探索全球前沿学术脉络

01.

arXiv (math.PR) 2026-06-24 DOI: arXiv:1705.03801

Conditionally Poissonian random digraphs

作者:

Christian M\"onch↗

arXiv:1705.03801v2 Announce Type: replace Abstract: We define a Poissonian model of directed random graphs which generalises the undirected Poissonian random graph process introduced by Norros and Reittu in Adv. Appl. Probab. 38 (2006), 59–75. Its loopless simple projection is a rank-one independent-arc inhomogeneous digraph of the type studied by Cao and Olvera-Cravioto, Random Struct. Alg. 56 (2020), 722–774. For the Poissonian multigraph itself, we discuss the relation to Norros-Reittu graphs, characterise limiting degree distributions, and record explicit exploration estimates. In particular, we give fixed-depth directed local weak limits, stopped branching-process couplings with weight-mass collision budgets, a comparison with the simple projection, and a rare-event concentration criterion. These estimates are intended as graph-side structural inputs for later dynamics on the graph.

阅读与讨论 → 访问原文 →

02.

arXiv (CS.CV) 2026-06-11 DOI: arXiv:2602.08735

From Correspondence to Actions: Human-Like Multi-Image Spatial Reasoning in Multi-modal Large Language Models

作者:

Masanari Oi↗Koki Maeda↗Ryuto Koike↗Daisuke Oba↗Nakamasa Inoue↗Naoaki Okazaki↗

While multimodal large language models (MLLMs) have made substantial progress in single-image spatial reasoning, multi-image spatial reasoning, which requires integration of information from multiple viewpoints, remains challenging. Cognitive studies suggest that humans address such tasks through two mechanisms: cross-view correspondence, which identifies regions across different views that correspond to the same physical locations, and stepwise viewpoint transformation, which composes relative viewpoint changes sequentially. However, existing studies incorporate these mechanisms only partially and often implicitly, without explicit supervision for both. We propose Human-Aware Training for Cross-view correspondence and viewpoint cHange (HATCH), a training framework with two complementary objectives: (1) Patch-Level Spatial Alignment, which encourages patch representations to align across views for spatially corresponding regions, and (2) Action-then-Answer Reasoning, which requires the model to generate explicit viewpoint transition actions before predicting the final answer. Experiments on three benchmarks demonstrate that HATCH consistently outperforms baselines of comparable size by a clear margin and achieves competitive results against much larger models, while preserving single-image reasoning capabilities.

阅读与讨论 → 访问原文 →

03.

arXiv (CS.CL) 2026-06-24 DOI: arXiv:2606.24841

Matching Tasks to Objectives: Fine-Tuning and Prompt-Tuning Strategies for Encoder-Decoder Pre-trained Language Models

作者:

Ahmad Pouramini↗Hesham Faili↗

Prompt-based learning has emerged as a dominant paradigm in natural language processing. This study explores the impact of diverse pre-training objectives on the performance of encoder-decoder pre-trained language models across generation and question answering tasks, with a focus on commonsense knowledge retrieval and completion. We highlight the benefits of incorporating multiple objectives during both pre-training and fine-tuning stages. We introduce the Match Task to Objective (MTO) framework and methods for determining the appropriate objective for a given task. This framework offers automated methods to prepare task-related data for adaptation through unsupervised training, based on the identified objective. In the fine-tuning stage, we design novel templates that align with the objectives of the pre-training and adaptation stages. When aligned with task requirements, these strategies can achieve a performance gain of over 120\% compared to conventional methods in few-shot settings. They significantly outperform related works in few-shot settings and exceed the baseline even in full-dataset scenarios. Furthermore, we extend this approach to include prompt-tuning methodologies, providing guidance for more effective soft prompt engineering and optimization. Our strategies significantly enhance prompt-tuning performance as well. These insights hold substantial value, precisely guiding the selection and optimization of models customized for specific tasks. Code is available at https://github.com/puraminy/MTO/

阅读与讨论 → 访问原文 →

04.

arXiv (CS.CV) 2026-06-17 DOI: arXiv:2507.05169

Critique of World Model: A Generative Latent Prediction Architecture for World Modeling

作者:

Eric Xing↗Mingkai Deng↗Jinyu Hou↗

World Model, the algorithmic simulator of the real-world environment which biological agents experience and act upon, has been an emerging topic in recent years due to the rising need to develop virtual agents with artificial (general) intelligence. There has been much discussion on what a world model really is, how to build it, how to use it, and how to evaluate it. In this essay, starting from the imagination in the famed Sci-Fi classic Dune, and drawing inspiration from the concept of ``hypothetical thinking'' in psychology literature, we argue the primary goal of a world model to be {\it simulating all actionable possibilities of the real world for purposeful reasoning and acting}. We examine the key design dimensions of world modeling: data, representation, architecture, learning objective, and usage, surveying existing approaches and analyzing their tradeoffs. Building on this examination, we propose a new Generative Latent Prediction (GLP) architecture for a general-purpose world model, based on stateful, hierarchical, multi-level, and mixed continuous/discrete representations, and a generative and self-supervised learning framework, with an outlook of a Physical, Agentic, and Nested (PAN) AGI system enabled by such a model.

阅读与讨论 → 访问原文 →

05.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2606.16950

Latent space mapping of interpretable structural coordinates from stochastic single-molecule signals

作者:

Matteo Cartiglia↗Sandro Kuppel↗Wouter Botermans Wannes Peeters↗Natan Biesmans↗Liam Vandekerckhove↗Eric Beamish↗Koen Ongena↗Wouter Renckens↗Pol Van Dorpe↗Sanjin Marion↗

arXiv:2606.16950v1 Announce Type: cross Abstract: Nanopores are versatile single-molecular sensors, but their utility is fundamentally constrained by stochastic translocation dynamics warping any encoded information. We resolve it by shifting from time-domain analysis to a learned latent-space mapping via a contrastive encoder trained exclusively on simulated signals from a physics-informed model. This encoder maps solid-state nanopore signals of engineered DNA barcodes into an interpretable molecular coordinate system. The learned representation is responsive to structural barcode parameters while remaining invariant to acquisition conditions and translocation conformation, allowing data pooling across devices. Molecule identification requires a single pass through the encoder, reducing computational cost by three orders of magnitude relative to alignment-based methods. We experimentally validate through mixture quantification, rare-variant detection, consensus barcode reconstruction, and real-time signal acquisition. This shift from temporal analysis to mapping structural coordinates into a latent space changes the paradigm behind analyzing stochastic sensor signals by linking classification to interpretable encoded molecular information.

阅读与讨论 → 访问原文 →

06.

medRxiv (Medicine) 2026-06-16 DOI: HASH:d70b878eee22fdd87c04cb2d50e52564

Ranking-optimized survival models can underperform fixed-horizon clinical prediction: a SUPPORT2 reanalysis of machine learning, attending-physician judgment, and the original SUPPORT model at 60- and 180-day mortality

作者:

Truong↗Q. H↗Hoang↗D. C↗Luu↗D. T↗

Machine-learning survival models are increasingly proposed for intensive-care mortality prediction and are almost always selected and reported using the concordance index, a ranking metric averaged over follow-up. Yet most bedside decisions hinge on a probability at a specific time, such as 60- or 180-day mortality. We asked whether ranking-optimized models remain competitive at fixed clinical horizons against two reference points clinicians actually rely on: unaided attending-physician judgment and the original 1995 SUPPORT logistic model. Reanalyzing the SUPPORT2 cohort (9,105 critically ill adults from five United States centers, 1989-1994) under a stratified 70/15/15 split, we compared a gradient-boosted survival model, the physician's recorded prognosis, and the 1995 model at 60 and 180 days, alongside several alternative learners. The survival model achieved competitive ranking concordance (0.705) yet underperformed both comparators at fixed horizons: at 60 days its area under the ROC curve was 0.750, against 0.808 for physicians on the matched sample and 0.827 for the 1995 model, a gap that held across eight independent data splits and remained statistically reliable after multiplicity correction. The shortfall was not miscalibration, since post-hoc recalibration left discrimination unchanged, nor limited capacity, since neural networks, a deep ranking model, and two timepoint-aware discrete-time models also failed to close it; replacing the ranking objective with timepoint-matched binary training recovered roughly half the gap, pointing to an objective-horizon mismatch. Discrimination was equitable across sex, race, and age, but leave-one-disease-out validation exposed severe failure for disease groups absent from training, and the physician advantage was conditional on a physician electing to provide an estimate. We recommend reporting timepoint-specific discrimination alongside concordance, timepoint-matched training when fixed-horizon predictions drive care, leave-one-subgroup validation, and distribution-free prediction intervals to support selective deployment.

阅读与讨论 → 访问原文 →

07.

arXiv (CS.LG) 2026-06-24 DOI: arXiv:2606.23871

Federated Survival Analysis in Healthcare: A Multi-Model Evaluation on Cross-Institutional Heterogeneous Breast Cancer Data

作者:

Natalia Moreno-Blasco↗Anusha Ihalapathirana↗Pekka Siirtola↗Miguel Fernandez-de-Retana↗

arXiv:2606.23871v1 Announce Type: new Abstract: Survival analysis is central to clinical decision-making, yet reliable time-to-event models require large, diverse cohorts that are rarely available at a single institution, while privacy regulations restrict the centralization of patient data. Federated learning (FL) offers a privacy-preserving alternative by training shared models without exchanging raw data, but its effectiveness for survival modeling under realistic, heterogeneous conditions remains insufficiently understood. This paper presents a systematic, multi-model evaluation of federated survival analysis on a cross-institutional breast cancer cohort with naturally heterogeneous distributed clients. Three representative survival models, the Cox Proportional Hazards model, DeepSurv, and Random Survival Forest (RSF), are compared across centralized, local, and federated training, and three federated optimization strategies (FedAvg, FedProx, and FedAdam) are assessed for the gradient-based models. Results show that FL consistently outperforms local training and approaches, and occasionally exceeds, centralized performance, while RSF offers the best overall balance of discrimination, calibration, and robustness across heterogeneous clients. We further find that performance depends on the diversity of client distributions, and that FedAvg and FedProx are stronger and more stable than FedAdam. Based on these findings, we derive practical, decision-oriented guidelines mapping data, privacy, interpretability, and resource constraints to recommended model and training-paradigm choices for federated survival modeling in healthcare.

阅读与讨论 → 访问原文 →

08.

arXiv (quant-ph) 2026-06-16 DOI: arXiv:2510.08129

Adaptively secure unitary designs with constant non-Clifford cost

作者:

Lennart Bittel↗Lorenzo Leone↗

arXiv:2510.08129v2 Announce Type: replace Abstract: Randomness is a fundamental resource in quantum information, with crucial applications in cryptography, algorithms, and error correction. A central challenge is to construct unitary $k$-designs that closely approximate Haar-random unitaries while minimizing the costly use of non-Clifford operations. In this work, we present a protocol able to generate unitary $k$-designs on $n$ qubits, secure against any adversarial quantum measurement, with a system-size-independent number of non-Clifford gates. Our construction applies a $k$-design only to a subsystem of size $\Theta(k)$, independent of $n$. This ``seed'' design is then ``diluted'' across the entire $n$-qubit system by sandwiching it between two random Clifford operators. The resulting ensemble forms an $\varepsilon$-approximate unitary $k$-design on $n$ qubits. We prove that this construction achieves full quantum security against adaptive adversaries using only $\tilde{O}(k^2 \log\varepsilon^{-1})$ non-Clifford gates. If one requires security only against polynomial-time adaptive adversaries, the non-Clifford cost decreases to $\tilde{O}(k + \log^{1+c} \varepsilon^{-1})$. This is optimal, since we show that at least $\Omega(k)$ non-Clifford gates are required in this setting. Compared to existing approaches, our method significantly reduces non-Clifford overhead while strengthening security guarantees to adaptive security as well as removing artificial assumptions between $n$ and $k$. These results make high-order unitary designs practically attainable in near-term fault-tolerant quantum architectures.

阅读与讨论 → 访问原文 →

09.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2606.06007

Diffusion Models for Adaptive Sequential Data Generation

作者:

Haoyang Cao↗Minshuo Chen↗Yinbin Han↗Renyuan Xu↗

arXiv:2606.06007v2 Announce Type: replace Abstract: Generating realistic synthetic sequential data is critical in real-world applications across operations research, finance, healthcare, energy systems, and scientific computing, where time-indexed observations are used for prediction, simulation, risk assessment, and data-driven decision-making. While diffusion models have achieved remarkable success in generating static data, their direct extensions to sequential settings often fail to capture temporal dependence and information structure. Designing diffusion models that can simulate sequential data in an adapted manner, and hence without anticipation of future information, therefore remains an open challenge. In this work, we propose a sequential forward-backward diffusion framework for adapted time series generation. Our approach progressively injects and removes noise along the sequence, conditioning on the previously generated history to ensure adaptiveness. A novel score-matching objective is introduced for efficient parallel training. We derive rigorous statistical guarantees under a generic framework, then establish score approximation, score estimation, and distribution estimation results with ReLU networks serving as a concrete instance. Empirically, we validate our method on synthetic data, including ARMA models and Gaussian processes, and demonstrate its effectiveness in constructing mean-variance optimal portfolios.

阅读与讨论 → 访问原文 →

10.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2606.14776

Deep Learning-Based Lunar Crater Terrain Relative Navigation

作者:

Batu Candan↗Simone Servadio↗

arXiv:2606.14776v1 Announce Type: cross Abstract: Accurate position estimation is crucial for the successful implementation of future lunar landings using autonomous vehicles, especially in dangerous environments with sparse terrain features. In this paper, we propose a terrain relative navigation (TRN) algorithm combining our deep-learning crater detector, which was designed specifically for the NASA Crater Detection Challenge problem, and an Extended Kalman Filter (EKF). Our detector analyzes crater features from the monocular images acquired from orbit, and their matches with craters from a global database are identified via a Hungarian assignment approach followed by the consensus-based outliers removal method. The estimated measurements are then used to refine an EKF, where spacecraft pose estimation in the Lunar-Centered Lunar-Fixed (LCLF) frame of reference, augmented with altitude aiding information, constrains radial drift. The simulation results indicate that even if the spacecraft is off from its actual location up to 5 km, TRN could recover from this situation, achieving navigation error reduction to a few hundred meters. It should be noted that in order to maintain crater feature correspondences, it is important to match the image resolution and the scales within the scene to the detector training set distribution.

阅读与讨论 → 访问原文 →

11.

arXiv (CS.CV) 2026-06-16 DOI: arXiv:2601.19506

Bridging Information Asymmetry: A Hierarchical Framework for Blind Face Restoration with Reduced Uncertainty

作者:

Zhengjian Yao↗Jiakui Hu↗Kaiwen Li↗Hangzhou He↗Xinliang Zhang↗Shuang Zeng↗Lei Zhu↗Yanye Lu↗

Blind face restoration remains a persistent challenge due to the inherent ill-posedness of reconstructing holistic structures from severely constrained observations. Current generative paradigms, while capable of synthesizing realistic facial details, remain limited by the under-constrained nature of blind restoration, where severely degraded inputs can be mapped to plausible yet identity-inconsistent outputs. To address this issue, we present Pref-Restore, a hierarchical framework for BFR with reduced restoration uncertainty. Our design is organized around three complementary principles: (1) Semantic Information Augmentation, where an auto-regressive semantic branch converts image and text cues into structured tokens that provide a stable high-level anchor; (2) Texture-level Fidelity Alignment, where the diffusion generator is trained under this anchor to recover identity-relevant details; and (3) Fidelity-constrained Preference Optimization, where a face-aware reward refines the diffusion trajectory while controlling the quality–fidelity trade-off. Extensive experiments on synthetic and real-world benchmarks show that Pref-Restore achieves state-of-the-art performance, with stronger identity-sensitive fidelity and lower restoration uncertainty across repeated sampling. Systematic ablations further attribute these gains to the proposed hierarchical design, showing the necessity of staged training, the robustness and quality dependence of the text pathway, and the benefit of fidelity-constrained preference optimization.

阅读与讨论 → 访问原文 →

12.

arXiv (math.PR) 2026-06-24 DOI: arXiv:2606.24793

Random sequential nearest-neighbor coloring on trees

作者:

Anne-Laure Basdevant↗Arvind Singh↗

arXiv:2606.24793v1 Announce Type: new Abstract: We study a nearest-neighbor coloring process in which vertices are revealed in random order and inherit the color of the closest vertex revealed before them. This model is a discrete analogue of coloring processes previously studied by Preater (2009) and Aldous (2018) in Euclidean spaces. We focus here on regular trees and analyze the associated genealogy of color inheritance. In contrast with the Euclidean case, the genealogical graph on an infinite regular tree is not connected: it has infinitely many infinite one-ended components, each with a distinct asymptotic direction, while every vertex has only finitely many descendants. We also describe how this structure is modified in the presence of finitely many initial seeds. Finally, we study local limits of the coloring on finite regular trees as their height tends to infinity, for two natural seed configurations: two fixed seeds, and one blue seed at the root with red seeds at the leaves.

阅读与讨论 → 访问原文 →

13.

arXiv (CS.AI) 2026-06-11 DOI: arXiv:2606.11400

Steering Where to Listen: Instruction-Based Activation Steering Redirects Temporal Attention in Large Audio-Language Models

作者:

Tsung-En Lin↗Hung-Yi Lee↗

arXiv:2606.11400v1 Announce Type: cross Abstract: Large Audio-Language Models (LALMs) excel at audio understanding but expose little about where in an audio signal they attend. We introduce instruction-based vector steering, which constructs a steering vector by contrasting activations from differently instructed prompts while keeping the audio fixed. Through a systematic probe of LALM attention, we find that - unlike standard prompting or audio-based steering - this intervention significantly redistributes the temporal attention allocated to audio tokens, concentrating it on acoustically relevant regions. We then show that this attention shift is behaviorally meaningful: in a controlled three-event setting, reading out the temporal position of maximal steering-induced attention change recovers the location of a queried sound event without any training, attaining 60.87% and 68.72% overlap with ground-truth intervals on Qwen2-Audio and Audio Flamingo 3, far above direct prompting (31.84%, 46.75%) and random baselines (27.74%). Our results characterize a mechanistic property of instruction-based steering in LALMs and provide a training-free probe for the latent temporal structure these models encode.

阅读与讨论 → 访问原文 →

14.

arXiv (math.PR) 2026-06-16 DOI: arXiv:2606.16636

Free energy of non-convex multi-species spin glasses with centered Ising spins

作者:

Hong-Bin Chen↗Victor Issa↗Jean-Christophe Mourrat↗

arXiv:2606.16636v1 Announce Type: new Abstract: We identify the limit free energy of all multi-species spin glasses with centered $\pm 1$ spins. The result was previously known only under a convexity assumption on the covariance function of the Hamiltonian. We also obtain a one-species reduction of the formula for balanced multi-species models.

阅读与讨论 → 访问原文 →

15.

arXiv (quant-ph) 2026-06-11 DOI: arXiv:2606.12301

An iterative Ising decoder for quantum error correction codes

作者:

Yuanqi Liu↗Weilei Zeng↗Peixiang Li↗Yantong Liu↗Guangyao Huang↗Yingwen Liu↗Dongyang Wang↗Junjie Wu↗Lingling Lao↗

arXiv:2606.12301v1 Announce Type: new Abstract: The Ising framework maps the decoding problem in quantum error correction onto ground-state optimization of a classical Hamiltonian, in which $X$-$Z$ error correlations enter as cross terms. Under phenomenological depolarizing noise, the exact joint formulation contains up to 8-body interactions for the toric code and 10-body for the $6.6.6$ color code. These high-order terms degrade solver convergence, inflate runtime, and raise the auxiliary spin overhead when embedding into native 2-body Ising hardware. In this work, we propose the iterative low-order decoding (ILOD) algorithm, which alternates between $X$- and $Z$-type sub-Hamiltonians, approximating cross-type correlations through Bayesian priors that reweight each type's couplings using the other type's inferred error configuration. This halves the maximum body count of interaction terms in the Hamiltonian, accelerating the solver, restoring convergence at larger code distances, and reducing the total spin count for 2-body embedding by a factor of $2.5$. For the toric code, ILOD attains a threshold of $4.73%$ versus $4.83%$ for the joint formulation, with the empirical runtime ratio scaling as $(0.81)^d$. For the $6.6.6$ color code, their thresholds agree within statistical uncertainty for small code distances, and ILOD remains convergent for larger distances where the joint formulation fails to converge despite a larger annealing budget.

阅读与讨论 → 访问原文 →

16.

arXiv (CS.CV) 2026-06-11 DOI: arXiv:2511.16672

EvoLMM: Self-Evolving Large Multimodal Models with Continuous Rewards

作者:

Omkar Thawakar↗Shravan Venkatraman↗Ritesh Thawkar↗Abdelrahman Shaker↗Hisham Cholakkal↗Rao Muhammad Anwer↗Salman Khan↗Fahad Khan↗

Recent advances in large multimodal models (LMMs) have enabled impressive reasoning and perception abilities, yet most existing training pipelines still depend on human-curated data or externally verified reward models, limiting their autonomy and scalability. In this work, we strive to improve LMM reasoning capabilities in a purely unsupervised fashion (without any annotated data or reward distillation). To this end, we propose a self-evolving framework, named EvoLMM, that instantiates two cooperative agents from a single backbone model: a Proposer, which generates diverse, image-grounded questions, and a Solver, which solves them through internal consistency, where learning proceeds through a continuous self-rewarding process. This dynamic feedback encourages both the generation of informative queries and the refinement of structured reasoning without relying on ground-truth or human judgments. When using the popular Qwen2.5-VL as the base model, our EvoLMM yields consistent gains upto $\sim$3\% on multimodal math-reasoning benchmarks, including ChartQA, MathVista, and MathVision, using only raw training images. We hope our simple yet effective approach will serve as a solid baseline easing future research in self-improving LMMs in a fully-unsupervised fashion. Our code and models are available at https://github.com/mbzuai-oryx/EvoLMM.

阅读与讨论 → 访问原文 →

17.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2606.14817

Combining Retrieval-Augmented Text Generation with LLMs for Reading Content Recommendations

作者:

Sooyeon Kim↗Piotr S. Maci\k{a}g↗

arXiv:2606.14817v1 Announce Type: cross Abstract: This work presents the design, implementation, and evaluation of a system for generating personalized reading content using Large Language Models (LLMs) combined with Retrieval-Augmented Generation (RAG). The proposed architecture consists of four modules: Input, RAG, Generation, and Judging and enables users to specify both a question and a target reading content complexity. RAG is employed to retrieve relevant information from the Internet, enriching and grounding the content produced by three modern LLMs: Meta LLaMA 4 Scout, LLaMA 3.1 8B Instant, and Google Gemma2 9B. Reading materials are generated using three prompting strategies (Chain-of-Thought, zero-shot, and few-shot), and the LLM-as-a-Judge module automatically evaluates answer quality and alignment with the desired readability level. Experimental results show that RAG consistently improves system performance across all models and prompting techniques, increasing relevance and particularly groundedness by up to 26-35 percentage points. Overall, the findings demonstrate that the RAG-augmented architecture effectively produces reading content tailored to user queries and desired textual complexity.

阅读与讨论 → 访问原文 →

18.

arXiv (CS.AI) 2026-06-24 DOI: arXiv:2507.16696

FISHER: A Foundation Model for Multi-Modal Industrial Signal Comprehensive Representation

作者:

Pingyi Fan↗Anbai Jiang↗Shuwei Zhang↗Xinhu Zheng↗Zhiqiang Lv↗Bing Han↗Wenrui Liang↗Junjie Li↗Wei-Qiang Zhang↗Yanmin Qian↗Xie Chen↗Jia Liu↗…

arXiv:2507.16696v3 Announce Type: replace-cross Abstract: Industrial signal analysis is hindered by severe data heterogeneity, which we characterize as the M5 problem. Existing solutions rely on specialized models that lack robustness and scalability, while large-scale pre-training has rarely been investigated in this area. In this work, we derive a prioritized roadmap for the M5 problem and propose FISHER, a Foundation model for multi-modal Industrial Signal compreHEnsive Representation. To address the foremost multi-sampling-rate problem, FISHER utilizes a novel sub-band modeling approach that treats sampling rate increments as concatenated sub-band information, enabling the adaptive usage of full signal bandwidth without resampling. FISHER is pre-trained by teacher-student self-distillation over external audio and music data. We also establish the RMIS benchmark, comprising 19 datasets across four modalities. In the experiment, FISHER outperforms 24 state-of-the-art series encoders (up to 2B) with much smaller sizes (up to 16x), showcasing groundbreaking diagnostic accuracy and remarkable versatility. We further demonstrate that 1) seamless adaptation to variable sampling rates is the key to generalization 2) audio and music data provide better temporal variability, which is essential for pre-training. Both FISHER and RMIS are open-sourced.

阅读与讨论 → 访问原文 →

19.

arXiv (CS.CL) 2026-06-16 DOI: arXiv:2509.24494

Why Tree-Style Branching Matters for Thought Advantage Estimation in GRPO

作者:

Hongcheng Wang↗Yinuo Huang↗Sukai Wang↗Guanghui Ren↗Hao Dong↗

Group Relative Policy Optimization (GRPO) trains Chain-of-Thought reasoning with verifiable rewards, but estimating thought-level advantages without value functions often suffers from high variance. Although tree-style branching is used in practice to reduce variance, it lacks a theoretical explanation of why it works and whether it is important or potentially necessary. We study thought-level advantage estimation in GRPO from a variance perspective under a minimal tree-style setting where multiple continuations are sampled for each thought. Using the multivariate delta method, we reveal a sampling-dimension asymmetry. Increasing sampled thoughts ($K$) leaves a strictly positive estimation-variance floor, whereas increasing continuations per thought ($M$) drives the leading-order estimation variance to zero at rate $1/M$. This implies that, within the fixed-temperature GRPO-style estimator without value models studied here, accurate thought-level advantage estimation cannot be achieved by scaling thought sampling alone, making continuation-level branching a principled and potentially necessary mechanism rather than a heuristic. Experiments further provide empirical evidence for its effectiveness and potential necessity, demonstrating improved optimization stability, training efficiency, and final performance not only in math but also across vision domains and under different model architectures and sizes.

阅读与讨论 → 访问原文 →

20.

arXiv (CS.LG) 2026-06-11 DOI: arXiv:2512.22219

MPK: A Compiler and Runtime for Mega-Kernelizing Tensor Programs

作者:

Xinhao Cheng↗Zhihao Zhang↗Yu Zhou↗Jianan Ji↗Jinchen Jiang↗Zepeng Zhao↗Ziruo Xiao↗Zihao Ye↗Yingyi Huang↗Ruihang Lai↗Hongyi Jin↗Bohan Hou↗…

arXiv:2512.22219v2 Announce Type: replace-cross Abstract: We introduce Mirage Persistent Kernel (MPK), the first compiler and runtime system that automatically transforms multi-GPU model inference into a single high-performance mega-kernel. MPK introduces an SM-level graph representation that captures data dependencies at the granularity of individual streaming multiprocessors (SMs), enabling cross-operator software pipelining, \rev{fine-grained overlap of computation and communication, and other optimizations that are infeasible under the conventional kernel-per-operator execution model}. The MPK compiler lowers tensor programs into optimized SM-level task graphs and generates fast CUDA implementations for each task, while the MPK in-kernel parallel runtime executes these tasks within a single persistent mega-kernel using decentralized scheduling across SMs. Together, these components provide end-to-end kernel fusion with minimal developer effort, while preserving the flexibility of existing programming models. Our evaluation shows that MPK significantly outperforms existing kernel-per-operator LLM serving systems, achieving up to 1.7$\times$ lower end-to-end inference latency and pushing LLM inference performance close to the limits of the underlying hardware. MPK is publicly available at https://github.com/mirage-project/mirage.

阅读与讨论 → 访问原文 →

21.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2606.15123

Data-Centric Benchmarking of Exploit Generation in LLMs: Understanding the Impact of Fine-Tuning

作者:

Yiwei Chen↗Lichi Li↗Kai Cheung↗Vinny Parla↗Ganesh Sundaram↗

arXiv:2606.15123v1 Announce Type: cross Abstract: We study the task of CVE-conditioned exploit generation, where a model drafts proof-of-concept (PoC) exploits given software vulnerability context. We adopt a data-centric approach, constructing a high-quality dataset via multi-stage preprocessing and introducing a scalable evaluation framework with LLM-as-judge and fine-grained rubrics. Under this unified setup, we benchmark 17 large language models across 8 evaluation criteria, providing systematic insights into their zero-shot capabilities. We further show that a compact 8B open-weight model, when fine-tuned on curated data, achieves over 42.5% improvement in exploit quality and rivals some proprietary models when combined with simple test-time rejection strategies. Our results highlight the importance of data quality, structured supervision, and evaluation design for reliable exploit generation, suggesting that these factors can be as critical as model scale in adapting LLMs to cybersecurity tasks.

阅读与讨论 → 访问原文 →

22.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2602.00781

Fast Non-Episodic Finite-Horizon RL with K-Step Lookahead Thresholding

作者:

Jiamin Xu↗Kyra Gan↗

arXiv:2602.00781v2 Announce Type: replace Abstract: Online reinforcement learning in non-episodic, finite-horizon MDPs remains underexplored and is challenged by the need to estimate returns to a fixed terminal time. Existing infinite-horizon methods, which often rely on discounted contraction, do not naturally account for this fixed-horizon structure. We introduce a modified Q-function: rather than targeting the full-horizon, we learn a K-step lookahead Q-function that truncates planning to the next K steps. To further improve sample efficiency, we introduce a thresholding mechanism: actions are selected only when their estimated K-step lookahead value exceeds a time-varying threshold. We provide an efficient tabular learning algorithm for this novel objective, proving it achieves fast finite-sample convergence: it achieves minimax optimal constant regret for $K=1$ and $\mathcal{O}(\max((K-1),C_{K-1})\sqrt{SAT\log(T)})$ regret for any $K \geq 2$. We numerically evaluate the performance of our algorithm under the objective of maximizing reward. Our implementation adaptively increases K over time, balancing lookahead depth against estimation variance. Empirical results demonstrate superior cumulative rewards over state-of-the-art tabular RL methods across synthetic MDPs and RL environments: JumpRiverswim, FrozenLake and AnyTrading. Code is provided on \href{https://github.com/jamie01713/K-Step-Lookahead}{github}.

阅读与讨论 → 访问原文 →

23.

arXiv (CS.AI) 2026-06-16 DOI: arXiv:2504.11320

Optimizing LLM Inference: Fluid-Guided Online Scheduling with Memory Constraints

作者:

Ruicheng Ao↗Gan Luo↗David Simchi-Levi↗Xinshang Wang↗

arXiv:2504.11320v4 Announce Type: replace-cross Abstract: Large language models now serve millions of users daily, with providers incurring costs exceeding $700,000 per day. Each request requires token-by-token inference, making GPU scheduling central to latency, capacity, and cost. The difficulty is endogenous memory growth: generated tokens expand the Key-Value (KV) cache, and overflow can evict in-progress requests and waste prior computation. We formulate inference as a multi-stage online scheduling problem with endogenous memory growth, linear iteration times, and GPU-resident KV-cache constraints. We introduce a fluid model that characterizes equilibrium batch composition, memory requirement, and stability region. Guided by the fluid model, we design WAIT (Waiting for Accumulated Inference Threshold), a threshold-based admission rule for known output lengths, and Nested WAIT, which extends the rule to unknown output lengths by regulating how requests advance across decode-stage segments. Both algorithms approximate the fluid benchmark asymptotically under the stated memory conditions. Nested WAIT uses an additional safety buffer of moderate scale to hedge against memory-overflow-induced evictions under unknown output lengths. In Vidur simulations configured for Llama-2-7B on an A100 GPU, with supplemental real-GPU validation reported in the appendix, the policies enlarge the empirically observed stable operating range relative to widely used baseline algorithms and reduce latency especially in near-overloaded and overloaded regimes.

阅读与讨论 → 访问原文 →

24.

arXiv (CS.CL) 2026-06-16 DOI: arXiv:2606.16748

MyPCBench: A Benchmark for Personally Intelligent Computer-Use Agents

作者:

Lawrence Keunho Jang↗Andrew Keunwoo Jang↗Jing Yu Koh↗Ruslan Salakhutdinov↗

Current benchmarks for computer-use agents evaluate models in impersonal environments. This leaves a gap between evaluation and deployment where personal assistants are expected to work across a user's whole digital life, including their context, historical data, and logged-in accounts. This gap is widest on web tasks, where live web evaluations cannot exercise sites that require logging in or personal information, the kind of site a real personal assistant has to drive. We introduce MyPCBench, which tests computer-use agents as personal assistants on a Linux desktop populated with 17 simulated real-world web applications and a full desktop stack, all seeded for one canonical persona, Michael Scott from The Office. We define 184 tasks in this environment, each inspired by a real request drawn from the OpenClaw community, and benchmark six closed and open-weight models with a uniform computer+bash tool surface. We find that the best model, Claude Opus 4.6, fully solves 55.4\% of the tasks, the only model above 50\%. Model failures cluster on tasks that span many applications and on long trajectories, where personalization stresses an assistant the most. We release the environment, task set, and agent harness at https://mypcbench.com.

阅读与讨论 → 访问原文 →

25.

arXiv (CS.LG) 2026-06-16 DOI: arXiv:2606.15682

ReQAT: Achieving Full-Precision Reasoning Accuracy with 4-bit Floating-Point Quantization-Aware Training

作者:

Janghwan Lee↗Sihwa Lee↗Jinseok Kim↗Yongjik Kim↗Jieun Lim↗Jinwook Oh↗Jungwook Choi↗

arXiv:2606.15682v1 Announce Type: new Abstract: Large Reasoning Models (LRMs) achieve strong problem-solving through long chain-of-thought, but their deployment is constrained by the high cost of full-precision inference and growing KV cache footprints. Microscaled FP4 formats enable efficient FP4 deployment; however, fully quantizing weights, activations, and KV caches (W4A4KV4) causes severe reasoning degradation that existing PTQ and QAT fail to recover. We identify that FP4 failures concentrate on low-entropy tokens–precise symbolic commitments such as digits and operators–where quantization noise inflates sampling errors that cascade through reasoning traces. Based on this insight, we propose ReQAT, a reasoning-centric FP4 training framework with three components: (i) Trace-Aligned QAT (TAQ), which revisits identical reasoning traces to focus updates on critical low-entropy decisions; (ii) Selective Entropy Minimization (SEM), which reinforces confidence at low-entropy positions; and (iii) Q-FIT, a quantization-friendly initialization that jointly calibrates RoPE-consistent KV cache transformations to stabilize QAT. Under the same training budget, ReQAT not only recovers but surpasses BF16 fine-tuning accuracy, while delivering up to 3.9x throughput speedup on NVIDIA DGX Spark and 3.1x on B200.

阅读与讨论 → 访问原文 →