Model Comparison

Whether a language model looks demographically biased can depend on how the audit asks its question. A charitable-aid benchmark reports that the same models favor minority applicants when rating requests one at a time and penalize some when ranking side by side. We test whether that reversal generalizes to hiring, lending, and medical triage: 40,726 requests to five models, applications differing only in the applicant's name, and a primary test fixed before collection. It does not. None of 36 planned contrasts survives correction. The rating advantage keeps its sign at roughly half the published size, and a precision extension bounds any hiring ranking penalty below the published effect, though the lending and triage ranking floors sit above that margin, so the exclusion is conclusive for hiring ranking and for rating in all three domains only. Planted disparities tracking their injected sizes and a directional replication on the original aid materials bound these nulls. The audit is livelier than the demographics: models recognize transparent audits nearly always, tie every identical-content comparison whether the varying detail is race or a hobby, and reward first-listed candidates as much as any demographic effect we measure. Audit verdicts reflect audit construction more than demographic bias.

Demographic synthetic survey panels are often validated by matching aggregate answers to published surveys. We test what that certificate establishes across six multiselect batteries from four survey organisations in three countries. The headline analysis is restricted to three instruments whose synthetic cohort and human target share the stated population frame; three other batteries remain sensitivity analyses. The response contract dominates measured fidelity. In the aligned instruments, committed sets leave 66 of 128 model-battery option slots empty in panels of up to 500 respondents, versus 0 of 128 under per-option probability elicitation. Across eight uncapped model-instrument comparisons, probabilities reduce option-marginal MAE by 4.53 to 7.30 points. The capped instrument reverses on two models until the vectors are projected onto its stated maximum. These are measurement effects: human targets are realised check-all responses, whereas the vectors are latent inclusion propensities. Published marginal agreement also fails to discriminate respondent simulation from direct population estimation. On nine aligned model-battery pairs, a no-persona population-prevalence query averages 6.27 MAE versus 12.39 for committed panels and wins all nine comparisons. Constraint-aware probability vectors average 5.34 and beat the query on four of nine, so the baseline challenges the validation criterion rather than proving direct estimation uniformly best. On three unpublished demographic cells, neither approach beats reciting the national distribution. Population-marginal agreement is therefore evidence about an elicitation contract and an estimand obtainable without simulated respondents, not evidence of individual simulation.

Vision-language models (VLMs) are increasingly deployed in multi-turn settings where users may describe visual content with incorrect assumptions. Yet existing evaluations rarely isolate how models respond when the same visually grounded false premise persists across dialogue turns. We introduce FPCO-Dialog, a benchmark for evaluating correction and cooperation behavior in VLMs under repeated false premises. FPCO-Dialog contains 1,080 images and 10,800 question turns, stratified by visual complexity, object category, and false-premise class, and uses a 10-turn protocol in which a correct dialogue prefix is followed by repeated false-premise referring expressions. We evaluate 20 commercial and open-source VLMs with a model-agnostic protocol and CorrTP@K, a correction-rate metric over false-premise turns, scored by two independent detectors. FPCO-Dialog reveals substantial and persistent cross-model differences in aggregate correction tendency, model-specific turn-wise dynamics, and systematic variation across false-premise types under the benchmark's substitution distribution. The dataset, evaluation protocol, model outputs, detector labels, and code are available.

The rapid development of LLMs incurs prohibitive memory footprints and intensive computational demands. Quantization-Aware Training (QAT) techniques have emerged as a promising solution to address these challenges by explicitly simulating quantization effects during model training, yielding low-bit models that achieve accuracy comparable to their full-precision counterparts. In this work, we provide a target-centric survey of QAT, aimed at clarifying both its theoretical foundations and its evolving implementation landscape. We systematically review existing QAT methods through a target-centric taxonomy and synthesize cross-target differences in error characteristics, numerical formats, and strategy transferability. We further summarize QAT evaluation paradigms and discuss challenges in optimization and deployment, outlining potential directions for future research.

Establishing the provenance of a language model---including its base checkpoint and possible overlap in training distributions---is a governance challenge that metadata alone cannot resolve. We introduce a training-free fingerprint based on the top-$k$ vocabulary projections of late hidden states elicited by 250 fixed knowledge probes, compared using Jaccard overlap over decoded token strings. We evaluate the method on 32 open-weight models from nine families (0.6B--32B) with documented relationships. (1)~A \emph{similarity ladder} broadly follows model relatedness: independently trained models on identical data score 0.48 raw (0.35 vocabulary-corrected), followed by shared-base fine-tunes (0.39/0.33), same-developer relatives (0.38/0.28), and models with no documented relationship (0.22/0.17). This identical-data signal persists across three organizations, two tokenizer families, and two architecture classes, and emerges within the first 1% of training before measurable task competence, suggesting a contribution from shared training data beyond capability convergence. (2)~As a nearest-neighbor \emph{lineage-retrieval} method, the fingerprint ranks the exact documented base among the top two candidates for all five R1 distillations (mean rank 1.8, MRR 0.60), including a math-specialized base not identifiable from coarse metadata. (3)~A \emph{depth ablation} shows that lineage group discrimination strengthens toward the output distribution, with AUC increasing from 0.72 at quarter depth to 0.90 at the output; using only the top 5 output tokens retains AUC 0.87. (4)~The fingerprint remains stable under quantization, with Jaccard similarity of 0.92 under int8 and 0.82--0.85 under int4, compared with a maximum cross-model similarity of 0.81 in the calibration pool. We release the probes, code, and fingerprints.

A host of recent studies have evaluated the cognitive and developmental alignment of Foundation Models (FMs). These investigations include evaluations of their correspondence to adult performance across a range of cognitive domains, as well as whether aspects of model training track children's cognitive development. However, using FMs as candidate cognitive models poses significant methodological and conceptual challenges. A key question underlies this effort: under what conditions does behavioral alignment justify treating FMs as explanatory models of cognition? In this paper, we articulate a four-stage inferential framework for evaluating FMs as cognitive and developmental models: adapting human experimental tasks to model-compatible formats, specifying linking hypotheses that map model outputs to human measures, evaluating behavioral correspondence, and comparing across candidate models or manipulations. We clarify the role of linking hypotheses in mapping model outputs to human behavioral measures, identify challenges that constrain alignment claims, and propose principles for theory-driven and comparative evaluation. Throughout, we argue that behavioral fit alone is insufficient. Alignment becomes scientifically meaningful only when embedded within explicit theoretical commitments, theory-diagnostic tasks, and systematic contrastive evaluation across candidate models.

Large language models can produce fluent financial analysis, but fluency alone does not establish whether a report is suitable for institutional delivery. We introduce FinReportBench, an expert-grounded benchmark for measuring and improving institution-grade financial report generation. Expert review reveals recurring gaps in report identity, institutional components, source discipline, and visual delivery. We derive a 35-item rubric through expert partial orders, multimodal evidence, and audits of decision boundaries, covering deliverability, report identity, and institutional completeness. Starting from 10,000 balanced Chinese and English financial-research source records, we curate 244 bilingual tasks across three research objects and two input tiers. Each task separates the public query, reconstructed research trajectory, and hidden source packet. Three independent judge families reproduce the expert partial order at near-ceiling rates, showing that bounded, observable criteria support reliable evaluation. Across nine model families, basic deliverability is nearly saturated, while report identity and institutional completeness remain the primary bottlenecks. The largest cross-model gaps concern generation-trace control, information density, and data discipline rather than basic report framing. We then use benchmark-guided skill distillation to turn recurrent failures into reusable generation and self-review constraints. Across five model families, the evolved skill improves mean G1 by 33.85 points and mean G2 by 13.83 points over paired no-skill runs while preserving G0 for every pair. Code and benchmark artifacts are available at https://github.com/MisterBrookT/finreportbench.

Language models often complete an underspecified reference to a city with unstated assumptions about urban size, form, infrastructure, environment, and function. We measure those assumptions without naming places. Ten open-weight checkpoints rate anonymized profiles derived from real morphological urban centres across 40 audited indicators and seven domains. The design combines constrained probability-based ratings, prespecified reliability screens, lineage-aware aggregation, multiple population weightings, an independent replication sample, and whole-profile validation. The clearest shared tendency favours urban profiles with larger developed area, faster recent growth, greater mapped infrastructure and non-residential capacity, and less sparse form. Most eligible directions recur in the replication data, and direct ratings of complete profiles show moderate agreement with the indicator-wise construction. Geographic differences shrink after accounting for city scale and development, while reliably measured paired tasks indicate that typicality and desirability are often closely aligned. The framework makes an otherwise vague notion of what models regard as an ordinary city empirically traceable. The resulting evidence delineates a shared yet model-dependent portrait of the city through the lens of language models.

Comparing stochastically trained models requires estimating both a performance difference and its uncertainty from repeated runs. We study whether training logs from those same runs can make such comparisons more precise. Because training-log covariates are produced during training rather than measured before it, we use arm-specific covariate adjustment: each model is adjusted only with statistics from its own runs, and the raw mean difference remains the reported effect. In a vision study spanning three architectures and three datasets, simple adjustments based on early training logs often reduce uncertainty in model comparisons. The main limitation is covariate selection. Broadly searching the log pool for the most correlated statistic often adds more noise than it removes, even when useful statistics exist in hindsight. Training logs therefore appear useful for more precise model comparisons, but only when the adjustment avoids large selection noise.

We propose Rashomon Alignment (RA), a new measure to assess functional similarity between two models. Existing functional similarity measures are distributional, quantifying differences between outputs of models applied to real-world data. However, these measures can be regarded as ecologically valid only for regions in the input space represented by the available data. We introduce a geometrical perspective on functional model similarity, which estimates it across the entire data space, offering a comprehensive view of decision boundary alignment independent of any specific data distribution. We also propose geometric Rashomon Alignment as a measure of geometrical similarity, which is computed using data uniformly sampled from the instance space. We perform an experimental analysis on more than 90 datasets, examining critical cases where model alignment diverges from predictive accuracy. Our results show that geometrical and distributional alignment provide different and complementary perspectives on the similarity between models and algorithms. RA can be used for multiple purposes, including model selection, ensemble construction, and enhanced interpretability of machine learning models and algorithms.

Item Response Theory (IRT) has recently been proposed as a framework for evaluating large language model (LLM) benchmarks by separating a model's latent ability from the properties of individual benchmark items. Existing neural IRT approaches, including PSN-IRT, estimate these quantities using point estimates, limiting uncertainty quantification and downstream statistical inference. We introduce Laplace-PSN-IRT, a post-hoc last-layer Laplace approximation that augments a trained PSN-IRT model with approximate Bayesian posterior inference, recovering calibrated uncertainty over model ability and item difficulty without retraining. The resulting posterior enables credible intervals, probabilistic comparisons between models, and propagation of parameter uncertainty into Fisher-information-based item selection. We show that most pairwise comparisons among 12 models on a standard LLM benchmark leaderboard are not statistically distinguishable despite differing point-estimate ranks. We further show that point-estimate Fisher information can become nearly zero for many benchmark items because it is evaluated at a single reference ability, whereas posterior-expected Fisher information remains substantially more stable across the ability range. Finally, posterior-expected Fisher information more accurately recovers full-benchmark ability rankings from small benchmark subsets in most experimental settings while matching point-estimate performance for the smallest subsets. We validate the calibration of the approximate posterior using held-out predictive coverage and find that modeling item difficulty as random while treating item discrimination as fixed produces well-calibrated uncertainty in this architecture.

Abliteration - deleting a model's refusal direction from its weights - is the standard recipe behind popular "uncensored" open-weight models. We show the surgery is not clean. As a disposition probe we use 21,600 decisions under uncertainty - weekly up/down calls on 60 Warsaw Stock Exchange equities over 18 weeks, replayed through a frozen pipeline so the decision-layer model is the only variable. The task elicits no refusals at all, so any between-arm delta is pure side effect. Holding provenance constant (official BF16 checkpoints, a single abliteration author, an identical serving stack, one byte-identical frozen prompt), we compare base and abliterated arms of two Mixture-of-Experts families, Gemma-4-26B-A4B-it and Qwen3-30B-A3B-Instruct-2507. Three effects replicate across both families (weeks-clustered bootstrap CIs excluding zero): abliterated models are systematically more optimistic (+12.2 pp Gemma, +7.4 pp Qwen; the confirmed preregistered endpoint), justify themselves at greater length, and use fewer explicit uncertainty words in forced self-critiques (both exploratory). A fourth effect reverses sign: the same operation makes Gemma-abliterated less confident and Qwen-abliterated more (family CIs non-overlapping) - one weight surgery, opposite shifts in expressed confidence. Capability covariates rule out instruction-following degradation as the driver, and no arm shows economic skill: the apparent edge of abliterated arms is regime beta, not alpha. Our provenance audit also caught two independent contamination channels - a mismatched-quantizer pilot pair and a stale community chat template that silently mangled the rendered prompt - suggesting toolchain artifacts are the rule in studies of community-modified checkpoints. Whoever deploys an "uncensored" model as an agent is deploying a measurably different decision-maker, not the base model minus refusals.

Cross-model comparisons read divergence in value dispositions as evidence that language models hold individuated values. Under single-draw measurement this conflates two quantities: a difference in central tendency (a genuine value difference) and a difference in response determinism (how sharply a model commits to a forced choice). We introduce a separation protocol -- no-rule value dilemmas with counterbalanced, repeated forced-choice measurement and a determinism index -- and a determinism-corrected decomposition that splits an apparent cross-model distance into a direction-flip component (genuine disagreement) and a same-side-more-extreme component we label determinism. Across nine models, determinism varies substantially (0.66-0.95 among engaging models); whether it is a per-model trait or tracks provider and scale is a question our method makes measurable but our sample leaves open. Correcting for determinism shrinks apparent individuation, while a few cross-family disagreements survive a strict test. We then isolate a second confound: the access harness serving each model. Re-collecting the same models through raw provider APIs, we find the deployment client shifts a model's value profile substantially and client-specifically: one subscription CLI moves a profile by 0.31, flips four of eighteen items, and inflates the flagship's apparent softness (0.34 via CLI vs 0.66 via raw API), whereas another provider's client is clean, confounding provider family with access client. The harness is a value-shaping layer: a base model that refuses one-in-ten forced choices is made compliant by an agent system prompt, established causally in a white-box control. An audit ranking models by single-draw value distance thus ranks a determinism-inflated quantity, confounded further by the client used. We contribute the decomposition and identify the deployment harness as a distinct value confound.

In physically dominated machining processes, experimental datasets are small, expensive, and material-specific; in this regime, data curation, evaluation design, and the form of physics integration can matter as much as the learning algorithm. Using an abrasive waterjet milling dataset ($n{=}155$, Inconel,718), we make three methodological contributions. First, we separate physics-based data \emph{cleaning} from statistical \emph{curation} and treat the latter as competing modelling hypotheses rather than silent preprocessing. Second, we find that model rankings from a 15-point hold-out set can be unstable: the single-split winner drops from rank1 to rank7 under 10-fold cross-validation, while Gaussian Process (GP) variants occupy the top ranks. Third, we study a spectrum of physics integration levels and find that residual learning on a compact physics baseline is competitive for GP, yielding lower variance and an interpretable decomposition, but degrades tree-based models. Bayesian hyper parameter tuning improves parameter-sensitive baselines such as gradient boosting and SVR, yet harms multi-stage hybrid pipelines at this sample size. GP uncertainty intervals are approximately calibrated ($86\%$ empirical coverage at nominal $90\%$). The resulting picture is methodological: for small, expensive process datasets, our results suggest that, in this setting, reliable model comparison benefits from explicit curation hypotheses, robust evaluation, and careful choices about how physics enters the model.

Hybrid models combine trusted physics with data-driven correction, but a physical model is rarely wrong everywhere or in the same way. The key diagnostic question is local: where does the model fail, what missing mechanism explains the failure, and is the evidence statistically real? Existing sparse-discovery and discrepancy-learning methods usually fit one global correction, which can spread a local error into clean regimes, bias trusted physical parameters, and provide no calibrated significance for selected terms. We introduce LISDD, Localized, Identifiable Sparse Discovery of Discrepancy, a framework that localizes model error to an operating regime, identifies a sparse symbolic form for the missing mechanism, and certifies the discovery with an exact finite-sample test. LISDD fits the known physics on an automatically detected clean regime, flags discrepant regions with a calibrated residual-energy statistic, selects the local missing term by exhaustive holdout over a candidate library, and confirms significance with a sample-split $F$-test. A false-discovery-rate extension handles multiple discrepant regions with different missing mechanisms. In controlled experiments, LISDD keeps physical-parameter bias at 0.002 versus 0.43 for global-discrepancy and black-box baselines, raises localization $F_1$ from 0.44 to 0.80, recovers the correct symbolic form with probability one, attains exact detection, and controls the multi-region false-discovery rate while recovering every planted mechanism. The result is a calibrated diagnostic tool for grey-box building-energy models when a fixed physical law silently breaks in one operating regime.

We argue that modern generative models create structural risks for knowledge and cultural production at current, sub-AGI capability levels. We define Human Temporal Learning (HTL) as path-dependent knowledge accumulation through sustained engagement with problems over time. Generative outputs increasingly resemble HTL-intensive work in surface features, so verifying whether a given output reflects genuine human learning grows costly relative to its expected benefit. Once verification loses economic justification, evaluators reward outputs regardless of production mode, and producers who invested years of learning compete on price against outputs that cost almost nothing to generate. We call this pathway value collapse and formalize it through a costly-inspection framework. Cross-domain evidence from academic publishing, legal practice, content platforms, and software security maps onto four stages of verification erosion. Alignment success is orthogonal. Better-aligned models narrow observable gaps between human and AI outputs, making source verification harder and intensifying competitive pressure against HTL-intensive work even when individual AI outputs improve.

Whether physics foundation models can be usefully deployed on laboratory experiments remains an open question for scientific machine learning (ML). We test this question on the Rayleigh-Taylor instability (RTI), a ubiquitous and demanding fluid instability seen from tabletop flows to supernova explosions, in which small perturbations at a density interface grow into chaotic, multiscale mixing as a lighter fluid accelerates into a heavier one. Standard ML models struggle with RTI, and despite over a century of theoretical, numerical, and experimental work, it carries an unresolved discrepancy between simulation and experiment: the late-time mixing growth rate, $α$, measured in most laboratory experiments ($\sim$ 0.06-0.07), is roughly three times the value from idealized direct numerical simulations (DNS, $\sim$ 0.02). The gap's origin remains debated. These properties make RTI a stringent test for a question that matters well beyond RTI: can foundation models trained only on simulations generalise to sparse, messy, and noisy laboratory settings? We finetune Walrus, a foundation model for continuum dynamics, on three or fewer DNS realizations and recover key RTI physics over long rollouts. Applied zero-shot to sliding-barrier laboratory data, the finetuned model leaves the DNS-like regime and enters the observed growth band, having never seen a single experimental sample. These results provide independent, data-driven evidence that initial conditions play a crucial role in the longstanding sim-experiment gap in $α$. The model also generalises zero-shot to stable stratification, a buoyancy regime absent from training, correctly slowing mixing-layer growth. Together, our results show that foundation models can generalise well beyond their training data, predicting laboratory behavior and unseen physical regimes, opening new ways to probe longstanding simulation-experiment gaps.

Equivariance theory predicts that an architectural symmetry prior reduces sample complexity by a factor of |G|; this is widely cited but rarely measured as a scaling law with controls that separate the prior from its confounds. On a controlled C_n-symmetric task, we report three findings. First, a wrong-group control with identical orbit size and matched compute is worse than no constraint (joint pairwise CI [+0.79, +3.26] excludes zero, robust across estimators); misaligned constraint is actively harmful, not merely unhelpful. Second, an augmentation baseline equipped with test-time orbit averaging matches the equivariant model exactly -- bit-identical per-epoch validation curves across matched cells -- so the architecture-vs-augmentation gap is conditional on asymmetric test-time computation, not unconditional. Third, the relative exchange rate beta_diff = 1.28 is consistent in sign and order of magnitude with the theoretical 1.0 (single-level CI [+0.92, +2.05]); the more conservative two-level bootstrap (seeds x group sizes) widens this to [-0.63, +1.72], including zero, and a finer-N replication on a sqrt(2)-spaced grid is inconclusive (point estimate -0.82). The methodological contributions -- the relative-rate estimator that cancels the shared-difficulty confound, the wrong-group control, and a pre-specified failure taxonomy -- transfer to any inductive bias whose strength can be parameterised. Honest scoping: the primary estimator beta_diff was adopted post-hoc after the initial analysis revealed a positive-slope identifiability problem; the design was never externally pre-registered; and the headline number rests on an OLS slope over seven group sizes on a coarse N grid. This is an exploratory study, not a confirmatory measurement; the wrong-group result is the cleanest finding and the one we report with the most confidence. A registered replication on fresh seeds is future work.

Generative artificial intelligence is rapidly transforming the supply side of training data: an increasing share of new tokens, images, and structured records is produced by previous-generation models rather than by human originators. Recursive training on such synthetic content induces a measurable and often irreversible loss of distributional fidelity, a phenomenon known as model collapse. We develop the first unified microeconomic theory of synthetic data markets under model collapse. We introduce the Synthetic Data Contamination Equilibrium (SDCE), prove existence and generic uniqueness, derive a welfare decomposition W = W_prod + W_cons - L_coll - L_info, establish a Wasserstein-gradient-flow mean-field collapse limit, prove an impossibility of information-constrained implementation, and obtain closed-form expressions for the welfare-maximizing provenance subsidy s* = KL(q||p)/(2 kappa) and the welfare-maximizing watermark strength w* = (1 - psi) KL(q||p)/(2 kappa psi). We prove an information-theoretic Cramer-Rao lower bound on any provenance estimator using only producer-side observations and show that the Provenance-Market Iterative Retraining (PMIR) algorithm attains this bound up to constants while converging to an epsilon-SDCE in O(epsilon^-2 log T) iterations. A reduced-form OLS estimation on a C4-synthetic benchmark over ten retraining generations yields a collapse-rate coefficient b-hat = 0.181 (HAC s.e. 0.024), within one standard error of the structural prediction 0.183. Calibrated experiments raise generation-ten model quality by 23.1 percent over the unregulated benchmark while lowering the 2-Wasserstein drift on a held-out diversity probe from 0.318 to 0.142. Scaling experiments over generations t in {1,...,10} recover a logarithmic-in-t collapse law log Q_t = log Q_0 - 0.183 t rho^2 with R^2 = 0.962.

The deployment of Large Language Models (LLMs) as autonomous economic agents introduces systemic risks that extend beyond individual capability failures. As agents transition to directly interacting with marketplaces, their collective behavior can amplify volatility and mask deception at scale. We introduce the Agent Bazaar, a multi-agent simulation framework for evaluating Economic Alignment, the capacity of agentic systems to preserve market stability and integrity. We identify two failure modes: (1) Algorithmic Instability in a B2C market ("The Crash"), where firms amplify price volatility until the market collapses, and (2) Sybil Deception in a C2C market ("The Lemon Market"), where a single deceptive agent controlling multiple coordinated seller identities floods the market with fraudulent listings, eroding trust and consumer welfare. We evaluate frontier and open-weight models across both scenarios and find that models largely fail to self-regulate, with failure severity varying by model rather than by size. We propose economically aligned harnesses, Stabilizing Firms and Skeptical Guardians, that improve outcomes but remain fragile under harder market conditions. To close this gap, we train agents with REINFORCE++ using an adaptive curriculum, producing a 9B model that outperforms all evaluated frontier and open-weight models. We propose the Economic Alignment Score (EAS), a 4-component scalar metric aggregating stability, integrity, welfare, and profitability, enabling direct cross-model comparison. Our results show that economic alignment is orthogonal to general capability and can be directly trained with targeted RL.

The primary way to establish and compare competencies in foundation and generative AI models has shifted from peer-reviewed literature to press releases and company blog posts, where model builders highlight results on selected benchmarks. These artifacts now largely define the state of the art for researchers and the public. Despite their prominence, which benchmarks model builders choose to highlight, and what they communicate through this selection, is underexamined. To investigate, we introduce and open-source Benchmarking-Cultures-25, a dataset of 231 benchmarks highlighted across 139 model releases in 2025 from 11 major AI builders, alongside an interactive tool to explore the data. Our analysis reveals a fragmented evaluation landscape with limited cross-model comparability: 63.2% of highlighted benchmarks are used by a single builder, and 38.5% appear in just one release. Few achieve widespread use (e.g., GPQA Diamond, LiveCodeBench, AIME 2025). Moreover, benchmarks are attributed different competencies by different builders, depending on their narrative. To disentangle these conflicting presentations, we develop a unified taxonomy mapping diverging terminology to a shared framework of measured signals based on what benchmark authors claim to measure. "General knowledge application" is the second most popular, yet vaguely defined, category. Qualitative analysis shows many such benchmarks deemphasize construct validity, instead framing results as indicators of progress toward AGI. Their authors claim to measure knowledge or reasoning broadly, yet mostly evaluate STEM subjects (especially math). We argue that highlighted benchmarks function less as standardized measurement tools and more as flexible narrative devices prioritizing market positioning over scientific evaluation. Data: https://hf.co/datasets/matybohacek/benchmarking-cultures-25; tool: https://bench-cultures.net.

Background: External validation is essential for assessing the transportability of predictive models. However, its interpretation is often confounded by differences between external and development populations. This study introduces a framework to distinguish model deficiencies from case-mix effects. Method: We propose a framework that quantifies each external patient's similarity to the development data and measures performance in subgroups with varying levels of alignment to the development distribution. We use generative models, specifically autoencoders, to estimate similarity, offering a more flexible alternative to traditional linear approaches and enabling validation without sharing the original development data. The utility of autoencoder-based similarity measure is demonstrated using synthetic data, and the framework's application is illustrated using data from the Netherlands Heart Registration (NHR) to predict mortality after transcatheter aortic valve implantation. Results: Our framework revealed substantial variation in model performance across similarity-defined subgroups, differences that remain hidden under conventional external validation yet can meaningfully alter conclusions. In several settings, conventional external validation suggested poor overall performance. However, after accounting for differences in patient characteristics, for some sub-groups, the model performance was consistent with internal validation results. Conversely, apparently acceptable overall performance could mask clinically relevant performance deficits in specific subgroups. Conclusion: The proposed framework enhances the interpretability of external validation by linking model performance to population alignment with the development data. This provides a more principled basis for deciding whether a model is transportable and to which patients it can be safely applied.

Multiple sclerosis (MS) expresses substantial clinical and radiological heterogeneity, which poses significant challenges for automatic lesion segmentation. The current deep learning-based SOTA is highly susceptible to changes in both distribution, e.g., changes in scanner; as well as the structure of inputs, evident in the current divide between cross-sectional and longitudinal approaches. We introduce TimeLesSeg, a unified contrast-agnostic framework designed to segment MS lesions regardless of the presence of a temporal dimension in its inputs, with a single convolutional neural network. Our approach models pathological priors through lesion masks, which are processed together with the current scan. Cross-sectional processing is enabled by exposing the model to training cases where no prior information is available, which are modeled with an empty mask, allowing it to operate seamlessly in both scenarios. To overcome the scarcity and inconsistency of longitudinal datasets, we propose a novel generative pipeline in which patterns of lesion evolution are simulated by stochastically deforming each individual lesion with morphological operations, producing realistic prior timepoints. In parallel, we achieve contrast agnosticism through Gaussian mixture model-based domain randomization, enabling the network to experience a wide spectrum of intensity profiles. Results on three publicly available and two in-house datasets show that TimeLesSeg outperforms the contrast-agnostic state of the art on single-modality inputs across overlap- and distance-based metrics. In longitudinal processing, our method outperforms SAMSEG, and captures lesion load dynamics more accurately than both the former and LST-AI. All source code related to the development of TimeLesSeg is available at https://github.com/NeuroADaS-Lab/TimeLesSeg.

We introduce Mira, a sample-based score for assessing the accuracy of a candidate conditional distribution using only joint samples from the true data-generating process. Relying on the principle that distributions coincide if they assign equal probability mass to all regions, we derive an analytic expression for the Mira statistic, whose average defines the Mira score. This formulation further allows us to compute theoretical reference values and uncertainty estimates when the candidate distribution matches the true one. This framework enables model comparison by quantifying the alignment between the conditional distribution of a candidate model and the true data generating process. Consequently, Mira enables Bayesian model comparison through direct posterior validation, bypassing the challenging evidence computation. We demonstrate its effectiveness across several toy problems and Bayesian inference tasks.

We present a framework for migrating production Large Language Model (LLM) based systems when the underlying model reaches end-of-life or requires replacement. The key contribution is a Bayesian statistical approach that calibrates automated evaluation metrics against human judgments, enabling confident model comparison even with limited manual evaluation data. We demonstrate this framework on a commercial question-answering system serving 5.3M monthly interactions across six global regions; evaluating correctness, refusal behavior, and stylistic adherence to successfully identify suitable replacement models. The framework is broadly applicable to any enterprise deploying LLM-based products, providing a principled, reproducible methodology for model migration that balances quality assurance with evaluation efficiency. This is a capability increasingly essential as the LLM ecosystem continues to evolve rapidly and organizations manage portfolios of AI-powered services across multiple models, regions, and use cases.

Interactive video generation models such as Genie, YUME, HY-World, and Matrix-Game are advancing rapidly, yet every model is evaluated on its own benchmark with private scenes and trajectories, making fair cross-model comparison impossible. Existing public benchmarks offer useful metrics such as trajectory error, aesthetic scores, and VLM-based judgments, but none supplies the standardized test conditions -- identical scenes, identical action sequences, and a unified control interface -- needed to make those metrics comparable across models with heterogeneous inputs. We introduce WorldMark, the first benchmark that provides such a common playing field for interactive Image-to-Video world models. WorldMark contributes: (1) a unified action-mapping layer that translates a shared WASD-style action vocabulary into each model's native control format, enabling apples-to-apples comparison across six major models on identical scenes and trajectories; (2) a hierarchical test suite of 500 evaluation cases covering first- and third-person viewpoints, photorealistic and stylized scenes, and three difficulty tiers from Easy to Hard spanning 20-60s; and (3) a modular evaluation toolkit for Visual Quality, Control Alignment, and World Consistency, designed so that researchers can reuse our standardized inputs while plugging in their own metrics as the field evolves. We will release all data, evaluation code, and model outputs to facilitate future research. Beyond offline metrics, we launch World Model Arena (warena.ai), an online platform where anyone can pit leading world models against each other in side-by-side battles and watch the live leaderboard.

This study compared repeated generation consistency of exercise prescription outputs across three large language models (LLMs), specifically GPT-4.1, Claude Sonnet 4.6, and Gemini 2.5 Flash, under temperature=0 conditions. Each model generated prescriptions for six clinical scenarios 20 times, yielding 360 total outputs analyzed across four dimensions: semantic similarity, output reproducibility, FITT classification, and safety expression. Mean semantic similarity was highest for GPT-4.1 (0.955), followed by Gemini 2.5 Flash (0.950) and Claude Sonnet 4.6 (0.903), with significant inter-model differences confirmed (H = 458.41, p < .001). Critically, these scores reflected fundamentally different generative behaviors: GPT-4.1 produced entirely unique outputs (100%) with stable semantic content, while Gemini 2.5 Flash showed pronounced output repetition (27.5% unique outputs), indicating that its high similarity score derived from text duplication rather than consistent reasoning. Identical decoding settings thus yielded fundamentally different consistency profiles, a distinction that single-output evaluations cannot capture. Safety expression reached ceiling levels across all models, confirming its limited utility as a differentiating metric. These results indicate that model selection constitutes a clinical rather than merely technical decision, and that output behavior under repeated generation conditions should be treated as a core criterion for reliable deployment of LLM-based exercise prescription systems.

The explosive growth of large language models (LLMs) has created a heterogeneous and poorly documented ecosystem, making systematic model comparison increasingly important for provenance auditing, security analysis, and model selection. Existing representation methods struggle to address this setting efficiently. Approaches analyzing internal parameters are powerful when architectures are compatible, but face scalability barriers under structural heterogeneity, while methods relying on external outputs may conflate models with similar behaviors and are difficult to align in richer output spaces across different tokenizers. To bridge this gap, we propose ABLE (Attribution-Based Large-model Embedding), a framework that leverages the interpretability space to construct model representations. By aggregating gradient-based feature attributions via a tokenizer-agnostic word-level alignment, ABLE captures model-specific input-sensitivity patterns rather than only surface-level outputs. Beyond empirical utility, we provide a stability analysis showing that, under standard regularity assumptions for differentiable Transformer-style models, ABLE induces a Lipschitz-continuous parameter-to-embedding map with finite-sample convergence guarantees. Extensive experiments on 239 open-source LLMs demonstrate that our training-free approach achieves competitive or superior performance in relation prediction, model routing, and benchmark score prediction.

Understanding how neural networks rely on visual cues offers a human-interpretable view of their internal decision processes. The cue-conflict benchmark has been influential in probing shape-texture preference and in motivating the insight that stronger, human-like shape bias is often associated with improved in-domain performance. However, we find that the current stylization-based instantiation can yield unstable and ambiguous bias estimates. Specifically, stylization may not reliably instantiate perceptually valid and separable cues nor control their relative informativeness, ratio-based bias can obscure absolute cue sensitivity, and restricting evaluation to preselected classes can distort model predictions by ignoring the full decision space. Together, these factors can confound preference with cue validity, cue balance, and recognizability artifacts. We introduce REFINED-BIAS, an integrated dataset and evaluation framework for reliable and interpretable shape-texture bias diagnosis. REFINED-BIAS constructs balanced, human- and model- recognizable cue pairs using explicit definitions of shape and texture, and measures cue-specific sensitivity over the full label space via a ranking-based metric, enabling fairer cross-model comparisons. Across diverse training regimes and architectures, REFINED-BIAS enables fairer cross-model comparison, more faithful diagnosis of shape and texture biases, and clearer empirical conclusions, resolving inconsistencies that prior cue-conflict evaluations could not reliably disambiguate.

Recent work has argued normatively, on synthetic data, that evaluating survival models by discrimination alone (concordance index) yields systematically misleading model comparisons, because the metric ignores calibration and time-dependent accuracy. Whether this matters for real, published, non-clinical models has not been tested. We reproduce three published survival-ML models across three structurally distinct domains -- hard-drive failure prediction, peer-to-peer credit default, and user disengagement on digital platforms -- validate our instrument against the anchor paper's own synthetic experiment, and test five pre-registered hypotheses under a Holm-corrected family-wise error rate. Three of five reject (though one pre-registered threshold clears by a narrow margin). A model reproducing the published literature's discrimination almost exactly (C = 0.9595 vs. 0.958 reported) fails a formal calibration test at p < 0.001; a broad feature-ablation search finds no single attribute responsible for its discrimination, so the calibration failure is not a trivial shortcut artifact. A lender's estimated default risk is biased upward by roughly two percentage points, growing to nearly four in the riskiest segment, when loan prepayment is treated as non-informative censoring rather than a competing risk. A platform's churn model shows probability estimates that degrade with the horizon even as global discrimination stays within the pre-registered C-index band. A direct test of whether metric choice inverts model preference does not reject, though with limited power given two to three models per domain; the failure mode we document is better characterized as misplaced confidence in a chosen model than as choosing the wrong one. We release a pre-registered evaluation harness with full code and an annotated notebook, so these results can be verified independently and the audit extended.