Ax Rupal Nigam, Niket Parikh, Hamid Osooli, Mikihisa Yuasa, Jacob Heglund, Huy T. Tran 4/1/2026

Zero-Shot Coordination in Ad Hoc Teams with Generalized Policy Improvement and Difference Rewards

Zero-shot coordination method for ad hoc multi-agent teams leveraging all pretrained policies for transfer without retraining.

Ax Wonduk Seo, Juhyeon Lee, Junseo Koh, Wonseok Choi, Hyunjin An, Jian Park, Seunghyun lee, Haihua Chen, Yi Bu 4/1/2026

MA-SAPO: Multi-Agent Reasoning for Score-Aware Prompt Optimization

Multi-agent reasoning framework for systematic prompt optimization in LLMs with interpretable score-aware improvement guidance.

Ax Floris-Jan Willemsen, Niki van Stein, Ben van Werkhoven 4/1/2026

Automated Algorithm Design for Auto-Tuning Optimizers

Automated algorithm design paradigm for auto-tuning optimizers across diverse irregular search spaces in high-performance computing.

Ax Hyeongkyun Kim, Orestis Oikonomou 4/1/2026

ZeroFlood: Flood Hazard Mapping from Single-Modality SAR Using Geo-Foundation Models

Framework using geo-foundation models for flood hazard mapping from SAR satellite imagery in data-scarce regions.

Ax Taiyi Su, Jian Zhu, Yaxuan Li, Chong Ma, Jianjun Zhang, Zitai Huang, Hanli Wang, Yi Xu 4/1/2026

Towards High-Consistency Embodied World Model with Multi-View Trajectory Videos

Embodied world model using multi-view trajectory videos to improve consistency in action-to-movement translation for robotic prediction.

Ax Minyoung Hwang, Alexandra Forsey-Smerek, Nathaniel Dennler, Andreea Bobu 4/1/2026

Masked IRL: LLM-Guided Reward Disambiguation from Demonstrations and Language

Inverse reinforcement learning approach using LLM guidance and language to disambiguate robot rewards from demonstrations.

Ax Alberto Compagnoni, Marco Morini, Sara Sarto, Federico Cocchi, Davide Caffagni, Marcella Cornia, Lorenzo Baraldi, Rita Cucchiara 4/1/2026

ReAG: Reasoning-Augmented Generation for Knowledge-based Visual Question Answering

Knowledge-based VQA system augmenting multimodal LLMs with external document retrieval and reasoning for domain-specific visual questions.

Ax Weiqi Li, Quande Zhang, Ruifeng Zhai, Liang Lin, Guangrun Wang 4/1/2026

VLA Models Are More Generalizable Than You Think: Revisiting Physical and Spatial Modeling

One-shot adaptation framework improving vision-language-action model generalization across novel camera viewpoints and visual perturbations.

Ax Pranav Pushkar Mishra, Kranti Prakash Yeole, Ramyashree Keshavamurthy, Mokshit Bharat Surana, Fatemeh Sarayloo 4/1/2026

A Systematic Framework for Enterprise Knowledge Retrieval: Leveraging LLM-Generated Metadata to Enhance RAG Systems

Systematic framework for enterprise knowledge retrieval using LLM-generated metadata to enhance RAG system performance.

Ax Hongyuan Tao, Bencheng Liao, Shaoyu Chen, Haoran Yin, Qian Zhang, Wenyu Liu, Xinggang Wang 4/1/2026

InfiniteVL: Synergizing Linear and Sparse Attention for Highly-Efficient, Unlimited-Input Vision-Language Models

Vision-language model combining linear and sparse attention for efficient unlimited-input processing with high-frequency visual perception.

Ax Mingzhi Chen, Taiming Lu, Jiachen Zhu, Mingjie Sun, Zhuang Liu 4/1/2026

Stronger Normalization-Free Transformers

Alternative activation functions surpassing Dynamic Tanh for stable training in normalization-free transformer architectures.

Ax Allen Liu 4/1/2026

Provably Extracting the Features from a General Superposition

Learning theoretic study on extracting features in superposition from complex ML models, advancing interpretability research.

Ax San Kim, Gary Geunbae Lee 4/1/2026

Merging Triggers, Breaking Backdoors: Defensive Poisoning for Instruction-Tuned Language Models

Defense mechanism against backdoor attacks in instruction-tuned LLMs using defensive poisoning to merge and break adversarial triggers.

Ax Junichiro Niimi 4/1/2026

The Mouth is Not the Brain: Bridging Energy-Based World Models and Language Generation

Architecture separating energy-based world models from language generation in LLMs to improve semantic understanding beyond fluent text production.

Ax Runxiang Cheng, Michele Tufano, Jos\'e Cambronero, Renyao Wei, Sherry Shi, Grant Uy, Pat Rondon, Franjo Ivan\v{c}i\'c 4/1/2026

Dynamic Cogeneration of Bug Reproduction Test in Agentic Program Repair

Agentic APR system that dynamically generates bug reproduction tests alongside AI-generated code fixes to increase developer confidence.

Ax Yi-Kai Zhang, Zhiyuan Yao, Hongyan Hao, Yueqing Sun, Qi Gu, Hui Su, Xunliang Cai, De-Chuan Zhan, Han-Jia Ye 4/1/2026

$V_0$: A Generalist Value Model for Any Policy at State Zero

Generalist value model for policy evaluation in LLM training with actor-critic methods, reducing computational cost of critic networks.

Ax Austin Veselka 4/1/2026

How to Train Your Long-Context Visual Document Model

Large-scale study of training vision-language models up to 344K context length with open-weight model recipes and data pipeline analysis.

Ax Sen Ye, Mengde Xu, Shuyang Gu, Di He, Liwei Wang, Han Hu 4/1/2026

Understanding vs. Generation: Navigating Optimization Dilemma in Multimodal Models

Analysis of trade-off between generative and understanding capabilities in multimodal models, proposing R3 framework to address conflict.

Ax Aleksei Liuliakov, Luca Hermes, Barbara Hammer 4/1/2026

DGPO: RL-Steered Graph Diffusion for Neural Architecture Generation

Reinforcement learning-steered diffusion model for neural architecture search on directed acyclic graphs, improving NAS efficiency.

Ax Sreecharan Sankaranarayanan 4/1/2026

Mitigating "Epistemic Debt" in Generative AI-Scaffolded Novice Programming using Metacognitive Scripts

Study of how LLM-scaffolded programming affects novice skill acquisition, proposing metacognitive guardrails to prevent cognitive outsourcing.

Ax Bian Sun, Zhenjian Wang, Orvill de la Torre, Zirui Wang 4/1/2026

When Metrics Disagree: Automatic Similarity vs. LLM-as-a-Judge for Clinical Dialogue Evaluation

Study comparing automatic similarity metrics versus LLM-as-a-judge evaluation for clinical dialogue, using domain-adapted Llama-2-7B with LoRA.

Ax Benjamin M. Chen, Hong Bao 4/1/2026

Training for Technology: Adoption and Productive Use of Generative AI in Legal Analysis

Randomized experiment on law students studying targeted training interventions to improve productive use of LLMs in legal analysis tasks.

Ax Wisdom Ikezogwo, Mehmet Saygin Seyfioglu, Ranjay Krishna, Karim Bouyarmane 4/1/2026

When Rubrics Fail: Error Enumeration as Reward in Reference-Free RL Post-Training for Virtual Try-On

Error enumeration as reward approach for reference-free RL post-training in virtual try-on tasks with multiple valid outputs.

Ax Alvaro Paredes Amorin, Andre Python, Christoph Weisser 4/1/2026

Not All News Is Equal: Topic- and Event-Conditional Sentiment from Finetuned LLMs for Aluminum Price Forecasting

Study on using finetuned lightweight LLMs for topic-conditional sentiment extraction from news to forecast aluminum commodity prices.

Ax Zhongming Yu, Naicheng Yu, Hejia Zhang, Wentao Ni, Mingrui Yin, Jiaying Yang, Yujie Zhao, Jishen Zhao 4/1/2026

Multi-Agent Memory from a Computer Architecture Perspective: Visions and Challenges Ahead

Position paper framing multi-agent memory as computer architecture problem, proposing three-layer memory hierarchy and identifying protocol gaps for collaborative agents.

Ax Zekun Wu, Adriano Koshiyama, Sahan Bulathwela, Maria Perez-Ortiz 4/1/2026

AgentDrift: Unsafe Recommendation Drift Under Tool Corruption Hidden by Ranking Metrics in LLM Agents

AgentDrift reveals unsafe recommendation drift in tool-augmented LLM agents when tool outputs are corrupted, exposing gaps in ranking-metric-based evaluation.

Ax Mikoto Kudo, Takumi Tanabe, Akifumi Wachi, Youhei Akimoto 4/1/2026

Sample-Efficient Hypergradient Estimation for Decentralized Bi-Level Reinforcement Learning

Sample-efficient hypergradient estimation for decentralized bi-level reinforcement learning in strategic decision-making problems.

Ax Yeounoh Chung, Rushabh Desai, Jian He, Yu Xiao, Thibaud Hottelier, Yves-Laurent Kom Samo, Pushkar Khadilkar, Xianshun Chen, Sam Idicula, Fatma \"Ozcan, Alon Halevy, Yannis Papakonstantinou 4/1/2026

100x Cost & Latency Reduction: Performance Analysis of AI Query Approximation using Lightweight Proxy Models

Analysis of AI query approximation using lightweight proxy models to reduce cost and latency of LLM-based SQL queries on structured and unstructured data.

Ax Jian Yang, Wei Zhang, Jiajun Wu, Junhang Cheng, Shawn Guo, Haowen Wang, Weicheng Gu, Yaxin Du, Joseph Li, Fanglin Xu, Yizhi Li, Lin Jing, Yuanbo Wang, Yuhan Gao, Ruihao Gong, Chuan Hao, Ran Tao, Aishan Liu, Tuney Zheng, Ganqu Cui, Zhoujun Li, Mingjie Tang, Chenghua Lin, Wayne Xin Zhao, Xianglong Liu, Ming Zhou, Bryan Dai, Weifeng Lv 4/1/2026

InCoder-32B: Code Foundation Model for Industrial Scenarios

InCoder-32B is a 32B code foundation model optimized for industrial scenarios requiring hardware semantics, specialized constructs, and strict resource constraints.

Ax Dharshan Kumaran, Arthur Conmy, Federico Barbero, Simon Osindero, Viorica Patraucean, Petar Velickovic 4/1/2026

How do LLMs Compute Verbal Confidence

Research investigation into how LLMs internally compute verbal confidence scores and whether they are generated just-in-time or cached during answer generation.

Ax Queenie Luo, Gary King, Michael Puett, Michael D. Smith 4/1/2026

Inducing Sustained Creativity and Diversity in Large Language Models

Method for inducing sustained creativity and diversity in LLM outputs during exploratory search tasks through iterative refinement techniques.

Ax Tian Wen, Zhiqin Yang, Yonggang Zhang, Xuefeng Jiang, Hao Peng, Yuwei Wang, Bo Han 4/1/2026

FedRG: Unleashing the Representation Geometry for Federated Learning with Noisy Clients

FedRG addresses performance degradation in federated learning with noisy annotations using representation geometry for improved noise detection.

Ax Zijian Lu, Yiping Zuo, Yupeng Nie, Xin He, Weibei Fan, Lianyong Qi, Shi Jin 4/1/2026

ContractSkill: Repairable Contract-Based Skills for Multimodal Web Agents

ContractSkill framework converts web agent skills into executable artifacts with explicit structure for detection and local repair, improving agent reliability.

Ax Michael Hind, Basel Shbita, Bo Wu, Farhan Ahmed, Chad DeLuca, Nathan Fulton, David Cox, Dan Gutfreund 4/1/2026

LLMON: An LLM-native Markup Language to Leverage Structure and Semantics at the LLM Interface

LLMON is a markup language designed to convey structured and semantic information to LLMs beyond plain text, improving prompt clarity and LLM interface usability.

Ax Zhi Sun, Wenming Zhang, Yi Wei, Liren Yu, Zhixuan Zhang, Dan Ou, Haihong Tang 4/1/2026

KARMA: Knowledge-Action Regularized Multimodal Alignment for Personalized Search at Taobao

KARMA framework addresses knowledge-action gap when fine-tuning LLMs for personalized search tasks at scale, proposing regularization methods for industrial recommendation systems.

Ax Toluwani Aremu, Daniil Ognev, Samuele Poppi, Nils Lukas 4/1/2026

Robust Safety Monitoring of Language Models via Activation Watermarking

Activation watermarking technique for detecting adversarial attacks on LLMs that evade safety monitoring while eliciting unsafe outputs.

Ax Eyal Weiss 4/1/2026

Cost-Sensitive Neighborhood Aggregation for Heterophilous Graphs: When Does Per-Edge Routing Help?

GNN layer with per-edge routing for heterophilous graphs, comparing cost-sensitive aggregation against uniform spectral approaches.

Ax Guifeng Deng, Pan Wang, Jiquan Wang, Shuying Rao, Junyi Xie, Wanjun Guo, Tao Li, Haiteng Jiang 4/1/2026

SleepVLM: Explainable and Rule-Grounded Sleep Staging via a Vision-Language Model

Vision-language model for sleep staging from polysomnography waveforms generating AASM-compliant clinical rationales with auditable reasoning.

Ax Lorcan McLaren, James Cross, Zuzanna Krakowska, Robin Rauner, Martijn Schoonvelde 4/1/2026

Magic Words or Methodical Work? Challenging Conventional Wisdom in LLM-Based Political Text Annotation

Controlled evaluation of how LLM model choice, size, and prompting strategies affect political text annotation, challenging conventional wisdom.

Ax Tanvir Hossain, Muhammad Ifte Khairul Islam, Lilia Chebbah, Charles Fanning, Esra Akbas 4/1/2026

Cross-attentive Cohesive Subgraph Embedding to Mitigate Oversquashing in GNNs

GNN technique using cross-attention and cohesive subgraph embedding to address oversquashing problem in graph neural networks.

Ax Edward J. Yoon 4/1/2026

ITQ3_S: High-Fidelity 3-bit LLM Inference via Interleaved Ternary Quantization with Rotation-Domain Smoothing

3-bit weight quantization method for LLMs using rotation-domain smoothing via Fast Walsh-Hadamard Transform, improving precision in extreme quantization.

Ax Koki Maeda, Naoaki Okazaki 4/1/2026

JaWildText: A Benchmark for Vision-Language Models on Japanese Scene Text Understanding

Benchmark dataset for evaluating vision-language models on Japanese scene text understanding, addressing multilingual complexity challenges.

Ax Chang Zong, Sicheng Lv, Si-tu Xue, Huilin Zheng, Jian Wan, Lei Zhang 4/1/2026

Building evidence-based knowledge graphs from full-text literature for disease-specific biomedical reasoning

EvidenceNet framework uses LLM-assisted pipeline to extract structured, evidence-grounded biomedical findings from full-text literature into knowledge graphs.

Ax Huanxuan Liao, Zhongtao Jiang, Yupu Hao, Yuqiao Tan, Shizhu He, Ben Wang, Jun Zhao, Kun Xu, Kang Liu 4/1/2026

ResAdapt: Adaptive Resolution for Efficient Multimodal Reasoning

Adaptive resolution framework for multimodal LLMs that reduces visual token overhead through input-side compression before encoding.

Ax Yuma Ichikawa, Keiji Kimura, Akihiro Yoshida, Yudai Fujimoto, Hiroki Tokura, Yamato Arai, Yoshiyuki Ishii, Yusei Kawakami, Genki Shikada, Achille Jacquemond, Yoshihiko Fujisawa, Katsuki Fujisawa, Takumi Honda, Akira Sakai 4/1/2026

OneComp: One-Line Revolution for Generative AI Model Compression

Framework for post-training compression of generative AI models with single-line implementation, addressing quantization and calibration challenges.

Ax Ivan Pasichnyk 4/1/2026

Beta-Scheduling: Momentum from Critical Damping as a Diagnostic and Correction Tool for Neural Network Training

Derives time-varying momentum schedule for neural network training from physics principles, eliminating need for manual tuning.

Ax Mingju Liu, Jiaqi Yin, Alvaro Velasquez, Cunxi Yu 4/1/2026

Differentiable Initialization-Accelerated CPU-GPU Hybrid Combinatorial Scheduling

Hybrid CPU-GPU framework combining differentiable optimization with ILP solving for combinatorial scheduling.

Ax Andrea Carbonati, Mohammadsina Almasi, Hadis Anahideh 4/1/2026

Multi-Agent LLMs for Adaptive Acquisition in Bayesian Optimization

Multi-agent LLM framework for Bayesian optimization exploring exploration-exploitation trade-off through implicit reasoning.

Ax Siva Kumar Sastry Hari, Vignesh Balaji, Sana Damani, Qijing Huang, Christos Kozyrakis 4/1/2026

Improving Efficiency of GPU Kernel Optimization Agents using a Domain-Specific Language and Speed-of-Light Guidance

LLM agents for GPU kernel optimization using domain-specific language and speed-of-light guidance to reduce design space.

Ax Tushar Dhananjay Pathak 4/1/2026

ARCS: Autoregressive Circuit Synthesis with Topology-Aware Graph Attention and Spec Conditioning

Amortized analog circuit generation system combining graph VAE and flow-matching models with SPICE validation.