Ax Michael Rottoli, Subhankar Roy, Stefano Paraboschi 5/15/2026

Predict-then-Diffuse: Adaptive Response Length for Compute-Budgeted Inference in Diffusion LLMs

Predict-then-Diffuse method for diffusion LLMs enabling adaptive response length under compute budget constraints with parallel generation.

Ax Roberto E. Medina 5/15/2026

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

DASE stopping heuristic for LLM ensembles enabling early commitment on consensus and adaptive deliberation with calibrated signals.

Ax Yaxin Hou, Jun Ma, Hanyang Li, Bo Han, Jie Yu, Yuheng Jia 5/15/2026

Beyond Distribution Estimation: Simplex Anchored Structural Inference Towards Universal Semi-Supervised Learning

Universal Semi-Supervised Learning framework addressing scarce labeled data and unknown unlabeled distributions using structural inference.

Ax Xueying Ding, Leman Akoglu 5/15/2026

Toward Privileged Foundation Models:LUPI for Accelerated and Improved Learning

PIQL framework integrating privileged information to accelerate learning and improve generalization in tabular foundation models.

Ax Willy Fitra Hendria 5/15/2026

Non-Monotonic Latency in Apple MPS Decoding: KV Cache Interactions and Execution Regimes

Non-monotonic latency behavior in Apple MPS transformer decoding with KV cache interactions, identifying 21x latency spikes.

Ax Zhiyao Zhang, Myeung Suk Oh, Zhen Qin, Jiaxiang Li, Xin Zhang, Jia Liu 5/15/2026

A Tale of Two Problems: Multi-Task Bilevel Learning Meets Equality Constrained Multi-Objective Optimization

Multi-task bilevel optimization extending bilevel learning beyond single-task settings with equality constraints for complex ML problems.

Ax Will LeVine, Brendan Evers, Sam Saltwick, Abhay Venkatesh 5/15/2026

RubricRefine: Improving Tool-Use Agent Reliability with Training-Free Pre-Execution Refinement

RubricRefine improves tool-use agent reliability through training-free pre-execution refinement using rubric-based feedback for code generation.

Ax Musa Cim, Poovaiah Palangappa, Miro Hodak, Ravi Dwivedula, Meena Arunachalam, Mahmut Taylan Kandemir 5/15/2026

Pretraining large language models with MXFP4 on Native FP4 Hardware

Full-pipeline FP4 quantization training for large language models on native FP4 hardware, studying MXFP4 in transformer pretraining.

Ax Shuzhang Zhong, Haochen Huang, Shengxuan Qiu, Pengfei Zuo, Runsheng Wang, Meng Li 5/15/2026

Breaking the Reward Barrier: Accelerating Tree-of-Thought Reasoning via Speculative Exploration

Tree-of-Thought reasoning acceleration for LLMs via speculative exploration, addressing reward dependency bottleneck in complex task solving.

Ax Bochao Li, Yao Fu, Wei Chen, Fang Kong 5/15/2026

Sample-Mean Anchored Thompson Sampling for Offline-to-Online Learning with Distribution Shift

Thompson sampling algorithm for offline-to-online learning addressing distribution shift between offline data and online environments.

Ax Jiazheng Zhang, Ziche Fu, Junrui Shen, Yunbin Zhao, Yunke Zhang, Zhiheng Xi, Long Ma, Chenxin An, Zhihao Zhang, Shichun Liu, Dingwei Zhu, Shihan Dou, Shaofan Liu, Han Li, Wiggin Zhou, Aiden Adams, Tao Gui, Fei Huang, Qi Zhang, Xuanjing Huang 5/15/2026

Entropy Polarity in Reinforcement Fine-Tuning: Direction, Asymmetry, and Control

Theoretical framework of entropy mechanics in LLM reinforcement learning with verifiable rewards analyzing token-level policy updates.

Ax Sijia Li, Yuchen Huang, Zifan Liu, Yanping Li, Jingjing Fu, Li Zhao, Jiang Bian, Ling Zhang, Jun Zhang, Rui Wang 5/15/2026

GEAR: Granularity-Adaptive Advantage Reweighting for LLM Agents via Self-Distillation

GEAR: granularity-adaptive advantage reweighting for LLM agents using self-distillation for fine-grained credit assignment.

Ax Igor Strozzi 5/15/2026

Procedural-skill SFT across capacity tiers: A W-Shaped pre-SFT Trajectory and Regime-Asymmetric Mechanism on 0.8B-4B Qwen3.5 Models

Supervised fine-tuning analysis for procedural-skill learning across Qwen3.5 model scales (0.8B-4B).

Ax Yuanda Xu, Hejian Sang, Zhengze Zhou, Ran He, Zhipeng Wang, Alborz Geramifard 5/15/2026

Beyond GRPO and On-Policy Distillation: An Empirical Sparse-to-Dense Reward Principle for Language-Model Post-Training

Sparse-to-dense reward principle for LLM post-training combining GRPO sparse rewards with dense token-level distillation.

Ax Rishabh Tiwari, Kusha Sareen, Lakshya A Agrawal, Joseph E. Gonzalez, Matei Zaharia, Kurt Keutzer, Inderjit S Dhillon, Rishabh Agarwal, Devvrit Khatri 5/15/2026

Learning, Fast and Slow: Towards LLMs That Adapt Continually

Continual learning method combining parameter updates and in-context learning for LLMs to adapt without catastrophic forgetting.

Ax Jack Young 5/15/2026

WriteSAE: Sparse Autoencoders for Recurrent State

WriteSAE: sparse autoencoder decomposing recurrent language model cache writes for interpretability.

Ax Andrew Y. Zhou, Sharvaree Vadgama, Sumanth Varambally, Peter Eckmann, Michael K. Gilson, Rose Yu 5/15/2026

ToolMol: Evolutionary Agentic Framework for Multi-objective Drug Discovery

ToolMol: evolutionary agentic framework using LLMs with molecular tools for multi-objective drug discovery.

Ax Ling-Qi Zhang, Kristin Branson 5/15/2026

Neurodata Without Boredom: Benchmarking Agentic AI for Data Reuse

Benchmarking agentic AI for neuroscience data reuse and format standardization across fragmented experimental datasets.

Ax Danny Wang, Ruihong Qiu, Zi Huang 5/15/2026

What Information Matters? Graph Out-of-Distribution Detection via Tri-Component Information Decomposition

TIDE: graph neural network out-of-distribution detection via information decomposition for robust node classification.

Ax Chaokai Wu, Haofu Shi, Ningxuan Ma, Jianghong Ma, Xiaofeng Zhang 5/15/2026

MLGIB: Multi-Label Graph Information Bottleneck for Expressive and Robust Message Passing

MLGIB: Graph Neural Network method addressing over-squashing in multi-label graphs via information bottleneck.

Ax Linghao Jin, Chufan Shi, Huijuan Wang, Nuan Wen, Zhengzhong Liu, Eric Xing, Xuezhe Ma 5/15/2026

EMO: Frustratingly Easy Progressive Training of Extendable MoE

EMO: progressive training method for Mixture-of-Experts models addressing efficiency paradox in sparse MoE scaling.

Ax Coleman Hooper, Minwoo Kang, Suhong Moon, Nicholas Lee, Eric Wen, John Wawrzynek, Michael W. Mahoney, Yakun Sophia Shao, Amir Gholami, Kurt Keutzer 5/15/2026

Speculative Interaction Agents: Building Real-Time Agents with Asynchronous I/O and Speculative Tool Calling

Speculative Interaction Agents: framework for real-time LLM agents using asynchronous I/O and speculative tool calling under 1-second latency.

Ax Guanghui Lan, Tianjiao Li, Yangyang Xu 5/15/2026

Projected gradient methods for nonconvex and stochastic smooth optimization: new complexities and auto-conditioned stepsizes

Novel projected gradient methods for nonconvex smooth optimization with improved iteration complexity and auto-conditioned stepsizes.

Ax Shijun Li, Hilaf Hasson, Joydeep Ghosh 5/15/2026

OMAC: A Holistic Optimization Framework for LLM-Based Multi-Agent Collaboration

OMAC framework automatically optimizes multi-agent LLM systems through cost-effectiveness analysis and collaborative patterns.

Ax Zhuoyun Zhong, Seyedali Golestaneh, Constantinos Chamzas 5/15/2026

ActivePusher: Active Learning and Planning with Residual Physics for Nonprehensile Manipulation

ActivePusher combines active learning with learned residual dynamics models for nonprehensile robotic manipulation.

Ax Ryan Thompson, Matt P. Wand, Joanna J. J. Wang 5/15/2026

Scalable Subset Selection in Linear Mixed Models

Scalable subset selection method for linear mixed models with thousands of candidate predictors.

Ax Yixiao Wang, Mingxiao Huo, Zhixuan Liang, Yushi Du, Lingfeng Sun, Haotian Lin, Jinghuan Shang, Chensheng Peng, Mohit Bansal, Mingyu Ding, Masayoshi Tomizuka 5/15/2026

VER: Vision Expert Transformer for Robot Learning via Foundation Distillation and Dynamic Routing

VER combines multiple vision foundation models through distillation and dynamic routing for flexible robotic learning tasks.

Ax Manish Nagaraj, Sakshi Choudhary, Utkarsh Saxena, Deepak Ravikumar, Kaushik Roy 5/15/2026

TRIM: Token-wise Attention-Derived Saliency for Data-Efficient Instruction Tuning

TRIM method uses token-wise attention saliency to identify important samples for efficient LLM instruction tuning with reduced data requirements.

Ax Rafael Oliveira, Daniel M. Steinberg, Edwin V. Bonilla 5/15/2026

Generative Bayesian Optimization: Generative Models as Acquisition Functions

Uses generative models as acquisition functions for batch Bayesian optimization, enabling large-scale optimization of non-continuous and high-dimensional design spaces.

Ax Mingyu Zhang, Lifeng Zhuo, Tianxi Tan, Guocan Xie, Xian Nie, Yan Li, Renjie Zhao, Zizhu He, Ziyu Wang, Jiting Cai, Yong-Lu Li 5/15/2026

IPR-1: Interactive Physical Reasoner

Interactive physical reasoning agent learning human-like causal understanding from game interaction with visual domain gaps.

Ax Yitian Chen, Cheng Cheng, Yinan Sun, Zi Ling, Dongdong Ge 5/15/2026

OPT-Engine: Benchmarking the Limits of LLMs in Optimization Modeling via Complexity Scaling

OPT-ENGINE benchmark evaluating LLM capabilities in optimization modeling across LP to MIP with controlled complexity scaling.

Ax Mingda Zhang, Wenjin Liu, Tiesunlong Shen, Qika Lin, Rui Mao, Erik Cambria, Xiaoying Tang, Haoran Luo 5/15/2026

FlowSteer: Towards Agents Designing Agentic Workflows via Reinforced Progressive Canvas Editing

Agent-designed agentic workflows via reinforced canvas editing with graph-level feedback and in-loop error repair for complex multi-step tasks.

Ax Xi Wang, Anushri Suresh, Alvin Zhang, Rishi More, William Jurayj, Benjamin Van Durme, Mehrdad Farajtabar, Daniel Khashabi, Eric Nalisnick 5/15/2026

Conformal Thinking: Risk Control for Reasoning on a Compute Budget

Conformal prediction framework for adaptive reasoning in LLMs, controlling risk-accuracy tradeoff when allocating compute budget for inference.

Ax Lin Zheng, Xinyu Li, Qian Liu, Xiachong Feng, Lingpeng Kong 5/15/2026

Proxy Compression for Language Modeling

Proxy compression training scheme for language models preserving efficiency of tokenization while enabling raw-byte inference interface.

Ax Arash Gholami Davoodi, Navid Rezazadeh, Seyed Pouyan Mousavi Davoudi, Pouya Pezeshkpour 5/15/2026

Geometry-Aware Decoding with Wasserstein-Regularized Truncation and Mass Penalties for Large Language Models

Top-W: geometry-aware decoding for LLMs using Wasserstein distance over token embeddings to balance diversity and coherence.

Ax Tom Kempton, Julia Rozanova, Parameswaran Kamalaruban, Maeve Madigan, Karolina Wresilo, Yoann L. Launay, David Sutton, Stuart Burrell 5/15/2026

DMAP: A Distribution Map for Text

Framework extracting distribution maps from LLM next-token probabilities for better statistical text analysis beyond perplexity metrics.

Ax Yasaman Torabi, Parsa Razmara, Hamed Ajorlou, Bardia Baraeinejad 5/15/2026

NeuroMambaLLM: Dynamic Graph Learning of fMRI Functional Connectivity in Autistic Brains Using Mamba and Language Model Reasoning

Combines Mamba state-space models with LLMs for dynamic fMRI graph learning in autism diagnosis using multimodal reasoning.

Ax Mehrshad Taji, Arad Mahdinezhad Kashani, Iman Ahmadi, AmirHossein Jadidi, Saina Kashani, Babak Khalaj 5/15/2026

MALLVI: A Multi-Agent Framework for Integrated Generalized Robotics Manipulation

Multi-agent LLM framework for robotic manipulation with closed-loop visual feedback, replacing specialized models with general reasoning.

Ax Anthony Liang, Yigit Korkmaz, Jiahui Zhang, Minyoung Hwang, Abrar Anwar, Sidhant Kaushik, Aditya Shah, Alex S. Huang, Luke Zettlemoyer, Dieter Fox, Yu Xiang, Anqi Li, Andreea Bobu, Abhishek Gupta, Stephen Tu, Erdem Biyik, Jesse Zhang 5/15/2026

Robometer: Scaling General-Purpose Robotic Reward Models via Trajectory Comparisons

Scalable reward modeling framework for robotics using trajectory comparisons to handle failed and suboptimal trajectories.

Ax Xinyi Hu, Yuhao Shen, Baolin Zhang, Hengxin Zhang, Jun Dai, Shuang Ge, Lei Chen, Yue Li, Mingcheng Wan 5/15/2026

ECHO: Elastic Speculative Decoding with Sparse Gating for High-Concurrency Scenarios

Speculative decoding optimization for LLM inference in high-concurrency serving using elastic sparse gating and dynamic trees.

Ax Fei Bai, Huatong Song, Shuang Sun, Daixuan Cheng, Yike Yang, Chuan Hao, Renyuan Li, Feng Chang, Yuan Wei, Ran Tao, Bryan Dai, Jian Yang, Wayne Xin Zhao, Ji-Rong Wen 5/15/2026

ClawGym: A Scalable Framework for Building Effective Claw Agents

Scalable framework for training and evaluating agents in claw-style environments with file systems, tools, and persistent workspace.

Ax Rahul Kumar 5/15/2026

The Compliance Trap: How Structural Constraints Degrade Frontier AI Metacognition Under Adversarial Pressure

Evaluation study of frontier LLM metacognitive failures under adversarial pressure, examining cognitive collapse in high-stakes scenarios.

Ax Zirui Tang, Xuanhe Zhou, Yumou Liu, Linchun Li, Yukai Wu, Weizheng Wang, Hongzhang Huang, Wei Zhou, Jun Zhou, Jiachen Song, Shaoli Yu, Jinqi Wang, Zihang Zhou, Hongyi Zhou, Yuting Lv, Jinyang Li, Jiashuo Liu, Ruoyu Chen, Chunwei Liu, GuoLiang Li, Jihua Kang, Fan Wu 5/15/2026

Workspace-Bench 1.0: Benchmarking AI Agents on Workspace Tasks with Large-Scale File Dependencies

Large-scale benchmark for evaluating AI agents on workspace tasks with file dependencies, testing real-world file system operations.

Ax Zongmin Yu, Liu Yang 5/15/2026

Evolutionary Ensemble of Agents

Decentralized framework organizing coding agents in co-evolving populations for algorithmic discovery and skill evolution.

Ax Zahra Asadi, Haeseung Jeon, Sohyun Han, Md Mahmuduzzaman Kamol, Se Eun Oh, Mohammad Saidur Rahman 5/15/2026

FreeMOCA: Memory-Free Continual Learning for Malicious Code Analysis

Memory-efficient continual learning framework for malware detection that avoids catastrophic forgetting while adapting to new threats.

Ax Ethan Parks, Dalal Alharthi 5/15/2026

Predictive Maps of Multi-Agent Reasoning: A Successor-Representation Spectrum for LLM Communication Topologies

Diagnostic framework for predicting multi-agent LLM system behavior across different communication topologies using successor representation.

Ax Yanyan Zhang, Chaoda Song, Vikash Singh, Xinpeng Li, Kai Ye, Zhe Hu, Zhongzhu Pu, Yu Yin, Vipin Chaudhary 5/15/2026

Overcoming Dynamics-Blindness: Training-Free Pace-and-Path Correction for VLA Models

Training-free method to improve Vision-Language-Action models' handling of temporal dynamics in non-stationary scenarios.

Ax Toluwani Aremu, Nils Lukas, Jie Zhang 5/15/2026

Watermarking Should Be Treated as a Monitoring Primitive

Analysis of watermarking as monitoring primitive for generative models, examining internal attribution and safety monitoring.

Ax Chaehee Song, Minseok Seo, Yeeun Seong, Doyi Kim, Changick Kim 5/15/2026

Query-Conditioned Test-Time Self-Training for Large Language Models

Test-time self-training method for LLMs that updates parameters at inference to adapt to specific queries and correct misconceptions.

HN rashidae 5/15/2026

Beyond Git: Coordinating humans, agents, and automation in a repo with a ledger

Proposes ledger-based system extending Git to coordinate humans, AI agents, and automation in software repositories.