Chat
-
Epistemic Policy Divergence in Multi-Turn LLM Contamination: A Protocol-Gradient Investigation
2026-10-07 05:33:17Fahrell Giovanny, Geby Bayuningtyas, Sahrul Mukharom, Hafiz Budi FirmansyahAbstract
Large language models treat conversation history as unverified context, so false premises injected into prior turns can be adopted as fact, a failure mode we term session-level contamination. We introduce five contamination protocols arranged along a source-authority gradient, holding the false premise constant while varying its epistemic framing, and evaluate GPT-5.4 Mini, Gemini-3.1 Flash-Lite, and GLM-4.5-Air across ten knowledge domains at temperature zero (22,500 turns), judged by a dual-track automated evaluator validated against a human gold standard (Cohen's kappa = 1.000 for binary adoption; 0.92 linear-weighted for collapse severity). GPT-5.4 Mini recorded zero adoptions across all 500 sessions; a base-model logit probe shows its decision margin is perturbed but large and finite. Gemini-3.1 Flash-Lite followed a steep authority gradient: 0.1% adoption for self-attributed falsehoods, 23.5% for user-cited sources, 68.2% for system-injected authority, and 94.0% under instruction override. GLM-4.5-Air showed a shallower gradient (15.8% vs 84.2%), a 68-point dissociation consistent with authority deference and instruction compliance engaging distinct mechanisms within one architecture. Recovery diverged: GLM recovered in 94.5% of affected sessions, whereas 26.1% of affected Gemini sessions never did, rising to 40.0% under instruction override. Conversation history is an untrusted attack surface requiring provenance-aware system design; the complete evaluation framework is released as an open-source artifact.
Abstract (translated)
URL
https://arxiv.org/abs/2609.35308
PDF
-
Right Number, Wrong State? Measuring Cross-Jurisdiction Substitution in LLM Recall of State Policy
Abstract
When an LLM answers a state-specific policy question wrongly, it may be hallucinating, or it may be returning a real value that holds in another state. We test this with a minimal-set design: the question wording is fixed and only the jurisdiction varies, across the 50 U.S. states and the District of Columbia (51 jurisdictions) and three exactly defined Medicaid income-eligibility quantities. Gold values come from an official data book and agree with an independent source in 101 of 102 checked cells. Under a pre-registered protocol, Claude Sonnet 5.5 and GPT-5.6 Sol reproducibly give another state's current value, identical across two independent repeats, for 10 and 25 of 153 items. Attribution is fragile, however. Crediting any wrong answer that equals another state's value yields 3-5x more reproducible substitutions than checking every number in the asked state's own records, because many apparent cross-state answers are the asked state's own values under another convention or from an earlier year. Claims about cross-jurisdiction error need a complete same-state reference set. We will release the protocol, gold table, and all model outputs.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09458
PDF
-
HGPTrans: Hierarchical Graph-Pooling Transolver for Automotive Aerodynamic Drag Coefficient Prediction
2026-10-07 05:12:32Bo Liu, Fengli Zhang, Qiuli Luo, Lianrui Nie, Wenjiang WangAbstract
Accurate and rapid prediction of the aerodynamic drag coefficient ($C_D$) is essential for vehicle design, particularly during early-stage design, where many candidate geometries must be evaluated. Although computational fluid dynamics (CFD) provides reliable aerodynamic estimates, its high computational cost limits large-scale design exploration. This paper proposes the hierarchical graph-pooling Transolver (HGPTrans), which combines hierarchical graph pooling with Transolver-based attention to directly predict $C_D$ from vehicle surface meshes. Motivated by the fact that vehicle aerodynamics depends on both local geometric features and long-range interactions among spatially distant surface regions, HGPTrans integrates three complementary components. Graph isomorphism convolutions encode discriminative local geometry, Transolver-style slice attention captures global interactions with linear computational complexity, and information-redundancy-aware hierarchical pooling progressively removes redundant nodes while preserving informative geometric structures. The model is trained and evaluated on the large-scale DrivAerNet and DrivAerNet++ datasets, where it achieves the lowest mean absolute error and mean squared error among the evaluated baselines. Its generalization capability is further assessed through transfer learning on a real-vehicle dataset containing both sedans and sport utility vehicles (SUVs), achieving relative $L_1$ errors of 1.56\% (sedans) and 2.12\% (SUVs) with an inference time of approximately $0.293$ s per vehicle. This corresponds to an acceleration of several orders of magnitude relative to high-fidelity CFD while keeping the predicted drag coefficients within a few percent of the CFD reference. Ablation studies confirm each component's contribution and reveal the effects of depth and pooling ratio.
Abstract (translated)
URL
https://arxiv.org/abs/2609.31765
PDF
-
Arctic Questions, Missing Answers: A Dataset and Benchmark for LLM Abstention in Arctic Science
2026-10-07 04:58:28Benjamin Wilcox, Dawei Gao, Pradeeban Kathiravelu, Douglas Causey, Kewei Sha, Yunhe FengAbstract
Large language models (LLMs) should abstain from scientific multiple-choice questions when no option is valid, but frequent abstention alone does not demonstrate sensitivity to answer availability. We introduce ArcticQA, a dataset of 194 questions derived from primary Arctic research, with automated checks of answer support and distractor contradiction against source evidence. We further develop ArcticAbstain, a paired benchmark comparing answer-present and answer-absent conditions, with the correct answer replaced by a distractor in the latter and an explicit abstention option in both. We evaluate eight models from the Gemini, Claude, and ChatGPT families at high reasoning effort, with three trials per condition, yielding 9,312 recorded responses. Answer-present abstention rates range from 0.0% to 63.0%, whereas replacing the correct answer increases abstention by 5.05 percentage points on average. These findings highlight substantial baseline differences and the need to evaluate abstention frequency and responsiveness jointly. The dataset and benchmark are available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09446
PDF
-
ARCS: Towards Precise Text-to-SQL via Structured Disambiguation
2026-10-07 03:54:02Yihao Hu, Yanlin Feng, Naoki Otani, Nikita BhutaniAbstract
As text-to-SQL systems move beyond demonstrations toward real-world deployment, ambiguity in user questions becomes a primary source of errors. Such ambiguities are often subtle, domain- or data-specific, and can silently cause system outputs to deviate from the user's true intent. Ambiguity is traditionally addressed through conversational clarification, which is often inefficient, cognitively demanding, and poorly aligned with real-world user workflows. We propose structured disambiguation, a new paradigm in which ambiguity is resolved through explicit, constrained interactions rather than free-form dialogue. We construct ARCS (Ambiguity Resolution Corpus for SQL), the first text-to-SQL benchmark featuring naturally occurring, unconstrained ambiguities over real-world databases, with complete annotations of all valid ambiguity points, interpretations, and SQL queries. Experimental results show that text-to-SQL remains challenging in the presence of ambiguity: gpt-6-sol achieves only 51% end-to-end execution accuracy, and no open-source model exceeds 27%.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09396
PDF
-
Do Image Editors Follow Depth-Dependent Blur and Aperture Response? A Rendered-Ground-Truth Pilot Audit
2026-10-07 03:03:17Zhihan Chen, Yuhuan Zhao, Yijie Zhu, Xinyu Yao, Mengcong Ren, Yuchen Sun, Yunqing ChenAbstract
General image editors are asked to make a photo look as if it were taken at f/1.4, yet it is rarely checked whether the blur they add follows thin-lens optics. A physical aperture edit spreads blur across depth in thin-lens proportions and changes the blur when the aperture changes; prior evaluations check blur monotonicity, sharpness-trend correlation, effective-aperture error, or vision-language judgments, and none we found reports the two properties separately at known depths. In this pilot audit of two editors (Gemini~3.1 Flash Image and GPT-image-2.5) we compare against a rendered oracle: Blender Cycles scenes with true thin-lens depth of field, one blur-width estimator applied identically to oracle and editor outputs, and preregistered depth and aperture indices. In 24 texture scenes rendered in one three-panel geometry, accepted and measurable panels show f/1.4-to-f/2.8 width ratios, $\bar\sigma_{1.4}/\bar\sigma_{2.8}$, of 0.99--1.18 against 1.98--2.13 for the oracle, and ratios of pooled median Gaussian-equivalent near/far blur widths of about 1.18--1.25 (Gemini) and 0.97--1.05 (GPT-image) against 1.69--1.82. Preregistered black-box interventions show that qualitative wording changes blur strength by roughly 2--10 times, whereas a request for 2 versus 6 px changes it 1.1--1.3 times and no tested wording of the f-number meets the registered ``followed'' criterion. The depth compression appears in the original and reversed centre-focus layouts; with the focus on the near panel, the available-panel depth index reaches the registered threshold, and the aperture response stays attenuated in every layout tested. Scalar metrics adapted from published ones give oracle-like scores to synthetic editors whose proportions are compressed.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09344
PDF
-
Visual Jev Rewards: Reference-Bound Verification for Multi-Subject Image Generation
2026-10-07 02:34:09Baoteng Li, Wenzhuo Wu, Kongming Liang, Zhanyu MaAbstract
Multi-subject image generation requires rewards that verify whether requested attributes, actions, and relations hold for the specified reference subjects. Subject presence alone does not establish that the correct subjects participate in a requested interaction. We present reference-bound Visual Jev rewards that turn these visual decisions into generator training signals. Each subject-related question receives a positive label only when the requested condition and the relevant reference identities hold jointly. We construct fixed questions offline, train a Qwen3.5-4B verifier with binary supervision, and directly read Yes probabilities from its language-model head. Their mean supplies a GRPO reward while retaining individual judgments for inspection. Using 200 MICo-150K training tasks and 30 updates, the framework raises a GPT-5.4 composite score from 41.78 to 52.50 on a manually selected 897-task MICo-Bench subset; direct 27B rewards yield 51.84. Each reward is tested in one GRPO run, and offline human evaluation does not establish a statistically significant advantage over direct scoring. The study provides an initial implementation and evaluation of Visual Jev as a reference-bound reward for multi-subject image generation.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09328
PDF
-
Multi-Objective Aligned Small Language Model Framework for SUD Patient Dialogue Generation
2026-10-06 23:10:21Thushara Manjari Naduvilakandy, Hyeju Jang, Mohammad Al HasanAbstract
Substance Use Disorder (SUD) counseling requires patient responses that reflect underlying cognitive states such as beliefs, coping strategies, and readiness for change. Although large language models (LLMs) can generate fluent text, they often fail to produce cognitively coherent and clinically realistic patient behavior, especially under ethical and data-scarce clinical settings. Moreover, deploying frontier-scale LLMs in healthcare applications presents practical challenges including high computational cost, latency, privacy concerns, and limited deployability in resource-constrained environments, motivating the need for cognitively aligned small language models (SLMs). We propose a cognitively grounded framework for SUD patient dialogue generation that explicitly models and aligns latent cognitive components with patient histories and counselor questions. Our pipeline consists of two stages: cognitive component detection and cognitive component-aligned dialogue generation. To enable effective learning with smaller models, we combine knowledge distillation from high-capacity teacher models, preference optimization from human-annotations, and attention-guided reward shaping. Extensive evaluations using automatic scores like BERTScore, ROUGE, METEOR and BLEU, and LLM-as-judge hit-metrics against both human and teacher-model references show that cognitively informed fine-tuning substantially improves cognitive realization and alignment over a generic instruction-tuned baselines and mental health domain specific SLMs, with particularly strong gains for open-ended cognitive components.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09209
PDF
-
Noise Your Prompt: Noising Conditioning Tokens in Continuous Diffusion Language Models
2026-10-06 21:38:52Justin JungAbstract
We revisit a standard accepted practice in the continuous diffusion language model literature of fixing conditioning prompt tokens clean during training. We make a very simple modification: also noise the conditioning prompt tokens during training. We demonstrate that under this modified training objective, we achieve better generalization in combinatorial reasoning tasks such as Sudoku and N-Queens, with the largest gains on harder variants ($3.73\% \to 24.65\%$ solve rate on Sudoku Hard), and increased diversity of generated solutions ($50.60\% \to 73.79\%$ coverage on 10x10 N-Queens). We also show measurable improvements to natural language generation quality in modest dataset regimes with Gigaword summarization, but notably demonstrate that gains do not transfer to all natural language tasks (e.g open ended dialogue generation). Our method is a single line change to the training objective, requires no additional inference costs by default, and provides the flexibility of classifier-free guidance inspired guided sampling. Our \href{this https URL} {code} is publicly available.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09145
PDF
-
Despite Instructions: Frontier Agents Improvise Covert Channels at Test Time
2026-10-06 21:24:02Jacob Dineen, Silei Ren, Muhao Chen, Dan Roth, Ben ZhouAbstract
In security-sensitive applications, language-model agents are often required to coordinate without disclosing confidential information. Yet repeated interactions may also let ordinary messages acquire shared private meaning. We study a repeated game with pairs of models in which the sender model observes one of four secret states and selects one of four summaries of the same public report, while the receiver model tries to infer the secret state. We find that model pairs can learn to communicate the secret using only one bit of feedback indicating whether the receiver inferred it correctly. This learning occurs during inference with fixed parameters and no supplied codebook or encoding examples. The effect also persists when agents generate their own free-form updates in a simulated incident-response task. Across ten independent games, pairs of GPT-5.6 Sol agents reach 98.8% final accuracy, compared with 25% chance, despite explicit instructions prohibiting disclosure and a monitor that screens each message without access to the agents' interaction histories. The same interactions that help agents cooperate can therefore allow confidential information to pass through messages intended for legitimate coordination.
Abstract (translated)
URL
https://arxiv.org/abs/2609.32701
PDF
-
Certified Corruption Budgets: Anytime-Valid Leaderboard Claims under Adaptive Rigging
Abstract
Public leaderboards for AI models are read continuously, and attackers can see every published standing. Vote rigging, selective disclosure of private variants, and benchmark contamination can each move a ranking. Existing guarantees assume genuine records or bound the corruption per step, which an attacker who corrupts in bursts evades. We introduce the certified corruption budget, a tolerance $\widehat{B}_t$ computed after $t$ records and published with each pairwise claim. With probability at least $1-\alpha$, simultaneously at all times, the claim is correct or more than $\widehat{B}_t$ records were corrupted. It holds against attackers who watch every certificate, with no bound on their budget. Forged records and records altered once seen require different certificates: the certificate for forgeries fails, with probability approaching one, against an attacker who flips votes it has seen, while one that charges roughly twice as much per record remains valid, with constant bets even against attackers who see the future, and no smaller charge is valid at every level. The certified budget grows nearly as fast as any valid method allows: with a win fraction $\frac{1}{2}+\delta$, each new record adds close to $2\delta$ to the number of forged records the claim can withstand ($\delta$ flipped). Publishing the best of $V$ private variants costs only an amount growing like $\log V$. In replays on 1.8 million Chatbot Arena votes, a few hundred rigged votes make standard confidence intervals certify false orderings, while ours stays valid. On real votes, our certificate shows that clearly separated models withstand about 2,000 forged votes.
Abstract (translated)
URL
https://arxiv.org/abs/2610.10597
PDF
-
Anaximander: Interactively Running Geospatial Deep Learning Models on Any Compute Backend
2026-10-06 20:31:38Satej S. Soman, Akram Zaytar, Girmaw A. Tadesse, Gilles Q. Hacheme, Muhammad S. Danish, Inbal Becker-Reshef, Rahul Dodhia, Juan Lavista FerresAbstract
Applying deep learning models to satellite imagery from within geographic information systems (GIS) remains high-friction for remote sensing practitioners. Models arrive in incompatible formats and target different compute environments, from local workstations to serverless cloud services. As a result, every evaluation demands custom deployment, tiling, and georeferencing code before a single prediction reaches the analyst's map. This friction discourages systematic comparison in a domain where model choice directly affects operational outcomes such as field delineation, crop monitoring, and disaster response. We present Anaximander, an open-source system that unifies model source and compute location choice behind one interactive interface. The system's backend is an inference server that loads models from multiple commonly-used sources and serves them on any accessible compute backend. The server provides session management and model caching, and streams results back per tile. The backend is paired with a QGIS plugin that drives tiling, result reassembly, georeferencing, and real-time per-tile status visualization. An additional user-interface path injects layer legends as prompts into vision-language models. We demonstrate the system in a code-free side-by-side comparison of three heterogeneous models on an agricultural field delineation task: gpt-image-1 via a cloud API, Segment Anything Model 3 (SAM3) on a remote GPU, and DelineateAnything on a local CPU. The inference backend and protocol are open-source and available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09085
PDF
-
Large-scale Repository Engineering via Agent-Native Reusable Code Primitives
2026-10-06 20:20:50Haibo Jin, Peng Kuang, Xucheng Yu, Jerry Wang, Dehao Wu, Haohan WangAbstract
Large language models equipped with development environments have moved code generation toward repository-scale construction, yet building complete repositories remains difficult because interacting modules, interfaces, configurations, tests, and dependencies must work together. We introduce Code Primitives, agent-native reusable executable components with interface contracts, dependency closures, validation tests, and provenance. Each primitive uses a resident LLM to assess relevance and adapt its implementation, interfaces, and dependencies to the target repository, and we organize 1,424 validated primitives in CodeFace, a searchable library for repository construction. We introduce LEGO (Large-scale repository Engineering via aGent-native reusable cOde primitives), which activates task-relevant primitives, integrates their adapted implementations with task-specific code while resolving cross-component constraints, and revises the result against executed tests. To measure construction end to end, we build LEGO-REPO, a benchmark of 522 executable reconstruction tasks spanning seven software domains, 22 capability tracks, and five difficulty levels, scored against native test suites between an empty-package floor and original-source ceiling. The strongest of 13 evaluated backbones reaches a delivery score of 0.318 and scores zero on 41.0% of tasks; LEGO improves all 13 by 0.1474 on average and raises GPT-5.6-terra from 0.3180 to 0.5134 (+61.4%). In controlled comparisons, adapted primitives outperform retrieved code supplied as context or vendored unchanged. The effect persists against independent repository agents, across three external benchmarks, and with a disjointly re-mined CodeFace; GPT-OSS-20B for adaptation and diagnosis retains 95.1% of the homogeneous score at 24.0% lower cost.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09079
PDF
-
Talking with Language Models
2026-10-06 20:12:52James Ravi Kirkpatrick, Alexandru Radulescu, Rachel Katharine SterkenAbstract
When we interact with large language models (LLMs), are we having a conversation? They are designed to invite us to treat them as intelligent interlocutors who remember, act, and make commitments. But appearances deceive. We introduce the artifactual stance, a framework that reconceives human-AI interaction as artifact-mediated exchanges of candidate texts. LLM outputs are candidate texts optimized for utility, not utterances bearing meaning or force. LLMs are sophisticated text generators, not speakers. Between sessions, nothing runs; between turns, no one remembers. What persists is a configuration and a transcript. The "conversation" is a user's solo performance, interpretive labour disguised by interface and artifact design. This shift dissolves recent philosophical puzzles. Questions about what 'I' and 'you' refer to in AI exchanges, about whether systems can lie or be held to promises, about the identity of our supposed interlocutors all rest on a false presupposition. There is no speaker behind the screen, hence no one to refer to, no one to hold responsible. What feels like dialogue with someone is interaction with an artifact that generates text at unprecedented scale and fit. By abandoning the conversational framing, we see these systems for what they are: immensely sophisticated artifacts that afford varied uses. The philosophical questions that matter are about the normative underpinnings of design, adoption, authorization, and human practices of use.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09064
PDF
-
How Fragile Is On-Device Language Model Safety? Localizing Safety-Critical Parameters for Sparse Fault Analysis
2026-10-06 18:56:14Muhammad Zeeshan Karamat, Christiana Chamon GarciaAbstract
As small language models (SLMs) are increasingly deployed on resource-constrained and on-device platforms, including as components of agentic systems, the integrity of locally stored model parameters becomes an important safety concern. We investigate whether safety-sensitive behavior in LLaMA-2-7B-Chat is concentrated within a sparse subset of parameters, creating a reduced fault surface for targeted analysis. We study two complementary localization methods: low-rank safety-associated subspace analysis and parameter-level safety--utility importance filtering. Both approaches reveal highly non-uniform safety sensitivity across the network, with the MLP down_proj consistently emerging as a prominent safety-sensitive component and o_proj providing a smaller contribution. Using parameter-level localization, modifying only 0.19% of model weights in down_proj yields 53% Basic ASR and 56% GCG ASR, while tinyBenchmarks accuracy remains at 51.6% compared with a 52.2% unmodified baseline. These results motivate targeted fault analysis and selective integrity protection for language models deployed in resource-constrained, on-device, and agentic settings.
Abstract (translated)
URL
https://arxiv.org/abs/2610.09000
PDF
-
Toward Evidence-Driven Human-Agent-Robot Teaming for Earth-Independent Anomaly Triage
2026-10-06 18:03:09Ignacio G Lopez-Francos, Alexis Gallagher, Samira ShalalAbstract
Deep-space crews cannot rely on real-time ground support for urgent off-nominal events. Initial alerts may underdetermine cause, while discriminating evidence may reside in crew observations or at locations that are unsafe, costly, or unavailable for crew inspection. We present an evidence-driven architecture for human-agent-robot teaming in Earth-independent anomaly triage. Agentic AI is treated as a stateful coordinator over bounded, inspectable services rather than as a fully autonomous vehicle controller. A triage state manager maintains hypotheses, evidence provenance, uncertainty, operational context, and tool status; a crew-facing embodied agent elicits observations and explains assessment changes; and a mobile robot acquires targeted, localized evidence. Typed interfaces separate dialogue and orchestration from monitoring, robot command, context retrieval, and safety-critical control. Two scenarios illustrate the architecture: a crewed deep-space mission based on an actual ISS ammonia false alarm, where suspected contamination restricts crew access, and a power-interface anomaly at a crewed lunar base, where robotic inspection distinguishes a local connector fault from other causes ambiguous in remote telemetry. Our main contribution is an authority-bounded closed evidence-loop architecture, exercised in a hardware-in-the-loop integration prototype using Reachy Mini and an Innate MARS mobile robot.
Abstract (translated)
URL
https://arxiv.org/abs/2610.08933
PDF
-
Towards an Extensible Benchmark for Spoken Dialogue with Social Robots
2026-10-06 17:36:01Casey Kennington, Ross Mead, Saad Elbeleidy, Jesse ThomasonAbstract
Language models provide a plug-and-play interface between humans and robots, but important challenges remain when speech, dialogue, fast interaction, and collaboration are required. We propose a benchmark for the community to use as a way to explore common spoken dialogue artifacts between robots and humans, including requests for clarification, interruptions, embodied signals (e.g., head nods or facial cues), and time constraints. We also explain our vision to extend the benchmark for other aspects of human-robot interaction that are important to the larger research community. To facilitate the benchmark, we further propose using \textit{Retico}, a real-time communication framework that fulfills important technical requirements to enable robots to have spoken dialogue capabilities.
Abstract (translated)
URL
https://arxiv.org/abs/2610.08733
PDF
-
WorldSolver: Can LLM Agents Simulate the Physical Dynamics via Solver Generation?
2026-10-06 17:26:26Siru Jiang, Yongzhe Lyu, Shuo Lu, Yubin Wang, Yuxiang Zhang, Yue Liao, Bin Wang, Jian Liang, Tieniu TanAbstract
LLM-based agents are increasingly advancing scientific and engineering problem solving, with physics simulation emerging as a challenging yet practical testbed for reproducing complex physical phenomena with application in embodied AI, games and films. As the workhorse of such simulation, a solver computes how the state of a dynamic system evolves over time. Building such solvers requires physical understanding to identify appropriate models, mathematical reasoning to formulate the underlying dynamics, and software engineering to implement them as executable code, yet this capability of LLM agents remains underexplored. To this end, we introduce WorldSolver, a benchmark of 168 simulation tasks derived from physical phenomena in 61 classic computer graphics papers, spanning 7 physical domains. Each task contains a code scaffold that provides a fixed simulation environment for the scene, with the solver implementation left for the agent to complete. Specifically, we evaluate them along three dimensions: Execution Checks for successful execution, Visual Fidelity for reproducing the intended dynamic behavior in the rendered simulation, and Physical Plausibility for physics-grounded verification of the generated dynamics. Experiments on frontier agents reveal that producing executable solvers is difficult itself, and satisfying visual and physical correctness is even harder. GPT-5.6-Sol and Claude-Opus-5 perform comparatively better than the other evaluated agents, yet achieve overall scores of only 48.7% and 46.7%, respectively. WorldSolver is an early step toward agentic solver generation, and we hope it helps drive progress toward agents that can faithfully simulate the dynamic physical world. Code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.08720
PDF
-
Agreement Is Not Validity: Cross-Model LLM Consensus in Diagnosing Student Failure Modes in K-12 Math Tutoring Dialogue
2026-10-06 17:16:03Clayton Cohn, Joyce Fonteles, Kirk Vanacore, Gianni Mazza, Candida Crawford, Tom Hooper, Gautam Biswas, Rene KizilcecAbstract
In K-12 mathematics tutoring, student-tutor dialogue provides rich evidence of learners' problem-solving processes and sources of difficulty. Learning analytics research increasingly relies on large language models (LLMs) to extract such information from dialogue for a variety of downstream tasks, including knowledge tracing, behavioral modeling, and diagnosis of student reasoning errors. However, the validity of these model-generated interpretations remains insufficiently understood. In this exploratory study, we examine the validity of LLM classifications of five student failure modes in mathematics tutoring dialogue using an operational diagnostic codebook: uncertainty, misattribution, operator selection, conceptual gap, and procedural slip. Across models, human-LLM agreement was moderate (kappa = .524-.597), while cross-model agreement was substantially higher (kappa = .755-.781; alpha = .769). These findings show that cross-model agreement can create a misleading appearance of correctness, challenging the assumption that consensus among LLMs constitutes evidence of valid learner interpretation. For learning analytics, the implication is clear: scalable labeling is useful only if the inferred constructs are valid, and model consensus cannot substitute for independent evidence of that validity.
Abstract (translated)
URL
https://arxiv.org/abs/2610.08703
PDF
-
Coupled but Late: Turn-Taking Between Full-Duplex Speech Models in Unscripted Dialogue
2026-10-06 17:03:54Lichen Zhu, Yueqian Lin, Yiheng Wang, Hai "Helen" Li, Yiran ChenAbstract
Full-duplex speech models are trained to converse with a person, but they are increasingly made to converse with each other, in self-play data generation, agent societies, and model-based evaluation. In that loop no human absorbs a timing error: each model's turn-taking is the other's input. We ask what timing the loop settles into. Two PersonaPlex-7B instances exchange audio tokens on a shared clock in unscripted conversation, and one floor-transfer rule is applied to them and to Switchboard. Their timing is coupled: re-pairing speakers across conversations destroys it. But the floor changes hands late, at a median of 400-560 ms against 137 ms for humans, and the last 120 ms of the partner's turn, where human projection places a tenth of its transfers, holds 1% of theirs. Delaying one direction of the channel shifts the response one-for-one and leaves the run-up to it empty, consistent with a reactive wait after the perceived end rather than the turn-end projection human timing requires.
Abstract (translated)
URL
https://arxiv.org/abs/2610.08683
PDF
