Ax Naveen George, Naoki Murata, Yuhta Takida, Konda Reddy Mopuri, Yuki Mitsufuji 23d ago

TILDE: TILt-based Distributional Erasure for Concept Unlearning

TILDE method for concept unlearning in text-to-image diffusion models balancing unwanted concept removal with model quality retention.

Ax Yoav Baron, Sara Dorfman, Roni Paiss, Daniel Cohen-Or, Or Patashnik 23d ago

Analysis-by-Proxy: Localization Signals in VLMs Operating as Condition Encoders

Analysis of localization signals in vision-language models used as condition encoders for diffusion-based image editing.

Ax Taeyun Roh, Eunha Lee, Wonjune Jang, Sohyun Chung, Junha Jung, Jaewoo Kang 23d ago

From Voting to Agent Collaboration: Answer-Type-Aware LLM Pipelines for BioASQ 14b

Question-type-aware LLM pipeline framework for biomedical question answering in BioASQ 14b using agent collaboration and evidence grounding.

Ax So Hasegawa, Shailaja Keyur Sampat, Lei Liu, Wei-Peng Chen 23d ago

Data Analysis in the Wild: Benchmarking Large Language Models Against Real-World Data Complexities

DataGovBench benchmark evaluating LLMs on real-world data analysis tasks including multi-tabular datasets and exploratory insight discovery.

Ax Sambaran Bandyopadhyay, Ananth Muppidi 23d ago

RSF-GLLM: Bridging the Semantic Gap in Multi-Hop Knowledge Graph QA via Recurrent Soft-Flow and Decoupled LLM Generation

RSF-GLLM framework for multi-hop knowledge graph QA using differentiable graph reasoning with Recurrent Soft-Flow module and decoupled LLM generation.

Ax Shervin Khalafi, Igor Krawczuk, Sergio Rozada, Charilaos Kanatsoulis, Antonio G Marques, Alejandro Ribeiro 23d ago

Graph Convolutional Attention: A Spectral Perspective on Graph Denoising and Diffusion

arXiv paper analyzing graph attention mechanisms for denoising from spectral perspective in graph diffusion models.

Ax Tianjiao Yu, Xinzhuo Li, Yifan Shen, Onkar Susladkar, Yuanzhe Liu, Xiaona Zhou, Ismini Lourentzou 23d ago

ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation

arXiv paper on ELSA3D unified 3D foundation model with elastic semantic anchoring for improved text-3D interaction and explicit structural representation.

Ax Jianyi Zhang, Hao Frank Yang, Ang Li, Xin Guo, Pu Wang, Haiming Wang, Yiran Chen, Hai Li 23d ago

MLLM-LLaVA-FL: Multimodal Large Language Model Assisted Federated Learning

arXiv paper proposing federated learning framework using multimodal LLMs to address data heterogeneity across distributed clients.

Ax Satiyabooshan Murugaboopathy, Connor T. Jerzak, Adel Daoud 23d ago

Platonic Representations for Poverty Mapping: Unified Vision-Language Codes or Agent-Induced Novelty?

arXiv paper investigating socioeconomic indicators in satellite imagery and LLM-generated text for poverty mapping in African neighborhoods.

Ax Constantin Venhoff, Iv\'an Arcuschin, Philip Torr, Arthur Conmy, Neel Nanda 23d ago

Base Models Know How to Reason, Thinking Models Learn When

arXiv paper discovering what thinking models learn via Sparse Autoencoders on reasoning traces and constructive model diffing analysis.

Ax Gil Pasternak, Dheeraj Rajagopal, Julia White, Dhruv Atreja, Matthew Thomas, George Hurn-Maloney, Ash Lewis 23d ago

Beyond Reactivity: Measuring Proactive Problem Solving in LLM Agents

arXiv paper introducing PROBE benchmark for measuring proactive problem-solving in LLM agents across longer time horizons and multiple information sources.

Ax Claire Yang, Claire Jie Zhang, Maya Cakmak, Max Kleiman-Weiner 23d ago

When Assisting One Disempowers Another

arXiv paper formalizing bystander disempowerment: how AI agents optimizing for one user unintentionally erode bystanders' agency in shared environments.

Ax Zeyu Xia, Jinzhe Ma, Congjie Zheng, Zhongyao Wang, Shufei Zhang, Yuqiang Li, Hang Su, P. Hu, Changshui Zhang, Xingao Gong, Wanli Ouyang, Lei Bai, Dongzhan Zhou, Mao Su 23d ago

VASP Agent: An Agentic Framework for Autonomous First-principles Calculations

arXiv paper presenting VASP Agent, LLM-based coding agent for autonomous first-principles materials computation with workspace-state management.

Ax Joyjit Roy, Samaresh Kumar Singh 23d ago

Agentic AI for Commercial Insurance Underwriting with Adversarial Self-Critique

arXiv paper on agentic AI system for insurance underwriting with adversarial self-critique mechanism ensuring reliability in regulated environments.

Ax Yufeng Wang 23d ago

Doing What They Say, Not What They Reason: Locating the Faithfulness Gap in LLM Agents

arXiv paper measuring faithfulness gap in LLM agents: whether agents act on stated reasoning vs conclusions in controlled Texas Poker simulator.

Ax Ziyuan Li, Yueru Sun, Yimeng Zhang 23d ago

EVA-Net: Subject-Independent EEG Motor Decoding with Video-Derived Motor Priors

arXiv paper on EEG motor decoding using video-derived motor priors for cross-subject brain-computer interface generalization without calibration.

Ax Wouter W. L. Nuijten, Mykola Lukashchuk, Thijs van de Laar, Bert de Vries 23d ago

What Type of Inference is Active Inference?

arXiv paper proving active inference can be rewritten as VFE minimization with entropy-correction terms, unifying goal-directed and information-seeking behavior.

Ax Guojun Liao 23d ago

A Three-Layer Framework for AI in Scientific Discovery

Three-layer framework for AI in scientific discovery: search/retrieval, model formation/evolution, and execution via optimization/simulation.

Ax Jasmine Brazilek, Joel Christoph, Maheep Chaudhary, Oliver Tullio, Carol Kline, Miles Tidmarsh, Arturs Kanepajs 23d ago

Your AI Travel Agent Would Book You a Bullfight: An Agentic Benchmark for Implicit Animal Welfare in Frontier AI Models

TAC benchmark: first agentic benchmark assessing animal welfare in AI agent behavior, extending beyond question-answer evaluations.

Ax Pu Li, Zhigang Lin, Qiang Wu, Yongxuan Lv, Fei Wang, Shan You 23d ago

Reward as An Agent for Embodied World Models

Framework using reward functions as agents to enable broader exploration in embodied world models beyond conservative training distributions.

Ax Yichen Guo, Kai Tang, Fenglai Lin, Yiding Sun, Dongxu Zhang, Wenya Wang, Lin William Cong, Shanghang Zhang 23d ago

FADE: Mitigating Hallucinations by Reducing Language-Prior Dominance in Large Vision-Language Models

FADE: mechanism for reducing language-prior dominance in vision-language models to mitigate hallucinations through information flow analysis.

Ax Shei Pern Chua, Hao Wu, Fangzhao Wu 23d ago

HARC: Coupling Harmfulness and Refusal Directions for Robust Safety Alignment

HARC: approach coupling harmfulness and refusal directions in residual streams to improve LLM safety alignment robustness.

Ax Akhiad Bercovich, Talor Abramovich, Daniel Afrimi, Shay Aharon, Nir Ailon, Vladimir Anisimov, Omer Ullman Argov, Maor Ashkenazi, Tomer Asida, Nave Assaf, Tomer Bar Natan, Alexander Bukharin, Grzegorz Chlebus, Marcin Chochowski, Eric Chung, Mohammad Dabbah, Carlo del Mundo, Ewa Dobrowolska, Ido Galil, Yaniv Galron, Amnon Geifman, Yonatan Geifman, Izik Golan, Alex Gronskiy, Tomasz Grzegorzek, Netanel Haber, Lior Kadoch, Grzegorz Karch, Tomer Keren, Abhinav Khattar, Amir Klein, Tugrul Konuk, Roi Koren, Daniel Korzekwa, Shaun Kotek, Konstantinos Krommydas, Itay Levy, Ofri Masad, Yoav Miron, Pavlo Molchanov, Shahar Mor, Zach Moshe, Saurav Muralidharan, Najeeb Nabwani, Besmira Nushi, Mostofa Patwary, Omri Puny, Johannes Rausch, Tomer Ronen, Sepehr Sameni, Itamar Schen, Elad Segal, Daniel Serebrenik, Ido Shahaf, Soumye Singhal, Daniil Sorokin, Sharath Turuvekere Sreenivas, Marta Stepniewska-Dziubinska, Ali Taghibakhshi, Nima Tajbakhsh, Oren Tropp, Dor Tzur, Anna Warno, Yi-Fu Wu, Michal Zawalski, Jiaqi Zeng, Yian Zhang, Ran Zilberstein, Amit Zuker, Ran El-Yaniv 23d ago

Nemotron-Labs-3-Puzzle-75B-A9B: Compressing Hybrid MoE LLMs

Nemotron-Labs-3-Puzzle-75B-A9B: compressed hybrid MoE LLM variant optimized for interactive deployment with 2x throughput improvement.

Ax Andrew Zhang, Chengzhan Li 23d ago

Agent Step Value: Probing the Observer Effect in Black-Box Traces

Agent Step Value: replay framework for evaluating which agent transitions helped or harmed task traces using LLM evaluators.

Ax Jacky Kwok, Shulu Li, Pranav Atreya, Yuejiang Liu, Yixing Jiang, Chelsea Finn, Marco Pavone, Ion Stoica, Azalia Mirhoseini 23d ago

LLM-as-a-Verifier: A General-Purpose Verification Framework

LLM-as-a-Verifier: framework for verification as a scaling axis, providing fine-grained feedback for agentic tasks without ground truth.

Ax Wenhao Wang, Yifan Sun, Zongxin Yang, Zhengdong Hu, Zhentao Tan, Yi Yang 23d ago

Replication in Visual Diffusion Models: A Survey and Outlook

Survey of memorization and replication phenomena in visual diffusion models, covering privacy, security, and copyright concerns.

Ax Shengsheng Zhou, Shuai Wang, Liang Ding, Yibing Zhan, Yong Luo, Zheng He, Fu Lin, Dapeng Tao 23d ago

Leveraging Metamemory Agent for Enhanced Data-Free Code Generation in Large Language Models

Metamemory Agent approach for data-free code generation in LLMs without requiring manually curated reference examples.

Ax Marthe Ballon, Andres Algaba, Vincent Ginis 23d ago

The relationship between reasoning and performance in large language models--o3 (mini) thinks harder, not longer

Analysis of o3 reasoning token usage versus accuracy gains in LLMs, showing improved efficiency rather than longer reasoning chains.

Ax Eric Tang, Jing Liu, Marcel B\"ohme 23d ago

Empirical Computation: Prompting versus Programming

Vision paper exploring LLMs solving computational problems via prompting rather than programming, examining challenges and opportunities.

Ax Christiaan Meijer, E. G. Patrick Bos 23d ago

Explainable embeddings with Distance Explainer

Distance Explainer: novel post-hoc method for interpreting embeddings by explaining distances between data points using saliency-based attribution.

Ax Alina Wernick, Kristof Meding 23d ago

Position: EU AI Act's Research Exemptions Can Break the Publication Norms of Major AI Conferences

EU AI Act's research exemptions and their potential impact on publication norms at major AI conferences.

Ax Lorenzo Steccanella, Joshua B. Evans, \"Ozg\"ur \c{S}im\c{s}ek, Anders Jonsson 23d ago

Learning The Minimum Action Distance

Framework learning minimum action distance from state trajectories for environment structure representation in MDPs.

Ax Rohitash Chandra, Jiyong Choi, Jayesh Sonawane 23d ago

Detoxify: A framework for abusive text transformation using LLMs

Detoxify: LLM framework for transforming abusive text to non-abusive while preserving meaning.

Ax David Chanin, Adri\`a Garriga-Alonso 23d ago

Sparse but Wrong: Incorrect L0 Leads to Incorrect Features in Sparse Autoencoders

Research on sparse autoencoders for LLM interpretability showing L0 hyperparameter effects on feature quality.

Ax Amin Omidvar 23d ago

SmartMixed: A Two-Phase Training Strategy for Adaptive Activation Function Learning in Neural Networks

SmartMixed: two-phase training strategy enabling per-neuron adaptive activation functions in neural networks.

Ax Juntao Cheng, Wanyue Zhang, Zhiwei Yu, Shuo Ren, Zheqi He, Shaoxuan Xie, Guocai Yao, Jieru Lin, B\"orje F. Karlsson, Jiajun Zhang 23d ago

SWITCH: Benchmarking Modeling and Handling of Tangible Interfaces in Long-horizon Embodied Scenarios

SWITCH: Benchmark for embodied agents handling tangible control interfaces in long-horizon scenarios.

Ax Scott Thornton 23d ago

SecureCode: A Production-Grade Multi-Turn Dataset for Training Security-Aware Code Generation Models

SecureCode: Production-grade dataset of 2,185 multi-turn examples for training security-aware code generation models.

Ax Wenyuan Huang, Zhenyu Zhang, Zhao Wang, Zhou Wei, Ting Huang, Fang Zhao, Jian Yang 23d ago

OpenGround: Planning-based Online Perception for Open-World 3D Visual Grounding

OpenGround: Planning-based perception for zero-shot 3D visual grounding in open-world scenarios without predefined object lookups.

Ax Gang Liao, Hongsen Qin, Ying Wang, Alicia Golden, Michael Kuchnik, Yavuz Yetim, Jia Jiunn Ang, Chunli Fu, Yihan He, Samuel Hsia, Zewei Jiang, Dianshi Li, Uladzimir Pashkevich, Varna Puvvada, Feng Shi, Matt Steiner, Ruichao Xiao, Liyuan Li, Nathan Yan, Xiayu Yu, Zhou Fang, Roman Levenstein, Kunming Ho, Haishan Zhu, Alec Hammond, Richard Li, Ajit Mathews, Kaustubh Gondkar, Abdul Zainul-Abedin, Ketan Singh, Hongtao Yu, Wenyuan Chi, Barney Huang, Sean Zhang, Noah Weller, Zach Marine, Wyatt Cook, Carole-Jean Wu, Gaoxiang Liu 23d ago

KernelEvolve: Scaling Agentic Kernel Coding for Heterogeneous AI Accelerators at Meta

KernelEvolve: Agentic framework for automated kernel coding targeting heterogeneous AI accelerators at scale for DLRM.

Ax Hunzalah Hassan Bhatti, Firoj Alam, Shammur Absar Chowdhury 23d ago

Multi-Task Instruction Tuning via Data Scheduling for Low-Resource Arabic SpeechLLMs

Multi-task instruction tuning for Arabic-English audio LLM covering ASR, speech/text summarization via data scheduling.

Ax Gloria Felicia (University of Virginia), Zitha Sasindran (Indian Institute of Science Bangalore), Jinfeng He (Cornell University), Michael Eniolade (University of the Cumberlands), Hemant Kumar (University of Arizona), Milan Hussain Angati (California State University Northridge) 23d ago

StepShield: When, Not Whether to Intervene on Rogue Agents

StepShield: First agent safety benchmark measuring detection timeliness for rogue agents via Early Intervention Rate metric.

Ax Stefano Woerner, Seong Joon Oh, Christian F. Baumgartner 23d ago

Universal Algorithm-Implicit Learning

Theoretical framework for universal meta-learning with formal definitions of practical universality across task distributions.

Ax Joshua S. Schiffman 23d ago

Transformers converge to invariant algorithmic cores

Analysis of transformer algorithm cores: extracting compact, necessary subspaces that recur across independent training runs.

Ax Rahul Marchand, Art O Cathain, Jerome Wynne, Philippos Maximos Giavridis, Sam Deverett, John Wilkinson, Jason Gwartz, Harry Coppock 23d ago

Quantifying Frontier LLM Capabilities for Container Sandbox Escape

SandboxEscapeBench: Open benchmark measuring LLM agent capabilities to escape container sandbox environments, addressing agentic AI security risks.

Ax Yibin Liu, Yaxing Lyu, Daqi Gao, Zhixuan Liang, Weiliang Tang, Shilong Mu, Xiaokang Yang, Yao Mu 23d ago

From Passive Observer to Active Critic: Reinforcement Learning Elicits Process Reasoning for Robotic Manipulation

PRIMO R1: Reinforcement learning framework enabling video MLLMs to evaluate robotic manipulation trajectories against task goals.

Ax Sohwi Lim, Lee Hyoseok, Jungjoon Park, Tae-Hyun Oh 23d ago

CLAY: Conditional Visual Similarity Modulation in Vision-Language Embedding Space

CLAY: Adaptive similarity computation for vision-language models that incorporates multiple conditions for flexible image retrieval.

Ax Chi-Sheng Chen, En-Jui Kuo 23d ago

Do Quantum Transformers Help? A Systematic VQC Architecture Comparison on Tabular Benchmarks

Empirical comparison of variational quantum circuit architectures including quantum transformers on classical tabular benchmarks.

Ax Mofei Li, Taozhi Chen, Guowei Yang, Jia Li 23d ago

Learning from Execution: Self-Evolving Memory for Private-Library Code Generation

Self-evolving memory system for LLMs to improve code generation on private enterprise libraries through execution-based learning and RAG.

Ax Evelyn Turri, Davide Bucciarelli, Sara Sarto, Lorenzo Baraldi, Marcella Cornia 23d ago

Few Channels Draw The Whole Picture: Revealing Massive Activations in Diffusion Transformers

Analysis of massive activations in Diffusion Transformers revealing sparse channels controlling image semantics in text-to-image generation.

Ax Michael S. Lee, Yash Maurya, Drew Rein, Bert Herring, Jonathan Nguyen, Kyungho Song, Udari Madhushani Sehwag, Jiyeon Cho, Kaustubh Deshpande, Yeongkyun Jang, Jiyeon Joo, Minn Seok Choi, Evi Fuelle, Christina Q. Knight, Joseph Brandifino, Max Fenkell 23d ago

ROK-FORTRESS: Measuring the Effect of Geopolitical Transcreation for National Security and Public Safety

ROK-FORTRESS bilingual benchmark measuring LLM safety risks across geopolitical and cultural contexts for national security scenarios.