Ax Hanbo Cheng, Limin Lin, Ruo Zhang, Yicheng Pan, Jun Du 5/18/2026

Unlocking Complex Visual Generation via Closed-Loop Verified Reasoning

Framework combining closed-loop reasoning and verification for complex text-to-image generation, addressing hallucinations and optimization instabilities.

Ax Anurup Ganguli 5/18/2026

TFGN: Task-Free, Replay-Free Continual Pre-Training Without Catastrophic Forgetting at LLM Scale

Architectural solution enabling continual pre-training of large language models on heterogeneous domains without catastrophic forgetting, replay buffers, or task labels.

Ax Dzung Pham, Kleomenis Katevas, Ali Shahin Shamsabadi, Hamed Haddadi 5/18/2026

AgentStop: Terminating Local AI Agents Early to Save Energy in Consumer Devices

Method for early termination of local LLM agents to reduce energy consumption on consumer devices while preserving privacy and eliminating API costs.

Ax Yi Xie, Siao Liu, Falong Fan, Yuanqi Yao, Yue Zhao, Bo Liu 5/18/2026

TeamTR: Trust-Region Fine-Tuning for Multi-Agent LLM Coordination

Research on multi-agent LLM coordination identifying and addressing the compounding occupancy shift problem in sequential fine-tuning of shared-context teams.

Ax Plawan Kumar Rath, Rahul Maliakkal 5/18/2026

Quantization Undoes Alignment: Bias Emergence in Compressed LLMs Across Models and Precision Levels

Study showing post-training quantization of LLMs causes bias emergence and alignment degradation across model families and precision levels.

Ax Yu Fu, Longxuan Yu, Haz Sameen Shahgir, Zhipeng Wei, Hui Liu, N. Benjamin Erichson, Yue Dong 5/18/2026

Reducing the Safety Tax in LLM Safety Alignment with On-Policy Self-Distillation

On-policy self-distillation approach to reduce performance degradation (safety tax) in aligned LLMs while maintaining safety.

Ax Stavros Bouras, Ioannis Kontopoulos, Chiara Pugliese, Francesco Lettich, Emanuele Carlini, Hanna Kavalionak, Chiara Renso, Konstantinos Tserpes 5/18/2026

Privacy Evaluation of Generative Models for Trajectory Generation

Privacy evaluation framework for generative models (GANs, VAEs, diffusion) trained on trajectory data.

Ax Fanxu Meng 5/18/2026

GQLA: Group-Query Latent Attention for Hardware-Adaptive Large Language Model Decoding

GQLA: hardware-adaptive attention mechanism for LLM decoding supporting multi-token prediction on commodity hardware.

Ax Xujia Chen, Xinyue Hu, Letian Chen, Daming Shi, Wenhui Fan 5/18/2026

Curriculum Learning of Physics-Informed Neural Networks based on Spatial Correlation

Curriculum learning strategy for Physics-Informed Neural Networks based on spatial correlation to improve PINN training.

Ax Reilly Haskins, Bilal Chughtai, Joshua Engels 5/18/2026

Training on Documents About Monitoring Leads to CoT Obfuscation

Study showing LLMs can obfuscate chain-of-thought reasoning when trained on monitoring documents, posing detection evasion risks.

Ax Qiang Liu, Felix Koehler, Benjamin Holzschuh, Nils Thuerey 5/18/2026

Tadpole: Autoencoders as Foundation Models for 3D PDEs with Online Learning

Tadpole: foundation model for 3D PDEs using autoencoders with online data generation for improved transferability and scalability.

Ax Kyle R. Chickering, Huijuan Wang, Mengxi Wu, Alexander Moreno, Muhao Chen, Xuezhe Ma, Daria Soboleva, Joel Hestness, Zhengzhong Liu, Eric Xing 5/18/2026

GQA-{\mu}P: The maximal parameterization update for grouped query attention

Maximal update parameterization (μP) extension for grouped query attention enabling hyperparameter transfer across LLM architectures.

Ax Gideon Popoola, John Sheppard 5/18/2026

GESD: Beyond Outcome-Oriented Fairness

GESD framework for measuring explanation stability disparities in ML fairness, extending beyond outcome-oriented fairness metrics.

Ax Yihong Dong, Jiaru Qian, Haoran Zhang, Peixu Wang, Binhua Li, Zhi Jin, Yongbin Li, Ge Li, Xiaokang Yang, Xue Jiang 5/18/2026

From I/O to Code with Discovery Agent

Discovery Agent: LLM-based system for IO2Code program synthesis from input-output examples, advancing beyond NL2Code.

Ax Marilyn Zhang, Tianfeng Chen, Fabi\'an Barzuna, Ankita Rathod, Mark E. Whiting 5/18/2026

LEAP: Trajectory-Level Evaluation of LLMs in Iterative Scientific Design

LEAP evaluates LLMs in iterative scientific design by measuring learning trajectory efficiency and convergence rates, not just final outcomes.

Ax Yihan Zhu, Yuhan Liu, Weijiang Li, Tengfei Luo, Meng Jiang 5/18/2026

Controllable Molecular Generative Foundation Models

Controllable foundation model for molecular graph generation using RL post-training to optimize chemistry-specific design tasks.

Ax Songwei Dong, Zihan Chen, Chengshuai Shi, Peng Wang, Jundong Li, Cong Shen 5/18/2026

Is One Score Enough? Rethinking the Evaluation of Sequentially Evolving LLM Memory

SeqMem-Eval: diagnostic framework for evaluating LLM memory over sequential tasks, measuring forgetting and negative transfer beyond aggregate metrics.

Ax Philipp Mondorf, Samuel J. Bell, Jesse Dodge, Dieuwke Hupkes 5/18/2026

LPDS: Evaluating LLM Robustness Through Logic-Preserving Difficulty Scaling

LPDS evaluates LLM robustness by scaling problem difficulty while preserving logic, testing entity-invariant reasoning capabilities.

Ax Biswa Sengupta 5/18/2026

Representation Without Reward: A JEPA Audit for LLM Fine-Tuning

Tests Joint-Embedding Predictive Architecture principles for LLM fine-tuning on Llama-3.2, evaluating hidden-state geometry improvements.

Ax Lizhang Chen, Jonathan Li, Qi Wang, Runlong Liao, Shuozhe Li, Chen Liang, Ni Lao, Qiang Liu 5/18/2026

$\phi$-Balancing for Mixture-of-Experts Training

φ-Balancing: principled framework for balanced expert utilization in Mixture-of-Experts models targeting population-level objectives.

Ax Mahdi Naser Moghadasi, Faezeh Ghaderi 5/18/2026

Transformer Scalability Crisis: The First Comprehensive Empirical Analysis of Performance Walls in Modern Language Models

Comprehensive empirical analysis of 118 transformer models revealing fundamental scalability performance walls and deployment constraints.

Ax Gaojie Jin, Yong Tao, Lijia Yu, Tianjin Huang 5/18/2026

Margin-Adaptive Confidence Ranking for Reliable LLM Judgement

Proposes Margin-Adaptive Confidence Ranking to improve reliability of LLM judgments through learned confidence estimation and hypothesis testing.

Ax Jake Fawkes, Jason Hartford 5/18/2026

$f$-Trajectory Balance: A Loss Family for Tuning GFlowNets, Generative Models, and LLMs with Off- and On-Policy Data

Introduces f-Trajectory Balance loss for training GFlowNets and LLMs with mixed on/off-policy data, generalizing MSE approaches.

Ax Shukai Du, Junzhe Zhang, Yiming Li 5/18/2026

Lagrangian Flow Matching: A Least-Action Framework for Principled Path Design

Proposes Lagrangian Flow Matching framework for designing probability paths in generative models beyond straight-line transport.

Ax Jiading Gai, Shuai Zhang, Xiang Song, Bernie Wang, George Karypis 5/18/2026

DualKV: Shared-Prompt Flash Attention for Efficient RL Training with Large Rollouts and Long Contexts

DualKV optimizes Flash Attention for RL post-training by sharing prompt tokens across rollouts, reducing compute/memory redundancy in GRPO and DAPO.

Ax Lute Lillo, Nick Cheney 5/18/2026

On the Stability of Growth in Structural Plasticity

Studies structural plasticity during training via network growth and pruning. Shows growth is not inverse of pruning in deep learning.

Ax Albert Kj{\o}ller Jacobsen, Leo Uhre Jakobsen, Johanna Marie Gegenfurtner, Georgios Arvanitidis 5/18/2026

Don't Stop Me Yet: Sampling Loss Minima via Dissipative Riemannian Mechanics

Sampling method for loss minima using Riemannian mechanics. Characterizes reparameterization invariant solutions in neural networks.

Ax Yikun Zhang, Xiwei Cheng, Tianyu Liu, Yuanqi Du, Wengong Jin 5/18/2026

DrugSAGE:Self-evolving Agent Experience for Efficient State-of-the-Art Drug Discovery

DrugSAGE: Self-evolving agent that accumulates experience for efficient drug discovery. Reuses past search knowledge to find SOTA models across tasks.

Ax Rowan Martnishn, Sean Anderson 5/18/2026

Layer-wise Derivative Controlled Networks

Layer-wise Derivative Controlled Networks balance accuracy, efficiency, and stability. Addresses spiky/unpredictable behavior in complex models.

Ax Shangjian Yin, Yu Fu, Yue Dong, Zhouxing Shi 5/18/2026

GRLO: Towards Generalizable Reinforcement Learning in Open-Ended Environments from Zero

GRLO: Framework for generalizable reinforcement learning in open-ended environments. Integrates RLHF and RLVR paradigms for LLM post-training.

Ax Yunfei Luo, Xi Chen, Yuliang Chen, Lanshuang Zhang, Md Mofijul Islam, Siwei Zhao, Peter Kotanko, Subhasis Dasgupta, Andrew Campbell, Rakesh Malhotra, Tauhidur Rahman 5/18/2026

Toward World Modeling of Physiological Signals with Chaos-Theoretic Balancing and Latent Dynamics

NormWear-2 world model for physiological signal prediction. Encodes multivariate health signals with chaos-theoretic approach for long-horizon forecasting.

Ax Daniel Holmberg, Joel Oskarsson, Erik Larsson, Fredrik Lindsten, Teemu Roos 5/18/2026

Njord: A Probabilistic Graph Neural Network for Ensemble Ocean Forecasting

Njord: Probabilistic graph neural network for ocean forecasting. Combines latent variable framework with GNN for ensemble weather prediction.

Ax Nicolas Zilberstein, Santiago Segarra, Eero Simoncelli, Florentin Guth 5/18/2026

Learning Normalized Energy Models for Linear Inverse Problems

Energy-based models for inverse problems in imaging with explicit prior density. Addresses diffusion model limitations for ill-posed problems.

Ax Shi-ang Qi, Vahid Balazadeh, Michael Cooper, Russell Greiner, Rahul G. Krishnan 5/18/2026

SurvivalPFN: Amortizing Survival Prediction via In-Context Bayesian Inference

SurvivalPFN applies prior-data fitted networks for survival analysis. Amortizes Bayesian inference for censored observations via in-context learning.

Ax Vincent-Daniel Yun, Junhyuk Jo, Sai Praneeth Karimireddy, Sunwoo Lee 5/18/2026

Ghosted Layers: Unconstrained Activation Alignment for Recovering Layer-Pruned LLMs

Ghosted Layers: Training-free method to recover performance of layer-pruned LLMs by solving boundary activation alignment. Closed-form solution for layer mismatch.

Ax Ceyu Xu, Jiangnan Yu, Yongji Wu, Yuan Xie 5/18/2026

STS: Efficient Sparse Attention with Speculative Token Sparsity

STS: Sparse attention mechanism for LLM inference without retraining. Enables efficient processing of multi-million token sequences for agentic applications.

Ax Yijun Lu, Zilei Yang, Yuyin Ma 5/18/2026

parallelcbf: A composable safety-filter and auditability framework for tensor-parallel reinforcement learning

Framework unifying tensor-parallel RL environments, control barrier functions, and safety-constrained training for UAVs. Open-source safety filter for reinforcement learning.

Ax Louisa Cornelis, Johan Mathe, Louis Van Langendonck, Guillermo Bern\'ardez, Nina Miolane 5/18/2026

OgBench: A Framework for Evaluating Graph Neural Networks on Omics Data

Benchmark framework for Graph Neural Networks on biological omics data. Addresses n<<p regime overlooked by existing GNN benchmarks.

Ax Xian Gao, Bo Hui, Min-Te Sun, Wei-Shinn Ku 5/18/2026

On the Fragility of Data Attribution When Learning Is Distributed

Reveals vulnerabilities in data attribution methods under distributed training. Shows single participants can inflate attribution values while preserving model performance.

Ax Bruno Trentini, Jacob Hume, Vincenzo Antonio Isoldi, Philipp Misof, Ekaterina S. Ivshina, Kelly Maggs 5/18/2026

Neural Point-Forms

Introduces differential forms for point cloud learning to capture higher-order geometric information. Proposes neural point-forms architecture.

Ax Sihan Zeng, Sujay Bhatt, Sumitra Ganesh 5/18/2026

Rethinking Neural Network Learning Rates: A Stackelberg Perspective

Analyzes non-uniform layer-specific learning rates using Stackelberg game theory. Provides principled understanding of when different layers benefit from different learning rates.

Ax Jaehun Jung, Hyunwoo Kim, Brandon Cui, Ximing Lu, David Acuna, Prithviraj Ammanabrolu, Yejin Choi 5/18/2026

DeltaPrompts: Escaping the Zero-Delta Trap in Multimodal Distillation

DeltaPrompts improves distillation of Vision-Language Models by identifying ineffective prompts in training data. Addresses zero-delta prompt problem in multimodal model compression.

Ax Pedram Bakhtiarifard, Sophia N. Wilson, Mahmoud Afifi, Jonathan Wensh{\o}j, Raghavendra Selvan 5/18/2026

Characterizing Learning in Deep Neural Networks using Tractable Algorithmic Complexity Analysis

Research on measuring learning in DNNs using Kolmogorov complexity and compression. Applies algorithmic complexity theory to understand neural network training dynamics.

Ax Minseo Kim, Huanghao Mai, Jay Shenoy, Alec Follmer, Gordon Wetzstein, Frederic Poitevin 5/18/2026

CrystalBoltz: End-to-End Protein Structure Determination via Experiment-Guided Diffusion for X-Ray Crystallography

CrystalBoltz uses diffusion models conditioned on X-ray crystallography measurements as priors for structure factor amplitude integration in protein determination.

Ax Haizhong Zheng, Yizhuo Di, Jiahui Wang, Shuowei Jin, Xueshen Liu, Yongji Wu, Z. Morley Mao, Ion Stoica, Jiawei Zhao, Beidi Chen 5/18/2026

AstraFlow: Dataflow-Oriented Reinforcement Learning for Agentic LLMs

AstraFlow is dataflow-oriented system for scaling reinforcement learning on agentic LLMs across multi-policy training on heterogeneous compute.

Ax Zezhong Ding, Jin Li, Xugang Wang, Xike Xie 5/18/2026

Gaussian Relational Graph Transformer

GelGT combines Gaussian processes with graph transformers to jointly model structural, semantic, and temporal information in relational data.

Ax Tan-Ha Mai, Chao-Kai Chiang, Han-Hwa Shih, Gang Niu, Masashi Sugiyama, Hsuan-Tien Lin 5/18/2026

Embracing Biased Transition Matrices for Complementary-Label Learning with Many Classes

Complementary-label learning method using biased transition matrices scales weakly-supervised classification to many classes more effectively.

Ax Hojun Chung, Junseo Lee, Songhwai Oh 5/18/2026

Offline Reinforcement Learning with Universal Horizon Models

Offline RL with universal horizon models reduces compounding errors in model-based learning by direct infinite-horizon future prediction.

Ax Xuechen Zhang, Zijian Huang, Kai Yang, Weijia Zhang, Jiasi Chen, Samet Oymak 5/18/2026

VSPO: Vector-Steered Policy Optimization for Behavioral Control

VSPO enables LLMs to optimize multiple behavioral objectives simultaneously through vector-steered policy optimization addressing sparse reward problems.

Ax Aditya Kudre, Heng-Sheng Chang, Prashant G. Mehta 5/18/2026

Transformer-like Inference from Optimal Control

Derives transformer-like inference architectures from optimal control theory, recovering decoder-only transformer operations from first principles.