Academic Intelligence · Curated Daily

探索全球前沿学术脉络

AcademicHub 汇聚顶级期刊与预印本平台的实时文献。定制您的专属科研雷达,利用大语言模型自动生成交叉领域文献分析简报。

01.
arXiv (CS.AI) 2026-06-19

ENPIRE: Agentic Robot Policy Self-Improvement in the Real World

arXiv:2606.19980v1 Announce Type: new Abstract: Achieving dexterous robotic manipulation in the real world heavily relies on human supervision and algorithm engineering, which becomes a central bottleneck in the pursuit of general physical intelligence. Although emerging coding agents can generate code to automate algorithm search, their successes remain largely confined in digital environments. We conjecture that the missing abstraction to automate robotics research is a repeatable feedback loop for real-world policy improvement: reset the scene, execute a policy, verify the outcome, and refine the next iteration. To bridge this gap, we introduce ENPIRE, a harness framework for coding agents that instantiates this physical feedback routine with four core modules: an Environment module (EN) for automatic reset and verification, a Policy Improvement module (PI) that launches policy refinement, a Rollout module (R) to evaluate policies with one or multiple physical robots operating in parallel, and an Evolution module (E) in which coding agents analyze logs, consult literature, improve training infrastructure and algorithm code to address failure modes. This closed-loop system transforms real-world manipulation learning into a controllable optimization procedure, minimizing human effort while allowing fair ablations across training recipe and agent variants. Powered by ENPIRE, frontier coding agents can autonomously train a policy to achieve a 99% success rate on challenging, dexterous manipulation tasks, such as organizing a pin box, fastening a zip tie, and tool use, a process that further accelerates when we dispatch an agent team on a robot fleet. Our results suggest a practical and scalable path toward deploying coding agents to autonomously advancing robotics in the physical world.

02.
bioRxiv (Bioinfo) 2026-06-18

Robust Conditional Diffusion with Noisy Templates for Antibody Sequence-Structure Design

Antibodies specifically recognize antigens and play a central role in therapeutic discovery. Designing antibodies for a given antigen remains challenging because antigen-antibody complex data are limited, whereas the sequence and conformational spaces of complementarity-determining regions (CDRs) are large. Retrieved CDR templates from databases or candidate libraries can narrow the design space and improve controllability, but retrieval for novel antigens is often sparse and imperfect; treating retrieved templates as hard conditions can bias the denoising process and cause negative transfer. To address this problem, we propose Robust Conditional Diffusion with Noisy Templates for antibody sequence-structure design (NT-ABDiff), a joint diffusion framework that treats candidate CDR-only templates as optional and potentially unreliable conditions. NT-ABDiff uses reliability-aware template modulation to estimate the context-conditioned usefulness of each candidate and to adaptively reweight and fuse multiple templates during conditioning. We further train the model with mixed-quality and corrupted templates as conditional perturbation regularization, encouraging the denoiser to exploit informative templates while remaining stable when templates are uninformative. Experiments under controlled template shifts and a train-set retrieval evaluation show that NT-ABDiff improves CDR-H3 sequence recovery and structural accuracy over strong baselines, while retaining robustness to missing, mismatched, and corrupted templates. Under a stringent random-template CDR-H3 evaluation, NT-ABDiff improves amino-acid recovery (AAR) from 30.03% to 39.47% and reduces RMSD from 3.160 to 2.915A; with train-set retrieval candidates, it achieves 39.50% AAR and 2.76 {ring} A RMSD. Code, processed splits, {ring} configuration files, and evaluation scripts are available at https://github.com/ShiDeng7rz/NT-ABDiff.

03.
arXiv (CS.LG) 2026-06-19

Pseudo-Formalization for Automatic Proof Verification

arXiv:2605.20531v2 Announce Type: replace-cross Abstract: Reliable verification of proofs remains a bottleneck for training and evaluating AI systems on hard mathematical reasoning. Fully formal proofs, in languages like Lean, are easy to verify because they are unambiguous and modular. Most proofs, particularly those written by AI systems, have neither property, and translating them into formal languages remains challenging in many frontier math settings. We propose Pseudo-Formalization (PF), a proof format that captures the modularity and precision of formal proofs while retaining the flexibility of natural language. A Pseudo-Formal proof is decomposed into self-contained modules, each stating its premises, conclusion, and proof in natural language. To verify the correctness of a regular natural language proof, an LLM translates it to Pseudo-Formal and then verifies each module independently, an algorithm we call Block Verification (BV). We evaluate PF+BV on two benchmarks spanning olympiad and research-level mathematics, where it pareto-dominates LLM-as-judge baselines on error-finding precision and recall. To support future work, we release our research-level proof verification benchmark ArxivMathGradingBench.

04.
arXiv (quant-ph) 2026-06-11

A post-selected quantum model of cosmic acceleration

arXiv:2606.12297v1 Announce Type: cross Abstract: The origin of cosmic acceleration remains a central problem in cosmology, commonly attributed to a cosmological constant within the $\Lambda$CDM model or to dynamical dark energy. Here, we develop an alternative approach in which acceleration emerges from quantum post-selection, a standard feature of quantum theory that is not usually incorporated into cosmological modelling. While quantum theory admits both pre-selected and post-selected ensembles, quantum cosmological models are almost exclusively formulated in terms of initial conditions. Building on previous work on post-selected quasiclassical dynamics, we construct a minimal predictive cosmological model in which post-selection and coarse-graining generate effective late-time acceleration without introducing a cosmological constant, dark energy, or modifications of general relativity. The resulting expansion history is highly constrained theoretically and depends on at most two parameters beyond standard Friedmann evolution. Confrontation with type Ia supernova and cosmic chronometer data yields statistically competitive fits while naturally avoiding the coincidence problem. The model also reproduces the standard radiation- and matter-dominated behaviour at early times and predicts a present-day jerk parameter significantly different from the $\Lambda$CDM value. These results suggest that cosmic acceleration may arise as a macroscopic quantum cosmological effect rather than from additional cosmological fluids or modified gravitational dynamics.

05.
arXiv (CS.CV) 2026-06-16

CoIRL-AD: Collaborative-Competitive Imitation-Reinforcement Learning in Latent World Models for Autonomous Driving

End-to-end autonomous driving models trained with imitation learning (IL) often generalize poorly, particularly in long-tail scenarios where expert demonstrations are sparse. Reinforcement learning (RL) can provide complementary task-level supervision, but applying RL to real-world autonomous driving is challenging in offline settings without interactive simulators, where datasets are dominated by expert actions and provide limited behavioral diversity. We propose CoIRL-AD, a competitive dual-policy framework that integrates IL and RL under a unified offline training regime. CoIRL-AD decouples imitation and reward optimization into separate actors to alleviate objective conflicts, uses imagined future rollouts for long-horizon reward estimation, and introduces a competition mechanism that selectively transfers beneficial behaviors while keeping RL anchored to expert-like driving. Experiments on the nuScenes benchmark show that CoIRL-AD consistently improves robustness over strong IL-based baselines, with especially large gains in cross-city generalization and long-tail scenarios. Code is available at: https://github.com/SEU-zxj/CoIRL-AD.

06.
arXiv (CS.CL) 2026-06-16

Compositional Reasoning Depth Predicts Clinical AI Failure: Empirical Evidence Consistent with Transformer Compositionality Limits in Electronic Health Record Question Answering

作者:

Aggregate accuracy benchmarks conceal a systematic structure in how large language models fail at electronic health record (EHR) question answering: questions requiring more inferential steps produce disproportionately more errors. Motivated by theoretical results on transformer compositionality limits, we introduce a pre-specified hop-count taxonomy – the number of distinct reasoning steps required to answer a clinical question from an EHR – as a principled predictor of model failure. We annotate 313 clinician-generated MedAlign EHR question-answer pairs across four hop levels and evaluate 301 questions in a within-model ablation (claude-sonnet-4-6, zero-shot vs. extended thinking) and cross-architecture replications (gpt-4o and gpt-5.4-2026-03-05, zero-shot). All three models, spanning two providers and two OpenAI generations (GPT-4 and GPT-5), show monotone accuracy decline with hop count: Claude Sonnet zero-shot falls from 30.6% (hop=1) to 17.6% (hop=4) (Cochran-Armitage z=-2.30, p=0.011; OR per hop 0.72, 95% CI [0.56,0.92], p=0.008); GPT-4o replicates this (37.8% to 14.7%; OR 0.58 [0.45,0.75], p

07.
arXiv (CS.LG) 2026-06-17

Toward Controllable Catalyst Inverse Design via Large-Scale Autoregressive Pretraining

arXiv:2606.17445v1 Announce Type: new Abstract: Inverse design of heterogeneous catalysts remains challenging because catalyst surfaces exhibit substantial structural complexity with coupled surface-adsorbate interactions across a vast chemical space that is difficult to explore efficiently through conventional screening alone. Although machine learning-based high-throughput screening has accelerated catalyst discovery, its efficiency inevitably declines as the search space grows, motivating the development of generative models that can directly construct catalysts with target properties. Here, we present a conditional catalyst generative model based on the Generative Pretrained Transformer architecture with a numerical embedding layer that enables the generation of catalyst structures conditioned on both categorical and continuous properties within a single autoregressive framework. The model was pretrained on 133 million catalyst structures and subsequently fine-tuned on approximately 460,000 optimized structures with associated categorical properties and binding energies for conditional generation. The resulting model achieved 98% structural validity, 95% optimization validity, and high categorical condition fidelity, with a 93 % joint match rate for adsorbate type and composition. For binding energy conditioning, the match rate of approximately 20% represents a four-fold improvement over the baseline training distribution, and the generated distributions shift systematically toward the target values, enabling a 1.5 to 4-fold improvement in screening efficiency for reaction-targeted catalyst discovery without additional fine-tuning. These results show that large-scale autoregressive pre-training, combined with explicit property conditioning, provides a practical route toward controllable catalyst generation and accelerated catalysts discovery.

08.
arXiv (quant-ph) 2026-06-19

Thermodynamic Value of XOR-Game-Induced Side Information in a Szilard Engine

arXiv:2605.12044v3 Announce Type: replace Abstract: We introduce a Szilard-type thermodynamic valuation of side-information channels induced by Bell-type correlations. In each round, a two-level working system is thermalized with a degenerate Hamiltonian, so that its physical microstate is a uniform classical bit. A trusted referee embeds this bit into a finite two-player XOR game, and a correlation resource produces a compressed controller bit. The controller uses only this compressed bit as side information for feedback. The construction is formulated first for arbitrary finite XOR games. The referee encoding makes the game-winning event equivalent to correct prediction of the physical microstate. Consequently, the induced side-information channel is binary symmetric, with success probability equal to the XOR-game winning probability of the supplied behaviour. The reversible Szilard feedback value is therefore fixed by the mutual information between the microstate and the controller record. Optimizing over local, quantum, and nonsignalling behaviour sets turns the corresponding game values into local, quantum, and nonsignalling thermodynamic ceilings. The construction is an effective-channel valuation, not a claim that Bell nonlocality is thermodynamic fuel. The controller receives only the compressed prediction bit, not the auxiliary variables that define the game. The thermodynamic costs of the referee, the correlation resource, and the preprocessing are not included. When controller-memory reset is included in a full cycle, the net work is non-positive, consistently with the second law.

09.
arXiv (math.PR) 2026-06-18

Metastability for the Curie-Weiss-Potts model with unbounded random interactions

arXiv:2505.11260v2 Announce Type: replace Abstract: We analyse the metastable behaviour of the disordered Curie–Weiss–Potts (DCWP) model subject to a Glauber dynamics. The model is a randomly disordered version of the mean-field $q$-spin Potts model (CWP), where the interaction coefficients between spins are general independent random variables. These random variables are chosen to have fixed mean (for simplicity taken to be $1$) and well defined cumulant generating function, with a fixed distribution not depending on the number of particles. The system evolves as a discrete-time Markov chain with single spin flip Metropolis dynamics at finite inverse temperature $\beta$. We provide a comparison of the metastable behaviour of the CWP and DCWP models, when $N \to \infty$. First, we establish the metastability of the CWP model and, using this result, prove metastability for the DCWP model (with high probability). We then determine the ratio between the metastable transition time for the DCWP model and the corresponding time for the CWP model. Specifically, we derive the asymptotic tail behavior and moments of this ratio. Our proof combines the potential-theoretic approach to metastability with concentration of measure techniques, the latter adapted to our specific context.

10.
bioRxiv (Bioinfo) 2026-06-12

Deciphering cross-omics complexity of tissues via diagonal integration of unpaired spatial multi-omics data

Recent spatial multi-omics technologies enable the simultaneous in situ profiling of multiple omics modalities on the same tissue section; however, they face challenges in experimental complexity and high costs. This technical limitation can be circumvented by diagonal integration methods, which integrate omics data from different modalities. However, existing single-cell diagonal integration approaches overlook spatial information, causing unreliable anchoring across omics layers. Here, we introduce STAMO, a graph attention neural network model for spatially aware integration of unpaired spatial slices from different omics. Systematic benchmarking on spatial epigenome-transcriptome slices proves that STAMO outperforms the state-of-the-art methods in generating aligned embeddings and identifying consensus spatial domains across omics. We apply STAMO to integrate unpaired data from diverse spatial omics types (transcripts, epigenetics, DNA, and proteins), including slices from spatial RNA and four different epigenomic modalities, spatial ATAC and RNA slices across embryonic stages, spatial protein and RNA slices, and spatial DNA and RNA slices. In addition, the integration capability of STAMO can be further used to achieve cross-omics generation, offering a solution for exploring spatial region-specific gene regulatory mechanisms.

11.
arXiv (CS.AI) 2026-06-17

Like a Hammer, It Can Build, It Can Break: Large Language Model Uses, Perceptions, and Adoption in Cybersecurity Operations on Reddit

arXiv:2604.09998v2 Announce Type: replace-cross Abstract: Large language models (LLMs) have recently emerged as promising tools for augmenting Security Operations Center (SOC) workflows, with vendors increasingly marketing autonomous AI solutions for SOCs. However, there remains a limited empirical understanding of how such tools are used, perceived, and adopted by real-world security practitioners. To address this gap, we conduct a mixed-methods analysis of discussions in cybersecurity-focused forums to learn how a diverse group of practitioners use and perceive modern LLM tools for security operations. More specifically, we analyzed 892 posts between December 2022 and September 2025 from three cybersecurity-focused forums on Reddit, and, using a combination of qualitative coding and statistical analysis, examined how security practitioners discuss LLM tools across three dimensions: (1) their stated tools and use cases, (2) the perceived pros and cons of each tool across a set of critical factors, and (3) their adoption of such tools and the expected impacts on the cybersecurity industry and individual analysts. Overall, our findings reveal nuanced patterns in LLM tools adoption, highlighting independent use of LLMs for low-risk, productivity-oriented tasks, alongside active interest around enterprise-grade, security-focused LLM platforms. Although practitioners report meaningful gains in efficiency and effectiveness in LLM-assisted workflows, persistent issues with reliability, verification overheads, and security risks sharply constrain the autonomy granted to LLM tools. Based on these results, we also provide recommendations for developing and adopting LLM tools to ensure the security of organizations and the safety of cybersecurity practitioners.

12.
arXiv (quant-ph) 2026-06-12

Exploring Exotic Spin-Dependent Interactions Beyond the Standard Model: Theoretical Foundations and Experimental Investigations

arXiv:2606.13318v1 Announce Type: cross Abstract: New interactions mediated by novel particles propose solutions to several important questions in modern physics. Axions serve as examples of such particles; they are lightweight and interact weakly with ordinary matter. This category of particles, including those similar to axions-termed Axion-Like Particles (ALPs)-arises from diverse theoretical frameworks, such as the Peccei-Quinn mechanism addressing the strong CP problem, string theory, and spontaneous supersymmetry breaking. Given their light mass and weak coupling, ALPs are also possible candidates for cold dark matter. Introducing these new interactions mediated by novel particles not only tackles several challenges in modern physics but also raises a crucial question: Are there undiscovered interactions beyond the Standard Model? Many of the interactions predicted by these theories are spin-dependent, which is the primary focus of this review. In this review, we first outline the theoretical foundations for investigating exotic spin-dependent interactions, highlighting their importance in various models beyond the Standard Model. We examine the potential roles of new lightweight particles in mediating these interactions, which may enhance our understanding of dark matter. Relevant formulas derived from theoretical models are included to support experimental investigations. Following this theoretical framework, we conduct a detailed review of recent experimental efforts to detect these exotic interactions. A systematic review of current constraints on these interactions is presented, along with an assessment of various detection approaches.

13.
bioRxiv (Bioinfo) 2026-06-11

Revealing trajectories of multi-modal voxel-level changes in neurodegenerative diseases using latent event mapping

Neurodegenerative diseases are driven by pathological mechanisms that can be indirectly measured in vivo using multi-modal neuroimaging. However, current computational methods that aim to reconstruct trajectories of voxel-level changes in the brain are either not computationally scalable or fully interpretable, limiting their ability to reveal associations between disease progression and underlying mechanisms. Here we introduce Latent Event Mapping (LEMING), a generative unsupervised modelling technique that learns a latent map of disease events along a common pseudo-timeline of events. We apply LEMING to amyloid PET and structural MRI data from the Alzheimer's Disease Neuroimaging Initiative to reveal the first voxel-level trajectories of events in Alzheimer's disease. Notably, we show how LEMING can provide new insights into progression-dependent disease mechanisms. We find that acetylcholine receptor density is significantly positively associated with both late-stage amyloid and atrophy events, suggesting that either these receptors are targeted later in disease progression, or that amyloid does not play an active role. This has strong implications for therapeutics that target acetylcholine receptors, particularly for early-stage intervention strategies.

14.
arXiv (CS.AI) 2026-06-16

LLM-WikiRace Benchmark: How Far Can LLMs Plan over Real-World Knowledge Graphs?

arXiv:2602.16902v5 Announce Type: replace Abstract: We introduce LLM-Wikirace, a benchmark for evaluating planning, reasoning, and world knowledge in large language models (LLMs). In LLM-Wikirace, models must efficiently navigate Wikipedia hyperlinks step by step to reach a target page from a given source, requiring look-ahead planning and the ability to reason about how concepts are connected in the real world. We evaluate a broad set of open- and closed-source models, including Gemini-3, GPT-5, and Claude Opus 4.5, which achieve the strongest results on the easy level of the task and demonstrate superhuman performance. Despite this, performance drops sharply on hard difficulty: the best-performing model, Gemini-3, succeeds in only 23\% of hard games, highlighting substantial remaining challenges for frontier models. Our analysis shows that world knowledge is a necessary ingredient for success, but only up to a point, beyond this threshold, planning and long-horizon reasoning capabilities become the dominant factors. Trajectory-level analysis further reveals that even the strongest models struggle to replan after failure, frequently entering loops rather than recovering. LLM-Wikirace is a simple benchmark that reveals clear limitations in current reasoning systems, offering an open arena where planning-capable LLMs still have much to prove. Our code and leaderboard available at https:/llmwikirace.github.io.

15.
bioRxiv (Bioinfo) 2026-06-11

Machine Learning-Guided Discovery of Bacterial-Selective Membrane-Active Compounds Reveals Mechanistic Bias in Antibiotic Training Datasets

The rise of antibiotic resistance necessitates the discovery of antibacterial compounds with novel mechanisms of action (MoAs). Recent machine learning approaches have shown promise in antibacterial compound discovery, but often identify derivatives of known antibiotic classes rather than mechanistically novel compounds. Previous approaches applied Tanimoto similarity filters at the end of screening pipelines, but this method has substantial drawbacks: Tanimoto similarity can be misleading in chemical space, and post-hoc filtering does not influence what activity models learn to prioritize. Here, we present a machine learning pipeline that addresses chemical novelty upfront by employing an XGBoost-based MoA classifier to explicitly prioritize compounds predicted to have mechanisms distinct from known antibiotic classes, combined with graph neural networks for antibacterial activity and toxicity prediction. Applied to the Zinc20 database, our approach successfully identified non-toxic antibacterial compounds structurally distinct from known antibiotics. Notably, the majority of these hits exhibited membrane-targeting activity with selectivity for bacterial cells over mammalian cells, suggesting potential for next-generation membrane-active antibiotics. However, we did not identify compounds with novel protein targets. Systematic analysis revealed that this limitation stems from mechanistic bias in training data rather than model architecture. Specifically, our activity model learned to preferentially score compounds similar to specific groups in the training data, thus overrepresenting certain MoA classes including membrane-active compounds. Even substantial model architecture and training data enhancements did not overcome this constraint. Our findings demonstrate that the primary bottleneck for discovering mechanistically novel antibiotics is the scarcity of diverse, mechanistically-annotated training data. This work provides both a methodological framework for mechanism-aware screening and critical insights into data requirements for genuinely novel antibiotic discovery.

16.
arXiv (CS.LG) 2026-06-11

Querying Counterfactuals on Tissue Graphs with Supervised Disentanglement

arXiv:2606.08493v2 Announce Type: replace-cross Abstract: Tissue graph counterfactuals ask how a cell's expression would change under altered spatial neighbor contexts. Such queries are central to predicting cell behavior in tissues, but lack a unified definition, with existing methods targeting specific intervention types or treating cells as i.i.d. In this work, we first formalize tissue graph counterfactuals as a class of spatial interventions that either rewire connections between cells (edge perturbation) or modify the expression of their neighbors (node perturbation). We then introduce Cellina (https://cellina.readthedocs.io) - a framework that uses supervised disentanglement to decompose a cell's intrinsic state from its spatial context, using the latter as a conditioning input for counterfactual predictions. Across benchmarks spanning over 2.5 million spatially-resolved cells in colorectal cancer and mouse brain, Cellina outperforms spatially-informed and non-spatial competitors in in-silico graph perturbations, disentanglement, and scalability. Additionally, we show that Cellina reveals biologically distinct cancer subdomains in an unsupervised manner and enables targeted neighbor perturbation simulations.

17.
medRxiv (Medicine) 2026-06-10

Developmental Associations Linking Childhood Trauma and Early Cannabis Use to Adolescent DNA Methylation and Psychotic-Like Experiences

Background. Psychotic-like experiences (PLEs) index early risk for psychotic disorders and are consistently associated with childhood trauma, yet underlying biological mechanisms remain poorly understood. DNA methylation (DNAm) may capture the biological embedding of early adversity, while adolescent exposures such as cannabis use may modify these processes. We examined epigenome-wide associations of childhood trauma and PLEs, tested the moderating role of early cannabis use, and evaluated DNAm as a potential mediator. Methods. We analysed data from the Avon Longitudinal Study of Parents and Children (ALSPAC), a UK population-based birth cohort. Childhood trauma was assessed prospectively and retrospectively. Epigenome-wide DNAm was measured in peripheral blood at ~17 years using the Illumina 450K array, and PLEs were assessed at 18 using a structured interview. Epigenome-wide association studies were conducted for trauma-DNAm and DNAm-PLEs associations in the final sample (n = 1,457), adjusting for demographic, biological, and technical covariates. Differentially methylated regions (DMRs) were identified using DMRff, followed by functional enrichment analyses. Cannabis use at 15.5 was modelled as a moderator with multiple imputation for missing data. Mediation was tested using the Divide-Aggregate Composite-null Test (DACT). Results. Childhood trauma was associated with widespread DNAm differences, primarily at the regional level, with enrichment in pathways related to cellular stress responses. In contrast, DNAm associated with PLEs was more limited and implicated loci involved in epigenetic regulatory processes. These signatures were largely distinct, and there was no evidence supporting mediation after multiple testing correction. Incorporating cannabis use altered the pattern and extent of DNAm associations, with stronger and more significant signals observed at both CpG and regional levels, although these did not translate into evidence of mediation. Conclusion. Childhood trauma and PLEs show distinct DNAm signatures in adolescence, with trauma-related DNAm reflecting broad stress-related processes and PLE-associated DNAm implicating regulatory mechanisms. We found little evidence that DNAm mediates the trauma-PLE association. Instead, adolescent exposures, particularly cannabis use, may distinctly influence trauma-related epigenetic variation with limited detectable downstream effects on PLEs. These findings support a context-dependent model of epigenetic risk and highlight the need for larger longitudinal studies to clarify causal pathways linking early adversity to psychosis.

18.
arXiv (CS.CL) 2026-06-16

Attention, not scale, drives human-AI alignment in multimodal language prediction

Humans routinely draw on visual context to predict upcoming words. To what extent current vision-language models produce comparable behaviour is unclear. Here we placed five state-of-the-art pretrained systems side-by-side with 600 human participants in a web-based Visual-World Paradigm. On each of 100 six-second movie clips, models and participants received either text only or synchronised video and text and judged how likely a specified target word was to appear next; human eye movements were tracked throughout. Adding visual context increased model-human alignment in predictability ratings across all architectures (average Delta r = 0.18) with no impact of parameter size. When visual context was informative, transformer attention significantly increased alignment. Attention maps from two transformer models corresponded with human gaze, explaining up to 70% of the inter-participant variance when the scene contained informative cues. Notably, cross-modal attention reliably tracked anticipatory human fixations on semantic cues. These results suggest that current transformer-based vision-language models can approximate human behaviour exploiting visual context during language prediction - and that selective attention to informative cues, not sheer model scale, is the principal driver of this alignment.

19.
arXiv (CS.CL) 2026-06-12

Language Model Circuits Are Sparse in the Neuron Basis

The high-level concepts that a neural network uses to perform computation need not be aligned to individual neurons (Smolensky, 1986). Language model interpretability research has thus turned to techniques which decompose the neuron basis into more interpretable units of model computation, such as sparse autoencoders (SAEs). However, not all neuron-based representations are uninterpretable. For the first time, we empirically show that MLP neurons are as sparse a feature basis as SAEs. We use this finding to develop an end-to-end gradient-based attribution pipeline for circuit tracing on the MLP neuron basis, which surfaces causally effective neurons on a variety of tasks. On a standard subject-verb agreement benchmark (Marks et al., 2025), a circuit of $\approx 10^2$ MLP neurons is enough to control model behaviour. On the multi-hop city-state-capital task from (Lindsey et al., 2025), we find a circuit in which small sets of neurons encode specific latent reasoning steps (e.g. mapping a city to its state), and can be steered to change the model's output. This work thus advances automated interpretability of language models without imposing additional training costs.

20.
arXiv (CS.LG) 2026-06-17

CheckMIABench: Firm Foundations For Membership Inference Attacks on Language Models

arXiv:2606.17464v1 Announce Type: new Abstract: Membership inference attacks (MIAs) are a canonical way to assess a machine learning model's privacy properties. Although several attempts have been made to evaluate MIAs on language models, the extant literature has suffered numerous difficulties in constructing clean evaluations to test new techniques. In particular, subtle distribution shifts between member and non-member sets can undermine the statistical validity of MIAs; recent work has underscored this by showing that "blind" methods with no access to the underlying model can perform far better than published methods on the same benchmarks. This paper constructs a benchmark for principled evaluation of MIAs against LLMs, by leveraging the insight that training data before and after a fixed point during training are drawn from the same distribution. Therefore, all open-source models with intermediate checkpoints and public training data can be converted into MIA testbeds. We apply our framework to a half-dozen published attacks on the Pythia and OLMo family of models, from 70M to 7B parameters. To facilitate further privacy research, we open-source a modular library for designing and implementing attacks in this setting: https://github.com/safr-ai-lab/pandora_llm.

21.
arXiv (CS.LG) 2026-06-18

Be Your Own Teacher: Steering Protein Language Models via Unsupervised Reward Optimization

arXiv:2606.18961v1 Announce Type: new Abstract: Protein language models (PLMs) have emerged as powerful tools for controllable biomolecular design, yet their post-training adaptation typically relies on costly wet-lab validation or curated preference datasets. To overcome this supervision bottleneck, we introduce unsupervised reward optimization of PLMs, a comprehensive framework for steerable protein generation without ground-truth labels. Our key insight is that task-agnostic rewards, which combine intrinsic model uncertainty with extrinsic semantic consistency informed by protein representation models, exhibit strong correlation with controllability measures across base models and temperature regimes. Building upon this discovery, we propose two offline algorithms: Soft Reward Optimization (SRO) and Binarized Reward Optimization (BRO), which effectively maximize the classical RLHF objective induced by these proxy rewards. Extensive experiments on compositional out-of-distribution prompts demonstrate that both methods significantly outperform competitive baselines (DPO, KTO), while approaching oracle performance across multiple sampling temperatures, model scales and protein families. Moreover, PLMs fine-tuned with unsupervised rewards can achieve consistently higher coverage compared to their base model in pass@k evaluations. By enabling self-improvement of PLMs through their own generated experience, our framework provides a scalable pathway toward controllable biomolecular design in settings where labeled preferences or experimental feedback are scarce or unavailable.

22.
arXiv (CS.CV) 2026-06-16

Stepwise Token Selection for Efficient Multimodal Large Language Models

In multimodal large language models (MLLMs), inference cost is largely dominated by the visual token prefix rather than the language backbone, making token reduction a key factor for improving efficiency. Existing approaches typically assign independent importance scores to visual tokens and retain a fixed number of top-ranked tokens, implicitly assuming token independence and a uniform compression ratio across inputs. In this work, we reformulate visual token pruning as a sequential decision-making process. Specifically, we introduce a pointer-style selection mechanism that iteratively chooses informative tokens, conditioning each decision on previously selected ones, and dynamically determines when to stop via a learned termination action. This enables joint optimization of both the selected subset and its size. To enable end-to-end training under standard language modeling objectives, we design a differentiable relaxation based on a variance-preserving noise interpolation scheme, allowing gradients to propagate through the discrete selection process. Extensive experiments on LLaVA-v1.5-7B and Qwen2.5-VL-7B demonstrate that our approach consistently outperforms fixed-ratio baselines across different compression levels. Under aggressive pruning that removes 88.9% of visual tokens, our method preserves 94.6% of the original accuracy while achieving a 1.88x speed-up in prefill latency.

23.
arXiv (CS.AI) 2026-06-11

Position: Hippocampal Explicit Memory Is the Cornerstone for AGI

作者:

arXiv:2606.11245v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities across various tasks, raising expectations for Artificial General Intelligence (AGI). This position paper argues that integrating explicit memory is the cornerstone for advancing LLMs toward AGI. The key reason is that the underlying learning mechanism of LLMs is highly analogous to human implicit memory. However, higher-order cognitive functions necessary for AGI, such as long-term strategic planning, metacognition, and symbolic reasoning, heavily rely on hippocampal explicit memory and cannot arise solely from implicit statistical learning. Drawing on findings from neuroscience, I advance this perspective and complement it with computational requirements for artificial explicit memory systems, hoping to foster further research and lay the groundwork for explicit memory integration.

24.
arXiv (CS.LG) 2026-06-15

Neither Parallel Nor Sequential: How DiffusionGemma Actually Commits Tokens

arXiv:2606.14620v1 Announce Type: new Abstract: Open diffusion language models are marketed as parallel, non-autoregressive decoders, yet the order in which a shipped checkpoint actually commits its tokens is almost never measured. We instrument DiffusionGemma 26B, a masked discrete-diffusion mixture-of-experts model built on Gemma 4, hooking its sampler's accept step to record which canvas positions commit, when, and at what confidence. Across a 686-prompt, six-regime probe suite we find that its decoding is neither parallel nor block-autoregressive: it follows a partial left-to-right commit bias whose apparent strength depends almost entirely on the granularity at which you look. Order is weak token by token and strengthens smoothly as the analysis is coarsened, so the model's "block size" turns out to be an artifact of the measuring ruler rather than the architecture. The model commits in large simultaneous batches, leaving much of the within-batch order genuinely undefined rather than merely unobserved. The behaviour is regime-dependent: structured JSON is committed in essentially arbitrary order, and a position's commit confidence tracks correctness on mathematical reasoning but carries no signal on factual recall. Commitment is aggressive, finishing in a short late burst well inside the step budget, while task accuracy matches the model's autoregressive Gemma-4 sibling. Beyond these findings, our central contribution is methodological: measuring decoding order honestly demands handling trailing-EOS padding, within-regime confounding, commit non-monotonicity, block-size sensitivity, and large commit-batch ties, each of which can otherwise manufacture a decoding-order result that is not really there.

25.
arXiv (CS.AI) 2026-06-19

Bistable by Construction: Wall-Clock-Calibrated State Monitors Have No Moment-Detection Regime at Agent Cadence

arXiv:2606.19386v1 Announce Type: cross Abstract: Runtime monitors for autonomous agents commonly threshold an accumulated internal state - a behavioural baseline, a drift statistic, or, in our prior work, a modelled affective state. We previously reported a State Saturation Trap: threshold-on-state triggers over a continuous affect engine become near-constant alarms on SWE-bench debugging agents (Modgil 2026). A post-release audit found the engine received dt=0 between actions, so its exponential decay never operated: the published trap is a pure-accumulator result. We correct the record (erratum, v2) and treat the flaw as an experiment. The key variable it exposes is whether a monitor's dynamics are calibrated in sample time (per observation, as in CUSUM) or wall-clock time (half-lives in seconds, as in affect models and EMA baselines). On fixed-rate streams these coincide; on agent streams, where inter-action time varies by orders of magnitude, they do not. A pre-registered sweep over uniform intervals (dt in {0..600}s) on 20 trajectories shows the wall-clock level trigger has two regimes: at dt=60s silent. Every critical dt lies in (1,30]s. Real agent runs measure latency at median 1.53s (p90 2.33s); real coding cadence sits inside the trap regime, vindicating the empirical finding under a corrected mechanism. The structure is a property of the calibration class, not the engine: a minimal wall-clock accumulator over the raw error stream reproduces the same cliff, while a sample-time CUSUM over the identical stream is exactly dt-invariant (20/20). A rising-edge trigger with hysteresis fires 0-3 times per trajectory in every condition. We conclude that wall-clock-calibrated leaky-integrator monitors admit no regime in which they act as moment detectors on agent streams; transition detection escapes the trap at every cadence, but does not recover human intervention timing.