Search: “benchmark”

LaMSUM: Amplifying Voices Against Harassment through LLM Guided Extractive Summarization of User Incident Reports

arXiv:2406.15809v5 Announce Type: replace Abstract: Citizen reporting platforms help the public and authorities stay informed about sexual harassment incidents. However, the high volume of data shared

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

MMGait: Towards Multi-Modal Gait Recognition

DGX agent

arXiv:2604.15979v1 Announce Type: new Abstract: Gait recognition has emerged as a powerful biometric technique for identifying individuals at a distance without requiring user cooperation. Most existi

model-releasesarxiv-cs-cv

20 Apr 2026

Scalable Maximum Entropy Population Synthesis via Persistent Contrastive Divergence

arXiv:2603.27312v2 Announce Type: replace Abstract: Maximum entropy (MaxEnt) modelling provides a principled framework for generating synthetic populations from aggregate census data, without access t

Symbolic Guardrails for Domain-Specific Agents: Stronger Safety and Security Guarantees Without Sacrificing Utility

arXiv:2604.15579v1 Announce Type: cross Abstract: AI agents that interact with their environments through tools enable powerful applications, but in high-stakes business settings, unintended actions c

safetyarxiv-cs-ai

Tabular foundation models for in-context prediction of molecular properties

arXiv:2604.16123v1 Announce Type: new Abstract: Accurate molecular property prediction is central to drug discovery, catalysis, and process design, yet real-world applications are often limited by sma

applicationsarxiv-cs-lg

The Illusion of Equivalence: Systematic FP16 Divergence in KV-Cached Autoregressive Inference

arXiv:2604.15409v1 Announce Type: cross Abstract: KV caching is a ubiquitous optimization in autoregressive transformer inference, long presumed to be numerically equivalent to cache-free computation.

Hardware

The threat of analytic flexibility in using large language models to simulate human data

arXiv:2509.13397v3 Announce Type: replace-cross Abstract: Social scientists are now using large language models to create 'silicon samples': synthetic datasets intended to stand in for human responden

hardwarearxiv-cs-ai

Transformer Neural Processes - Kernel Regression

arXiv:2411.12502v4 Announce Type: replace-cross Abstract: Neural Processes (NPs) are a rapidly evolving class of models designed to directly model the posterior predictive distribution of stochastic p

When Cultures Meet: Multicultural Text-to-Image Generation

arXiv:2502.15972v2 Announce Type: replace-cross Abstract: Text-to-image generation models have achieved strong performance in culturally homogeneous settings, yet their ability to generate multicultur

Wisdom is Knowing What not to Say: Hallucination-Free LLMs Unlearning via Attention Shifting

arXiv:2510.17210v3 Announce Type: replace Abstract: The increase in computing power and the necessity of AI-assisted decision-making boost the growing application of large language models (LLMs). Alon

AgentGA: Evolving Code Solutions in Agent-Seed Space

arXiv:2604.14655v1 Announce Type: cross Abstract: We present AgentGA, a framework that evolves autonomous code-generation runs by optimizing the agent seed: the task prompt plus optional parent archiv

Auxiliary Finite-Difference Residual-Gradient Regularization for PINNs

arXiv:2604.14472v1 Announce Type: new Abstract: Physics-informed neural networks (PINNs) are often selected by a single scalar loss even when the quantity of interest is more specific. We study a hybr

Can LLMs Score Medical Diagnoses and Clinical Reasoning as well as Expert Panels?

arXiv:2604.14892v1 Announce Type: new Abstract: Evaluating medical AI systems using expert clinician panels is costly and slow, motivating the use of large language models (LLMs) as alternative adjudi

safetyarxiv-cs-lg

Chinese Language Is Not More Efficient Than English in Vibe Coding: A Preliminary Study on Token Cost and Problem-Solving Rate

arXiv:2604.14210v1 Announce Type: new Abstract: A claim has been circulating on social media and practitioner forums that Chinese prompts are more token-efficient than English for LLM coding tasks, po

ContractSkill: Repairable Contract-Based Skills for Multimodal Web Agents

arXiv:2603.20340v3 Announce Type: replace-cross Abstract: Self-generated skills for web agents are often unstable and can even hurt performance relative to direct acting. We argue that the key bottlen

ControlFoley: Unified and Controllable Video-to-Audio Generation with Cross-Modal Conflict Handling

model-releasesarxiv-cs-cv

arXiv:2604.15086v1 Announce Type: cross Abstract: Recent advances in video-to-audio (V2A) generation enable high-quality audio synthesis from visual content, yet achieving robust and fine-grained cont

FAIR Universe Weak Lensing ML Uncertainty Challenge: Handling Uncertainties and Distribution Shifts for Precision Cosmology

model-releasesarxiv-cs-cv

arXiv:2604.14451v1 Announce Type: cross Abstract: Weak gravitational lensing, the correlated distortion of background galaxy shapes by foreground structures, is a powerful probe of the matter distribu

How Embeddings Shape Graph Neural Networks: Classical vs Quantum-Oriented Node Representations

arXiv:2604.15273v1 Announce Type: new Abstract: Node embeddings act as the information interface for graph neural networks, yet their empirical impact is often reported under mismatched backbones, spl

IG-Search: Step-Level Information Gain Rewards for Search-Augmented Reasoning

arXiv:2604.15148v1 Announce Type: cross Abstract: Reinforcement learning has emerged as an effective paradigm for training large language models to perform search-augmented reasoning. However, existin

safetyarxiv-cs-cl

Internal Knowledge Without External Expression: Probing the Generalization Boundary of a Classical Chinese Language Model

arXiv:2604.14180v1 Announce Type: new Abstract: We train a 318M-parameter Transformer language model from scratch on a curated corpus of 1.56 billion tokens of pure Classical Chinese, with zero Englis

Research

Just Pass Twice: Efficient Token Classification with LLMs for Zero-Shot NER

arXiv:2604.05158v2 Announce Type: replace Abstract: Large language models encode extensive world knowledge valuable for zero-shot named entity recognition. However, their causal attention mechanism, w

researcharxiv-cs-cl

Knowing When Not to Answer: Evaluating Abstention in Multimodal Reasoning Systems

arXiv:2604.14799v1 Announce Type: new Abstract: Effective abstention (EA), recognizing evidence insufficiency and refraining from answering, is critical for reliable multimodal systems. Yet existing e

Language Model as Planner and Formalizer under Constraints

arXiv:2510.05486v2 Announce Type: replace Abstract: LLMs have been widely used in planning, either as planners to generate action sequences end-to-end, or as formalizers to represent the planning doma

safetyarxiv-cs-cl

Learning to Think Like a Cartoon Captionist: Incongruity-Resolution Supervision for Multimodal Humor Understanding

arXiv:2604.15210v1 Announce Type: cross Abstract: Humor is one of the few cognitive tasks where getting the reasoning right matters as much as getting the answer right. While recent work evaluates hum

safetyarxiv-cs-cl

MixAtlas: Uncertainty-aware Data Mixture Optimization for Multimodal LLM Midtraining

arXiv:2604.14198v1 Announce Type: cross Abstract: Domain reweighting can improve sample efficiency and downstream generalization, but data-mixture optimization for multimodal midtraining remains large

Physically-Induced Atmospheric Adversarial Perturbations: Enhancing Transferability and Robustness in Remote Sensing Image Classification

model-releasesarxiv-cs-cv

arXiv:2604.14643v1 Announce Type: new Abstract: Adversarial attacks pose a severe threat to the reliability of deep learning models in remote sensing (RS) image classification. Most existing methods r

Retrieve, Then Classify: Corpus-Grounded Automation of Clinical Value Set Authoring

arXiv:2604.14616v1 Announce Type: new Abstract: Clinical value set authoring -- the task of identifying all codes in a standardized vocabulary that define a clinical concept -- is a recurring bottlene

Time-RA: Towards Time Series Reasoning for Anomaly Diagnosis with LLM Feedback

arXiv:2507.15066v5 Announce Type: replace Abstract: Time series anomaly detection (TSAD) has traditionally focused on binary classification and often lacks the fine-grained categorization and explanat

VisRet: Visualization Improves Knowledge-Intensive Text-to-Image Retrieval

arXiv:2505.20291v4 Announce Type: replace-cross Abstract: Text-to-image retrieval (T2I retrieval) remains challenging because cross-modal embeddings often behave as bags of concepts, underrepresenting

VoxSafeBench: Not Just What Is Said, but Who, How, and Where

arXiv:2604.14548v1 Announce Type: cross Abstract: As speech language models (SLMs) transition from personal devices into shared, multi-user environments, their responses must account for far more than

safetyarxiv-cs-lg

Which bird does not have wings: Negative-constrained KGQA with Schema-guided Semantic Matching and Self-directed Refinement

arXiv:2604.14749v1 Announce Type: new Abstract: Large language models still struggle with faithfulness and hallucinations despite their remarkable reasoning abilities. In Knowledge Graph Question Answ

applicationsarxiv-cs-cl

Analog Optical Inference on Million-Record Mortgage Data

arXiv:2604.13251v1 Announce Type: new Abstract: Analog optical computers promise large efficiency gains for machine learning inference, yet no demonstration has moved beyond small-scale image benchmar

Autonomous Multi-objective Alloy Design through Simulation-guided Optimization

arXiv:2507.16005v2 Announce Type: replace-cross Abstract: Alloy discovery is constrained by vast compositional spaces, competing objectives, and prohibitive experimental costs. Although simulations an

CANVAS: Continuity-Aware Narratives via Visual Agentic Storyboarding

arXiv:2604.13452v1 Announce Type: new Abstract: Long-form visual storytelling requires maintaining continuity across shots, including consistent characters, stable environments, and smooth scene trans

Correct Prediction, Wrong Steps? Consensus Reasoning Knowledge Graph for Robust Chain-of-Thought Synthesis

arXiv:2604.14121v1 Announce Type: new Abstract: LLM reasoning traces suffer from complex flaws -- *Step Internal Flaws* (logical errors, hallucinations, etc.) and *Step-wise Flaws* (overthinking, unde

Document-tuning for robust alignment to animals

arXiv:2604.13076v1 Announce Type: new Abstract: We investigate the robustness of value alignment via finetuning with synthetic documents, using animal compassion as a value that is both important in i

Research

DroneScan-YOLO: Redundancy-Aware Lightweight Detection for Tiny Objects in UAV Imagery

arXiv:2604.13278v1 Announce Type: new Abstract: Aerial object detection in UAV imagery presents unique challenges due to the high prevalence of tiny objects, adverse environmental conditions, and stri

researcharxiv-cs-cv

EVE: A Domain-Specific LLM Framework for Earth Intelligence

arXiv:2604.13071v1 Announce Type: new Abstract: We introduce Earth Virtual Expert (EVE), the first open-source, end-to-end initiative for developing and deploying domain-specialized LLMs for Earth Int

Fast training of accurate physics-informed neural networks without gradient descent

arXiv:2405.20836v3 Announce Type: replace-cross Abstract: Solving time-dependent Partial Differential Equations (PDEs) is one of the most critical problems in computational science. While Physics-Info

FlexGuard: Continuous Risk Scoring for Strictness-Adaptive LLM Content Moderation

arXiv:2602.23636v3 Announce Type: replace Abstract: Ensuring the safety of LLM-generated content is essential for real-world deployment. Most existing guardrail models formulate moderation as a fixed

Research

HiProto: Hierarchical Prototype Learning for Interpretable Object Detection Under Low-quality Conditions

arXiv:2604.13981v1 Announce Type: new Abstract: Interpretability is essential for deploying object detection systems in critical applications, especially under low-quality imaging conditions that degr

researcharxiv-cs-cv

HiVLA: A Visual-Grounded-Centric Hierarchical Embodied Manipulation System

arXiv:2604.14125v1 Announce Type: new Abstract: While end-to-end Vision-Language-Action (VLA) models offer a promising paradigm for robotic manipulation, fine-tuning them on narrow control data often

applicationsarxiv-cs-cv

Tutorials

Interpretable Stylistic Variation in Human and LLM Writing Across Genres, Models, and Decoding Strategies

arXiv:2604.14111v1 Announce Type: new Abstract: Large Language Models (LLMs) are now capable of generating highly fluent, human-like text. They enable many applications, but also raise concerns such a

tutorialsarxiv-cs-cl

Leveraging LLM-GNN Integration for Open-World Question Answering over Knowledge Graphs

arXiv:2604.13979v1 Announce Type: new Abstract: Open-world Question Answering (OW-QA) over knowledge graphs (KGs) aims to answer questions over incomplete or evolving KGs. Traditional KGQA assumes a c

MAny: Merge Anything for Multimodal Continual Instruction Tuning

arXiv:2604.14016v1 Announce Type: new Abstract: Multimodal Continual Instruction Tuning (MCIT) is essential for sequential task adaptation of Multimodal Large Language Models (MLLMs) but is severely r

Mosaic: An Extensible Framework for Composing Rule-Based and Learned Motion Planners

model-releasesarxiv-cs-ro

arXiv:2604.13853v1 Announce Type: new Abstract: Safe and explainable motion planning remains a central challenge in autonomous driving. While rule-based planners offer predictable and explainable beha

Neural architectures for resolving references in program code

arXiv:2604.14073v1 Announce Type: new Abstract: Resolving and rewriting references is fundamental in programming languages. Motivated by a real-world decompilation task, we abstract reference rewritin

applicationsarxiv-cs-lg

RAG or Learning? Understanding the Limits of LLM Adaptation under Continuous Knowledge Drift in the Real World

arXiv:2604.05096v2 Announce Type: replace Abstract: Large language models (LLMs) acquire most of their knowledge during pretraining, which ties them to a fixed snapshot of the world and makes adaptati