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.
medRxiv (Medicine) 2026-06-15

Pulmonary extracellular vesicles drive alveolar macrophage dysfunction via microRNA transfer in Acute Respiratory Distress Syndrome

Background: Alveolar macrophage (AM) dysfunction contributes to Acute Respiratory Distress Syndrome (ARDS) pathogenesis. We investigated the role of extracellular vesicles (EVs) in mediating this dysfunction. Methods: Pulmonary EVs were isolated from broncho-alveolar lavage and non-directed bronchial lavage samples of ventilated sepsis patients with and without ARDS, and post-operative control patients via ultracentrifugation. AMs were isolated from lung tissue resections of lobectomy patients. AMs were treated with pooled EVs for 24 hours prior to functional, metabolic and autophagy profiling. EV cargo was profiled via small RNA transcriptomics and proteomics. Mechanistic role of EV microRNAs was assessed via mimic / antagomir transfection. Results: Pulmonary EVs from sepsis patients with ARDS impaired AM efferocytosis, and control EVs had no effect. ARDS EV treatment enhanced AM mitochondrial-linked respiration, but not glycolysis. ARDS EV treatment impaired LC3B-II and LAMP1 expression, indicating dysregulated AM autophagy-lysosomal machinery. Proteomics revealed downregulation of innate immune pathways in ARDS EVs. Transcriptomics revealed enrichment of 24 microRNAs in ARDS EVs; miR-652-3p was the most enriched, validated by RT-qPCR. EV miR-652-3p was associated with 90-day mortality (9.20 vs 0.59 RQ, p=0.0295) and inversely correlated with oxygenation (PaO2/FiO2). AM transfection with miR-652-3p mimic induced similar dysregulation of function and autophagy as ARDS EVs. Transfection of ARDS EVs with antagomirs to miR-652-3p prior to AM treatment partially rescued efferocytosis and autophagy. Conclusions: Targeting EV miR-652-3p may restore alveolar macrophage function and reduce excessive inflammation, thus offering a novel therapeutic strategy for patients with ARDS.

02.
arXiv (CS.AI) 2026-06-11

Noise-Aware Framework for Correcting Corrupted Labels

arXiv:2606.11695v1 Announce Type: cross Abstract: High-quality labeled data is essential for training reliable ML/DL models. However, real-world datasets often contain a considerable proportion of corrupted labels, which can severely degrade model performance. To address this problem, we propose CANOLA, a novel framework for correcting corrupted labels through noise-aware learning and iterative label refinement. CANOLA explicitly estimates the underlying noise distribution of the dataset and incorporates this information into the training of a noise-aware Deep Neural Network. By incorporating noise characteristics during learning, CANOLA enables the model to down-weight unreliable supervision signals and focus on trustworthy patterns, thereby improving robustness and generalization. Label correction is performed via cautious, iterative soft label refinement, in which model predictions are blended with observed labels to prevent premature or erroneous updates. This progressive refinement allows the dataset to be repaired in a stable and controlled manner. We evaluate CANOLA on six widely used datasets under realistic noisy labeling scenarios. Experimental results show that CANOLA consistently outperforms SOTA label correction methods, achieving relative improvements ranging from 19% to 52% in error reduction. Moreover, models trained on datasets corrected by CANOLA obtain substantial downstream performance gains. Even simple classifiers trained on CANOLA's corrected data can outperform complex model-centric approaches by margins of up to 67%.

03.
arXiv (CS.LG) 2026-06-11

Characterizing the Impact of NVFP4 Quantization for Low-Power Edge AI Deployment

arXiv:2606.06527v3 Announce Type: replace-cross Abstract: Energy-efficient neural-network inference at the edge requires reducing arithmetic cost, memory traffic, computation energy, and storage overhead while maintaining acceptable accuracy. This paper presents an ablation-focused study of NVFP4 quantization for edge-efficient neural networks, with emphasis on the relationship between activation precision, weight precision, block-size scaling, retraining, and model accuracy. NVFP4 activations are represented using 4-bit FP4 data, an FP8 block scale, and an FP32 tensor scale, enabling ultra-low precision inference while preserving activation dynamic range. A block-size ablation over six edge-efficient models shows that block size B = 16 provides a practical accuracy/storage trade-off, requiring only 4.5078 bits per input for N = 4096. A weight precision ablation further shows that FP8 and FP16 weights provide only modest gains over FP4 weights under the same NVFP4 activation path, suggesting that activation quantization and scaling dominate much of the accuracy behavior. To isolate the benefit of the NVFP4 data type, this work compares conventional unscaled FP4 activation inference and NVFP4 activation inference with and without retraining. The results show that conventional FP4 inference collapses accuracy for most compact models, while NVFP4 without retraining already recovers substantial accuracy by restoring activation dynamic range through FP8 block scaling and FP32 tensor scaling. When combined with retraining, NVFP4 achieves the best accuracy across the evaluated models, demonstrating the effectiveness of scaling-aware FP4 (NVFP4) inference. These findings provide general design guidance for hardware-software co-design of low power edge inference across a broad range of accelerator platforms, including GPUs, Tensor Cores, FPGAs, domain-specific AI accelerators, near-memory computing systems, and emerging edge-computing architectures.

04.
arXiv (CS.LG) 2026-06-19

Execution-State Capsules: Graph-Bound Execution-State Checkpoint and Restore for Low-Latency, Small-Batch, On-Device Physical-AI Serving

作者:

arXiv:2606.20537v1 Announce Type: new Abstract: Mainstream LLM serving systems reuse prefix work mainly through paged or radix key-value (KV) caches. This is highly effective for high-throughput, high-concurrency serving, but it manages only one positional fragment of execution state: the KV cache. We study the opposite regime: low-latency, small-batch, on-device physical-AI serving, where interactive LLM agents, speech systems, and robot policies repeatedly branch, reset, interrupt, and re-enter under tight responsiveness budgets. We introduce execution-state capsules, a graph-bound checkpoint and restore mechanism for the complete restorable state at a committed boundary. FlashRT is a white-box, backend-facing kernel runtime whose evaluated NVIDIA CUDA backend runs captured graph plans over contiguous static buffers with no block-table indirection. Because the live state is a closed set of named buffers, a capsule can snapshot, restore, fork, or roll back the whole execution boundary, including KV, recurrent state, convolution state, MTP state, and metadata. This moves reuse from token-addressed KV fragments to graph-bound execution-state boundaries. On an RTX 5090, capsule restore is byte-exact at the stored-state level and token-identical under greedy decode. A KV-only ablation diverges, showing that recurrent state is load-bearing. GPU-resident snapshot and restore are sub-millisecond, and TTFT speedup over cold prefill grows from 3.9x at 2k tokens to 27x at 16k tokens. On Jetson AGX Thor and DGX Spark, the same correctness and structural properties hold. Capsules are not a replacement for high-throughput KV-cache serving; they define a complementary latency-first serving point for explicit execution-state reuse.

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

Focus When Necessary: Adaptive Routing and Collaborative Grounding for Training-Free Visual Grounding

While Multimodal Large Language Models (MLLMs) excel in cross-modal reasoning, they often struggle to perceive fine-grained details in complex high-resolution images. Recent training-free methods address this through image scaling and localized cropping. However, applying these manipulations indiscriminately introduces computational redundancy for simple queries and can degrade accuracy by truncating essential global context or introducing irrelevant background noise. To this end, we propose LazyMCoT, a dynamic and training-free framework that adaptively allocates visual grounding efforts based on sample difficulty. The framework features an Adaptive Routing mechanism that evaluates predictive uncertainty using first-token statistics from a single forward pass. This efficiently bypasses confident cases while ensuring the recall of difficult samples via conformal calibration. For these challenging cases, a Collaborative Grounding module integrates the inherent cross-modal attention of the model with an external visual expert through a two-stage refinement process. This refinement process generates a precise localized display to recover small or occluded targets. Extensive experiments across diverse benchmarks demonstrate that LazyMCoT rivals training-based approaches by simultaneously improving reasoning accuracy and reducing average inference latency. Our code is availble at https://github.com/TencentBAC/LazyMCoT.

06.
arXiv (CS.CL) 2026-06-15

Benchmarking Web Agent Safety under E-commerce Deceptive Interfaces

As autonomous web agents are increasingly deployed to perform real-world tasks, ensuring their safety has become a critical concern. In this work, we study web agent behavior under realistic deceptive interfaces in the e-commerce domain. We introduce WebDecept, a lightweight and configurable plugin framework that enables controlled injection of deceptive interface patterns into existing web environments. Using WebDecept, we instantiate seven deceptive patterns commonly observed on the open web, including targeted advertisements, domain redirection, and shopping manipulation. By injecting these patterns into the frontend during task execution, we perform controlled evaluation of multiple multimodal web agents. Our results show that current web agents are highly susceptible to multiple classes of deceptive interfaces, and that prompt-based constraints are often insufficient to mitigate these failures. We further analyze how the design choices of deceptive patterns influence the success of such manipulations. These findings highlight safety challenges that should be addressed as web agents are scaled toward real-world deployment.

07.
arXiv (math.PR) 2026-06-16

Geometry of critical discrete structures: long-range percolation on the hierarchical lattice and the discrete torus

arXiv:2509.09589v2 Announce Type: replace Abstract: Consider (a) balls $\Lambda_n$ of growing volumes in the $d$-dimensional hierarchical lattice, and (b) the $d$-dimensional discrete torus $\mathbb{T}_n^d$ on $n^d$ vertices. Place edges independently between each pair of vertices $x\neq y\in\Lambda_n$ or $\mathbb{T}_n^d$ with probability $1-\exp(-\beta J(x, y) )$ where $J(x, y) \asymp \| x-y \|^{-\alpha}$ for some $0

08.
PLOS Computational Biology 2026-06-18

A comparison of contact patterns derived from the population structure in agent-based models and empirical contact survey data

作者:

by Janik Suer, Johannes Ponge, Michael Brüggemann, Jan Pablo Burgard, Vitaly Belik, Bernd Hellingrath, Alejandra Rincón Hidalgo, Andrzej K. Jarynowski, Richard Pastor, Huynh Thi Phuong, Steven Schulz, Ashish Thampi, Chao Xu, Marlli Zambrano, Rafael Mikolajczyk, André Karch, Veronika K. Jaeger, on behalf of the OptimAgent Consortium Agent-based models (ABMs) are powerful tools for simulating disease spread, relying on individual-level interaction rules from which emergent dynamics arise. An important component in ABMs is contact behaviour. To reduce computational complexity, contact behaviour in ABMs is often assumed as random mixing within structurally defined settings (as, e.g., workplaces). with setting composition typically based on empirical data such as census information. However, the validity of this approach to represent contacts remains unclear. To address this gap, we compare the contact structure derived through this approach in a large-scale ABM with empirical contact survey data with respect to age contact matrices for households, schools, workplaces, all remaining contact settings, and all contacts combined (based on difference matrices and sum of squared errors (SSE)). Our results demonstrate that random mixing in settings with known age compositions like households (SSE:0.7(95%CI0.4–0.9)), schools (SSE:0.7(95%CI:0.3–1.1)) and workplaces (SSE:0.5(95%CI:0.2-0.7)), captures basic interaction patterns but fails to account for age-related variation in contact numbers. The largest differences arise for contacts outside these settings (SSE:3.8(95%CI:1.2–6.5)), as ABMs typically use random regional contacts that do not capture age-structured behaviour observed in contact surveys. Applying contact matrices from both approaches to an age-structured compartmental model, leads to noticeable differences in simulated epidemic outcomes regarding reproduction numbers and spreading dynamics between age groups. Our results suggest that naïve approaches to represent contact behaviour in ABMs based on population structure can be valid in settings with defined age-structures while settings with low a priori structure require more advanced methods to represent contact behaviour observed in contact surveys.

09.
arXiv (CS.CL) 2026-06-16

Understanding the Behaviors of Environment-aware Information Retrieval

Recent retrieval-augmented generation (RAG) approaches have demonstrated strong capability in handling complex queries, yet current research overlooks a critical challenge: different retrievers require fundamentally different query formulation strategies for optimal performance. In this work, we present the first systematic analysis of how LLMs can learn to adapt their query formulation strategies for different retrievers via reinforcement learning (RL). Our empirical study reveals that RL effectively teaches an LLM to tailor its queries to specific retriever characteristics. We discover that different retrievers exhibit surprisingly distinct optimal query styles (e.g., descriptive vs. question-like), suggesting strategies learned for one retriever ineffective for another. We further show that performance can be enhanced by incorporating retriever-specific human guidance and by scaling model size. To facilitate learning over multi-retrieval-step trajectories, we introduce a branching-based rollout technique that improves training stability. Our work provides the first empirical evidence and actionable insights for building truly retriever-aware RAG systems. Code and resources are available at https://github.com/LCO-Embedding/Envs-aware-Information-Retrieval.

10.
arXiv (CS.CL) 2026-06-24

MERGE: Minimal Expression-Replacement GEneralization Test for Natural Language Inference

As many benchmarks have become saturated, it has become increasingly important to create new datasets that evaluate the generalization capacity of current state-of-the-art models in reasoning. However, designing high-quality reasoning datasets is challenging, as their manual construction is costly, and their automatic generation is unreliable, often leading to synthetic data with limited scope. In this paper, we propose the Minimal Expression-Replacement GEneralization (MERGE) test that evaluates the robustness of reasoning models against non-adversarial variants of existing evaluation datasets. We automatically obtain high-quality variants from the original instances with Minimal Expression REplacement (MERE) generation, which uses Masked Language Models (MLMs) and safeguarding filters. We apply the MERGE test to Natural Language Inference (NLI), a popular task of reasoning. We generate new NLI datasets from two widely used existing ones with the MERE generation and use them to evaluate multiple strong NLI models. The results indicate that both LLMs and fine-tuned NLI models generalize poorly: they struggle to consistently and correctly classify variants minimally different in form and reasoning from the original ones. Further, we also analyze how certain aspects in variant generation, such as the word class and the source MLMs, affect model performance.

11.
medRxiv (Medicine) 2026-06-12

Mathematical analysis of the overall survival after chemoradiotherapy of limited-stage small cell lung cancer and the effect of dose/fractionation

The purpose of this work is to analyze the 2-year overall survival (OS2y) of limited-stage small cell lung cancer (LS-SCLC) treated with chemoradiotherapy (CRT), aiming at characterizing the response of LS-SCLC, and in particular the /{beta} value and proliferation parameters. Through a systematic analysis of the literature, we collated a dataset containing 57 entries (3363 patients) of response of LS-SCLC treated with CRT. Radiotherapy schedules ranged from hyper- to hypofractionation. Four radiobiological models to describe the OS2y were investigated, with progressive levels of complexity including the effect of radiotherapy, chemotherapy, treatment year and toxicity. The Akaike Information Criterion (AIC) was used to compare models, and the profile likelihood methodology to compute confidence intervals. Model 4, which includes the effect of radiotherapy, chemotherapy, treatment year and dose-dependent toxicity, provided the best fits of the experimental data (lowest AIC value). While being the best model, model 4 still fails to provide a good prediction of the OS2y, in particular failing to predict the survival of the schedules achieving the lower/higher survivals. The radiobiological analysis of the dose-response of LS-SCLC to CRT does not allow to narrowly constrain the value of response parameters. We attribute this limitation to the large heterogeneity of this disease. Nonetheless, our analysis shows a large /{beta} value (>9 Gy, 95% CI), which implies a low fractionation effect in the radiotherapy of LS-SCLC. and an accelerated proliferation of tumor cells, {lambda}' > 1.6 Gy/day (95% CI), after a kick-off time of ~4-5 weeks, which supports the use of accelerated protocols to avoid the effect of tumor proliferation on the clinical outcome.

12.
arXiv (CS.LG) 2026-06-18

Unreduced Persistence Diagrams for Topological Machine Learning

arXiv:2507.07156v2 Announce Type: replace-cross Abstract: Supervised machine learning pipelines trained on features derived from persistent homology have been experimentally observed to ignore much of the information contained in a persistence diagram. Computing persistence diagrams is often the most computationally demanding step in such a pipeline, however. To explore this dynamic, we introduce several methods to generate topological feature vectors from unreduced boundary matrices and investigate their theoretical and computational properties. We compared the performance of pipelines trained on vectorizations of unreduced PDs to vectorizations of fully-reduced PDs across several data and task types. Our results indicate that models trained on PDs built from unreduced diagrams can perform on par and even outperform those trained on fully-reduced diagrams on some tasks. We also benchmarked the computational performance of an algorithm for computing unreduced diagrams, which was implemented as a heavily modified version of Ripser. These computations are parallelizable and required an order of magnitude less memory on average compared to computing full persistence diagrams. Our results suggest that machine learning pipelines which incorporate topology-based features may benefit in terms of computational cost and performance by utilizing information contained in unreduced boundary matrices.

13.
arXiv (CS.CL) 2026-06-11

A Controlled Study of Decoding-Time Truthfulness Methods on Instruction-Tuned LLMs

作者:

In this work, we introduce CHAIR (Classifier of Hallucination As ImproveR), a supervised framework for detecting hallucinations by analyzing internal logits from each layer of every token. Our method extracts a compact set of features such as maximum, minimum, mean, standard deviation, and slope-from the token logits across all layers, enabling effective hallucination detection without overfitting. Experiments on TruthfulQA and MMLU datasets demonstrate that CHAIR significantly improves detection accuracy, particularly in zero-shot scenarios, showcasing its robustness and generalizability. Beyond hallucination detection, CHAIR highlights the potential of using internal representations for designing advanced decoding strategies. By leveraging patterns in logits, we suggest that more sophisticated models and adaptive decoding methods could further reduce hallucinations and enhance text completion quality. CHAIR not only offers a practical solution for detecting hallucinations but also lays the groundwork for exploring richer representations in LLMs to improve their factuality and coherence.

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

All-Mem: Agentic Lifelong Memory via Dynamic Topology Evolution

Lifelong interactive agents are expected to assist users over months or years, which requires continually writing long term memories while retrieving the right evidence for each new query under fixed context and latency budgets. Existing memory systems often degrade as histories grow, yielding redundant, outdated, or noisy retrieved contexts. We present All-Mem, an online/offline lifelong memory framework that maintains a topology structured memory bank via explicit, non destructive consolidation, avoiding the irreversible information loss typical of summarization based compression. In online operation, it anchors retrieval on a bounded visible surface to keep coarse search cost bounded. Periodically offline, an LLM diagnoser proposes confidence scored topology edits executed with gating using three operators: Split, Merge, and Update, while preserving immutable evidence for traceability. At query time, typed links enable hop bounded, budgeted expansion from active anchors to archived evidence when needed. Experiments on LoCoMo and LongMemEval-s show improved retrieval and QA over representative baselines. The code is available at https://github.com/LvCan926/All-Mem.

15.
arXiv (CS.AI) 2026-06-17

Trustworthy Self-Composable Big-Data-as-a-Service: An LLM-Orchestrated Multi-Agent Framework for Automated Data Engineering, AutoML, MLOps Deployment, and Drift-Aware Lifecycle Optimization

arXiv:2606.17915v1 Announce Type: cross Abstract: Big-Data-as-a-Service (BDaaS) platforms require re liable automation across data ingestion, cleaning, feature engi neering, model development, deployment, and post-deployment monitoring. However, existing LLM-based data science agents and AutoML systems mainly focus on isolated workflow stages, leaving limited support for lifecycle-level orchestration, artifact governance, human oversight, and drift-aware adaptation. This paper proposes a trustworthy self-composable BDaaS frame work based on LLM-orchestrated multi-agent collaboration. The proposed architecture decomposes the BDaaS lifecycle into specialized agents for data ingestion, data cleaning, feature engineering, AutoML training, model evaluation, MLOps de ployment, monitoring, and drift detection. A central LLM or chestration layer coordinates agent execution, validates interme diate outputs, manages workflow context, and enables dynamic workflow composition. The framework also incorporates shared artifact governance, reproducibility support, human-in-the-loop checkpoints, and drift-aware feedback loops. A prototype-based evaluation is conducted using controlled tabular benchmark datasets with missing values, categorical variables, outliers, class imbalance, and simulated covariate drift. Compared with manual ML, AutoML-only, and single-agent LLM baselines, the pro posed multi-agent BDaaS pipeline achieves competitive predictive performance while improving lifecycle-level reliability, including workflow completion, artifact traceability, deployment readiness, reproducibility, and drift recovery. The results suggest that LLM-orchestrated multi-agent systems can extend conventional AutoML toward trustworthy, adaptive, and production-oriented BDaaS lifecycle automation.

16.
arXiv (CS.CL) 2026-06-12

Shopping Reasoning Bench: An Expert-Authored Benchmark for Multi-Turn Conversational Shopping Assistants

Conversational shopping assistants now serve hundreds of millions of customers, yet no existing benchmark jointly evaluates the open-ended multi-turn reasoning, domain expertise, and criterion-level quality that real shopping conversations demand. Shopping reasoning is unique among language model applications. Unlike factual question answering or verifiable code generation, it requires balancing subjective preferences, budget constraints, and cross-product trade-offs across multi-turn dialogue, capabilities absent from previous e-commerce and general-purpose benchmarks. We introduce the Shopping Reasoning Bench, an expert-authored benchmark of 525 missions (232 single-turn, 293 multi-turn) with 10863 importance-weighted binary rubrics authored by retail domain experts. These criteria are organized under a taxonomy of five reasoning categories and fifteen subcategories covering diverse demands such as preference refinement, trade-off analysis, and compatibility assessment. An evaluation of nine models across three families (GPT, Claude, Gemini) shows that pass rates reach only 57–77% overall. On multi-turn missions, all models score 13–29 points lower on optional above-and-beyond criteria than on required ones, and performance degrades 4–18 points as conversations progress. These gaps show that current models handle basic shopping assistance but fall short of expert-level advice, making Shopping Reasoning Bench a challenging testbed for future shopping assistant development.

17.
arXiv (CS.AI) 2026-06-24

Graph Alignment for Benchmarking Graph Neural Networks and Learning Positional Encodings

arXiv:2505.13087v2 Announce Type: replace-cross Abstract: We propose a novel benchmarking methodology for graph neural networks (GNNs) based on the graph alignment problem, a combinatorial optimization task that generalizes graph isomorphism by aligning two unlabeled graphs to maximize overlapping edges. We frame this problem as a self-supervised learning task and present several methods to generate graph alignment datasets using synthetic random graphs and real-world graph datasets from multiple domains. For a given graph dataset, we generate a family of graph alignment datasets with increasing difficulty, allowing us to rank the performance of various architectures. Our experiments prove that there is an optimal task difficulty for having a statistically relevant ranking of different models and that, even on a structure-only task, anisotropic models perform better compared to isotropic ones. To further prove that our synthetic task capture meaningful information, we show its effectiveness for self-supervised GNN pre-training: the learned node embeddings can be leveraged as positional encodings by transformers for graph regression or can be used to reconstruct the full structure of the graph with $98\%$ accuracy. To support reproducibility and further research, we provide an open-source Python package to generate graph alignment datasets and benchmark new GNN architectures. The source code is available at https://github.com/adrien-lagesse/graph-alignment-benchmark.

18.
arXiv (CS.LG) 2026-06-11

Robustness of Mixtures of Experts to Feature Noise

arXiv:2601.14792v2 Announce Type: replace Abstract: Despite their practical success, it remains unclear why Mixture of Experts (MoE) models can outperform dense networks beyond sheer parameter scaling. We study an iso-parameter regime where inputs exhibit latent modular structure but are corrupted by feature noise, a proxy for noisy internal activations. We show that sparse expert activation acts as a noise filter: compared to a dense estimator, MoEs achieve lower generalization error under feature noise, improved robustness to perturbations, and faster convergence speed. Empirical results on synthetic data and real-world language tasks corroborate the theoretical insights, demonstrating consistent robustness and efficiency gains from sparse modular computation.

19.
medRxiv (Medicine) 2026-06-10

Prediction of immunotherapy response using live tumor fragments from routine clinical biopsies

Functional ex vivo assays using live tumor tissues have demonstrated strong predictive accuracy for response to immune checkpoint inhibitors (ICIs) but are not scalable, requiring manual processing of large resections collected at academic centers. Here, an ex vivo live tumor fragment (LTF) platform was developed using standard-of-care biopsies from 228 patients with suspected malignancy collected across prospective, multicenter observational trials and biobanks. Hierarchical clustering of ICI-mediated changes in cytokine production identified two groups: responders and nonresponders. A binary classifier (elive index) using 8 cytokines achieved an AUC of 0.99 for cluster prediction. elive index correctly predicted clinical benefit in 93% (26/28) of patients (P = 3.2x10-5) and accurately identified 83% (10/12) of objective responders. Critically, elive responders were identified among biomarker-negative patients, highlighting the platform as a scalable approach that complements existing companion diagnostics and expands the population of patients identified to benefit from ICI therapy.

20.
arXiv (CS.LG) 2026-06-24

Sakana Fugu Technical Report

arXiv:2606.21228v2 Announce Type: replace Abstract: The capabilities of frontier Large Language Models (LLMs) continue to advance, with different providers increasingly specializing in distinct domains. This raises a natural next objective: how to combine the individual specializations of various LLMs into a collectively intelligent system. To this end, we report the development of Sakana Fugu, a family of orchestrator models that harness and amplify the capabilities of an LLM agent team. Fugu models are themselves language models trained to understand user queries and dynamically devise agentic scaffolds to solve them. Through these adaptive scaffolds, Fugu accesses performance beyond any individual LLM agent, achieving state-of-the-art results compared to other publicly accessible models across a range of challenging tasks, including SWE-Bench Pro, Terminal Bench, LiveCodeBench, GPQA-Diamond, Humanity's Last Exam, and CharXiv Reasoning. We release two models: Fugu, which balances performance with latency for everyday use, and Fugu-Ultra, which prioritizes answer quality on the hardest problems. We describe our training paradigm, which encompasses large-scale fine-tuning, evolutionary algorithms, and reinforcement learning approaches, along with the infrastructure and core design principles that turn these methods into a production system. We hope this report encourages further research into multi-agent systems and dynamic, query-adaptive agentic scaffolds as a path toward the next frontier of AI capabilities, accessed through collective intelligence.

21.
arXiv (CS.AI) 2026-06-24

Zero-Shot Test-Time Canonicalization using Out-of-Distribution Scoring

arXiv:2606.24178v1 Announce Type: cross Abstract: Pretrained vision models often misclassify inputs that are rotated, scaled, or sheared, even though these affine transformations leave the object class unchanged. Robustness is usually restored either by building equivariance into the architecture or by retraining with augmentation, both of which require changing or retraining the model. Test-time canonicalization instead leaves the classifier untouched. It undoes the transformation of each input, mapping it to a canonical form near the training distribution before classification. Existing canonicalizers, however, rely on a narrow set of logit-based energy scores and bespoke search procedures, leaving the design space of scoring functions and optimizers unexplored. We reframe canonicalization as out-of-distribution (OOD) detection, which lets any OOD score serve as the energy minimized over transformations. Across benchmarks ranging from handwritten characters and sketches to natural images and 3D point clouds, we systematically evaluate around twenty OOD scores and nine search algorithms, finding that distance-based scores paired with random search and local refinement perform best overall. Because canonicalizing an already-aligned input can hurt accuracy, we add a gated mechanism that transforms an input only when its OOD score indicates this is needed, preserving most in-distribution accuracy while retaining the robustness gains on transformed inputs. Code is available at github.com/johschm/its.

22.
arXiv (CS.LG) 2026-06-11

Re-evaluating Confidence Remasking in Masked Diffusion Language Models

arXiv:2606.12232v1 Announce Type: new Abstract: Masked diffusion language models (dLLMs) have recently emerged as a competitive alternative to autoregressive language models, with the promise of faster inference via parallel token generation. A notable limitation of the masked formulation, however, is that once a token has been unmasked it can no longer be revised, leaving dLLMs vulnerable to early sampling mistakes. To address this, a growing body of work has sought to extend masked dLLMs with self-correcting (remasking) capabilities. One appealing subset of these methods does so in a training-free, post-hoc manner based on token confidences, with encouraging early reported results. In this work, we revisit the empirical evaluation of a representative post-hoc remasking method, WINO [Hong et al., 2026], and find that under standard decoding settings (shorter block lengths) it brings little-to-no benefit over confidence-based unmasking alone [Wu et al., 2025]. Extending the evaluation to non-greedy decoding, we find that while confidence-based remasking can mitigate errors introduced by increased stochasticity to some extent, it also exacerbates the diversity collapse previously reported for confidence-based unmasking. Overall, our results show that the benefits of post-hoc confidence-based remasking are highly setting-dependent, underscoring the need for a more comprehensive evaluation framework.

23.
arXiv (CS.CL) 2026-06-17

Zone of Proximal Policy Optimization: Teacher in Prompts, Not Gradients

Knowledge distillation transfers a teacher's competence to a small student but is brittle in the small-student regime: forcing the student to imitate logits from a much larger teacher concentrates it on the teacher's sharpest modes, hurting generalization on benchmark families beyond the training corpus. Reinforcement learning (RL) avoids logit imitation by training on the student's own rollouts. However, on questions where every rollout fails-yielding zero advantage and being silently discarded-injecting a stronger teacher's response into the policy gradient breaks the on-policy assumption and induces drift. We introduce Zone of Proximal Policy Optimization (ZPPO), inspired by Vygotsky's zone of proximal development, which keeps the teacher inside the prompt rather than the policy gradient. On hard questions, ZPPO constructs two reformulated prompts: a Binary Candidate-included Question (BCQ) pairs one correct teacher response with one incorrect student response as anonymized candidates the student must discriminate, and a Negative Candidate-included Question (NCQ) aggregates the student's wrong rollouts into a single prompt to surface their shared failure modes. A prompt replay buffer recirculates each hard question until it either graduates-the student's mean rollout accuracy on it reaches half- or is FIFO-evicted under finite capacity, amplifying BCQ and NCQ inside the student's current zone of proximal development. On the Qwen3.5 family at four student scales (0.8B-9B) with a 27B teacher, post-trained as vision-language models and evaluated on a 31-benchmark suite (16 VLM, 10 LLM, 5 Video), ZPPO outperforms off/on-policy distillation and GRPO, with the largest gains at the smallest scale.

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

GAGPO: Generalized Advantage Grouped Policy Optimization

arXiv:2605.13217v1 Announce Type: cross Abstract: Reinforcement learning has become a powerful paradigm for post-training large language model agents, yet credit assignment in multi-turn environments remains a challenge. Agents often receive sparse, trajectory-level rewards only at the end of an episode, making it difficult to determine which intermediate actions contributed to success or failure. As a result, propagating delayed outcomes back to individual decision steps without relying on costly auxiliary value models remains an open problem. We propose Generalized Advantage Grouped Policy Optimization (GAGPO), a critic-free reinforcement learning method for precise, step-aligned temporal credit assignment. GAGPO constructs a non-parametric grouped value proxy from sampled rollouts and uses it to compute TD/GAE-style temporal advantages, recursively propagating outcome supervision backward through time. Combined with group-wise advantage normalization and an action-level importance ratio, GAGPO extracts stable, localized optimization signals directly from multi-turn trajectories. Experiments on ALFWorld and WebShop show that GAGPO outperforms strong reinforcement learning baselines. Further analyses demonstrate faster early-stage learning, improved interaction efficiency, and smoother optimization dynamics, suggesting that GAGPO offers a simple yet effective framework for multi-turn agentic reinforcement learning.

25.
arXiv (math.PR) 2026-06-16

On stability of outliers from the circular law

arXiv:2606.16609v1 Announce Type: new Abstract: This work investigates the stability of outliers from the circular law, via the convergence of their associated diagonal overlaps between eigenvectors - also known as the squared eigenvalue condition numbers. We consider and compare two paradigmatic cases, namely: 1) the Complex Ginibre Ensemble conditioned on the existence of an outlier, and 2) the outlier induced by a rank-one Hermitian perturbation of a Complex Ginibre matrix. In both cases, we prove almost sure convergence towards a specific constant that only depends on the radius of the outlier and its status - either conditioned or induced. These results can be generalized to other complex integrable ensembles with the same techniques, and complement our understanding of eigenvalue stability in non-Hermitian ensembles.