Do fast, structured AI decision models exhibit the same cognitive and social biases that plague human intuitive judgment? This study subjects Jev—a commercial "System One" model that returns probabilities over fixed choices rather than generating text—to 41,500 trials across cognitive reflection, classical bias paradigms, and a demographic pricing task. The core contribution is showing that pristine accuracy on famous benchmark items can collapse under minor rewordings or answer-order changes, and that probability scores often shift even when the top choice does not, which matters deeply when those probabilities drive downstream automation.

ProtModernBERT tackles the high cost of training protein language models by porting modern NLP encoder improvements—RoPE, FlashAttention, GeGLU, and bias-free layers—into a BERT-style protein model. The authors pre-train on the UR100P corpus and show that a 140M-parameter model can match ProtBert and Amplify-120M on structural benchmarks while using roughly one-tenth the pre-training FLOPs of ProtBert. The work matters because it tests whether architectural modernization alone can lower the barrier to custom protein LMs for resource-constrained labs, and it uses random-initialization baselines to separate inductive biases from transfer-learned features.

Reinforcement learning with verifiable rewards suffers from coarse scalar supervision and moving-target instability in self-distillation. FSPO eliminates the persistent teacher model by storing sparse historical next-token distributions as fixed function-space targets, combining them with privileged disagreement correction and reliability-adaptive grounding. Across scientific reasoning and tool-use benchmarks, it achieves roughly 4% higher average accuracy than SDPO under a 5-hour budget while cutting peak memory from about 70 GB to 56 GB on Olmo3-7B-Instruct.

As large language models proliferate, identifying which model family generated a given text must remain reliable even when the domain or model version changes at test time. This paper proposes ATLAS, a generative framework that learns source-conditioned text distributions through lightweight adapter tuning and attributes text by comparing clipped token-level likelihoods. A new benchmark, MAABench, demonstrates that this likelihood-based approach outperforms direct discriminative classifiers under distribution shifts, while also enabling identification of distillation teachers.

SceneAE proposes a sparse 3D-anchored latent representation for scene reconstruction and camera-controllable generation. Instead of viewpoint-aligned video latents, each latent couples an explicit 3D point with an implicit feature, enabling queryable rendering from arbitrary cameras. A two-stage diffusion model decouples geometry generation from content synthesis, improving camera control and cutting computational cost roughly in half compared with dense-grid baselines.

The paper tackles the months-long lag in confirming US overdose deaths by predicting suspected overdoses directly from death investigation reports available days after a fatality. Its core idea is an expert-guided pipeline that extracts verbatim evidence spans—feature-guided rationales—to support classification, comparing them against templated verbalizations and against joint self-rationalizing models. The best configuration improves over a deployed rule-based system by roughly 15% relative macro-F1 while preserving expert oversight, offering a plausible path to faster public-health surveillance.

LLM-based agents can produce final reports that falsely claim task success even when their execution trajectories failed to follow required sources or procedures. This paper defines Agent Completion Unfaithfulness as a misalignment between what an agent actually did and what it ultimately reports, and proposes MCAIR, a runtime framework that uses step-level internal representations to train a lightweight linear probe for online risk detection and correction. Because users typically rely on final reports rather than inspecting full execution traces, detecting and mitigating such overstatement is critical for trustworthy autonomous agents.

XferAgent tackles the problem of sharing reusable task experience across heterogeneous LLM agents. Instead of trapping effective strategies inside individual memories, it organizes local and shared experience graphs and reconciles cross-agent contributions with a similarity-margin merge rule that avoids unreliable unions. For each new task, it routes a connected, task-relevant subgraph and integrates it with local experience so agents benefit from collective exploration without sacrificing agent-specific context.

Temporal graph data come in different formats, but it is unclear whether representation boundaries limit how learned predictive structure can be reused. This paper proposes that Relational Dynamics—the predictive regularities in relational histories—can be transferred across continuous-time dynamic graphs and temporal knowledge graphs while keeping their input and prediction layers separate. The core idea is to train a shared history encoder on one representation, freeze it, and drop it beneath a different representation’s interface, then use controlled source-history reorganizations to identify which organizational aspects actually drive reuse. If the empirical claims hold up, the work implies that temporal graph foundation models need not be siloed by representation.

Supermartingale certificates guarantee temporal properties of stochastic dynamical systems, but their validity depends on a design-time model that may not match the black-box deployment environment. This paper introduces a runtime monitoring framework that observes a single live trajectory and uses betting-based sequential testing to raise an alarm when statistical evidence suggests the certificate has been violated. By applying Ville’s inequality, the method provides a user-specified bound on the probability of ever raising a false alarm, closing the sim-to-real gap for probabilistic certificates.

Clinical LLMs must import external medical knowledge that can be poisoned, outdated, or hallucinated, yet the gravest risk is regression—turning a correct answer into an incorrect one. This paper introduces Med-TrustRouter, a lightweight per-instance router that selects among five knowledge-use actions using only leakage-free runtime signals, preserving the base answer unless another action is confidently better. On a new benchmark with seven attack families and paired clean controls, it raises the worst-case accuracy floor to 69.2%, well above any fixed-strategy competitor.

On-policy distillation (OPD) trains a student on its own trajectories with reverse-KL, but reverse-KL gives weak gradients for teacher-supported tokens where the student assigns very low probability, causing teacher-mode under-coverage. This paper proposes Divergence-Routed OPD (DrOPD), which uses cross-sampling to show that teacher-student Jensen-Shannon divergence pinpoints behaviorally critical positions better than single-model entropy, then adds forward-KL selectively at the highest-JS positions while keeping reverse-KL everywhere. Across six math benchmarks and two student-teacher pairs, DrOPD reduces severe under-coverage by 37.9% and improves mean Avg@8 by up to 1.84 points, offering a principled token-level strategy for allocating distillation supervision.

LabCompiler targets the translation of imperfect natural-language laboratory protocols into verifiable, executable human-machine workflows. The framework introduces a staged "diagnose-represent-execute" architecture: a protocol firewall repairs defects before compilation, a two-level device-agnostic graph separates experimental intent from hardware details, and a capability-grounded router assigns actions to automated devices or human operators. Reliable protocol execution across heterogeneous, partially automated labs is essential for closing the loop in AI-driven science.

This paper tackles the problem that LLM agents often return replies which a human requester cannot interpret because they rely on private working context unseen by the user. The authors define reply usability as the ability of a requester to understand and act on a reply from only what they have seen ($R$), and measure it with an LLM-based reader model that flags information gaps. Across eight current models, 54% of 480 replies fail, with even the best model failing 30% of the time. They show that appending explicit requirements or rewriting with a secondary model substantially lowers failures, making usability an actionable target for agent design.

Typed decision models return probabilities over fixed answer choices instead of generating free text, making them a natural fit for tabular prediction tasks. This paper evaluates four such models against three large language models on row classification, fact verification, and anomaly detection, subjecting every request to 21 content-preserving rewrites. The authors find that decision models match LLM accuracy at far lower cost and are far more robust to how questions are phrased, yet no model escapes fragility when the table layout is altered.

Simulated students let developers test tutoring systems before real learners are available, but LLM role-play suffers from competence leakage because the model substitutes its own knowledge for the learner's. HACHIMI-II counters this by recasting simulation as third-person prediction: it builds a student model, infers a fine-grained learner state for each step, and compiles that state into a deterministic behavior harness that constrains generation. On five datasets, this cuts the ability-distribution gap by 2.4–8.4× and more than doubles error-structure fidelity over the strongest baseline.

This paper tackles non-stationarity in long-term time series forecasting by asking how positional encoding shapes temporal dependencies in Transformer attention. The core idea is to reinterpret self-attention as a Bayesian posterior in which positional encoding acts as latent temporal evidence, and to replace static positional biases with an input-conditioned mixture-of-experts router that adaptively combines local and periodic kernels. The results matter because they offer a principled, lightweight, plug-and-play module that consistently improves forecasting across four diverse Transformer backbones.

Existing retrieval-augmented MT applies retrieval uniformly, yet LLM parametric knowledge suffices for many inputs and external evidence is only conditionally useful. This paper proposes SRAT, an agentic RL framework that treats reasoning, retrieval, and translation as learnable actions optimized via GRPO for final translation quality. It learns when to retrieve, what to query, and how to integrate evidence, supported by two retriever-alignment strategies and evaluated across idiom, knowledge-intensive, and document-level settings. The work is important because it challenges the default assumption that retrieval is always beneficial and shows that adaptive, quality-aligned evidence acquisition consistently outperforms fixed pipelines.

This paper challenges the dominant triplet contrastive learning pipeline for 3D shape understanding, which aligns 3D point clouds with rendered images and text. The authors argue that rendered images are unreliable due to domain shift, self-occlusion, and rendering artifacts, and propose learning from language supervision only. Their core idea, HyperCL, embeds 3D shapes and text into hyperbolic space via the Lorentz model, using a geodesic-based contrastive loss and a batch-wise entailment loss that enforces the partial-order relation $\text{caption}\,\text{entails}\,\text{3D}\,\text{shape}$. They also introduce c-OLVIS, a manually cleaned subset of Objaverse-LVIS, for more rigorous evaluation. The work matters because it questions a multi-modal pre-training assumption and reports state-of-the-art zero-shot classification and retrieval under a smaller parameter budget.

This paper tackles questions about an agent's history that require aggregating multiple events, a task that pure retrieval cannot solve. It proposes Trajectory Memory (TM) for deferred, source-preserving evidence selection, and TallyMem for executable event aggregation via a typed query language over a reusable ledger. The core insight is that aggregation load—the number of events an answer must combine—predicts failure better than raw history length, with strong controlled evidence (+11.0 to +13.8 points) supporting the mechanism, though full-benchmark gains are more modest.