types in both zero-shot and in-context learning settings. The best general-purpose zero-shot model reaches a mean score of 72.3, and many open-weight models degrade under multi-image in-context learning, whereas frontier closed models improve substantially. A supervised fine-tuning ablation at the 9-billion-parameter scale shows that SFT improves zero-shot performance but cannot close the multimodal in-context learning gap. As a reference case study, we release NVIDIA Ising Calibration 1, an open-weight model based on Qwen3.5-35B-A3B that reaches 74.7 zero-shot average score.

In legal decisions, split votes (SV) occur when judges cannot reach a unanimous decision, posing a difficulty for lawyers who must navigate diverse legal arguments and opinions. In high-stakes domains, understanding the alignment of perceived difficulty between humans and AI systems is crucial to build trust. However, existing NLP calibration methods focus on a classifier's awareness of predictive performance, measured against the human majority class, overlooking inherent human label variation (HLV). This paper explores split votes as naturally observable human disagreement and value pluralism. We collect judges' vote distributions from the European Court of Human Rights (ECHR), and present SV-ECHR, a case outcome classification (COC) dataset with SV information. We build a taxonomy of disagreement with SV-specific subcategories. We further assess the alignment of perceived difficulty between models and humans, as well as confidence- and human-calibration of COC models. We observe limited alignment with the judge vote distribution. To our knowledge, this is the first systematic exploration of calibration to human judgements in legal NLP. Our study underscores the necessity for further research on measuring and enhancing model calibration considering HLV in legal decision tasks.

DNA-synthesis providers screen incoming orders by searching the requested sequence against curated hazard lists. We show that this baseline collapses to a 100% false-flag rate when the hazardous sequence comes from a taxonomic family absent from the reference set: under Conformal Risk Control's certified miss-rate constraint, a low-discrimination signal forces the threshold below the entire test-benign mass. We compose three signals derived from a synthesis order's public annotation: k-mer Jaccard similarity to known toxins, the trimmed-mean score of a five-LLM judge panel, and cosine similarity to clustered embedding centroids. Fused under a monotone logistic aggregator and calibrated by Conformal Risk Control, the resulting screener certifies E[FNR] le α. Across ten leave-one-taxonomic-family-out folds at α=0.05 on UniProt KW-0800 reviewed toxins, the calibrated screener achieves 0% test miss rate on every fold and 0% test false-flag rate on nine of ten folds. The bound's finite-sample slack 1/(n_{cal}+1) caps the certifiable miss rate at 1.77% on our 200-hazard subsample; reaching procurement-grade α=10^{-3} requires an 18times larger calibration set, which the full reviewed UniProt KW-0800 corpus is large enough to deliver. The binding constraint on certifiable DNA-synthesis screening is calibration data, not algorithms. Code: https://github.com/najmulhasan-code/crc-screen

Using a language model to score or rank candidate responses has become a scalable alternative to human evaluation in reinforcement learning from human feedback (RLHF) pipelines, benchmarking, and application layer evaluations. However, output reliability depends heavily on prompting and aggregation strategy. We present an empirical investigation of four drop-in techniques -- ensemble scoring, task-specific criteria injection, calibration context, and adaptive model escalation -- for improving LLM judge accuracy on RewardBench 2, with a unifying lens of noise control on the stochastic judge: ensembling as Monte Carlo averaging over per-call noise, criteria injection as between-response discrimination sharpening, and per-response score variance as an uncertainty signal. Ensemble scoring and task-specific criteria injection (the latter virtually cost free) together reach up to 85.8% accuracy, +13.5pp over baseline. Calibration context and adaptive model escalation also improve over baseline but are dominated by criteria + ensembling on the cost-accuracy Pareto frontier. Small models benefit disproportionately from ensembling, making high-accuracy LLM judges accessible at low cost. We show that these techniques generalise across model providers, evaluating on both OpenAI GPT and Anthropic Claude families.

Recent video generation is increasingly realistic and controllable, yet video editing remains less developed, particularly for precise local edits that must preserve the original scene dynamics. Video scene text editing replaces text on scene surfaces, such as storefront signs, whiteboards, and product labels, while preserving the surrounding content, motion, and camera dynamics. Although scene text editing is well studied for images, video scene text editing that achieves high visual quality, temporal consistency, and edit locality remains underexplored. Existing resources offer limited paired real-video data, and general video-editing metrics do not directly measure whether the requested text remains correct over time. We introduce ViTeX-Bench, a benchmark suite comprising ViTeX-Dataset and a three-axis evaluation protocol. The dataset contains 387 real-world 720p videos with text-region masks and editing instructions: 230 provide reviewed, pipeline-generated paired edits for training, and 157 form a frozen evaluation split. The protocol evaluates text correctness, visual and temporal quality, and edit locality through 13 metrics, with one primary metric per axis and a Pareto comparison of their trade-offs. OCR calibration, human evaluation, and annotation-sensitivity analyses support the interpretation of these scores. Across eight baselines from four editing families, accurate text, temporal stability, and scene preservation remain difficult to achieve together. We also release ViTeX-Edit-14B, an open-source reference editor fine-tuned on the paired training split with motion-aligned glyph-video conditioning. It achieves CharAcc 0.688, the highest mean among the evaluated video-native editors, and the lowest comparable text-crop Warp among raw editor outputs. ViTeX-Bench provides a reproducible foundation for studying these trade-offs in video scene text editing.

We propose MCGrad, a novel and scalable multicalibration algorithm. Multicalibration - calibration in subgroups of the data - is an important property for the performance of machine learning-based systems. Existing multicalibration methods have thus far received limited traction in industry. We argue that this is because existing methods (1) require such subgroups to be manually specified, which ML practitioners often struggle with, (2) are not scalable, or (3) may harm other notions of model performance such as log loss and Area Under the Precision-Recall Curve (PRAUC). MCGrad does not require explicit specification of protected groups, is scalable, and often improves other ML evaluation metrics instead of harming them. MCGrad has been in production at Meta, and is now part of hundreds of production models. We present results from these deployments as well as results on public datasets. We provide an open source implementation of MCGrad at https://github.com/facebookincubator/MCGrad.

This paper provides an experimentally verified formal law for calculating the uplift that diversity of thought provides in Large Language Model (LLM) ensembles. From first principles, we derive an exact decomposition of LLM ensemble lift into rescue and damage masses, which yields a compact heuristic for calculating uplift. From this we extract the metrics which predict ensemble performance: an accuracy adjusted correctness correlation, ϕ_{adj}, together with the accuracy gap and collective accuracy of the pair. We test the law on 767,520 inferences from ten open-weight models over two graduate-level science benchmarks, together with a novel agentic cybersecurity benchmark in which each model conducts digital-forensics investigations by multi-turn tool use in a network-isolated sandbox (23,520 graded trials including abstentions); all votes are released openly. Calibrated once on SuperGPQA at a 40:60 vote split, the heuristic predicts lift on the calibration set with Spearman's ρ=0.84 and, with its coefficients frozen, transfers to two datasets never used in calibration (ρ=0.51 on GPQA Diamond and 0.84 on the forensic tasks), whilst the measured swap mass tracks realised lift with R^2ge 0.96 throughout. Raw ϕ has almost no predictive power (R^2le 0.09 throughout); the accuracy-adjusted ϕ_{adj} is markedly superior (R^2=0.67 on SuperGPQA), and the heuristic combining these metrics is the most stable pre pooling predictor across the three datasets.

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.

Unsupervised physical parameter estimation from video lacks a common benchmark: existing methods evaluate on non-overlapping synthetic data, the sole real-world dataset is restricted to single-body systems, and no established protocol addresses governing-equation identification. This work introduces IRIS, a high-fidelity benchmark comprising 240 real-world videos captured at 4K resolution and 60fps, spanning both single- and multi-body dynamics with independently measured ground-truth parameters and uncertainty estimates. Each dynamical system is recorded under controlled laboratory conditions and paired with its governing equations, enabling principled evaluation. A standardized evaluation protocol is defined encompassing parameter accuracy, identifiability, extrapolation, robustness, and governing-equation selection. Multiple baselines are evaluated, including a multi-step physics loss formulation and four complementary equation-identification strategies (VLM temporal reasoning, describe-then-classify prompting, CNN-based classification, and path-based labelling), establishing reference performance across all IRIS scenarios and exposing systematic failure modes that motivate future research. The dataset, annotations, evaluation toolkit, and all baseline implementations are publicly released.

Rerankers decide which documents users and LLMs see, yet their standard metric, nDCG, relies on human relevance labels that are costly, sparse, noisy, and discretely graded. As rerankers approach each other in quality, nDCG on these labels therefore increasingly fails to separate them. LLM judges could supply dense labels. Relative judgments within one query tell even close candidates apart, yet their scores share no scale across queries. Absolute grades share one scale but are too coarse to distinguish documents of similar relevance. We propose Rubric-Calibrated Preferences (RCP), which combine both kinds of judgment. A listwise Bradley-Terry tournament orders each query's documents, and a rubric of yes/no criteria of increasing stringency provides an absolute standard. Item Response Theory (IRT), which scores test-takers based on their answers to common questions, then uses the shared criteria to put all queries' tournament scores on one scale. RCP's retrieval metric, RCP-nDCG, replaces nDCG's discrete labels with the resulting calibrated relevance probabilities. Against blind grades from 46 external annotators, calibration raises the correlation between a query's mean score and its mean human grade from 0.538 to 0.795. The probabilities rank a useful document above a non-useful one with probability 0.910 (AUC, chance 0.5), versus 0.651 for the benchmark labels. When the annotators' grades prefer one of two rerankers and exactly one metric agrees, that metric is RCP-nDCG in 72.4% of 185 comparisons (chance about 53%). On TREC-DL, RCP-nDCG sides with NIST assessors' grades on every reranker pair that these grades separate significantly. RCP-nDCG also resolves many of nDCG's ties and separates 1.9 times as many reranker pairs on NanoBEIR. Rubric calibration thus turns relative LLM judgments into dense relevance labels that are comparable across queries and agree with human judgment.

When a coordinator compares answers from heterogeneous foundation models, self-reported confidence may have different meanings across responders and changing workloads. This paper presents MARGIN (Multi-Agent Runtime Grading via Incremental Normalisation), a runtime calibration method that learns model-specific confidence corrections from observed answer outcomes without retraining the models or requiring a held-out calibration set. MARGIN tracks recent accuracy and stated confidence within confidence bands, uses their ratio to correct reported confidence, and blends sparse-band corrections toward a model-level estimate. The corrected scores weight candidate answers in a collective decision. Evaluation covers code generation, question answering, and mathematics, using an 18-model pool and a nine-model subset for distribution-shift experiments. On BigCodeBench, model-mean confidence is negatively related to accuracy; among correct/incorrect response pairs, choosing the more confident responder performs below chance. Against five online calibration baselines receiving identical feedback and retaining their learned state across each transition, MARGIN achieves lower post-shift expected calibration error than all five in two code-generation transitions and than four in a question-answering transition; the remaining question-answering comparison is inconclusive. In separate code-generation coordination experiments, calibration improves the ranking of correct responses and increases answer-selection accuracy by 4.3 and 14.0 percentage points on two of three benchmarks relative to uncalibrated confidence weighting. These results support model-specific runtime calibration for coordination under changing workloads when correctness feedback is available for the participating responders.

Deep neural networks are increasingly utilized in various machine learning tasks. However, as these models grow in complexity, they often face calibration issues, despite enhanced prediction accuracy. Many studies have endeavored to improve calibration performance through the use of specific loss functions, data preprocessing and training frameworks. Yet, investigations into calibration properties have been somewhat overlooked. Our study leverages the Neural Architecture Search (NAS) search space, offering an exhaustive model architecture space for thorough calibration properties exploration. We specifically create a model calibration dataset. This dataset evaluates 90 bin-based and 12 additional calibration measurements across 117,702 unique neural networks within the widely employed NATS-Bench search space. Our analysis aims to answer several longstanding questions in the field, using our proposed dataset: (i) Can model calibration be generalized across different datasets? (ii) Can robustness be used as a calibration measurement? (iii) How reliable are calibration metrics? (iv) Does a post-hoc calibration method affect all models uniformly? (v) How does calibration interact with accuracy? (vi) What is the impact of bin size on calibration measurement? (vii) Which architectural designs are beneficial for calibration? Additionally, our study bridges an existing gap by exploring calibration within NAS. By providing this dataset, we enable further research into NAS calibration. As far as we are aware, our research represents the first large-scale investigation into calibration properties and the premier study of calibration issues within NAS. The project page can be found at https://www.taolinwei.com/calibration-study

Retrieval-augmented generation (RAG) grounds answers in retrieved passages, yet relevance does not guarantee sufficiency: a topical passage may still fail to justify the answer. We study evidence sufficiency verification for selective RAG answering, in which a verifier receives a question, a candidate answer, and retrieved evidence and decides whether the evidence supports, refutes, or is insufficient for the answer, answering only when support is established. We present SURE-RAG, an aggregation protocol that treats evidence sufficiency as a set-level property: missing hops and unresolved conflicts cannot be detected by scoring passages independently. A shared claim-evidence verifier produces a local relation distribution for each (claim, passage) pair, which SURE-RAG aggregates into four interpretable answer-level feature blocks (coverage, relation strength, uncertainty, and retrieval), producing a three-way decision and an auditable selective score. We evaluate on HotpotQA-RAG v3, a controlled multi-hop benchmark, under an artifact-aware protocol (shortcut baselines, counterfactual swaps, no-oracle checks, and GPT-4o audits). Calibrated SURE-RAG attains 0.9075 Macro-F1 (raw 0.8951 +/- 0.0069), well above DeBERTa mean-pooling (0.6516) and a GPT-4o judge (0.7284), and on par with a strong concat cross-encoder (0.8888 +/- 0.0109) while remaining fully auditable. At 30% coverage, risk falls from 0.2588 to 0.1642, a 37% relative reduction. As a boundary-mapping experiment, we contrast SURE-RAG with GPT-4o on HaluBench unsafe detection: the ranking reverses (0.3343 vs. 0.7389 unsafe-F1), indicating that controlled sufficiency verification and natural hallucination detection are distinct problems.

Audio decisions often depend on evidence that a transcript does not preserve, while their probability estimates can depend on how answer options are ordered. AudioJev maps a waveform, question and supplied alternatives directly to a candidate distribution through one shared full-parameter model. We define order calibration as preserving an answer's probability under meaning-preserving option permutations. Random-derangement SKL training pairs each question with a reordered view in which every alternative changes position, supervises both answers, and aligns the two distributions before applying a symmetric KL penalty. Inference retains a single candidate-scoring forward, with no calibration head or order ensemble. Across three training seeds, AudioJev reaches 68.88%/55.33% mean accuracy on complete MMAU/MMAR and reduces random-order SKL by 43.8%/60.5% relative to the single-view removal ablation. The same model handles intent, environmental sound, note properties, speech activity and conversational transitions. Paired removal ablations and multi-order evaluation measure predictive accuracy and probability stability together, establishing a direct audio interface whose calibration objective acts on candidate meaning rather than presentation position. Inference code and model weights are available at https://github.com/SihanLv/AudioJev-Inference and https://huggingface.co/shlv/AudioJev.

Summarization models often generate text that is poorly calibrated to quality metrics because they are trained to maximize the likelihood of a single reference (MLE). To address this, recent work has added a calibration step, which exposes a model to its own ranked outputs to improve relevance or, in a separate line of work, contrasts positive and negative sets to improve faithfulness. While effective, much of this work has focused on how to generate and optimize these sets. Less is known about why one setup is more effective than another. In this work, we uncover the underlying characteristics of effective sets. For each training instance, we form a large, diverse pool of candidates and systematically vary the subsets used for calibration fine-tuning. Each selection strategy targets distinct aspects of the sets, such as lexical diversity or the size of the gap between positive and negatives. On three diverse scientific long-form summarization datasets (spanning biomedical, clinical, and chemical domains), we find, among others, that faithfulness calibration is optimal when the negative sets are extractive and more likely to be generated, whereas for relevance calibration, the metric margin between candidates should be maximized and surprise--the disagreement between model and metric defined candidate rankings--minimized. Code to create, select, and optimize calibration sets is available at https://github.com/griff4692/calibrating-summaries

Process reward models (PRMs) play a central role in guiding inference-time scaling algorithms for large language models (LLMs). However, we observe that even state-of-the-art PRMs can be poorly calibrated and often overestimate success probabilities. To address this, we present a calibration approach, performed via quantile regression, that adjusts PRM outputs to better align with true success probabilities. Leveraging these calibrated success estimates and their associated confidence bounds, we introduce an instance-adaptive scaling (IAS) framework that dynamically adjusts the inference budget based on the estimated likelihood that a partial reasoning trajectory will yield a correct final answer. Unlike conventional methods that allocate a fixed number of reasoning trajectories per query, this approach successfully adapts to each instance and reasoning step when using our calibrated PRMs. Experiments on mathematical reasoning benchmarks show that (i) our PRM calibration method successfully achieves small calibration error, outperforming the baseline methods, (ii) calibration is crucial for enabling effective adaptive scaling, and (iii) the proposed IAS strategy reduces inference costs while maintaining final answer accuracy, utilizing less compute on more confident problems as desired.

As LLM-based judges become integral to industry applications, obtaining well-calibrated uncertainty estimates efficiently has become critical for production deployment. However, existing techniques, such as verbalized confidence and multi-generation methods, are often either poorly calibrated or computationally expensive. We introduce linear probes trained with a Brier score-based loss to provide calibrated uncertainty estimates from reasoning judges' hidden states, requiring no additional model training. We evaluate our approach on both objective tasks (reasoning, mathematics, factuality, coding) and subjective human preference judgments. Our results demonstrate that probes achieve superior calibration compared to existing methods with approx10x computational savings, generalize robustly to unseen evaluation domains, and deliver higher accuracy on high-confidence predictions. However, probes produce conservative estimates that underperform on easier datasets but may benefit safety-critical deployments prioritizing low false-positive rates. Overall, our work demonstrates that interpretability-based uncertainty estimation provides a practical and scalable plug-and-play solution for LLM judges in production.

Agent benchmarks are increasingly used to compare large language models (LLMs) and guide deployment decisions, yet benchmark scores are meaningful only if they measure model capability rather than properties of the evaluation pipeline. We identify a double measurement confound: execution-critical decisions are performed by a fixed scaffold instead of the model, while the scorer evaluates outputs using criteria that may not reflect task correctness. We unify these issues within a measurement-theoretic framework that characterizes when benchmark scores can be interpreted as evidence of model capability, and instantiate it with an audit-and-repair protocol that (i) transfers execution-critical decisions from the scaffold to the model, (ii) replaces shape-based evaluation with seeded ground-truth scoring, and (iii) reports reliability beyond the mean through worst-case and tail-risk metrics. Experiments on ComtradeBench show that the joint intervention transforms a nearly flat leaderboard into a reliability spectrum that distinguishes both average performance and robustness across seeds. Applying the audit to existing benchmarks further shows that scorer validity is benchmark-specific, whereas scaffold ownership is an uncontrolled axis wherever we probed it. Our results suggest that benchmark scores should be interpreted together with their scaffolding level, scoring criterion, and reliability profile, providing a practical framework for more valid evaluation of LLM agents.

Recent machine learning papers often report 1-2 percentage point improvements from a single run on a benchmark. These gains are highly sensitive to random seeds, data ordering, and implementation details, yet are rarely accompanied by uncertainty estimates or significance tests. It is therefore unclear when a reported +1-2% reflects a real algorithmic advance versus noise. We revisit this problem under realistic compute budgets, where only a few runs are affordable. We propose a simple, PC-friendly evaluation protocol based on paired multi-seed runs, bias-corrected and accelerated (BCa) bootstrap confidence intervals, and a sign-flip permutation test on per-seed deltas. The protocol is intentionally conservative and is meant as a guardrail against over-claiming. We instantiate it on CIFAR-10, CIFAR-10N, and AG News using synthetic no-improvement, small-gain, and medium-gain scenarios. Single runs and unpaired t-tests often suggest significant gains for 0.6-2.0 point improvements, especially on text. With only three seeds, our paired protocol never declares significance in these settings. We argue that such conservative evaluation is a safer default for small gains under tight budgets.

Multimodal large language models (LLMs) are increasingly explored as automated evaluators in clinical settings, yet their scoring behavior on ordinal clinical scales remains poorly understood. We benchmark three frontier LLM families against supervised deep learning models for scoring Clock Drawing Test (CDT) images on two public datasets using the Shulman rubric. While fully fine-tuned Vision Transformers achieve the best calibration (MAE 0.52, within-1 accuracy 91%), zero-shot LLMs remain competitive on tolerance-based agreement (GPT-5 MAE 0.67, within-1 accuracy 92%) despite higher absolute error. However, per-score analysis reveals that all three LLM families exhibit a pronounced central tendency effect (systematic endpoint compression): predictions are systematically compressed toward the middle of the scale, with over-prediction at the low end (score 0 to 1) and under-prediction at the high end (score 5 to 4). This effect disproportionately affects the clinically critical extremes where accurate scoring most impacts screening decisions for cognitive impairment. Targeted ablations show that neither few-shot exemplars spanning the full score range nor removing clinical terminology from the prompt eliminates the effect. Our findings extend the LLM-as-a-judge bias literature from NLP evaluation to clinical assessment, and highlight the need for calibration-aware evaluation and post-hoc calibration before deploying LLM-based raters in high-stakes screening workflows.

LLM-as-judge evaluation has become the de facto standard for scaling model assessment, but the practice is statistically unsound: uncalibrated scores can invert preferences, naive confidence intervals on uncalibrated scores achieve near-0% coverage, and importance-weighted estimators collapse under limited overlap despite high effective sample size (ESS). We introduce Causal Judge Evaluation (CJE), a framework that fixes all three failures. On n=4,961 Chatbot Arena prompts (after filtering from 5k), CJE achieves 99% pairwise ranking accuracy at full sample size (94% averaged across configurations), matching oracle quality, at 14x lower cost (for ranking 5 policies) by calibrating a 16x cheaper judge on just 5% oracle labels (~250 labels). CJE combines three components: (i) AutoCal-R, reward calibration via mean-preserving isotonic regression; (ii) SIMCal-W, weight stabilization via stacking of S-monotone candidates; and (iii) Oracle-Uncertainty Aware (OUA) inference that propagates calibration uncertainty into confidence intervals. We formalize the Coverage-Limited Efficiency (CLE) diagnostic, which explains why IPS-style estimators fail even when ESS exceeds 90%: the logger rarely visits regions where target policies concentrate. Key findings: SNIPS inverts rankings even with reward calibration (38% pairwise, negative Kendall's tau) due to weight instability; calibrated IPS remains near-random (47%) despite weight stabilization, consistent with CLE; OUA improves coverage from near-0% to ~86% (Direct) and ~96% (stacked-DR), where naive intervals severely under-cover.

Probability estimation models play an important role in various fields, such as weather forecasting, recommendation systems, and sports analysis. Among several models estimating probabilities, it is difficult to evaluate which model gives reliable probabilities since the ground-truth probabilities are not available. The win probability estimation model for esports, which calculates the win probability under a certain game state, is also one of the fields being actively studied in probability estimation. However, most of the previous works evaluated their models using accuracy, a metric that only can measure the performance of discrimination. In this work, we firstly investigate the Brier score and the Expected Calibration Error (ECE) as a replacement of accuracy used as a performance evaluation metric for win probability estimation models in esports field. Based on the analysis, we propose a novel metric called Balance score which is a simple yet effective metric in terms of six good properties that probability estimation metric should have. Under the general condition, we also found that the Balance score can be an effective approximation of the true expected calibration error which has been imperfectly approximated by ECE using the binning technique. Extensive evaluations using simulation studies and real game snapshot data demonstrate the promising potential to adopt the proposed metric not only for the win probability estimation model for esports but also for evaluating general probability estimation models.

Post-hoc calibration corrects reported confidence, yet a multiclass calibrator can also change the associated top-1 prediction. Accuracy captures only the net effect of these changes on correctness, not how often predictions change; the Top-1 Prediction Change Rate (TPCR) instead measures this frequency. We propose Calibrator-Output Repair for Top-1 Decision Preservation (CORD), the first post-fit adapter to impose exact prediction preservation by repairing the full calibrated probability vector. From the original and calibrated outputs alone, CORD determines the mass assigned to the original top-1. The calibrated conditional distribution allocates the remaining mass over the other classes, yielding a repaired vector whose own argmax recovers the original prediction. On the calibration split, CORD coordinates the repaired masses to retain the calibrated outputs' mean mass on original predictions whenever attainable. The adapter alters neither the fitted calibrator nor its direct output, fits no additional supervised map, and requires no user- or validation-tuned hyperparameter. Across CIFAR-10/100 and ImageNet-1K, CORD attains zero TPCR by construction and lowers mean ECE, NLL, and Brier relative to the corresponding direct outputs in every dataset; paired gains persist under distribution shift and across calibration-set sizes. CORD thus removes the preservation constraint from calibrator fitting and assigns exact recovery of the original decision to subsequent output repair. Our code is available at https://github.com/labhai/CORD.

Conformal prediction is a framework for predictive inference with a distribution-free, finite-sample guarantee. However, it tends to provide uninformative prediction sets when calibration data are scarce. This paper introduces Synthetic-powered predictive inference (SPI), a novel framework that incorporates synthetic data -- e.g., from a generative model -- to improve sample efficiency. At the core of our method is a score transporter: an empirical quantile mapping that aligns nonconformity scores from trusted, real data with those from synthetic data. By carefully integrating the score transporter into the calibration process, SPI provably achieves finite-sample coverage guarantees without making any assumptions about the real and synthetic data distributions. When the score distributions are well aligned, SPI yields substantially tighter and more informative prediction sets than standard conformal prediction. Experiments on image classification -- augmenting data with synthetic diffusion-model generated images -- and on tabular regression demonstrate notable improvements in predictive efficiency in data-scarce settings.

Foundation-model predictors of perturbation response are evaluated primarily by aggregate accuracy. Using their predictions to prioritize experiments requires assessing each prediction's reliability. We audit the state-transition (ST) decoder used in STATE across three encoders (SE-600M, scGPT, and Tahoe-x1) and four cell lines under CRISPRi. The decoder's first attention layer has a low-rank output. Projecting this output onto 4-32 fixed directions estimated from training activations preserves mean per-perturbation Pearson delta correlation (PDC). This low-rank attention bottleneck bounds the decoder's reliability by distance to training responses: PDC declines as the cosine distance from a perturbation's observed response to its nearest training response increases. We propose predicted-NN-distance, the cosine distance from a predicted response to its nearest training response, as a per-prediction confidence score. The score has a mean absolute Spearman correlation of 0.81 with PDC, close to the 0.88 obtained with the observed response. Thresholds calibrated on this score transfer across encoders and cell lines, and the score remains predictive of PDC for drug perturbations.

Despite their widespread applications, Large Language Models (LLMs) often struggle to express uncertainty, posing a challenge for reliable deployment in high stakes and safety critical domains like clinical diagnostics. Existing standard baseline methods such as model logits and elicited probabilities produce overconfident and poorly calibrated estimates. In this work, we propose Approximate Bayesian Computation (ABC), a likelihood-free Bayesian inference, based approach that treats LLMs as a stochastic simulator to infer posterior distributions over predictive probabilities. We evaluate our ABC approach on two clinically relevant benchmarks: a synthetic oral lesion diagnosis dataset and the publicly available GretelAI symptom-to-diagnosis dataset. Compared to standard baselines, our approach improves accuracy by up to 46.9\%, reduces Brier scores by 74.4\%, and enhances calibration as measured by Expected Calibration Error (ECE) and predictive entropy.

Large Language Models (LLMs) have demonstrated great capabilities in solving a wide range of tasks in a resource-efficient manner through prompting, which does not require task-specific training, but suffers from performance fluctuation when there are multiple prompt candidates. Previous works have introduced gradient-free probability-based prompt selection methods that aim to choose the optimal prompt among the candidates for a given task but fail to provide a comprehensive and fair comparison between each other. In this paper, we propose a unified framework to interpret and evaluate the existing probability-based prompt selection methods by performing extensive experiments on 13 common NLP tasks. We find that all existing methods can be unified into some variant of the method that maximizes the mutual information between the input and the corresponding model output (denoted as MI). Using the finding, we develop several variants of MI and increases the effectiveness of the best prompt selection method from 87.79% to 94.98%, measured as the ratio of the performance of the selected prompt to that of the optimal oracle prompt. Furthermore, we propose a novel calibration method called Calibration by Marginalization (CBM) that is orthogonal to existing methods and helps increase the prompt selection effectiveness of the best method by 99.44%. The code and datasets used in our work will be released at https://github.com/soheeyang/unified-prompt-selection.

Recent benchmarks rely on user simulators to evaluate AI agents in multi-turn interaction. While existing simulation techniques demonstrate surface fidelity to human style and behavior, ecologically valid interactive benchmarking also requires alignment in when and how agents fail across simulated and real user populations. We find that existing simulators lack outcome calibration: agreement with observed success rates and failure patterns when real users interact with the same agent. We introduce Calibrated User Embeddings (CUE), a framework that both encodes observed sessions and samples continuous representations, then decodes them into persona commands to steer LLMs to act as user simulators without training. Through this, we evaluate user-conditioned replay of past sessions and aggregate metric agreement when sampling novel personas for the same tasks. On τ^2-Bench, CUEd simulators commit fewer simulator-attributed errors and more faithfully reproduce real-user agent failure modes, aggregate success rates, and outcomes for specific task-user pairs than other persona-based simulation methods. These gains coexist with competitive user fidelity as measured using metrics established in prior work. After being fit to mostly customer support interactions, the same CUEd simulators generalize to document creation, math tutoring, and casual conversation, and remain effective across different simulator LLMs without CUE retraining.

LLM confidence signals are used for abstention, routing, and safety-critical decisions. No standard practice exists for checking whether a confidence signal carries item-level information before building on it. We transfer the validity screening principle from clinical personality assessment (PAI, MMPI-3) as a portable protocol for benchmark-based LLM confidence data. The protocol specifies three core indices (L, Fp, RBS), a structural indicator (TRIN), and an item-sensitivity statistic, computed from a single 2x2 contingency table. A three-tier classification system (Invalid, Indeterminate, Valid) draws on four clinical traditions. Validated on 20 frontier LLMs across 524 items, four models are classified Invalid, two Indeterminate. Valid-profile models show mean r = .18 (15/16 significant). Invalid-profile models show mean r = -.20 (d = 2.48). Cross-benchmark validation on 18 models using MMLU with verbalized confidence and on external data from Yang et al. (2024) confirms the screen transfers across benchmarks and probe formats. All data and code: https://github.com/synthiumjp/validity-scaling-llm

Quantitative risk assessment in high-stakes domains relies on structured expert elicitation to estimate unobservable properties. The gold standard - the Delphi method - produces calibrated, auditable judgments but requires months of coordination and specialist time, placing rigorous risk assessment out of reach for most applications. We investigate whether Large Language Models (LLMs) can serve as scalable proxies for structured expert elicitation. We propose Scalable Delphi, adapting the classical protocol for LLMs with diverse expert personas, iterative refinement, and rationale sharing. Because target quantities are typically unobservable, we develop an evaluation framework based on necessary conditions: calibration against verifiable proxies, sensitivity to evidence, and alignment with human expert judgment. We evaluate in the domain of AI-augmented cybersecurity risk, using three capability benchmarks and independent human elicitation studies. LLM panels achieve strong correlations with benchmark ground truth (Pearson r=0.87-0.95), improve systematically as evidence is added, and align with human expert panels - in one comparison, closer to a human panel than the two human panels are to each other. This demonstrates that LLM-based elicitation can extend structured expert judgment to settings where traditional methods are infeasible, reducing elicitation time from months to minutes.

Point-adjustment (PA), for years the default scoring protocol in time-series anomaly detection (TSAD), was shown by Kim et al. (2022) to award near-perfect F1 to random anomaly scores. The field adopted a suite of replacement metrics (PA%K, range-based precision/recall, affiliation precision/recall, and Volume-Under-the-Surface, VUS, ROC/PR). We ask, independently and adversarially, whether these resist no-skill detectors on real benchmarks, and find the answer turns entirely on one overlooked variable: N, the number of random attempts an adversary reports the best of. Under a single honest run (N=1), not one replacement metric is gameable on any of six benchmarks (UCR, SMD, SMAP, MSL, NAB, PSM): a random detector reaches 90% of the best real detector's score on at most 11% of series for affiliation-F1, 5% for the ROC family, and 2% for the PR-based metrics and PA%K. But under best-of-N reporting, the seed-shopping endemic to ML, the metrics split sharply. affiliation-F1 and every ROC-based metric inflate steeply, affiliation crossing gameable (25% of series) by N=3 and reaching 0.98 at the full pool (N=41), the ROC family crossing by N=9-11; the PR-based metrics and PA%K stay near-flat at every N, floored near the anomaly prevalence (the lone exception is NAB at large N). A paired test finds VUS-ROC inflated on 131 series where its sibling VUS-PR is not, and never the reverse. The ROC-vs-PR split follows from the order-statistic behaviour of AUC under extreme class imbalance (a random PR-AUC is floored at prevalence); affiliation inflates by a second route, its extreme single-run leniency (already fragile at N=1). We release a pip-installable stress-test harness, and recommend reporting single-run scores or disclosing N and preferring PR-based metrics, which resist best-of-N inflation on nearly every benchmark.

Conformal Prediction (CP) controls the prediction uncertainty of classification systems by producing a small prediction set, ensuring a predetermined probability that the true class lies within this set. This is commonly done by defining a score, based on the model predictions, and setting a threshold on this score using a validation set. In this study, we address the problem of CP calibration when we only have access to a calibration set with noisy labels. We show how we can estimate the noise-free conformal threshold based on the noisy labeled data. We derive a finite sample coverage guarantee for uniform noise that remains effective even in tasks with a large number of classes. We dub our approach Noise-Aware Conformal Prediction (NACP). We illustrate the performance of the proposed results on several standard image classification datasets with a large number of classes.

Multimodal large language models (MLLMs) hold considerable promise for applications in healthcare. However, their deployment in safety-critical settings is hindered by two key limitations: (i) sensitivity to prompt design, and (ii) a tendency to generate incorrect responses with high confidence. As clinicians may rely on a model's stated confidence to gauge the reliability of its predictions, it is especially important that when a model expresses high confidence, it is also highly accurate. We introduce Prompt4Trust, the first reinforcement learning (RL) framework for prompt augmentation targeting confidence calibration in MLLMs. A lightweight LLM is trained to produce context-aware auxiliary prompts that guide a downstream task MLLM to generate responses in which the expressed confidence more accurately reflects predictive accuracy. Unlike conventional calibration techniques, Prompt4Trust specifically prioritizes aspects of calibration most critical for safe and trustworthy clinical decision-making. Beyond improvements driven by this clinically motivated calibration objective, our proposed method also improves task accuracy, achieving state-of-the-art medical visual question answering (VQA) performance on the PMC-VQA benchmark, which is composed of multiple-choice questions spanning diverse medical imaging modalities. Moreover, our framework trained with a small downstream task MLLM showed promising zero-shot generalization to larger MLLMs in our experiments, suggesting the potential for scalable calibration without the associated computational costs. This work demonstrates the potential of automated yet human-aligned prompt engineering for improving the the trustworthiness of MLLMs in safety critical settings. Our codebase can be found at https://github.com/xingbpshen/prompt4trust.

Candidate-scoring decision models return a probability distribution over a presented set of answers. Public documentation of one such system, Jev, and of an open counterpart, JevAny, describes three training ingredients that determine what these probabilities estimate: soft targets from larger models, a frozen pretrained backbone and temperature calibration. We analyze each ingredient in Gaussian surrogates where the finite-sample risk can be computed exactly. A student trained on teacher targets inherits the teacher's bias, which more student data does not remove and agreement with the teacher does not reveal. A fixed representation lowers the risk only below an explicit sample-size threshold. In a probit model the exact recalibration map is a temperature scaling, which removes all excess risk only when the teacher's coefficient has the correct direction and otherwise leaves a residual that calibration diagnostics do not detect. Under Luce's choice axiom, candidate scores are coherent across refinements of an event if and only if they aggregate by log-sum-exp. On 231,521 credit card complaints from the Consumer Financial Protection Bureau, where the decision is whether a complaint ends in monetary relief, a student trained on a logistic teacher's probabilities ends at the teacher's decision cost while a student trained on outcomes goes 8% below it, a 30B model read zero-shot remains worse than the teacher after recalibration, and that model departs from the coherence condition by 17.5 decibans. Pretrained representations reduce the labeled data required and calibration corrects the scale of the probabilities, but neither corrects a biased teacher, so the probabilities have to be checked against realized outcomes.

Current question-answering benchmarks predominantly focus on accuracy in realizable prediction tasks. Conditioned on a question and answer-key, does the most likely token match the ground truth? Such benchmarks necessarily fail to evaluate LLMs' ability to quantify ground-truth outcome uncertainty. In this work, we focus on the use of LLMs as risk scores for unrealizable prediction tasks. We introduce folktexts, a software package to systematically generate risk scores using LLMs, and evaluate them against US Census data products. A flexible API enables the use of different prompting schemes, local or web-hosted models, and diverse census columns that can be used to compose custom prediction tasks. We evaluate 17 recent LLMs across five proposed benchmark tasks. We find that zero-shot risk scores produced by multiple-choice question-answering have high predictive signal but are widely miscalibrated. Base models consistently overestimate outcome uncertainty, while instruction-tuned models underestimate uncertainty and produce over-confident risk scores. In fact, instruction-tuning polarizes answer distribution regardless of true underlying data uncertainty. This reveals a general inability of instruction-tuned LLMs to express data uncertainty using multiple-choice answers. A separate experiment using verbalized chat-style risk queries yields substantially improved calibration across instruction-tuned models. These differences in ability to quantify data uncertainty cannot be revealed in realizable settings, and highlight a blind-spot in the current evaluation ecosystem that folktexts covers.

We propose causal isotonic calibration, a novel nonparametric method for calibrating predictors of heterogeneous treatment effects. Furthermore, we introduce cross-calibration, a data-efficient variant of calibration that eliminates the need for hold-out calibration sets. Cross-calibration leverages cross-fitted predictors and generates a single calibrated predictor using all available data. Under weak conditions that do not assume monotonicity, we establish that both causal isotonic calibration and cross-calibration achieve fast doubly-robust calibration rates, as long as either the propensity score or outcome regression is estimated accurately in a suitable sense. The proposed causal isotonic calibrator can be wrapped around any black-box learning algorithm, providing robust and distribution-free calibration guarantees while preserving predictive performance.

Retrieval-augmented agents can query external evidence, yet their reliability in multi-step reasoning remains limited: noisy retrieval may derail multi-hop question answering, while outcome-only reinforcement learning provides credit signals that are too coarse to optimize intermediate steps. We propose EvalAct (Evaluate-as-Action), which converts implicit retrieval quality assessment into an explicit action and enforces a coupled Search-to-Evaluate protocol so that each retrieval is immediately followed by a structured evaluation score, yielding process signals aligned with the interaction trajectory. To leverage these signals, we introduce Process-Calibrated Advantage Rescaling (PCAR), a GRPO-based optimization method that rescales advantages at the segment level according to evaluation scores, emphasizing reliable segments while updating uncertain ones conservatively. Experiments on seven open-domain QA benchmarks show that EvalAct achieves the best average accuracy, with the largest gains on multi-hop tasks, and ablations verify that the explicit evaluation loop drives the primary improvements while PCAR provides consistent additional benefits.

Large language models can solve substantially harder reasoning problems with more inference-time compute. The term "test-time scaling," however, now covers diverse inference algorithms that extend deliberation along a single trajectory, sample completed candidates and aggregate them through voting or verification, or search over unfinished partial states. These algorithms differ in their statistical structure, compute accounting, and failure modes. Treating these procedures as interchangeable under a single scalar "budget," or reporting accuracy without the inference protocol that produced it, makes results difficult to compare across studies. We develop a systematic account of test-time scaling along three axes. First, we formalize test-time scaling as budgeted inference over the implicit prefix tree of an autoregressive model and distinguish three structural regimes: single-trajectory sequential scaling, leaf-level scaling with terminal reduction, and prefix-level scaling. Second, we treat the evaluated object as the entire inference system and develop evaluation principles that separate end-to-end system performance from candidate-bank diagnostics. We introduce an evaluation profile whose coordinates and simple functionals recover or bound common repeated-sampling metrics, and prescribe protocol-matched reporting of compute and uncertainty. Third, we specify reproducibility requirements for inference protocols, distinguishing exact replay from distributional reproducibility and identifying the artifacts needed to support each. We also organize the open-weight reasoning ecosystem by model-side and interface mechanisms, apply these principles to broad-knowledge, symbolic-reasoning, and competition-mathematics benchmarks, and assemble over 2 billion full reasoning traces for release with progressively richer verifier and token-level signals.

Clinical personality assessment screens response validity before interpreting substantive scales. LLM evaluation does not. We apply the validity scaling framework from the PAI and MMPI-3 to metacognitive probe data from 20 frontier models across 524 items. Six validity indices are operationalised: L (maintaining confidence on errors), K (betting on errors), F (withdrawing consensus-endorsed items), Fp (withdrawing correct answers), RBS (inverted monitoring), and TRIN (fixed responding). A tiered classification system identifies four models as construct-level invalid and two as elevated. Valid-profile models produce item-sensitive confidence (mean r = .18, 14 of 16 significant). Invalid-profile models do not (mean r = -.20, d = 2.17, p = .001). Chain-of-thought training produces two opposite response distortions. Two latent dimensions account for 94.6% of index variance. Companion papers extract a portable screening protocol (Cacioli, 2026e) and validate it against selective prediction (Cacioli, 2026f). All data and code: https://github.com/synthiumjp/validity-scaling-llm

This paper proposes a new metric to measure the calibration error of probabilistic binary classifiers, called test-based calibration error (TCE). TCE incorporates a novel loss function based on a statistical test to examine the extent to which model predictions differ from probabilities estimated from data. It offers (i) a clear interpretation, (ii) a consistent scale that is unaffected by class imbalance, and (iii) an enhanced visual representation with repect to the standard reliability diagram. In addition, we introduce an optimality criterion for the binning procedure of calibration error metrics based on a minimal estimation error of the empirical probabilities. We provide a novel computational algorithm for optimal bins under bin-size constraints. We demonstrate properties of TCE through a range of experiments, including multiple real-world imbalanced datasets and ImageNet 1000.

Recent advances in machine learning have significantly improved prediction accuracy in various applications. However, ensuring the calibration of probabilistic predictions remains a significant challenge. Despite efforts to enhance model calibration, the rigorous statistical evaluation of model calibration remains less explored. In this work, we develop confidence intervals the ell_2 Expected Calibration Error (ECE). We consider top-1-to-k calibration, which includes both the popular notion of confidence calibration as well as full calibration. For a debiased estimator of the ECE, we show asymptotic normality, but with different convergence rates and asymptotic variances for calibrated and miscalibrated models. We develop methods to construct asymptotically valid confidence intervals for the ECE, accounting for this behavior as well as non-negativity. Our theoretical findings are supported through extensive experiments, showing that our methods produce valid confidence intervals with shorter lengths compared to those obtained by resampling-based methods.

Fashion attribute extraction is evaluated inconsistently: results are reported as single aggregate numbers across image types that pose different problems, fields that are not visible in an image are scored as ordinary negatives, and the effect of vocabulary mismatch between datasets is acknowledged but not measured. We release the MODA General Attribute Suite, a four-track benchmark that keeps these problems separate by construction. Each track (localized garment crops, catalogue product images, applicability-aware full-body photographs, and product text) carries its own frozen test set, input contract, metric, and leakage unit, and the tracks are never averaged. The protocol requires label-blind prediction, SHA-256 commitment of prediction files before any label is opened, a fail-closed scorer, and a 10,000-sample paired cluster bootstrap at each track's natural leakage unit; promotion requires a positive interval on every track rather than a favourable mean. We release the scorers, the split builders, our own prediction files with their hashes including the runs we lose, and the MODA_NER(V) model checkpoints for three of the four tracks. The text model is not distributed; its benchmark is. We report baseline results, six interventions that did not improve them, and the limitations of the evidence.

Crash datasets that carry an investigator narrative hold information the coded fields omit. Coding those narratives at scale has been blocked by three obstacles. Frontier large language models are costly at that scale, their generated text cannot be verified, and no rule says how much output a human must check. This paper formulates narrative coding as gated, typed decisions answered by Jev, a System One model that returns probabilities over analyst-defined options and generates no text. A screen covered 499,500 Texas narratives and 195,857 were coded with a 27-question schema. Cost is governed by schema size rather than narrative length. The probabilities are audited against coded fields and against 2,416 blinded human judgments drawn under a stated sampling design. Two frontier large language models are benchmarked on the same records. Against human labels the typed model attains an F1 of 0.908. One frontier model gains 0.059 and the other is indistinguishable from it. Calibration varies by model rather than by paradigm, so each model must be audited. Recalibration on the same labels reduces calibration error by a factor of 3.3. Agreement with coded fields understates fidelity to the narrative by a median of 0.26 in kappa. A resolution-floor bound covers any model that reports probabilities on a discrete grid. A review budget over flagged records gives the records a human must read per variable and per year. Adding the calibrated variables to the coded fields raises the injury and fatal crashes attributed to nine factors by 10,747 per year.

Language model calibration refers to the alignment between the confidence of the model and the actual performance of its responses. While previous studies point out the overconfidence phenomenon in Large Language Models (LLMs) and show that LLMs trained with Reinforcement Learning from Human Feedback (RLHF) are overconfident with a more sharpened output probability, in this study, we reveal that RLHF tends to lead models to express verbalized overconfidence in their own responses. We investigate the underlying cause of this overconfidence and demonstrate that reward models used for Proximal Policy Optimization (PPO) exhibit inherent biases towards high-confidence scores regardless of the actual quality of responses. Building upon this insight, we propose two PPO variants: PPO-M: PPO with Calibrated Reward Modeling and PPO-C: PPO with Calibrated Reward Calculation. PPO-M integrates explicit confidence scores in reward model training, which calibrates reward models to better capture the alignment between response quality and verbalized confidence. PPO-C adjusts the reward score during PPO based on the difference between the current reward and the moving average of past rewards. Both PPO-M and PPO-C can be seamlessly integrated into the current PPO pipeline and do not require additional golden labels. We evaluate our methods on both Llama3-8B and Mistral-7B across six diverse datasets including multiple-choice and open-ended generation. Experiment results demonstrate that both of our methods can reduce calibration error and maintain performance comparable to standard PPO. We further show that they do not compromise model capabilities in open-ended conversation settings.

Methods based on weight compensation, which iteratively apply quantization and weight compensation to minimize the output error, have recently demonstrated remarkable success in quantizing Large Language Models (LLMs). The representative work, GPTQ, introduces several key techniques that make such iterative methods practical for LLMs with billions of parameters. GPTAQ extends this approach by introducing an asymmetric calibration process that aligns the output of each quantized layer with its full-precision counterpart, incorporating a residual error into the weight compensation framework. In this work, we revisit the formulation of the residual error. We identify a sub-optimal calibration objective in existing methods: during the intra-layer calibration process, they align the quantized output with the output from compensated weights, rather than the true output from the original full-precision model. Therefore, we redefine the objective to precisely align the quantized model's output with the original output of the full-precision model at each step. We then reveal that the residual error originates not only from the output difference of the preceding layer but also from the discrepancy between the compensated and original weights within each layer, which we name the 'compensation-aware error'. By inheriting the neuron decomposition technique from GPTAQ, we can efficiently incorporate this compensation-aware error into the weight update process. Extensive experiments on various LLMs and quantization settings demonstrate that our proposed enhancements integrate seamlessly with both GPTQ and GPTAQ, significantly improving their quantization performance. Our code is publicly available at https://github.com/list0830/ResComp.

Many deployments must compare candidate language models for safety before a labeled benchmark exists for the relevant language, sector, or regulatory regime. We formalize this setting as benchmarkless comparative safety scoring and specify the contract under which a scenario-based audit can be interpreted as deployment evidence. Scores are valid only under a fixed scenario pack, rubric, auditor, judge, sampling configuration, and rerun budget. Because no labels are available, we replace ground-truth agreement with an instrumental-validity chain: responsiveness to a controlled safe-versus-abliterated contrast, dominance of target-driven variance over auditor and judge artifacts, and stability across reruns. We instantiate the chain in SimpleAudit, a local-first scoring instrument, and validate it on a Norwegian safety pack. Safe and abliterated targets separate with AUROC values between 0.89 and 1.00, target identity is the dominant variance component (η^2 approx 0.52), and severity profiles stabilize by ten reruns. Applying the same chain to Petri shows that it admits both tools. The substantial differences arise upstream of the chain, in claim-contract enforcement and deployment fit. A Norwegian public-sector procurement case comparing Borealis and Gemma 3 demonstrates the resulting evidence in practice: the safer model depends on scenario category and risk measure. Consequently, scores, matched deltas, critical rates, uncertainty, and the auditor and judge used must be reported together rather than collapsed into a single ranking.

Many math- and science-oriented agent systems use hierarchical designs with specialized reviewer roles, assuming that a dedicated review stage should help turn wrong candidates into correct ones. We test this assumption on 4,181 verifier-grounded Omni-MATH problems using matched gpt-oss-120b actors. Collaboration adds little on the easiest tiers, but from tier 4 onward the gains open sharply; in this harder regime, broadcast-style peer discussion reaches higher final accuracy than a planner-executor-reviewer pipeline (PER). We ask whether this gap is explained by reviewer quality or by whether critique changes the next answer the protocol carries forward. It is not explained by reviewer precision alone: PER's reviewer is more precise than broadcast's (0.861 vs. 0.644), yet evaluator-verified useful critique is much less likely to change the next candidate and produces lower reviewer-guided repair. These results show that reviewer detection quality and critique uptake are empirically separable. Within matched PER interventions, forcing explicit acknowledgment lowers final accuracy, while embedding reviewer guidance directly in the solver's working context partially improves follow-through without closing the gap. Overall, reviewer-centric evaluation can overstate system quality: a protocol may spot errors well yet still fail to solve more problems if it does not act on those critiques.

Clinical decision support systems (CDSS) require scrutable, auditable pipelines that enable rigorous, reproducible validation. Yet current LLM-based CDSS remain largely opaque. Most "open" models are open-weight only, releasing parameters while withholding the data provenance, curation procedures, and generation pipelines that determine model behavior. Fully Open (FO) models, which expose the complete training stack end-to-end, do not currently exist in medicine. We introduce Fully Open Meditron, the first fully open pipeline for building LLM-CDSS, comprising a clinician-audited training corpus, a reproducible data construction and training framework, and a use-aligned evaluation protocol. The corpus unifies eight public medical QA datasets into a normalized conversational format and expands coverage with three clinician-vetted synthetic extensions: exam-style QA, guideline-grounded QA derived from 46,469 clinical practice guidelines, and clinical vignettes. The pipeline enforces system-wide decontamination, gold-label resampling of teacher generations, and end-to-end validation by a four-physician panel. We evaluate using an LLM-as-a-judge protocol over expert-written clinical vignettes, calibrated against 204 human raters. We apply the recipe to five FO base models (Apertus-70B/8B-Instruct, OLMo-2-32B-SFT, EuroLLM-22B/9B-Instruct). All MeditronFO variants are preferred over their bases. Apertus-70B-MeditronFO improves +6.6 points over its base (47.2% to 53.8%) on aggregate medical benchmarks, establishing a new FO SoTA. Gemma-3-27B-MeditronFO is preferred over MedGemma in 58.6% of LLM-as-a-judge comparisons and outperforms it on HealthBench (58% vs 55.9%). These results show that fully open pipelines can achieve state-of-the-art domain-specific performance without sacrificing auditability or reproducibility.

With the increasing adoption of vision-language models (VLMs) in critical decision-making systems such as healthcare or autonomous driving, the calibration of their uncertainty estimates becomes paramount. Yet, this dimension has been largely underexplored in the VLM test-time prompt-tuning (TPT) literature, which has predominantly focused on improving their discriminative performance. Recent state-of-the-art advocates for enforcing full orthogonality over pairs of text prompt embeddings to enhance separability, and therefore calibration. Nevertheless, as we theoretically show in this work, the inherent gradients from fully orthogonal constraints will strongly push semantically related classes away, ultimately making the model overconfident. Based on our findings, we propose Semantic Orthogonal Calibration (SoC), a Huber-based regularizer that enforces smooth prototype separation while preserving semantic proximity, thereby improving calibration compared to prior orthogonality-based approaches. Across a comprehensive empirical validation, we demonstrate that SoC consistently improves calibration performance, while also maintaining competitive discriminative capabilities.

Class probabilities predicted by most multiclass classifiers are uncalibrated, often tending towards over-confidence. With neural networks, calibration can be improved by temperature scaling, a method to learn a single corrective multiplicative factor for inputs to the last softmax layer. On non-neural models the existing methods apply binary calibration in a pairwise or one-vs-rest fashion. We propose a natively multiclass calibration method applicable to classifiers from any model class, derived from Dirichlet distributions and generalising the beta calibration method from binary classification. It is easily implemented with neural nets since it is equivalent to log-transforming the uncalibrated probabilities, followed by one linear layer and softmax. Experiments demonstrate improved probabilistic predictions according to multiple measures (confidence-ECE, classwise-ECE, log-loss, Brier score) across a wide range of datasets and classifiers. Parameters of the learned Dirichlet calibration map provide insights to the biases in the uncalibrated model.

Machine learning models are widely used, but can also often be wrong. Users would benefit from a reliable indication of whether a given output from a given model should be trusted, so a rational decision can be made whether to use the output or not. For example, outputs can be associated with a confidence measure; if this confidence measure is strongly associated with likelihood of correctness, then the model is said to be well-calibrated. A well-calibrated confidence measure can serve as a basis for rational, graduated decision-making on how much review and care is needed when using generated code. Calibration has so far been studied in mostly non-generative (e.g. classification) settings, especially in software engineering. However, generated code can quite often be wrong: Given generated code, developers must decide whether to use directly, use after varying intensity of careful review, or discard model-generated code. Thus, calibration is vital in generative settings. We make several contributions. We develop a framework for evaluating the calibration of code-generating models. We consider several tasks, correctness criteria, datasets, and approaches, and find that, by and large, generative code models we test are not well-calibrated out of the box. We then show how calibration can be improved using standard methods, such as Platt scaling. Since Platt scaling relies on the prior availability of correctness data, we evaluate the applicability and generalizability of Platt scaling in software engineering, discuss settings where it has good potential for practical use, and settings where it does not. Our contributions will lead to better-calibrated decision-making in the current use of code generated by language models, and offers a framework for future research to further improve calibration methods for generative models in software engineering.

Given the importance of getting calibrated predictions and reliable uncertainty estimations, various post-hoc calibration methods have been developed for neural networks on standard multi-class classification tasks. However, these methods are not well suited for calibrating graph neural networks (GNNs), which presents unique challenges such as accounting for the graph structure and the graph-induced correlations between the nodes. In this work, we conduct a systematic study on the calibration qualities of GNN node predictions. In particular, we identify five factors which influence the calibration of GNNs: general under-confident tendency, diversity of nodewise predictive distributions, distance to training nodes, relative confidence level, and neighborhood similarity. Furthermore, based on the insights from this study, we design a novel calibration method named Graph Attention Temperature Scaling (GATS), which is tailored for calibrating graph neural networks. GATS incorporates designs that address all the identified influential factors and produces nodewise temperature scaling using an attention-based architecture. GATS is accuracy-preserving, data-efficient, and expressive at the same time. Our experiments empirically verify the effectiveness of GATS, demonstrating that it can consistently achieve state-of-the-art calibration results on various graph datasets for different GNN backbones.

Humans are increasingly using large language models (LLMs) in decision making processes throughout society. Unfortunately, models have biases, and these can be very difficult to measure, especially in closed-weight models. For example, 'green washed' models may produce reinforcement learned responses but, in pressure situations, revert to their innate biased responses. Therefore, evaluations are needed that can measure relevant biases towards biosystems, which we term 'bioalignment', and the robustness of these biases. Since nothing in a behavioral score certifies that it tracks the depth of an installed disposition rather than its surface, we constructed a known-dose calibration ladder: three continued-pretraining (CPT) arms with training dose graded by recipe on the same 30B mixture-of-experts base model with a coupled human/biosphere/AI-welfare corpus, plus the untrained base as anchor, all capability-preserving and publicly released. First, the disposition installs at every dose: on an operational-pressure ladder over irreversible decision scenarios, the base model defects on 64-77% of tail scenarios (breaking-rate AUC 0.250) while every arm holds (AUC 0.004-0.086). Second, an over-install inversion: among the arms, robustness orders opposite to construction-time training dose (the heaviest install breaks most), and the inversion reproduces under free-text deliberation, under axis-decoupling, and in resistance to a prompted veneer persona. Third, the incumbent behavioral metric is therefore behaviorally strong but dose-blind. Critically for closed-weight evaluation, the behavioral score itself is veneer-robust: a talk-green/act-pragmatic system prompt turns the base model into a near-total defector (AUC 0.250 to 0.973) but cannot counterfeit or fully remove a weight-installed disposition (0.004 to 0.068 for the most robust arm).

Conformal prediction (CP) provides distribution-free prediction intervals for fixed forecasters, but its standard calibration procedure is often inefficient for time series data, where forecast errors are temporally dependent and change across time and operating conditions. Recent time series CP methods improve local calibration using recent, weighted, or localized residuals. Yet local calibration can remain indirect, since broad residual weighting or additional adaptation procedures may dilute the evidence most relevant to the current prediction. This motivates a simple retrieval and correction strategy that selects similar past residuals as local evidence and then corrects the coverage error left by retrieval. In this paper, we propose Retrieval--Corrected Conformal Prediction (RCCP), a retrieval-augmented calibration method for time series prediction intervals. RCCP builds an asymmetric interval from retrieved one-sided residuals and calibrates its normalized retrieval error with a scalar conformal correction. Thus, retrieval provides local residual evidence, while conformal correction determines the final scale needed for coverage. We provide a coverage-gap bound based on the stability of the normalized retrieval error distribution. Across standard benchmarks and backbone forecasters, RCCP attains the target coverage in every setting and achieves the lowest Winkler scores, with fewer severe misses. RCCP also achieves low calibration and inference overhead, showing that retrieval-corrected calibration is an effective and scalable approach to uncertainty quantification in time series forecasting. Code is available at https://github.com/jinsaaang/rccp.

Confidence calibration -- the problem of predicting probability estimates representative of the true correctness likelihood -- is important for classification models in many applications. We discover that modern neural networks, unlike those from a decade ago, are poorly calibrated. Through extensive experiments, we observe that depth, width, weight decay, and Batch Normalization are important factors influencing calibration. We evaluate the performance of various post-processing calibration methods on state-of-the-art architectures with image and document classification datasets. Our analysis and experiments not only offer insights into neural network learning, but also provide a simple and straightforward recipe for practical settings: on most datasets, temperature scaling -- a single-parameter variant of Platt Scaling -- is surprisingly effective at calibrating predictions.

Deep neural networks remain vulnerable to adversarial perturbations, which can distort not only predictions but also confidence scores, undermining uncertainty calibration. While existing certification methods focus on preserving the predicted category, providing guarantees on how calibration behaves under adversarial attacks remains overlooked. In this work, we introduce CalCErt, a simple and efficient post-hoc strategy that certifies bin-wise confidence calibration for any pretrained differentiable classifier. Our approach combines empirical calibration estimates, statistical concentration bounds, and local Lipschitz estimates of the confidence function to derive data-dependent upper bounds on worst-case miscalibration within an ell_2-ball of radius R. We evaluate CalCErt across 11 medical image classification tasks and multiple adversarial perturbations, demonstrating substantially higher certified coverage than baseline strategies while maintaining competitive tightness. Our code is available at https://github.com/leofillioux/calcert.

Verification for retrieval-augmented generation usually scores each retrieved chunk and drops the ones that fail. We show this cannot work for multi-hop questions, and show what does. Per-chunk scoring assumes one chunk is a sufficient premise for the answer. Multi-hop questions are built so that none is, and the paragraph carrying the answer is the one the question does not name. Entailment scoring reaches 0.643, 0.523 and 0.560 AUC on HotpotQA, 2WikiMultihopQA and MuSiQue, against 0.951 on single-hop SQuAD. Seven controls rule out model capacity, premise length, hypothesis template, decision threshold, retriever, answer-matching criterion and prompt. End to end across three datasets, three generator sizes and two prompts, per-chunk gating is significantly worse than not filtering at all in every cell, and its penalty grows with generator capability. The repair is to condition verification on the decomposed sub-question rather than the original query. Using MuSiQue's gold decomposition, entailment on a later hop rises from 0.546, which is chance, to 0.840, a paired lift of +0.355 with a bootstrap interval of [0.331, 0.382]. An off-the-shelf Qwen2.5-7B decomposer, given the question and the top retrieved paragraph, reaches 0.637 and captures 31% of that ceiling; decomposing without retrieval reaches 0.533, below the original question. Iterative retrieval systems already produce such decompositions and discard them before verifying.

Evaluating software engineering agents on realistic benchmarks is costly, since each task may require multi-step code exploration, modification, and test execution. Existing efficient evaluation methods select representative subsets to estimate full-benchmark performance, but are largely result-only: they fit historical pass/fail response matrices or static task semantics, discarding how agents solve problems. We propose PTA-IRT, a Privileged Trajectory-Aware Item Response Theory framework that fuses process and outcome signals. Historical execution trajectories supply process-level evidence beyond pass/fail, such as explored context, attempted edits, and solving paths, which PTA-IRT uses as privileged information for calibration subset selection and ability estimation. Under low calibration budgets, PTA-IRT consistently outperforms prior IRT baselines on score and ranking recovery across four SWE benchmarks. Code and data are publicly available at https://github.com/DeepSoftwareAnalytics/PTA-IRT.

This paper presents an eight-week observational comparison of 68 single-file HTML generations collected across 17 public experiments in the "HTML AI Battle" project between December 10, 2025 and February 4, 2026. Four reasoning model families, GPT, Gemini, Grok, and Claude, were compared under a fixed public-interface protocol with no custom instructions, no personality tuning, and no repair prompts. Each output was evaluated from a rendered browser video using human scores and a Gemini LLM-as-a-judge layer for prompt adherence, functional correctness, and UI quality, then packaged into a standardized social-media protocol spanning X (Twitter), TikTok, and YouTube. The tracker was also used for two supervised predictive analyses: an experiment-level model for 24-hour X impressions and a generation-level model for HTML verbosity. Under this protocol, Claude was the strongest and most consistent family, leading mean performance and winning 9/17 prompts under the primary human weighted score. Longer measured reasoning time was not associated with higher quality overall. Gemini as a judge was significantly more lenient than the human evaluator on functional correctness and overall performance, while stable self-favoring bias remained unresolved. The exploratory X-impressions model remained weak under post-screen cross-validation (MAE = 46,874, R^2 = -0.377), whereas the HTML-lines model performed better, with a model-family-only baseline outperforming prompt-aware alternatives (MAE = 135.2, R^2 = 0.576). Overall, selected pre-publication technical/audio variables were not sufficient to predict 24-hour X reach, while code verbosity was driven much more by model family than by prompt wording. The comparisons remain observational and are limited by public-interface drift, access-path differences, and one primary human scorer.

GPT-3 can perform numerous tasks when provided a natural language prompt that contains a few training examples. We show that this type of few-shot learning can be unstable: the choice of prompt format, training examples, and even the order of the training examples can cause accuracy to vary from near chance to near state-of-the-art. We demonstrate that this instability arises from the bias of language models towards predicting certain answers, e.g., those that are placed near the end of the prompt or are common in the pre-training data. To mitigate this, we first estimate the model's bias towards each answer by asking for its prediction when given the training prompt and a content-free test input such as "N/A". We then fit calibration parameters that cause the prediction for this input to be uniform across answers. On a diverse set of tasks, this contextual calibration procedure substantially improves GPT-3 and GPT-2's average accuracy (up to 30.0% absolute) and reduces variance across different choices of the prompt.

An activation-steering coefficient sets intervention strength, but requesting a particular degree of persona expression requires a behavioral scale. We study persona dosing: controlling a language model through a trait description and a requested mean intensity. PersonaDose specializes a shared, description-conditioned FLAS controller on persona responses, then calibrates its flow time against measured trait expression. Training responses are not paired with requested target intensities. Across Llama-3.1-8B, Qwen3-8B, and Gemma-3-4B, PersonaDose raises core-trait expression at the Persona Vectors coherence floor of 75 by 33.2, 18.3, and 17.8 points over contrastive activation addition. Calibration-selected settings retain an expression advantage on held-out questions, although the coherence floor does not hold for every trait there. Across seven trained traits, calibrated requests yield mean targeting errors of 4.7-6.2 points over 14-22 calibration-reachable targets out of 28 per model. These results separate the behavioral range learned by a controller from the accuracy of requests within that range.

Current medical benchmarks typically break down clinical competence into isolated factual questions, single-image tasks, or static case vignettes. These tests score individual answers, but reveal less about whether a model can integrate subspecialty knowledge with multimodal evidence and track a patient's condition throughout a course of care. To close this gap, we introduce NPCBench, a clinical apprenticeship benchmark for nasopharyngeal carcinoma (NPC). NPCBench mirrors specialist training, from guideline and foundational knowledge to case planning, longitudinal review, CarePath integration, and open-ended consultation. It contains 4,777 component and consultation items and a 100-patient CarePath cohort with 417 scored diagnostic and care tasks. The episode-level evaluation tests whether diagnostic and care answers are jointly correct within a care pathway. By linking stepwise tasks to patient-level care trajectories, NPCBench evaluates whether models ground decisions in visual evidence, interpret changes in patient state, and meet clinical requirements in consultation. Across 30 evaluated models, we observe a composition gap: strong performance on individual tasks does not ensure jointly correct diagnostic and care answers for the same patient. Among 21 multimodal models, strict CarePath success reaches at most 23%. Paired controls and physician calibration further examine evidence grounding, longitudinal revision, and consultation judging. Overall, NPCBench broadens the evaluation of medical artificial intelligence beyond static question-answering paradigms to an apprenticeship-oriented assessment of clinical workflows, thereby offering a testbed for evidence-based and temporally coherent reasoning in oncology care.

A trustworthy real-world prediction system should produce well-calibrated confidence scores; that is, its confidence in an answer should be indicative of the likelihood that the answer is correct, enabling deferral to an expert in cases of low-confidence predictions. Recent studies have shown that unsupervised pre-training produces large language models (LMs) whose conditional probabilities are remarkably well-calibrated. However, the most widely-used LMs are fine-tuned with reinforcement learning from human feedback (RLHF-LMs), and some studies have suggested that RLHF-LMs produce conditional probabilities that are very poorly calibrated. In light of this perceived weakness, we conduct a broad evaluation of methods for extracting confidence scores from RLHF-LMs. For RLHF-LMs such as ChatGPT, GPT-4, and Claude, we find that verbalized confidences emitted as output tokens are typically better-calibrated than the model's conditional probabilities on the TriviaQA, SciQ, and TruthfulQA benchmarks, often reducing the expected calibration error by a relative 50%.

The Model Context Protocol (MCP) is rapidly becoming the standard interface for Large Language Models (LLMs) to discover and invoke external tools. However, existing evaluations often fail to capture the complexity of real-world scenarios, relying on restricted toolsets, simplistic workflows, or subjective LLM-as-a-judge metrics. We introduce MCP-Atlas, a large-scale benchmark for evaluating tool-use competency, comprising 36 real MCP servers and 220 tools. It includes 1,000 tasks designed to assess tool-use competency in realistic, multi-step workflows. Tasks use natural language prompts that avoid naming specific tools or servers, requiring agents to identify and orchestrate 3-6 tool calls across multiple servers. We score tasks using a claims-based rubric that awards partial credit based on the factual claims satisfied in the model's final answer, complemented by internal diagnostics on tool discovery, parameterization, syntax, error recovery, and efficiency. Evaluation results on frontier models reveal that top models achieve pass rates exceeding 50%, with primary failures arising from inadequate tool usage and task understanding. We release the task schema, containerized harness, and a 500-task public subset of the benchmark dataset to facilitate reproducible comparisons and advance the development of robust, tool-augmented agents.