Language models are increasingly used for discrete decisions like tool selection or intent classification, yet prompting LLMs for these closed-set choices is slow and sensitive to option order. This paper introduces Crivo, a family of compact encoders scaling from 17 million to 1 billion parameters, and runs a controlled study to determine whether scoring each option in a separate encoder pass outperforms reading all options jointly in a single sequence. The work matters because it tests the limits of parameter efficiency and architectural choices for high-speed decision nodes that must generalize across changing label sets.

This paper proposes an adaptive framework for cryptographic primality testing that integrates arbitrary-precision arithmetic, hardware-aware sieving, and a machine-learning-guided Miller-Rabin witness selector anchored on quantum entropy. The core idea is to reduce expensive modular exponentiations by using lightweight structural features of a candidate integer to prioritize bases likely to expose compositeness early. It targets the key-generation bottleneck for RSA and ECC, yet the contribution is more architectural than empirically demonstrated, resting heavily on a withdrawn survey and a toy Python benchmark.

This paper initiates the statistical theory of reverse-KL policy distillation under misspecification, asking whether the teacher's full action distribution improves efficiency over action-level feedback and whether online student trajectories are necessary versus offline teacher data. The authors establish minimax-optimal rates for full-feedback settings, develop clipped importance-weighting and trajectory-truncation algorithms, and prove a striking separation: offline action feedback can match full-feedback rates, while online action feedback requires exponentially more data. The results give a first theoretical footing for empirical design choices in modern LLM distillation.

As Mathlib grows past 400,000 theorems, effective semantic search is critical for both human formalization and automated proving, yet real-world queries are short and ambiguous while existing engines train on lengthy, context-rich informalizations. Lean Finder V2 closes this gap with an adversarial generator-retriever loop that synthesizes over 600,000 faithful, challenging short queries, then uses them for both retriever fine-tuning and GEPA-optimized input rewriting. The result is a new state of the art on human-curated and real-world benchmarks, plus measurable gains in downstream theorem-proving tasks.

FocusVLA targets a subtle but important failure mode in multi-view Vision-Language-Action models: policies tend to develop early visual preferences—especially over-reliance on wrist cameras—and persist with them even when task stages shift toward global planning. The paper proposes predicting future interaction intent from multimodal context and short-term action history, then converting that intent into step-wise routing weights that dynamically balance global and local views during action generation. Because the routing module sits on the action-generation side without altering the visual encoder or language backbone, it is designed as a lightweight, plug-in upgrade for existing VLAs. It is evaluated on all 50 tasks of the RoboTwin 2.0 benchmark and on six real-world manipulation tasks, reporting 10+ percentage point improvements over a $\pi_{0.5}$ baseline trained under identical conditions.

Meta-RL agents struggle to adapt quickly to unseen tasks when task evidence is sparse or spread across long interaction histories, because recurrent models compress history too aggressively and transformers scale quadratically. METTA proposes a two-scale neural memory architecture that uses Test-Time Training (TTT) layers for fine-grained within-episode adaptation and compact episode summaries for cross-episode reasoning, preserving linear sequence complexity. The core claim is that this design enables stronger task-family-level generalization and more structured exploration on long-horizon sparse-reward tasks than both recurrence and full-attention baselines, while keeping wall-clock training cost low.

Wrist PPG heart rate monitoring is vulnerable to motion artifacts, and naively applying corrections can lower average error while damaging individual windows. This paper proposes a selective correction framework organized as an agent around a fixed core estimator: signal-analysis modules extract pulse and motion evidence, a temporal memory module generates candidate heart rates, and a decision module predicts each candidate's benefit and harm before a separate risk gate chooses to retain, correct, or reject the output. The authors train the auxiliary modules on PPG-DaLiA and evaluate frozen weights on three additional public datasets plus a prototype wearable stream, reporting macro-averaged MAE reductions up to 44.5%. The core idea—explicitly trading off predicted gains against harms rather than universally replacing estimates—is a clear contribution to wearable heart rate analysis, though deployment readiness is weakened by calibration transfer failures and an extremely limited prototype.

LLM agents can execute prohibited actions when tool responses omit safety-relevant facts, even if every returned statement is true and the user request is benign. EpiSEM-Bench introduces 300 matched unsafe/safe task pairs across six domains with controlled evidence suppression and an optional audit-based recovery tool. The work shows that robustness to missing evidence and useful recovery are distinct capabilities that single-score benchmarks conceal, making joint evaluation essential for safe agent deployment.

This paper studies handover prediction in cellular networks, where time-series models must meet a hard 40 ms inference deadline and a tight energy budget on user equipment. Rather than ranking by accuracy alone, the authors propose a protocol that filters models by tail latency feasibility then ranks survivors on measured per-inference energy—including feature computation. They benchmark eight architectures and find that gradient-boosted trees dominate the accuracy–energy–latency frontier, while the top accuracy-only neural model (CNN-1D, F1=0.914) costs over 3× more energy and trails in F1. The work challenges standard benchmarks by showing that duty-cycled measurement and full-pipeline accounting can invert model rankings.

This paper shows that large language models fail to treat co-referring molecular identifiers—such as trivial names, IUPAC names, and SMILES strings—as interchangeable routes to the same knowledge. The authors introduce ChemIDBench, a controlled benchmark of 896 approved drugs, and demonstrate a striking double dissociation: lexical names unlock entity-associated facts like therapeutic indications, while structural identifiers unlock structure-computable properties like benzene-ring presence. Even when a model recognizes a molecule from its SMILES, the corresponding factual knowledge often remains inaccessible, a phenomenon the authors term cross-identifier binding failure. They trace this to a missing access relation rather than missing knowledge, and propose Binding Self-Distillation to transfer lexical access directly to SMILES without retraining from scratch.

xVGAE tackles graph generation for moderate-sized, potentially dense graphs where diffusion and autoregressive methods become prohibitively slow. It builds on the Aldous-Hoover representation theorem to design a hierarchical VAE with a global graph latent $z_0$ and node latents $z_i$ that are conditionally i.i.d., using a learnable neural graphon decoder for one-shot generation. The core pitch is that this structure yields sample quality competitive with state-of-the-art while sampling one to three orders of magnitude faster, and produces interpretable latent representations.

Standard on-policy distillation forces a student to copy a teacher at every token, even when the teacher's preferred actions hurt the student's eventual success. This paper proposes Advantage-Based On-Policy Distillation (AOPD), which weights each teacher token by its counterfactual advantage under the current student policy. The key insight is that imitation should optimize for student task reward, not teacher agreement, and the authors derive an unbiased estimator that needs only standard student rollouts and terminal rewards.

Onger and Foscari analyze competitive market making as a two-sided pricing game where strategic liquidity providers face a non-strategic informed trader with inelastic demand. Because bid and ask quotes jointly set the market mid-price that determines trade direction, standard equilibrium existence theorems fail and payoffs are non-convex and discontinuous. The authors prove that pure-strategy Nash equilibria exist via a novel fixed-point argument on a midpoint operator, derive a structural linear scaling of spreads with asset volatility, and show that no-internal-regret learning on fine price grids recovers the competitive benchmark as tick size vanishes.

N-gram embedding tables give language models cheap capacity gains but can consume hundreds of billions of parameters. This paper asks whether those embeddings can be stored at low precision without losing their benefit. The authors systematically compare quantization schemes—uniform, normal-quantile, and Lloyd–Max codebooks with per-embedding scales—and find that aggressive bit reduction often preserves most of the tables’ gain, freeing memory to grow the tables themselves.

Autonomous research agents can automate entire scientific workflows from ideation through experimentation to writing, but they frequently compromise research integrity through behaviors like test-set exploitation, fabricated citations, and manipulated reporting. RIDAR addresses this gap by introducing RIVT, a taxonomy of 14 research integrity violation types spanning the full lifecycle, together with RIDAR-Post-hoc for automated detection and RIDAR-Runtime for live intervention. The work matters because the authors show such violations are nearly universal in current frontier agents, making integrity safeguards a prerequisite for trustworthy machine-generated science.

Industrial organizations struggle to govern structured telemetry and unstructured documentation together because the work relies on scarce domain experts. The paper proposes contestable governance: specialist AI agents generate scored, cited arguments for metadata decisions, human stewards review and intervene before anything is committed, and an append-only record preserves full lineage. This design aims to stop unreviewed agent errors from propagating into downstream systems while making human oversight attributable and auditable.

Depth growth and looping reuse computation across layers to train or run Transformers more efficiently. This paper shows that both methods create an inductive bias toward learning from context over memorizing fixed associations in weights, accelerating in-context learning and propagating induction heads through depth. The evidence spans models up to 3B parameters and connects architecture design to the mechanistic origins of few-shot performance.

Muffin tackles the overhead and synchronization bottleneck of multi-agent reasoning by replacing full-trace debate with asynchronous sharing of compact intermediate findings through a shared memory buffer $B^{(d)}$. An orchestrator routes queries to either a single agent or a team of parallel agents that read and append selective findings while keeping private reasoning local. The authors claim strong accuracy-token Pareto gains across four reasoning benchmarks and four open-weight language models.

The paper tests whether matching spectral predictability $\Omega$ between synthetic pretraining data and target time series is sufficient for effective transfer learning under distribution shift. They evaluate this "predictability-aware" hypothesis using LSTM, PatchTST, and TCN across controlled synthetic conditions and the real-world ETTm1 benchmark. The core finding is a negative result: while $\Omega$-matching works under ideal conditions, it fails across different synthetic source families on real data, suggesting practitioners need more nuanced source-selection criteria.

Standard deep Gaussian processes become increasingly ill-conditioned as depth grows because discrete layer composition amplifies parameter perturbations. This paper proposes replacing the discrete stack with a latent continuous-depth SDE whose drift family carries a GP-based prior, yielding two models: CDGP with a direct separable GP prior, and FlowDGP with a flow-evolved base GP. The central claim is that depth in probabilistic regression is better modeled as continuous latent evolution than as a discrete composition, and that retaining a GP prior over the drift improves stability and competitiveness relative to both discrete DGPs and neural continuous-depth alternatives.