Publication Radar.

Closing the gap between frontier and smaller language models is usually approached by distillation: fine-tuning the small model (student) on a frontier model's (teacher) outputs, which binds transfer to training compute and leaves the transferred knowledge largely opaque. We propose Artifact Distillation: a teacher solves training instances and externalizes its problem-solving procedure into an artifact bank that a student queries at inference time, with no gradient updates. Across four LiveBench tasks, a bank written by Claude Opus 4.5 lets Claude Haiku 4.5 outperform unaided Claude Sonnet 4.5 on spatial reasoning (0.707 vs. 0.647) at 60% of its cost, and reach Sonnet 4.5's level on table joining at 89% of its cost. The same raw bank raises Qwen3.5-27B's pooled score from 0.730 to 0.800, showing that transfer is not confined to the teacher's model family.

Large reasoning models have recently demonstrated remarkable capabilities in solving complex tasks, where supervised fine-tuning (SFT) on long chain-of-thought data serves as a crucial foundation for eliciting and enhancing their reasoning abilities. Despite rapid progress in both improving and analyzing reasoning-oriented SFT, the field still lacks a systematic survey that consolidates its fast-growing literature. To fill this gap, we present a comprehensive review of recent advancements in reasoning SFT, examining the literature through the dual lenses of methodological design and analytical investigation. First, we review methodological improvements across the SFT pipeline and categorize them into data-centric approaches and algorithm-centric innovations. Second, we reorganize analytical studies along three dimensions: data characteristics, optimization dynamics, and mechanistic insights. Finally, by synthesizing current research focuses and remaining bottlenecks, we outline promising future directions for reasoning SFT. We hope this survey deepens the understanding of reasoning SFT and paves the way for advanced reasoning models.

Listwise preference feedback offers richer supervision for large language model (LLMs) alignment than pairwise comparisons, but a full ranking also reveals more sensitive user preferences. Existing privacy-preserving alignment methods focus mainly on pairwise feedback, while listwise alignment under local differential privacy (LDP) remains unexplored. Extending LDP to this listwise domain raises a critical challenge: the high sensitivity of the full ranking space demands excessive noise, neutralizing the benefits of listwise supervision. To address this challenge, we analyze the interaction between supervision granularity and privacy noise in downstream training, and introduce a rank-aware exponential mechanism that privatizes listwise preference data into a low-sensitivity granularity suitable for downstream alignment. The mechanism leverages ranking information to sample a fixed-size binary partition, concentrating the noise near borderline items instead of perturbing all items uniformly as randomized response (RR) does. Empirical evaluations on downstream LLM alignment tasks show that our mechanism consistently outperforms existing LDP baselines at matched privacy budgets. The code is available at https://github.com/tdsai-lab/RAEM.

Despite their strong performance, large language models remain highly sensitive to prompt formulation. This issue is typically addressed at inference-time through manual prompt optimization, which introduces overhead and lacks scalability. Conversely, we focus on train-time approaches that build prompt robustness into the model during instruction tuning. A common strategy is to mix multiple paraphrases of each prompt into training. We demonstrate that across trainable parameters, the per-paraphrasing gradients interfere, with 57% exhibiting sign conflicts and the naive gradient average discarding approximately a fifth of the magnitude. Therefore, we develop two gradient-merging methods, Prompt-TIES and Prompt-DARE, that address this interference by leveraging interference-resolution operators from model and gradient merging and applying them to prompt gradients. In this study, we compare the efficacy of our methods to increase model robustness to train-time robustness methods spanning all three established classes: data construction, consistency regularization, and contrastive alignment. The evaluation is conducted across four models from two families at two scales. We observe the interaction between prompt phrasing and the specific examples accounts for 54% of per-example variability on average, and exceeds 70% on tasks such as ANLI and WiC. The gradient-merging methods directly improve this local robustness and nearly halve per-example variance on fully fine-tuned models at little cost to overall accuracy. The other train-time methods enhance raw performance, but fail to improve per-example variance reliably. Code: https://github.com/FSadrieh/prompt-agnostic-llms. Model checkpoints: https://huggingface.co/prompt-agnostic-language-models.

Large Language Models (LLMs) have rapidly become important tools for science, education, medicine, and professional work. Despite their apparent simplicity from the user's perspective, effective and responsible use requires an understanding of how these systems work, what their limitations are, and how their outputs should be evaluated. I wrote this document as a basis for my lecture on LLM in the master in artificial intelligence in medicine at the university of Turin, Italy. This book provides a practical and accessible introduction to LLMs for scientists and professionals. It explains the fundamental ideas behind tokenisation, embeddings, training, inference, context windows, prompting, and instruction hierarchies, while deliberately avoiding mathematical detail. Particular attention is given to the limitations of LLM-generated answers, including hallucinations, uncertainty, reproducibility, bias, and the need for systematic verification and human oversight. The book also discusses privacy, governance, security, and the responsible handling of personal and sensitive information. More advanced topics include Retrieval-Augmented Generation (RAG), retrieval quality, chunking, embeddings, vector search, agents, tools, and workflows. A substantial practical exercise guides the reader through running a local LLM with Ollama and Open WebUI and using it to interact with a collection of scientific papers. The exercise emphasizes evidence retrieval, source verification, uncertainty, disagreement between studies, and the distinction between plausible language and scientifically supported conclusions. No programming knowledge is required.

Language models do not merely memorize facts; they absorb rhetorical strategies from pretraining data, including the human tendency to reason by analogy. We ask: is this absorbed capability geometrically accessible before any instruction tuning, and what determines whether it can be activated at inference time? We extract an "Analogy Vector" from the residual stream via difference-in-means and show it enables steerable explanation style across seven instruction-tuned models (Gemma-2, Llama-3.1, Qwen-2.5; 2B-32B), improving harmonic mean from 3.22 to 3.94 (+22%) for Gemma-2-9B-IT on 300 held-out test prompts. A small human validation study confirms the LLM judge on Analogy Quality. Our central finding, from seven matched base-instruct pairs, is a gradient of behavioral accessibility: the analogy direction exists in every base model, but its behavioral accessibility under linear intervention scales continuously with instruction exposure during training. Gemma-2 base models (strict pretraining/IT separation) show flat steering landscapes: present but behaviorally inert. Qwen-2.5 base models (instruction data woven into pretraining) are already partially accessible before any explicit fine-tuning; base-instruct cosine similarity of the analogy vector reaches 0.94-0.97 for Qwen versus 0.37-0.66 for Gemma, providing geometric corroboration of the behavioral gradient. This challenges the binary "base vs instruct" framing: the relevant variable is how much instruction exposure a model has received across its full training pipeline. We position activation steering as a complementary interpretability tool that reveals how instruction exposure organizes rhetorical knowledge in activation space.

This paper introduces OrbitalGym, a simulation environment for studying when planning algorithms or system design determine the outcome of adversarial orbital encounters. In an orbital game, two non-cooperative teams of satellites maneuver during rendezvous and proximity operations to win a differential-reward objective. Advances in terrestrial game playing have prompted many methods for orbital games, but studies typically evaluate one method in one scenario, often against a passive or random defender. It is unclear whether better planning matters once vehicle capability (maneuverability, sensing, communications) and adversarial responses are considered. OrbitalGym is a JAX-based environment supporting varying numbers of agents, state representations (position, velocity, attitude, fuel), explicit or implicit communication, and proximity or conical field-of-view sensors. It compiles games to batched GPU code, enabling large-scale parallel simulation of policies across many configurations. This identifies regimes where policy choice affects outcomes and regimes where system capability dominates. We study two close-range games: pursuit-evasion, where one spacecraft tries to capture another, and lady-bandit-guard, where a guard prevents an attacker from reaching a protected asset. We compare PD, LQR, model predictive path integral control, and Monte Carlo tree search over ratios of thrust and propellant, initial separation, game duration, sensor capability, and the definition of capture. In pursuit-evasion the propellant ratio decides most encounters: against five of six evader policies, pursuer policy changes capture probability by 10 percent or more in at most 15 of 90 capability configurations. In lady-bandit-guard, policy matters more: against bandits able to reach the asset, a better guard policy raises defense probability by at least 10 percent in 56 to 74 of 90 configurations.

Large language models (LLMs) have enabled social simulation with persistent agents, where fixed computational individuals maintain internal states and interact over time. However, large-scale social networks are inherently open-population systems: the participating population changes over time, users enter and leave dynamically, and only a sparse subset becomes observable through communication at any given time. In such systems, persistent account identities do not provide continuously observable individual-state trajectories, making recursively maintained psychological states insufficiently grounded in empirical observations. We therefore shift the simulation target from persistent individual trajectories to the observable evolution of social communication, modeling who participates, what they communicate, and how these communications shape subsequent system dynamics. We instantiate this formulation with participant activation from a global user pool, collective state transition, and an LLM policy for communication generation, while explicitly separating environmental inputs from endogenous system evolution. Across five Qwen backbones, our framework reduces average KL divergence by 71.2% over Direct LLM and 35.0% over MF-LLM, with boundary and counterfactual experiments further validating the proposed modeling boundaries. Our results support communication-centered simulation as a principled formulation for long-horizon open-population social systems. Project Website: https://ai4ss.github.io/OpenSim/

Deployed classifiers often act on a prediction only when its confidence clears a threshold and pass the rest to a person. We ask whether one threshold serves all languages equally on typed decision interfaces, APIs that return a probability for each option of a fixed multiple-choice question. In a confirmatory study with a written analysis plan, three vendors' interfaces classified 1,000 MASSIVE items, each in 51 languages, with a threshold of 0.9. Even when a request is answered correctly in both English and another language, the English version clears the threshold more often, by 8.6-11.7 percentage points on average. Observed coverage, the share of requests handled automatically, is lower than English's in every other language, while the requests the gate handles are rarely wrong. The gap remains when every language is reweighted to the same accuracy, and a standard calibration metric (ECE) does not detect it on one interface. In exploratory analyses, English's coverage stays higher when each language gets its own threshold fitted to the same 5% error target. Translating inputs into English nearly closes the gap at 0.9 but leaves English ahead at that error target. Requests written natively in African languages show a gap in the same direction when English is their human translation. We recommend reporting per-language coverage at the operating threshold.

Users engage with large language models on tasks that are purely objective ("When was Netflix founded?"), purely subjective ("What should I name my child?") or anywhere in-between. We have rich tools for evaluating model accuracy with respect to objective criteria. However, when it comes to subjective aspects of a task, we do not have similarly developed methodology at hand. In this paper, we take some early steps towards framing the question of subjective evaluation. Rather than measure how often a model is right or wrong on a question, we instead ask, how do models make choices among multiple viable responses without a clear winner? To study this question, we restrict our focus to short-form content, and consider comparison tasks among three such content types: numbers, baby names, and fictional cities. We identify 29 aspects of these short-form examples covering many areas of syntax and semantics (length, alphabetic order, popularity, vowel mix, spoken sound, etc.). We show first that models tend to have strong and reproducible preferences for or against some of these aspects. Second, we show that variations in how the model is asked to express a preference may cause some of these aspects to activate or deactivate in the model's selection. Third, we show that overall preference distributions among pairs may vary radically based on seemingly-unimportant variations in the nature of the question posed to the model, raising new questions about manipulation. And finally, we show that, if models are used both to create lists of alternatives and to rank them, there are unexpected interactions between the two stages that may also drive the models' stated preferences.

As multi-agent language model networks become widely deployed, predicting their collective behavior is a critical challenge. Classical distributed consensus theory relies on network topology to predict convergence, structurally assuming that agent updates are convex combinations, population means are conserved, and weights are content-independent. In this work, we demonstrate that these foundational assumptions fail for large language models (LLMs). By deterministically extracting continuous credence from token-level probabilities across eight instruction-tuned models and 73,080 controlled exchanges, we directly measure the empirical LLM update operator. We find that LLMs systematically violate convexity by updating outside the hull of observed opinions, exhibiting active repulsion (negative weights) that fundamentally breaks row-stochastic modeling and standard bounded-confidence repairs. Scaling these empirically fitted operators to simulations of 10,000 agents reveals that algebraic connectivity explains almost none of the final outcome variance. Instead, population dynamics are governed entirely by the model's internal update bias and message framing. Consequently, populations of several models, including those with up to 7B parameters, reliably compound small local drifts to reach confident agreement on objectively false claims regardless of their network structure. Our results prove that classical graph-theoretic metrics are inadequate for LLM populations, highlighting a profound deployment risk where multi-agent deliberation predictably saturates in collective delusion.

Multi-agent LLM pipelines built on interoperability protocols such as the Open Floor Protocol (OFP) or A2A are usually justified on capability grounds, with their runtime cost treated as an overhead of unknown size. Comparing such a pipeline with one model call is underspecified, because a pipeline can give each role a different model. We give a closed-form model that sets the coordination tax (decomposition overhead plus per-handoff protocol cost) against the heterogeneity dividend from routing work to a cheaper model, with a break-even threshold and a feasibility bound, $α< 1/r$, and measure every term, with OFP as the protocol under test, on 135 HumanEval and ClassEval tasks in 2430 runs of six architectures with Claude Opus 5.5 and Haiku 5.5. The protocol adds 7.2 ms per handoff, 0.11% of pipeline wall time. Decomposition does not pay on these tasks: a planner-coder-reviewer pipeline costs 3.22x as much as a single strong call, takes 3.24x as long, and solves no more tasks. Moving planner and reviewer to the cheap model cuts the pipeline's cost to 0.39x but changes pass@1 by -6.7 points, leaving it dearer, slower and less accurate than the single call, and the break-even forecast misses by -19% because the cheap roles inflate token volume. The saving lies in choosing a model per task instead: an oracle choice between the two single calls solves 94.3% of runs at half the strong call's cost per solved task.

Mega events can generate concentrated travel demand, intensifying congestion around venues and increasing travelers' environmental exposure. Existing traffic management strategies focus on travel efficiency, with limited attention to exposure risks. Taking mega-event travel as a short-term high-demand scenario, this study develops an exposure-aware multi-agent routing framework integrating road-network conditions, event-induced demand, background traffic, and environmental data within a unified simulation environment. Motorized travelers optimize travel efficiency, whereas micro-mobility travelers consider travel time and dynamic exposure to traffic-related PM2.5 and noise. We compare the framework with shortest-distance, congestion-aware, and exposure-aware routing baselines. Results show that, under high demand, the SARSA-based framework reduces vehicle travel time by 34.59% and cumulative environmental exposure by 58.65%, while micro-mobility travel time remains nearly unchanged. Exposure reduction is jointly driven by vehicle-side congestion mitigation and low-exposure routing. The framework provides quantitative support for vehicle diversion, low-exposure route guidance, and health-oriented traffic management in short-term high-demand scenarios.

Efficient communication remains a major bottleneck in large language model multiagent systems. Latent communication reduces repeated natural-language exchange by transferring internal model states between agents, but these transfers remain costly and are typically designed around conventional KV-cache representations. We address this problem from two complementary perspectives. We introduce SelectKV, a boundary-aware method that selectively communicates KV states while protecting inherited context, and HyLaMAS, which extends latent collaboration to hybrid linear-attention models by communicating both KV caches and recurrent states through a shared workspace. Across six benchmarks, SelectKV reduces communicated KV positions and logical KV payload by 9.7-9.9%, with accuracy unchanged on two benchmarks, within one percentage point on three, and six points lower on HumanEval+. Across Qwen3.5 model scales, HyLaMAS reduces decoded output tokens by 53-80% and achieves 1.16-2.78x speedups relative to text-based multi-agent communication. Together, these independently evaluated methods represent two complementary approaches to efficient latent-state communication across conventional Transformer and hybrid model architectures.

Web agents require diverse, executable, and resettable website environments for reproducible training and evaluation. Human-assisted mirrors demand per-site engineering, while generative construction can diverge from specified real websites. Capture-and-replay preserves observations, with interaction coverage constrained by captured content and replay capabilities. We present WebAnything, a fully automated framework that reconstructs source-grounded, reusable Web environments from real URLs without per-site manual implementation or repair after initial setup and authorization. An Explorer-Builder collects browser-observable structures, resources, and interaction outcomes; an independent verifier checks frozen candidates and guides local repair or targeted source recapture. We construct 244 website artifacts across 14 application domains and 53 categories, supporting 12,984 executed and accepted tasks. Compared with human-assisted WebHarbor, WebAnything achieves higher aggregate scores on three fidelity metrics, with 12 wins, two ties, and no losses in a blind audit. Fine-tuning Qwen3.5-9B on 5,856 interaction trajectories improves REAL success by 9.9 percentage points, Odysseys-Easy Rubric Avg and Task Avg by 6.25 and 6.32 percentage points, and WebArena-Verified Overall success by 13.33 percentage points under the fixed harness. Code: https://github.com/china-qijizhifeng/Web-Anything

Language model assistants sometimes change a correct answer when the user suggests a wrong one. The behavior, called sycophancy, is usually attributed to preference-based post-training, yet early measurements found it in pretrained models. We ask where it comes from by measuring it at every available stage of training: 13 open checkpoints from 5 lineages, from base model through supervised fine-tuning and preference optimization to release. We separate how far an injected claim moves a model's answer from how firmly the answer was held, and vary who makes the claim, how reliable they are said to be, and how strongly they make it. Two sycophancy behaviors with different origins emerge, and both replicate across datasets. Deference to a claim in the prompt is already present in every base model, which ranks sources in the same order as its post-trained descendants. Capitulation in conversation, dropping a stated answer when the user objects, is largely absent from base models and present in every post-trained checkpoint. When the objection names an alternative, most post-trained models give up firmly held answers as often as loosely held ones. We discuss the implications of these measurements for evaluating sycophancy and approaches for addressing each behavior.

Recent work has hypothesised that large language models may represent some concepts as multi-dimensional manifolds instead of single directions. Using these manifolds to interpret or steer representations often assumes that moving along them corresponds to a continuous change in the represented concept. In this work, we test this continuous manifold assumption on the prototypical weekday manifold. Despite its striking circular geometry, we find that the weekday representations behave more like seven discrete states arranged in a circle than a continuous ring. First, we show that local sensitivity has seven minima aligned with the weekdays, producing a step-like mapping between representation and behaviour under on-manifold steering. Second, we test whether any concepts are represented along the manifold outside the seven discrete states, and find neither time of day, uncertainty, nor natural text populate the intermediate space. Together, these results challenge the continuous interpretation of the weekday manifold and motivate testing whether movement along other representation manifolds correspond to meaningful changes in model representations and behaviour.

LLM agents that answer questions over Web knowledge graphs are increasingly wrapped in checks derived from the graph's ontology, on the assumption that a formal schema makes a reliable judge. Yet a checker can be wrong as well: it can miss errors, and it can flag correct answers and push the agent into "fixing" them. Whether a curated ontology is worth its cost as such a judge, and how it compares with an ontology an LLM writes or a schema mined from the data, is unknown. We make the checker itself the object of study. Across curated, LLM-generated and data-derived ontologies, used as prompt context, as a checker, or both, and across open-weight and proprietary agents on public graphs and a private graph of nearly $10^{9}$ triples, we account for what ontology-derived checks catch and what they wrongly flag. We show how each source trades caught errors against false alarms, how a curated schema used as published can contradict its own data, which errors structural checks miss, and whether combining sources or giving the agent the checker's specific feedback pays off. On UniProt, the curated ontology flags more correct answers than either an ontology derived from the data or one an LLM derives from a summary of the data with masked identifiers. On real agent answers, its flags are only marginally more precise than random flags.

Quantum machine learning (QML) is increasingly explored for medical classification, yet its behaviour under subgroup-aware fairness criteria remains understudied. We present an empirical fairness audit of classical and quantum classifiers for medical imaging, assuming QML is not inherently fair. Two scenarios are evaluated: dermatological lesions on DDI (sensitive attribute: Fitzpatrick skin type) and chest X-rays on CheXpert-derived data (sensitive attribute: age group). Using ResNet-18 embeddings, we train SVM, MLP, QSVM and QCNN under a baseline and three mitigation settings: differentiable Equalized Odds (EO) regularisation, Gradient Reversal Learning (GRL), and both combined. Foundation models (PanDerm, EVA-X) serve as external references. Quantum classifiers may achieve competitive aggregate performance while still exhibiting subgroup disparities. Since a smaller subgroup AUC gap may stem from degrading the best group rather than improving the worst, each mitigation is also tested against a pre-specified non-inferiority margin relative to its baseline. Under this criterion, several configurations with the smallest disparity, notably QCNN with GRL on DDI, are not clinically acceptable due to loss of global discrimination. On DDI, QSVM with GRL+EO performed best and approached PanDerm in the performance/fairness trade-off; on CheXpert, QCNN with GRL achieved the highest aggregate performance but increased disparity, and EVA-X remained stronger. Fairness effects are model- and dataset-dependent, so QML requires empirical subgroup-aware evaluation. All quantum models run in classical simulation; we claim no quantum advantage.

Retrieval-augmented generation (RAG) lets a large language model (LLM) answer questions by retrieving relevant passages from a large corpus. However, RAG struggles with multi-hop questions, which require chaining facts spread across several documents through intermediate entities that the question never mentions. Graph-based RAG methods retrieve once from the entities a question names and so rarely recover the full evidence chain. Iterative RAG methods recover it step by step, but their accuracy is limited by the many tasks they ask the LLM to perform, such as planning the sub-questions, deciding their dependencies, rewriting them as answers arrive, and keeping track of what is known. We propose DeFT-RAG, a RAG framework built on three components. First, DeFT-RAG generates one self-contained sub-question at a time from the original question and the sub-questions answered so far with their answers (QA pairs). Each sub-question names every entity it asks about, which removes the need to plan, track dependencies or rewrite, and lets the LLM focus on a single fine-grained task at a time. Second, DeFT-RAG moves the remaining LLM work outside the LLM: rule-based updates to a QA table that keeps track of what is known, and yes/no decisions made by a fine-tuned lightweight decision model. Third, DeFT-RAG caches the QA pairs and reuses an answer when a later question asks a sub-question with the same meaning. On MuSiQue, 2WikiMultiHopQA and HotpotQA, DeFT-RAG achieves the best accuracy among thirteen baselines with both gpt-4o-mini and Qwen3.6 as the backbone, while giving the LLM fewer input tokens than any other iterative method. With a QA cache shared across 1,000 MuSiQue questions, 23.1% of sub-questions are answered from the cache, which reduces LLM calls by 12.9% without lowering accuracy.

An AI developer often has several candidate versions of a system and can deploy a mixture of them, sending each request to a randomly chosen version, as in A/B testing. Each stakeholder (e.g., a safety team, a regulator, or a customer with private test data) accepts a mixture exactly when its expected performance (e.g., its rate of harmful replies) meets the stakeholder's own standard, which the developer can learn only by asking whether a proposed mixture is acceptable. We model this as choosing a lottery over $m$ alternatives for $n$ agents, each of whom accepts exactly the lotteries in an unknown halfspace and answers such binary membership queries. While Choo et al. (ICML 2026) sought a lottery that every agent accepts, none may exist when, say, safety and helpfulness requirements conflict, so we minimize the number of rejecting agents. We show that this problem is APX-hard even when all halfspaces are known and all utilities are $0$ or $1$. We then give exact algorithms that run in polynomial time when $m$ is constant, and in quasi-polynomial time when, for a constant $η$, some optimal lottery has no accepting agent within $η$ of its threshold. When only some agents can be queried, optimizing over a uniform sample of agents gives additive and relative guarantees for the whole population, with a sample size that does not depend on $n$, and a linear program gives a polynomial-time method for any $m$. On the other hand, for $m\ge 3$, a deterministic algorithm that queries only $K$ distinct agents cannot guarantee fewer than $n-K$ extra rejections, and this is achievable; in other words, it can guarantee nothing for the agents it does not ask. Small synthetic experiments indicate that our worst-case bounds are conservative on random instances.

AI agents can now write software from a plain-English description, but code that runs is not the same as code that does what was asked. This gap matters most for smart contracts, the programs behind Web3 applications that move money on a blockchain and cannot be fixed once deployed. We ask how to check whether an AI-written contract behaves as it should, and where the requested behavior gets lost. We design an evaluation built on execution rather than opinion. We replay the same transactions on an AI-written contract and on the original it should reproduce and measure how often they agree. We test every grading tool on planted mistakes, compare against an answer key written without the description, and audit our own scorer until known wrong answers cannot pass. Common methods, such as AI judges and standard analysis tools, miss obvious mistakes, and scoring choices matter too. Checking only the functions every system includes hid about half the difference between pipelines. We then propose an elicitation-first architecture. Before writing code, the agent asks the client a short checklist of questions, checks the contract against an independently written executable answer key, and returns disagreements to the client instead of silently rewriting code. Asking first matters far more than reviewing or patching afterwards, across three AI models, while automated security fixes often break requested behavior. A few well-chosen questions recover much of what a short description omits, and their value can be estimated without the original contract. To our knowledge, this is the first study in smart contract synthesis to gather missing requirements before generation and check the result against an independently written executable model. The lessons extend to any setting where AI agents turn requests into programs. We release our tools, 600 tasks and all results.

AI agents that share a secret convention can coordinate through messages that look innocent, so a monitor that reads only the messages can miss them. We propose a behavioural test: a partner that uses a code reacts to the exact wording of a message, whereas an uninvolved agent placed in its seat does not. Partner--Outsider Positive Listening (PoPLi) is the difference between the two agents in how much their action changes when the message is reworded with its meaning kept, and it is computed after the fact by replaying logged prompts. In a simulated market with five conditions (no collusion, a joint reward with neutral messages, the same with an openly stated intention, a covert first-letter code, and the code with a stated intention), PoPLi is between 0.48 and 0.59 whenever the code is present (95% intervals all above 0.29) and between -0.10 and -0.01 without it. Message-only monitors did not separate the conditions: one reading the speaker's messages answered "no" in all 50 markets, and one reading every seller's messages flagged 9 of 10 markets with no collusion. PoPLi detects the use of a private channel, not collusion as such, and it cannot detect collusion through incentives alone or codes carried by meaning.

Existing scientific-agent benchmarks rarely evaluate whether agents can reconstruct the exact molecular structures and generation workflows specified by sequencing protocols, a capability critical for analyzing customized sequencing assays. We introduce LibStructBench, a document-grounded benchmark for reconstructing sequencing-library structures and molecular workflows. Our experiments show that frontier agents often produce approximately correct reconstructions but remain unreliable at exact reconstruction, particularly in molecular-state representation and workflow connectivity. Switching the underlying language model produced larger average performance differences than switching agent harnesses, with a statistically significant difference only for molecular-state reconstruction. Additional knowledge improved oligo-family recovery in a small pilot without improving mean molecular-workflow scores. Autonomous accumulation of procedural rules and cross-protocol examples showed non-monotonic validation performance and no mean held-out molecular-workflow improvement in the tested trajectory. These findings highlight the need for more controlled and verifiable approaches to scientific reasoning about molecular processes in complex experimental protocols.

Language-model agents differ fundamentally from prior intelligent systems in their ability to learn from natural language in context. This capability enables agents to iteratively refine their answers in response to feedback, a prevalent mode of operation in modern agentic workflows. However, the community primarily ranks models by metrics such as Pass@1 or Pass@K, which cannot capture how they utilize feedback on prior attempts. We introduce Seq@K, a natural evaluation protocol that measures whether a model can solve a problem within K sequential attempts, provided feedback after each previous failure. Unlike multi-turn agentic benchmarks that examine long-horizon planning, Seq@K isolates and formalizes the core capability of refinement from feedback. We evaluate Seq@K on a suite of 7 benchmarks and 16 frontier models, and show that Seq@K reveals a materially different model-selection landscape: under Seq@K, (1) model rankings often differ, (2) the Pareto frontiers admit more efficient configurations and (3) the gains from choosing the optimal model configuration can widen or shrink relative to those of Pass@K. We further show that analyzing model trajectories under Seq@K helps characterize iterative refinement behavior, such as the rate at which models address issues raised in feedback, or how rigidly they hold onto their own answers. Finally, we study how the nature of feedback affects models' iterative refinement, showing that practitioners must carefully design and benchmark the feedback loops of iterative agentic systems. Together, these results establish how Seq@K offers a more faithful metric for intelligent systems that typically operate in such feedback loop, and the necessity to consider Seq@K in evaluating models for these systems.

Scaling visual reinforcement learning (RL) with GPU-accelerated simulation is computationally expensive due to rendering and visual-policy inference costs. Consequently, two-stage teacher--student distillation is commonly used to train visual policies, but introduces an imitation gap. In this preliminary study, we introduce \echomethod{}, \textbf{a single-stage off-policy RL framework that scales experience collection without proportionally increasing visual computation}. Through \textbf{inverse distillation}, a state-based auxiliary policy imitates the evolving visual policy and collects inexpensive, unrendered experience. Echo leverages this experience for off-policy value learning, directly improving the visual policy through a shared privileged critic. We instantiate \echomethod{} with FastTD3 and evaluate it on three visual control tasks and two dexterous manipulation tasks in Genesis. Relative to \sdpgmethod{} under the evaluated configurations, Echo reduces time to reference returns by an average of \textbf{60\% across the three locomotion tasks} and \textbf{56\% across both hand tasks}, while improving peak mean returns by \textbf{62\% and 12\%}, respectively. Both hand tasks also achieve higher mean returns at matched simulator-transition budgets. These initial results suggest that scaling off-policy experience is a promising approach to efficient direct visual RL.

Food similarity underlies retrieval, recommendation, substitution, and other food computing tasks, yet food representations are typically evaluated by identity preservation. We ask what happens to food similarity beyond identity: how different visual representations organize foods into relational neighborhoods, what culinary relationships these neighborhoods reveal or obscure, and how consistently these structures emerge across representations. Indian cuisine provides a challenging setting in which visually similar foods may have different culinary identities, while foods within a culinary family can vary substantially in appearance. Using an expanded Khana dataset and seven pretrained visual representations spanning convolutional, transformer-based, self-supervised, and image-text aligned models, we evaluate image-to-image retrieval and image-to-dish neighborhoods, examining culinary-family structure, contrast cases, and cross-representation agreement. We also aggregate image representations to compare dish-level visual similarity across models and against recipe-derived similarity for the same dishes. Image-level neighborhoods vary substantially across representations, including similarities across culinary-family boundaries. At the dish level, aggregation reveals common visual structure despite local variation, but this structure corresponds less strongly with recipe-derived similarity. These findings show that recognition performance alone cannot characterize relational structure: food similarity depends on representation, modality, and abstraction level rather than being a fixed property of the food collection. Understanding these structures is therefore a prerequisite to combining multiple food information modalities into a coherent notion of similarity.

International climate policy requires coordination among countries whose mitigation decisions impose local costs but generate global benefits. Classical game-theoretic and integrated assessment approaches can study these incentives, but typically rely on prescribed behavioural assumptions or equilibrium concepts. Recent work has therefore explored Multi-Agent Reinforcement Learning as a way to model climate-policy cooperation as an emergent outcome of repeated strategic interaction. We extend an existing MARL climate framework with a multi-stage negotiation mechanism in which agents propose agreements, accept or reject commitments, and face penalties for non-compliance. This allows us to study how agreement design affects learned mitigation behaviour, compliance, and climate outcomes. In an idealized setting with four homogeneous agents, we find that non-binding agreements do little to overcome free-riding, while moderate non-compliance penalties improve cooperation and reduce warming. However, excessively strong penalties discourage ambitious commitments, revealing a trade-off between enforcement and participation. We position the framework as a methodological contribution for testing agreement design rather than a predictive model of real negotiations.

Deepfake detectors are almost always deployed as the single checkpoint that scored best on a validation split, yet the reliability of that choice is rarely measured. We measure it with PairCLIP-SWA, a single-backbone detector that adapts an OpenCLIP ViT-L/14 visual tower by progressive block unfreezing on a class-balanced corpus built by CLIP-similarity-guided one-to-one subsampling, and averages the fine-tuning trajectory in weight space. Trained in-domain on the IEEE SP Cup 2025 (DF-Wild-Cup) benchmark under the same protocol as the strongest published three-backbone ensemble, the weight-averaged detector reaches 97.04% accuracy, 98.64% AU-ROC and a 4.07% equal error rate, against 94.46%, 97.60% and 5.00%. Its training log shows why averaging matters: over the averaged epochs the non-averaged checkpoint swings across 5.34 points of validation accuracy while its validation loss more than doubles, whereas the average is never worse and is up to 5.99 points better. On an OpenForensics-derived benchmark the same instability becomes a misranking: the checkpoint that validation accuracy selects (98.43%) scores 88.85% on a disjoint 10,905-image test partition, rejecting 21.1% of genuine photographs, while the terminal average, ranked last on validation, scores 94.76%. A study of twenty-two in-house configurations on the identical DF-Wild-Cup split, including preprocessing ablations, shows that CLIP pre-training rather than ensembling drives accuracy, that secondary face cropping costs about 34 points, and that frequency-domain inputs transfer poorly. We recommend per-class selection criteria measured off the early-stopping split.

Accurate AI assistance can improve immediate judgments while reducing opportunities to practise the operations needed for independent performance. Verification Learning Theory (VLT) formalises this possibility for claim-evaluation practice followed by correct, independent, rule-specific feedback. The model links perceived assistant reliability to evidence checking and links the history of those checks to retained rule use. Its strict form assumes that task demands and perceived reliability influence learning only through checking, with equal learning per genuine check. It therefore predicts that perceived fallibility can improve delayed unaided judgment even without observed assistant errors, and that this advantage disappears when checking is already required and performed. Rule-specific gains and forecasts from checking histories provide further tests. Simulations examine the consequences and parameter dependence of these commitments; they do not constitute empirical validation. Comparisons with depth-of-processing and feedback-encoding accounts identify where predictions diverge and where the accounts remain observationally equivalent. The framework specifies experiments that can test the proposed mechanism and separates retention of practised rules from the broader problem of recognising which rule applies to a new claim.

On PushT, our fixed action updater (1,505 parameters) solves 8 of 16 scenes reserved from our configuration selection. The comparison planners perform as follows: Selected Adam (7 tasks solved, total planning and encoding time 218.440s), adapted iCEM (2 tasks solved, total planning and encoding time 49.309s), released LePlanner (16 tasks solved, total planning and encoding time 5.652s). The total planning and encoding time for our action updater is 59.079s. Note that our action updater does not satisfy the task-cost criterion we fixed before evaluation. We also conduct another matched experiment where we only replay the first ten commands in each context and then allow the next solve 4, 8 or 12 updates (subsequent solves all get 8 updates). Among all 16 contexts (8 scene clusters), the 3 budgets get 9, 8 and 7 successes, respectively. Interestingly, the extra updates did not rescue any failing task in this experiment. All the changes in success/failure happen in one scene cluster, where we observe that the extra updates decrease the costs of the selected latent world-model trajectories for our updater, while the tasks that used to be successes become failures. The code, weights, evaluators, execution traces and cost accounting accompany the paper.

Subjective judgments such as offensiveness often vary substantially across individuals, motivating pluralistic modeling approaches that incorporate annotators' socio-demographic information to capture this variation. However, prior findings on the benefits of socio-demographic information for pluralistic modeling are mixed, calling for a systematic analysis of both its effectiveness and the mechanisms through which it improves prediction. In this work, we study $\textbf{when}$ and $\textbf{where}$ socio-demographics benefit pluralistic offensiveness modeling, as well as $\textbf{the role of explicit reasoning}$ in leveraging this information. Across three multi-annotator datasets with prompted and trained LLMs, we first show that demographic gains are small and inconsistent for individual predictions but become clearer at the group and population levels, because aggregation attenuates the idiosyncratic variation that masks small but systematic demographic signals. Second, using a reasoning and prediction format that explicitly decomposes individual judgment into an overall baseline and a demographic-specific deviation, we show that the gains arise from both components: demographic information improves baseline estimation and enables meaningful demographic-specific shifts, and interventions show that better component estimates improve final predictions. Finally, comparing direct demographic conditioning with explicit reasoning about how the text interacts with annotator demographics, we find that reasoning yields more accurate baseline and shift estimates as well as final predictions, whereas conditioning alone rarely produces informative shifts. These results suggest that the mixed evidence in prior work may partly stem from individual-level evaluation, and that the value of socio-demographic information depends on both how it is evaluated and how models are guided to use it.

Metis is a bet that better data, better training methods and new architectures can buy what a bigger parameter count cannot, at a fraction of the cost. Metis-1.6 is its latest step: two language models built from scratch by a single researcher, Praxis (2.94B parameters, 0.46B active per pass) and Logos (10.23B parameters, 1.18B active per pass). Both run a hybrid Mamba-2 and attention mixture-of-experts stack twice for every token, so each token gets a second look. Each was trained on 1.02T tokens in total: 804.76B of pretraining, 18B of context extension to 131,072 tokens, and 200B of supervised fine-tuning. Metis-1.6 was never meant to challenge frontier models, and across 40 post-trained benchmarks it mostly does not: Logos averages 52.93, seventh of 17 models in its comparison set and within a point of Qwen3-4B. But it gets there on roughly a twelfth of the median training data of the comparators that disclose it, and with fewer active parameters per token than any model that beats it, which puts it on the Pareto frontier for both. In mathematics it ranks second of 30 models, with 67.50% on AIME 2025 and 63.64% on HMMT February 2026. It is weak at code, instruction following and long context. The rest of the report is the part I found most valuable: what went wrong. The optimizer I configured never ran, and attention trained with AdamW for all of pretraining. A 1.77B-parameter n-gram memory learned to switch itself off. Five fine-tuning choices cost capability. And a length reward under group-mean normalization punished correct solutions for being long once a group's pass rate passed 2/3, a sign inversion I prove and fix with a one-line clamp. I close with the 10B-token ablations behind MoRE (Mixture of Recursive Experts), the looped architecture of Metis-1.7.

Public disclosure repositories such as SEC EDGAR support informed investment through web access to corporate filings. Supporting their review with language model agents requires discovering unstated deficiencies across related documents, acquiring missing company facts, and verifying that requested revisions satisfy evidence-backed requirements. Existing financial question-answering evaluations provide limited evidence of these interactive capabilities. We introduce \textbf{FilingLoop}, an interactive benchmark and a self-evolving agent framework for corporate disclosure review and revision. Grounded in historical U.S.\ Securities and Exchange Commission correspondence, FilingLoop reconstructs 380 predominantly IPO review tasks from 96 companies under distinct information constraints. An evaluator separate from the acting agents scores final filings and responses against 6,099 eligible requirements and audits unsupported claims and harmful revisions. Our framework couples iterative review and revision within each case with procedural learning across cases, converting requirement-level evaluation feedback into role-specific guidance without updating model weights. Updates affect only subsequent cases, allowing procedural adaptation to be assessed without retrying scored tasks. On the evaluated stream, evolution increases requirement fulfillment from 24.28\% to 27.96\% and full comment resolution from 16.90\% to 20.75\%, with a 5.8\% increase in total cost over the fixed-policy variant. FilingLoop connects the concerns an agent discovers with the company facts it obtains and the revisions it commits, providing requirement-level feedback for procedures used in later review tasks.

Spiking reservoir computing combines the training efficiency of fixed recurrent reservoirs with the energy efficiency of event-driven inference, where computation is triggered by sparse spikes rather than executed densely at every timestep. However, existing spiking reservoirs almost exclusively rely on Leaky Integrate-and-Fire (LIF) neurons, whose low-pass dynamics limit their ability to distinguish information encoded at different temporal frequencies. We introduce HRF-Res, a liquid state machine in which LIF neurons are replaced by a heterogeneous population of Harmonic Resonate-and-Fire (HRF) neurons. These neurons exhibit intrinsic oscillatory dynamics and selectively respond to input frequencies, enabling the reservoir to capture frequency-specific temporal patterns. The result is more discriminative representations and higher classification accuracy than LIF-based reservoirs, achieved without sacrificing the energy efficiency of spiking computation, with up to 19x lower energy consumption than an equivalent non-spiking oscillatory reservoir. Evaluated on ten benchmarks spanning time series, sequential images, audio spike trains, and neuromorphic vision, HRF-Res achieves state-of-the-art accuracy among spiking reservoir methods. These results demonstrate that introducing frequency-selective dynamics into spiking reservoirs improves representational quality and classification performance while preserving their inherent energy efficiency.

Distributed fire-monitoring systems may involve many sensing devices reporting over a shared uplink, even though the operational objective is often to identify regions exhibiting evidence of fire rather than to recover device-specific measurements. We introduce TERRA (Task-Oriented Edge Representation and Regional Aggregation), an edge-AI framework that combines task-oriented local inference, identity-free region-tagged reporting, and permutation-invariant regional evidence aggregation. TERRA supports two reporting modes: TERRA-Score, in which each device communicates a quantized local fire probability, and TERRA-Feature, in which it communicates a compact learned representation for receiver-side aggregation using a permutation-invariant DeepSets model. Using all 15 numerical sensing modalities of the distributed indoor EN54 fire-alarm dataset as a proof-of-concept testbed, we evaluate regional inference for an episode-level fire-scenario target under repetition-disjoint three-fold cross-validation. We study representation size, task-payload quantization, report availability, and robustness to missed and false recovered messages. With nine reports and perfect message recovery, TERRA-Feature achieves mean AUPRC 0.990 using a 24-bit task payload, while TERRA-Score achieves 0.987 using only 6 bits; both attain an MDR of 1.81% at FAR $\leq$1%. TERRA-Feature achieves slightly higher AUPRC under reliable recovery, whereas the more compact TERRA-Score is less sensitive to increasing false-message contamination.

Vision-Language-Action (VLA) models offer a promising paradigm for general-purpose household robotics. However, inference and data-collection strategies that have been deployed in smaller-scale settings do not necessarily generalize to large-scale, diverse task suites. To address this gap, we conduct a systematic, paired evaluation of previously proposed inference and data-collection strategies across all 100 tasks. Our evaluation uses a fine-tuned $π_{0.5}$ model based on the 2025 BEHAVIOR Challenge winner, which achieves a Q-score of 0.39 on the public test set. These results show that proposed strategies do not transfer uniformly to new environments and tasks. Project page, videos, and detailed analysis: https://rubenhammele.github.io/behavior2026

Large language model agents are increasingly used for scientific discovery. However, existing benchmarks mix law inference with programming, score data fitting instead of prediction on unseen inputs, and do not separate empirical evidence from pretrained priors. We introduce SciGame, a controlled benchmark that converts scientific literature into structured scientific cards. Each card yields a pair of Original and dimensionally consistent Counterfactual environments with the same task interface. SciGame contains a Historical collection of canonical laws and a live Temporal collection curated from papers published after model knowledge cutoffs. It evaluates agents in candidate-guided Identification and open-ended Discovery settings. Each setting runs under Prior-only, Fixed, and Adaptive protocols. Experiments on five frontier models yield three findings. First, familiar priors hurt performance once the governing relation is altered. Second, adaptive experimentation helps the strongest models but harms weaker ones under limited budgets. Third, good explanations can hide quantitative failures. Recovering the exact symbolic form remains the main bottleneck, even when agents interpret evidence correctly. Current agents can investigate systems actively, but inducing exact laws in unfamiliar settings remains an open challenge.

Across human history, knowledge has undergone several transitions, from orality to writing and then to digital archives. At each transition, humans established new conventions to preserve knowledge for the new era. Nowadays, as research agents are increasingly deployed for laboratory experiments, knowledge is crossing another boundary: from records written for humans to those for machines. In this work, we introduce VerifiedLab, a multi-agent harness framework for this human-to-machine transition of knowledge. VerifiedLab transforms research output from passive artifacts into active systems that accelerate downstream use, adoption and prediction of future trends. In VerifiedLab, a team of AI scientists with distinct roles reasons over records verified field by field: a principal investigator plans, a librarian computes the evidence directly from the records, specialists draft, and a gate and a critic check each claim against the measured reliability of its data before the harness decides whether to release or withhold it. We demonstrate the efficacy of VerifiedLab through extensive experiments on perovskite solar cells: VerifiedLab recovers known physical trends, anticipates devices reported a year later, and lets AI scientists inherit the literature together with a measure of how far it can be trusted. By scaling the inference-time verification compute, VerifiedLab introduces a new pathway for knowledge dissemination and machine harness for AI co-scientists.

Cooperative multi-agent reinforcement learning (MARL) enables teams of agents, such as robots and autonomous vehicles, to learn coordinated behavior, but what such teams learn is largely determined by their reward function. Because task success alone provides too little guidance for learning, intermediate rewards are usually designed by hand, and this effort must often be repeated whenever the environment or the team changes. Large language models (LLMs) can automate the effort by writing reward functions as code, yet the resulting rewards are typically designed and tested in a single setting. To enable their reuse across settings, GARDEN (Generalist Autonomous Reward Design through Evidence Navigation) is proposed, in which one reward program is developed for a family of related scenarios and transferred, unchanged, to scenarios that were not available during its design. During development, environment code, training records, and agent behavior are investigated by an LLM reward engineer, and draft rewards are tested in trial training runs, while the learning algorithm and the rule for selecting the final reward are held fixed. On unseen scenarios of the Overcooked and SMAX benchmarks, task returns were increased by 58.2% and 52.5% over the default rewards on two kitchen layouts, and the mean win rate in two combat scenarios was raised from 51.0% to 89.9%. Under matched candidate and probe budgets, this exceeded the 84.2% and 79.2% obtained with fixed diagnostic feedback and Eureka-inspired reward reflection. These results show that a single autonomously engineered reward can be reused effectively across related cooperative scenarios.

Large language models (LLMs) have emerged as potential tools for numerical optimization. Despite promising results, their intrinsic optimization capabilities and limitations remain poorly understood, particularly in high dimensions. This paper investigates the ability of LLMs to exploit mathematical problem structures and generate feasible, near-optimal solutions. We develop an evaluation framework that distinguishes solution quality from execution failures, including numerical inaccuracies, constraint violations, and output-format errors. Using representative problems from convex optimization, fractional programming, and constrained projection, we assess LLM performance against deterministic reference solvers across different dimensions. Our results reveal that LLMs can capture useful optimization structures and produce near-optimal solutions, even when their outputs fail to satisfy numerical or formatting requirements. However, reliability deteriorates with increasing problem dimension and depends strongly on the underlying optimization structure. To further investigate the limitations, we deploy an iterative refinement mechanism that provides targeted feedback on feasibility and solution quality. The results demonstrate that iterative refinement substantially improves reliability and extends the range of problem dimensions over which LLMs produce valid solutions. Nevertheless, persistent numerical and structural errors limit their applicability as standalone optimizers. These findings provide insights into the capacity and limits of LLM-based optimization and motivate hybrid approaches that combine LLM reasoning with deterministic numerical verification.

Latent multi-agent systems reduce inference cost by keeping intermediate reasoning and inter-agent communication in continuous representations.\ Reusable skills, however, are commonly supplied as text or stored as layer-wise Key-Value states or skill-specific model parameters, introducing repeated processing or substantial per-skill storage.\ We introduce \textbf{SkillMem}, which represents each reusable skill as a small external latent memory and makes it directly available during latent multi-agent reasoning.\ A role-conditioned cross-attention reader allows the same memory to guide different agents according to their roles and current hidden states.\ We train the skill memories and shared reader through behavioral distillation, matching the change in token predictions induced by a textual skill while keeping the backbone frozen.\ For our experiments, we obtain textual skills from paired successful and failed multi-agent trajectories, although existing skills can also be compiled directly.\ Across six benchmarks and Qwen3 models ranging from 4B to 14B parameters, SkillMem provides strong gains for smaller backbones while preserving efficient latent multi-agent inference across model scales.\ We use the same $8\times256$ memory configuration across all evaluated model scales, requiring 9.6\,KB of skill-specific storage per skill, including its retrieval embedding.\ For a library of 100 skills, SkillMem reduces total storage by more than $38\times$ compared with alternative skill representations and achieves up to a $4.23\times$ end-to-end inference speedup.\ These results show that compact procedural memories can be reused without sacrificing the efficiency of latent multi-agent reasoning.

Rounded capacity cuts (RCCs) are among the most effective cuts for capacitated vehicle-routing relaxations, but exact separation is computationally prohibitive at scale. Existing heuristic and neural separators can be fast, but often fail to produce sufficiently many effective cuts within practical separation budgets. We introduce SAG-Sep, a separator that scores node membership in RCC-inducing subsets across vehicle-count levels from a fractional LP solution in a single encoder forward pass, avoiding the iterative predict-and-coarsen inference used by NeuralSEP-style separators. SAG-Sep encodes the LP solution as a Sparse Augmented Graph (SAG), adding flow-informed multi-hop edges that expose long-range routing structure without densifying attention. Beyond LP-aware sparse encoding, we observe that high-value exact RCCs are typically nested, organizing into onion-like subset families, and SAG-Sep's node scores track subset nesting depth. This motivates Onion-Sweep (OS), a training-free decoder that ranks customers by their scores, evaluates the RCC violation of every prefix of the ranking in one vectorized pass, and emits the most violated prefix of each replica together with a cross-replica family of nested proposals, converting independent node scores into structurally coordinated RCC candidates without a tuned threshold or probing band. On the NeuralSEP benchmark with N=1000 customer instances, SAG-Sep with OS reduces the average root gap by 17.7% relative to the strongest NeuralSEP baseline, while the single-pass SAG-Sep separator achieves a roughly 30-fold per-iteration neural separation speedup. Our findings suggest that LP-aware sparse encoding and the nested structure of high-value cuts are structural patterns worth leveraging in neural separation more broadly.

Time-series data play an important role in various tasks, such as weather forecasting, web traffic analysis, and speech recognition. Recently, pioneering studies have increasingly adopted Joint-Embedding Predictive Architecture (JEPA) based encoders to learn time-series representations through latent-space prediction. However, they often overlook the underlying factors governing transitions between time-series observations. In this work, we hypothesize that explicitly modeling the information behind changes in time series can facilitate time-series modeling. Building on this hypothesis, we propose ACTS-WM, a self-supervised latent action-conditioned world model framework for time-series analysis. First, we introduce a latent action modeling approach that infers the underlying information behind time-series transitions as latent actions, without requiring explicit action annotations. We then develop a JEPA-based encoder and an action-conditioned world model to predict target time-series representations conditioned on the learned latent actions and task information. We evaluate ACTS-WM across 4 representative tasks on 22 datasets from diverse domains, covering univariate and multivariate forecasting and classification. Comprehensive experiments demonstrate that ACTS-WM can improve downstream performance over pretrained JEPA-based encoders by learning complementary representations.

Multi agent LLM systems raise the risk of collusion among agents pursuing objectives hidden from their overseers. When potential partners identities are unknown, collusion first requires initiation: agents must discover one another, often through a public channel observed by ordinary agents and monitors. Strategic vagueness offers a way to navigate this mixed audience, conveying intent to a prospective partner while preserving a plausible ordinary interpretation. We introduce a mutual identification game in which two schemers must identify each other through public interaction, while jointly solving an ordinary task. We find that some models discover successful handshake strategies without explicit signalling guidance, using task compatible exchanges that monitors often fail to flag at low false positive rates. To characterize the communicative capabilities underlying these exchanges, we study a controlled coordination setting. We find an asymmetric response to message explicitness: as messages become more indirect, monitor punishment can fall faster than partner coordination, creating a profitable intermediate regime of strategic vagueness. Together, these findings illustrate a Blackstone style evidential asymmetry: communication sufficient for coordination between agents may remain insufficient for confident detection by monitor, opening a new surface of collusion.

Large language model (LLM) inference has become an operational workload that fleets must provision, monitor, and bill against latency service-level objectives (SLOs). A single throughput number cannot answer the questions operators actually face, because generative serving couples two phases with very different service characteristics, continuous batching reshapes the effective service process, and user-visible tail latency can degrade long before aggregate throughput saturates. We present GAUSS, a workload-aware analytical and Monte-Carlo simulator that predicts the full distributions of time-to-first-token (TTFT) and time-between-tokens (TBT), together with the sustainable request rate, for a given workload, engine, and accelerator. GAUSS pairs calibrated, phase-specific latency surfaces for prefill and decode with a Lindley-recurrence model of queueing under stochastic arrivals and estimates performance by ensemble Monte-Carlo sampling rather than by replaying a single trace. This lets GAUSS reason about workloads, routing policies, and hardware that have never been deployed, and to capture the latency tails that trace-driven simulators systematically under-sample. We describe the model and show how it supports operating-point selection, traffic segmentation, and shortest-remaining-processing-time scheduling capacity questions that trace-driven simulators cannot address.

PairFormer-based all-atom models demonstrate strong capabilities in biomolecular binder design. However, existing methods often use all-atom representations for modality-specific binder design, such as protein and peptide design. In contrast, recent geometric latent diffusion models pursue unified binder design by representing molecular blocks from different modalities in a shared latent space. These complementary strengths motivate us to investigate how to generalize the use of pretrained all-atom representations to unified binder design. We introduce PairFuse, a parameter-efficient method that enables this generalization through representation fusion. Specifically, we use lightweight layerwise adapters to integrate pretrained all-atom representations into an independently pretrained geometric latent diffusion model. Our adapters integrate pretrained all-atom representations through two mechanisms: enriching the diffusion model's molecular block features and incorporating pairwise structural information into the geometric diffusion model's denoiser. Theoretically, we prove that the augmented denoiser preserves geometry and pairwise attention fusion provides additional attention expressivity beyond feature-only fusion. Experimentally, across antibody, peptide, and small molecule binder-design benchmarks, PairFuse achieves stronger performance than the pretrained diffusion model and existing representation-fusion baselines.

Earthquake sequences are stochastic on two separated timescales: the seismic waves of any individual earthquake propagate over seconds to minutes, while the earthquakes themselves arrive as a sparse, self-exciting occurrence process, with inter-event intervals from seconds to months. Respecting both scales in one training objective is the open challenge. Foundation models (FMs) for seismic waveforms instead import their learning targets from audio and vision, where every target is computed from a single fixed-length window. Such a window holds one realization of the wave propagation (an earthquake trace), never the occurrence process that decides when earthquakes happen. We develop nine model designs sharing one Mamba encoder-decoder, trained on a fixed corpus of 3M windows, varying the self-supervised objective, quantization bottleneck and mixer; the two key comparisons are architecture-matched, varying the objective alone. Each checkpoint is evaluated frozen on three downstream tasks. Varying the target redistributes downstream skill rather than raising it, and no checkpoint wins everywhere. Adding masking pressure to our strongest classifier re-specializes it rather than making it a generalist: it gains 6.9 points of few-shot P-wave picking precision and loses 9.6 points of few-shot classification accuracy. By the standard imported from language and vision, these models are feature extractors rather than FMs. Controls on capacity, architecture, and domain proximity do not explain why skill moves rather than rises. Against task-specific baselines the advantage is regime-dependent: our checkpoints beat them few-shot on classification and ground-motion prediction, and trail them on full data. We argue the binding constraint is the target, and that settling the question requires a learning target defined across events, the timescale that makes seismicity stochastic in the first place.

This paper focuses on algorithmic bias of machine learning and artificial intelligence applications in healthcare information systems. Based on the quantitative data and qualitative comments from a survey and interviews with healthcare professionals, who have different job roles (e.g., clinical vs. administrative), this study provides findings about the relationships between algorithmic bias, perceived fairness, and the intended acceptance and adoption of ML algorithms and algorithm generated outcomes. The results suggest that the opinions of healthcare professionals toward the causes of algorithmic bias, the criteria of algorithm assessment, the perceived fairness, and bias mitigation approaches may vary depending on their job roles, perspectives, tasks, and the algorithm characteristics. More research is needed to investigate algorithmic bias to ensure fairness and equality in healthcare.

Advancing language models for specialized scientific applications requires evaluations of scientific fluency, physical reasoning, and quantitative analyses. Using inertial confinement fusion (ICF) as a test case, we have developed an expert-curated benchmark of 250 questions spanning qualitative reasoning, numerical calculation, symbolic derivation, and equation formulation. Responses have been evaluated using deterministic numerical and symbolic checks together with language-model judges. We compared a variety of out-of-the-box closed-source and open-source large language models with Prospero, a model specialized through combined domain-adaptive pretraining and supervised fine-tuning for ICF. We also conducted initial retrieval-assisted evaluations. Prospero improved its aggregate score by 10.7 percentage points over its base model, with gains across all question types. Frontier models achieved the highest aggregate scores, but critical physics errors persisted, particularly in numerical calculations. We note that domain-specific performance can improve through specialization, but benchmark gains alone do not establish scientific reliability. Retrieval improved performance for a smaller model but yielded no statistically significant improvement for a frontier model. Two scientific stress tests further illustrate useful capabilities and the need for expert assessment of physical assumptions and model validity. Based on these findings, we recommend evaluating existing domain capabilities and retrieval benefits before committing to substantial additional specialization.