Search: “benchmark”

Test-Time Compute Games

arXiv:2601.21839v2 Announce Type: replace-cross Abstract: Test-time compute has emerged as a promising strategy to enhance the reasoning abilities of large language models (LLMs). However, this strate

The Convergence Gap: Instruction-Tuned Language Models Stabilize Later in the Forward Pass

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

arXiv:2605.07282v1 Announce Type: new Abstract: Final outputs hide when a checkpoint commits to its next-token prediction. We introduce the convergence gap, a model-diffing diagnostic that decodes eac

model-releasesarxiv-cs-lg

11 May 2026

The Position Curse: LLMs Struggle to Locate the Last Few Items in a List

model-releasesarxiv-cs-cl

arXiv:2605.07127v1 Announce Type: cross Abstract: Modern large language models (LLMs) can find a needle in a haystack (locating a single relevant fact buried among hundreds of thousands of irrelevant

Theory of Optimal Learning Rate Schedules and Scaling Laws for a Random Feature Model

arXiv:2602.04774v2 Announce Type: replace-cross Abstract: Setting the learning rate (LR) for a deep learning model is a critical part of successful training. Choosing LRs is often done empirically wit

Think-with-Rubrics: From External Evaluator to Internal Reasoning Guidance

arXiv:2605.07461v1 Announce Type: new Abstract: Rubrics have been extensively utilized for evaluating unverifiable, open-ended tasks, with recent research incorporating them into reward systems for re

ThinKV: Thought-Adaptive KV Cache Compression for Efficient Reasoning Models

arXiv:2510.01290v2 Announce Type: replace Abstract: The long-output context generation of large reasoning models enables extended chain of thought (CoT) but also drives rapid growth of the key-value (

Toeplitz MLP Mixers are Low Complexity, Information-Rich Sequence Models

arXiv:2605.06683v1 Announce Type: cross Abstract: Transformer-based large language models are in some respects limited by the quadratic time and space computational complexity of attention. We introdu

Tools as Continuous Flow for Evolving Agentic Reasoning

arXiv:2605.07339v1 Announce Type: new Abstract: Large Language Models (LLMs) have demonstrated remarkable capabilities in orchestrating tools for reasoning tasks. However, existing methods rely on a s

TRACE: Tourism Recommendation with Accountable Citation Evidence

arXiv:2605.07677v1 Announce Type: cross Abstract: Tourism is a high-stakes setting for conversational recommender systems (CRS): a plausible-sounding suggestion can waste real money and trip time once

researcharxiv-cs-ai

Traffic Scenario Orchestration from Language via Constraint Satisfaction

model-releasesarxiv-cs-ro

arXiv:2605.06966v1 Announce Type: new Abstract: Autonomous vehicles (AVs) require extensive testing in simulation, but test case generation for driving scenarios is laborious. The desired scenarios ar

Training-Free Multimodal Large Language Model Orchestration

arXiv:2508.10016v3 Announce Type: replace Abstract: Building interactive omni-modal assistants often relies on end-to-end multimodal alignment to fuse heterogeneous modalities, which incurs substantia

Tree SAE: Learning Hierarchical Feature Structures in Sparse Autoencoders

arXiv:2605.07922v1 Announce Type: new Abstract: Learning hierarchical features in Sparse Autoencoders (SAEs) is essential for capturing the structured nature of real-world data and mitigating issues l

tutorialsarxiv-cs-lg

TRUST: Test-Time Refinement using Uncertainty-Guided SSM Traverses

arXiv:2509.22813v2 Announce Type: replace Abstract: State Space Models (SSMs) have emerged as efficient alternatives to Vision Transformers (ViTs), with VMamba standing out as a pioneering architectur

tutorialsarxiv-cs-cv

UNA: A Unified Supervised Framework for Efficient LLM Alignment Across Feedback Types

arXiv:2408.15339v4 Announce Type: replace-cross Abstract: RL alignment methods, including RLHF and DPO, are primarily based on pairwise preference data. Although scalar or score-based feedback has bee

UNCOM: Zero-shot Context-Aware Command Understanding for Tabletop Scenarios

arXiv:2410.06355v3 Announce Type: replace-cross Abstract: This paper presents UNCOM, a novel hybrid framework for interpreting natural human commands in tabletop scenarios. The system integrates multi

Applications

Uncovering and Shaping the Latent Representation of 3D Scene Topology in Vision-Language Models

arXiv:2605.07148v1 Announce Type: new Abstract: Decades of cognitive science establish that humans navigate environments by forming cognitive maps, defined as allocentric and topology-preserving repre

applicationsarxiv-cs-cv

Understanding Robustness of Model Editing in Code LLMs

arXiv:2511.03182v2 Announce Type: replace-cross Abstract: Large language models (LLMs) for code are increasingly used in software development, but they remain static after pretraining while APIs and s

Uneven Evolution of Cognition Across Generations of Generative AI Models

arXiv:2605.06815v1 Announce Type: new Abstract: The pursuit of artificial general intelligence necessitates robust methods for evaluating the cognitive capabilities of models beyond narrow task perfor

UniD-Shift: Towards Unified Semantic Segmentation via Interpretable Share-Private Multimodal Decomposition

arXiv:2605.07356v1 Announce Type: new Abstract: Semantic segmentation of large-scale 3D point clouds is crucial for applications such as autonomous driving and urban digital twins. However, the sparse

safetyarxiv-cs-cv

UniV2D: Bridging Visual Restoration and Semantic Perception for Underwater Salient Object Detection

arXiv:2605.07146v1 Announce Type: new Abstract: Underwater salient object detection (USOD) plays a vital role in marine vision tasks but remains fundamentally challenging due to severe visual degradat

tutorialsarxiv-cs-cv

Local Ai

Unlocking High-Fidelity Molecular Generation from Mass Spectra via Dual-Stream Line Graph Diffusion

arXiv:2605.07048v1 Announce Type: cross Abstract: De novo molecular generation from tandem mass spectra is a challenging inverse problem whose core difficulty lies in the circular dependency between a

local-aiarxiv-cs-ai

Unsolvability Ceiling in Multi-LLM Routing: An Empirical Study of Evaluation Artifacts

arXiv:2605.07395v1 Announce Type: cross Abstract: Efficient routing across multiple LLMs enables cost-quality tradeoffs by directing queries to the cheapest capable model. Prior work attributes routin

VecCISC: Improving Confidence-Informed Self-Consistency with Reasoning Trace Clustering and Candidate Answer Selection

arXiv:2605.08070v1 Announce Type: new Abstract: A standard technique for scaling inference-time reasoning is Self-Consistency, whereby multiple candidate answers are sampled from an LLM and the most c

researcharxiv-cs-ai

VISD: Enhancing Video Reasoning via Structured Self-Distillation

arXiv:2605.06094v2 Announce Type: replace-cross Abstract: Training VideoLLMs for complex reasoning remains challenging due to sparse sequence level rewards and the lack of fine grained credit assignme

safetyarxiv-cs-ai

Visual Text Compression as Measure Transport

arXiv:2605.06708v1 Announce Type: cross Abstract: Visual text compression (VTC) promises efficient long-context processing by rendering text into an image and re-encoding it with a vision-language mod

VLA-GSE: Boosting Parameter-Efficient Fine-Tuning in VLA with Generalized and Specialized Experts

model-releasesarxiv-cs-ro

arXiv:2605.06175v2 Announce Type: replace Abstract: Vision-language-action (VLA) models inherit rich visual-semantic priors from pre-trained vision-language backbones, but adapting them to robotic con

Applications

Weblica: Scalable and Reproducible Training Environments for Visual Web Agents

arXiv:2605.06761v1 Announce Type: new Abstract: The web is complex, open-ended, and constantly changing, making it challenging to scale training data for visual web agents. Existing data collection at

applicationsarxiv-cs-ai

When Routine Chats Turn Toxic: Unintended Long-Term State Poisoning in Personalized Agents

model-releasesarxiv-cs-cl

arXiv:2605.06731v1 Announce Type: cross Abstract: Personalized LLM agents maintain persistent cross-session state to support long-horizon collaboration. Yet, this persistence introduces a subtle but c

Why do Large Language Models Fail in Low-resource Translation? Unraveling the Token Dynamics of Large Language Models for Machine Translation

arXiv:2605.07533v1 Announce Type: new Abstract: Large Language Models (LLMs) have recently demonstrated strong performance in machine translation (MT). However, most prior work focuses on improving or

researcharxiv-cs-cl

Why Self-Inconsistency Arises in GNN Explanations and How to Exploit It

arXiv:2605.07527v1 Announce Type: cross Abstract: Recent work has observed that explanations produced by Self-Interpretable Graph Neural Networks (SI-GNNs) can be self-inconsistent: when the model is

WiCER: Wiki-memory Compile, Evaluate, Refine Iterative Knowledge Compilation for LLM Wiki Systems

arXiv:2605.07068v1 Announce Type: cross Abstract: The LLM Wiki pattern, to compile and provide domain knowledge into a persistent artifact and serve it to LLMs via KV cache inference, promises context

Your Language Model is Its Own Critic: Reinforcement Learning with Value Estimation from Actor's Internal States

arXiv:2605.07579v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) for Large Reasoning Models hinges on baseline estimation for variance reduction, but existing ap

Applications

A Foundation Model for Zero-Shot Logical Rule Induction

arXiv:2605.04916v1 Announce Type: cross Abstract: Inductive Logic Programming (ILP) learns interpretable logical rules from data. Existing methods are transductive: their learned parameters are bound

applicationsarxiv-cs-lg

A large-scale heterogeneous 3D magnetic resonance brain imaging dataset for self-supervised learning

arXiv:2506.14432v3 Announce Type: replace-cross Abstract: We present FOMO260K, a large-scale, heterogeneous dataset of 260,927 brain Magnetic Resonance Imaging (MRI) scans from 77,589 MRI sessions and

researcharxiv-cs-cv

Adapt to Thrive! Adaptive Power-Mean Policy Optimization for Improved LLM Reasoning

arXiv:2605.04066v1 Announce Type: new Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) is an essential paradigm that enhances the reasoning capabilities of Large Language Models (LLMs).

Adaptive Consensus in LLM Ensembles via Sequential Evidence Accumulation: Automatic Budget Identification and Calibrated Commit Signals

arXiv:2605.04236v1 Announce Type: new Abstract: Large Language Model ensembles improve reasoning accuracy up to a performance boundary; beyond it, additional deliberation degrades accuracy. Static-bud

researcharxiv-cs-lg

Adaptive Policy Selection and Fine-Tuning under Interaction Budgets for Offline-to-Online Reinforcement Learning

arXiv:2605.05123v1 Announce Type: new Abstract: In offline-to-online reinforcement learning (O2O-RL), policies are first safely trained offline using previously collected datasets and then further fin

safetyarxiv-cs-lg

Advancing Analytic Class-Incremental Learning through Vision-Language Calibration

arXiv:2602.13670v2 Announce Type: replace Abstract: Class-incremental learning (CIL) with pre-trained models (PTMs) faces a critical trade-off between efficient adaptation and long-term stability. Whi

safetyarxiv-cs-lg

Aes3D: Aesthetic Assessment in 3D Gaussian Splatting

model-releasesarxiv-cs-cv

arXiv:2605.05155v1 Announce Type: new Abstract: As 3D Gaussian Splatting (3DGS) gains attention in immersive media and digital content creation, assessing the aesthetics of 3D scenes becomes important

Agentic Vulnerability Reasoning on Windows COM Binaries

arXiv:2605.05000v1 Announce Type: cross Abstract: Windows Component Object Model (COM) services run with elevated privileges and are widely accessible to authenticated users, making race conditions in

ANDRE: An Attention-based Neuro-symbolic Differentiable Rule Extractor

arXiv:2605.04193v1 Announce Type: cross Abstract: Inductive Logic Programming (ILP) aims to learn interpretable first-order rules from data, but existing symbolic and neuro-symbolic approaches struggl

tutorialsarxiv-cs-lg

Local Ai

Angle-I2P: Angle-Consistent-Aware Hierarchical Attention for Cross-Modality Outlier Rejection

arXiv:2605.04541v1 Announce Type: new Abstract: Image-to-point-cloud registration (I2P) is a fundamental task in robotic applications such as manipulation,grasping, and localization. Existing deep lea

local-aiarxiv-cs-cv

ARTA: Adversarial-Robust Multivariate Time--Series Anomaly Detection via Sparsity-Constrained Perturbations

arXiv:2603.25956v2 Announce Type: replace Abstract: Time-series anomaly detection (TSAD) is a critical component in monitoring complex systems, yet modern deep learning-based detectors are often highl

ArtiFixer: Enhancing and Extending 3D Reconstruction with Auto-Regressive Diffusion Models

arXiv:2603.00492v2 Announce Type: replace Abstract: Per-scene optimization methods such as 3D Gaussian Splatting provide state-of-the-art novel view synthesis quality but extrapolate poorly to under-o

researcharxiv-cs-cv

Assessing Cognitive Effort in L2 Idiomatic Processing: An Eye-Tracking Dataset

model-releasesarxiv-cs-cl

arXiv:2605.04857v1 Announce Type: new Abstract: This paper presents the development and validation of an eye-tracking dataset designed to investigate how second-language (L2) learners process idiomati

Awaking Spatial Intelligence in Unified Multimodal Understanding and Generation

arXiv:2605.04128v1 Announce Type: cross Abstract: We present JoyAI-Image, a unified multimodal foundation model for visual understanding, text-to-image generation, and instruction-guided image editing

researcharxiv-cs-cl

Back to Blackwell: Closing the Loop on Intransitivity in Multi-Objective Preference Fine-Tuning

arXiv:2602.19041v2 Announce Type: replace Abstract: A recurring challenge in preference fine-tuning (PFT) is handling extit{intransitive} (i.e., cyclic) preferences. Intransitive preferences often ste

Balanced Aggregation: Understanding and Fixing Aggregation Bias in GRPO

arXiv:2605.04077v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) has become a central paradigm for improving reasoning and code generation in large language mode