Ax Yinsicheng Jiang, Yeqi Huang, Liang Cheng, Cheng Deng, Xuan Sun, Luo Mai 5/7/2026

ContextPilot: Fast Long-Context Inference via Context Reuse

System for accelerating long-context LLM inference through context reuse for retrieval-augmented generation and agent memory.

Ax Leyan Xue, Changqing Zhang, Kecheng Xue, Xiaohong Liu, Guangyu Wang, Zongbo Han 5/7/2026

MULTIBENCH++: A Unified and Comprehensive Multimodal Fusion Benchmarking Across Specialized Domains

Comprehensive multimodal fusion benchmark across specialized domains to address evaluation gaps in current fusion methods.

Ax Bing Liu, Boao Kong, Limin Lu, Kun Yuan, Chengcheng Zhao 5/7/2026

Row-stochastic matrices can provably outperform doubly stochastic matrices in decentralized learning

Theoretical analysis showing row-stochastic matrices outperform doubly stochastic matrices in decentralized learning with heterogeneous node weights.

Ax Dingwei Zhu, Zhiheng Xi, Shihan Dou, Yuhui Wang, Sixian Li, Junjie Ye, Honglin Guo, Shichun Liu, Chenhao Huang, Yajie Yang, Junlin Shang, Senjie Jin, Ming Zhang, Jiazheng Zhang, Caishuang Huang, Yunke Zhang, Yuran Wang, Tao Gui 5/7/2026

DVPO: Distributional Value Modeling-based Policy Optimization for LLM Post-Training

DVPO algorithm for stable LLM post-training via distributional value modeling, handling noisy supervision in RL settings.

Ax Stefan Nielsen, Edoardo Cetin, Peter Schwendeman, Qi Sun, Jinglue Xu, Yujin Tang 5/7/2026

Learning to Orchestrate Agents in Natural Language with the Conductor

RL-trained Conductor model that orchestrates multiple LLM agents by discovering coordination strategies and optimizing communication topologies.

Ax Andrea Napoli, Paul White 5/7/2026

Variance Matters: Improving Domain Adaptation via Stratified Sampling

Variance-reduced domain adaptation method via stratified sampling to address domain shift in real-world ML deployment.

Ax Daniel Rose, Roxane Axel Jacob, Johannes Kirchmair, Thierry Langer 5/7/2026

NEAT: Neighborhood-Guided, Efficient, Autoregressive Set Transformer for 3D Molecular Generation

Transformer-based autoregressive model for 3D molecular generation with permutation invariance properties.

Ax Daphne Theodorakopoulos, Marcel Wever, Marius Lindauer 5/7/2026

Dynamic Hyperparameter Importance for Efficient Multi-Objective Optimization

Method for dynamic hyperparameter importance analysis in multi-objective optimization to improve model selection efficiency.

Ax Xueyan Niu, Bo Bai, Wei Han, Weixi Zhang 5/7/2026

On the Non-decoupling of Supervised Fine-tuning and Reinforcement Learning in Post-training

Analysis of interaction between supervised fine-tuning and reinforcement learning during LLM post-training, examining their non-decoupling effects.

Ax Leon G\"otz, Lars Frederik Peiss, Erik Sauer, Andreas Udo Sass, Thorsten Bagdonat, Stephan G\"unnemann, Leo Schwinn 5/7/2026

A Scalable Multi-Task Model for Virtual Sensors

Multi-task learning approach for virtual sensors using time series foundation models to predict signals from available measurements.

Ax Sidney Bender, Marco Morik 5/7/2026

Visual Disentangled Diffusion Autoencoders: Scalable Counterfactual Generation for Foundation Models

Framework integrating foundation models with disentangled autoencoders to mitigate spurious correlations and improve robustness in vision tasks.

Ax Mingda Liu, Zhenghan Zhu, Ze'an Miao, Katsuki Fujisawa 5/7/2026

Norm Anchors Make Model Edits Last

Identifies norm-feedback loops causing sequential model edits to fail, proposes norm anchors to stabilize edit composition.

Ax Xiaoyuan Cheng, Wenxuan Yuan, Boyang Li, Yuanchao Xu, Yiming Yang, Hao Liang, Bei Peng, Robert Loftin, Zhuo Sun, Yukun Hu 5/7/2026

How Does the Lagrangian Guide Safe Reinforcement Learning through Diffusion Models?

Augmented Lagrangian-guided diffusion for safe offline reinforcement learning with multimodal action distributions.

Ax Haocheng Xi, Shuo Yang, Yilong Zhao, Muyang Li, Han Cai, Xingyang Li, Yujun Lin, Zhuoyang Zhang, Jintao Zhang, Xiuyu Li, Zhiying Xu, Jun Wu, Chenfeng Xu, Ion Stoica, Song Han, Kurt Keutzer 5/7/2026

Quant VideoGen: Auto-Regressive Long Video Generation via 2-Bit KV-Cache Quantization

Quant VideoGen: 2-bit KV-cache quantization enabling autoregressive video generation on memory-constrained hardware.

Ax Yujuan Pang, Jiaxin Li, Xin Sheng, Ran Peng, Yong Ma 5/7/2026

Beyond Variance: Prompt-Efficient RLVR via Rare-Event Amplification and Bidirectional Pairing

Prompt-efficient RLVR via rare-event amplification and bidirectional pairing for improved optimization and transfer in reasoning tasks.

Ax Dingwei Zhu, Zhiheng Xi, Shihan Dou, Jiahan Li, Chenhao Huang, Junjie Ye, Sixian Li, Mingxu Chai, Yuhui Wang, Yajie Yang, Ming Zhang, Jiazheng Zhang, Shichun Liu, Caishuang Huang, Yunke Zhang, Yuran Wang, Tao Gui, Xipeng Qiu, Qi Zhang, Xuanjing Huang 5/7/2026

DFPO: Scaling Value Modeling via Distributional Flow towards Robust and Generalizable LLM Post-Training

DFPO: distributional flow method for robust LLM post-training via distributional RL, improving OOD generalization with fine-grained value modeling.

Ax Jonathan von Rad, Yong Cao, Andreas Geiger 5/7/2026

UniComp: A Unified Evaluation of Large Language Model Compression via Pruning, Quantization and Distillation

UniComp: unified evaluation framework comparing LLM compression techniques (pruning, quantization, distillation) across performance, reliability, efficiency.

Ax Binyu Zhao, Wei Zhang, Xingrui Yu, Zhaonian Zou, Ivor Tsang 5/7/2026

Advancing Analytic Class-Incremental Learning through Vision-Language Calibration

Addresses class-incremental learning with pre-trained models through vision-language calibration to balance adaptation and stability.

Ax Yu Huang, Zixin Wen, Yuejie Chi, Yuting Wei, Aarti Singh, Yingbin Liang, Yuxin Chen 5/7/2026

The Implicit Curriculum: Learning Dynamics in RL with Verifiable Rewards

Theoretical analysis of training dynamics in reinforcement learning with verifiable rewards for reasoning models, showing natural curriculum emergence.

Ax Jiahao Zhang, Lujing Zhang, Keltin Grimes, Zhuohao Yu, Gokul Swamy, Zhiwei Steven Wu 5/7/2026

Back to Blackwell: Closing the Loop on Intransitivity in Multi-Objective Preference Fine-Tuning

Addresses intransitive preferences in LLM preference fine-tuning by modeling multi-objective trade-offs with principled scalarization.

Ax Sarthak Munshi, Manish Bhatt, Vineeth Sai Narajala, Idan Habler, Ammar Al-Kahfah, Ken Huang, Blake Gatto 5/7/2026

Manifold of Failure: Behavioral Attraction Basins in Language Models

Framework for mapping unsafe regions in LLMs using quality-diversity search to understand failure modes and improve AI safety.

Ax Seungju Back, Dongwoo Lee, Naun Kang, Taehee Lee, S. K. Hong, Youngjune Gwon, Sungjin Ahn 5/7/2026

Understanding LoRA as Knowledge Memory: An Empirical Analysis

Empirical analysis of LoRA as parametric knowledge memory for continuous LLM updating, comparing with ICL and RAG approaches.

Ax Yanbo Wang, Jiaxuan You, Chuan Shi, Muhan Zhang 5/7/2026

Relational In-Context Learning via Synthetic Pre-training with Structural Prior

RDB-PFN: first relational foundation model trained on synthetic data to overcome data scarcity in relational database pre-training.

Ax Jingwei Li, Xinran Gu, Jingzhao Zhang 5/7/2026

Capacity-Aware Mixture Law Enables Efficient LLM Data Optimization

Introduces compute-efficient pipeline for LLM data mixture optimization using capacity-aware scaling laws to improve downstream performance.

Ax Yihong Chen, Zhouchen Lin, Quanming Yao 5/7/2026

Attention Sinks Induce Gradient Sinks: Massive Activations as Gradient Regulators in Transformers

Analyzes attention sinks and massive activations in Transformers through backpropagation perspective, explaining gradient regulation mechanisms.

Ax Hadi Hojjati, Narges Armanfard 5/7/2026

ARTA: Adversarial-Robust Multivariate Time--Series Anomaly Detection via Sparsity-Constrained Perturbations

ARTA: Adversarially robust time-series anomaly detection via sparsity-constrained perturbation training for improved detector robustness.

Ax Jicheng Ma, Yunyan Yang, Juan Zhao, Liang Zhao 5/7/2026

Geometric Evolution Graph Convolutional Networks: Enhancing Graph Representation Learning via Ricci Flow

GEGCN: Graph convolutional networks enhanced with Ricci flow for dynamic geometric representation learning.

Ax Ahmer Raza, Hudson Smith 5/7/2026

Massively Parallel Exact Inference for Hawkes Processes

Massively parallel O(N) algorithm for Hawkes process inference using sparse matrices, exploiting GPU parallelization.

Ax Vikram Krishnamurthy, Luke Snow 5/7/2026

Malliavin Calculus for Counterfactual Gradient Estimation in Adaptive Inverse Reinforcement Learning

Malliavin Calculus for Adaptive Inverse Reinforcement Learning: Novel passive Langevin-based algorithm for recovering loss functions from gradients.

Ax Rishab Balasubramanian, Pin-Jie Lin, Rituraj Sharma, Anjie Fang, Fardin Abdi, Viktor Rozgic, Zheng Du, Mohit Bansal, Tu Vu 5/7/2026

The Master Key Hypothesis: Unlocking Cross-Model Capability Transfer via Linear Subspace Alignment

Master Key Hypothesis: Post-trained capabilities transfer across models via linear subspace alignment without retraining, enabling cross-model transfer.

Ax Linggang Kong, Lei Wu, Yunlong Zhang, Xiaofeng Zhong, Zhen Wang, Yongjie Wang, Yao Pan 5/7/2026

CausalGaze: Unveiling Hallucinations via Counterfactual Graph Intervention in Large Language Models

CausalGaze: Counterfactual graph intervention to detect and unveil LLM hallucinations by exposing underlying causal mechanisms.

Ax Sandro Andric 5/7/2026

When Reasoning Models Hurt Behavioral Simulation: A Solver-Sampler Mismatch in Multi-Agent LLM Negotiation

Solver-Sampler Mismatch in multi-agent LLM negotiation: Reasoning models hurt behavioral simulation; stronger reasoning ≠ better sampling.

Ax Sumeet Ramesh Motwani, Chuan Du, Aleksander Petrov, Christopher Davis, Philip Torr, Antonio Papania-Davis, Weishi Yan 5/7/2026

AutoOR: Scalably Post-training LLMs to Autoformalize Operations Research Problems

AutoOR: Synthetic data and RL pipeline training LLMs to autoformalize operations research problems for industrial optimization.

Ax Zhaokun Wang, Jinyu Guo, Jingwen Pu, Hongli Pu, Meng Yang, Xunlei Chen, Jie Ou, Wenyi Li, Guangchun Luo, Wenhong Tian 5/7/2026

CAP: Controllable Alignment Prompting for Unlearning in LLMs

CAP: Controllable Alignment Prompting for unlearning sensitive information from LLMs without parameter access, enabling regulatory compliance.

Ax Wei Jiang, Wei Wang 5/7/2026

Sub-Token Routing in LoRA for Adaptation and Query-Aware KV Compression

Sub-Token Routing in LoRA: Fine-grained compression via sub-token routing combined with KV compression for efficient transformer adaptation.

Ax Audrey Cherilyn, Houman Safaai 5/7/2026

Supernodes and Halos: Loss-Critical Hubs in LLM Feed-Forward Layers

Supernodes and Halos: Study showing loss sensitivity concentrates in ~1% of channels in transformer FFNs, enabling targeted compression.

Ax Wenshuo Wang 5/7/2026

Knowledge Distillation Must Account for What It Loses

Knowledge Distillation Must Account for What It Loses: Position paper arguing student models should preserve teacher reliability beyond task metrics.

Ax Xinshuai Dong, Haifeng Chen, Xuyuan Liu, Shengyu Chen, Haoyu Wang, Shaoan Xie, Kun Zhang, Zhengzhang Chen 5/7/2026

The Power of Order: Fooling LLMs with Adversarial Table Permutations

LLMs vulnerable to semantically-invariant row/column permutations in tabular data, demonstrating fragility in Table Question Answering tasks.

Ax Shihong Ding, Fangyu Du, Cong Fang 5/7/2026

Near-optimal and Efficient First-Order Algorithm for Multi-Task Learning with Shared Linear Representation

Near-optimal first-order algorithm for multi-task learning with shared linear representations, addressing non-convex matrix factorization challenges.

Ax Rohit Agarwal, Joshua Lin, Mark Braverman, Elad Hazan 5/7/2026

AI Alignment via Incentives and Correction

AI Alignment via Incentives and Correction: Framework applying law-and-economics deterrence models to prevent misconduct in agentic AI systems.

Ax Anamika Paul Rupa, Anietie Andy 5/7/2026

Probe-Geometry Alignment: Erasing the Cross-Sequence Memorization Signature Below Chance

Probe-Geometry Alignment method surgically removes internal memorization traces from unlearned LLMs without capability loss using cross-sequence probes.

Ax Mario Koddenbrock, Christoph Lange, Robin Legner, Martin J\"ager, Martin K\"ogler, Mariano N. Cruz Bournazou, Peter Neubauer, Felix Biessmann, Erik Rodner 5/7/2026

RamanBench: A Large-Scale Benchmark for Machine Learning on Raman Spectroscopy

RamanBench: First large-scale benchmark for machine learning on Raman spectroscopy data with standardized datasets and evaluation protocols.

Ax Soyeon Kim, Seongwoo Lim, Kyowoon Lee, Jaesik Choi 5/7/2026

Manifold-Aligned Guided Integrated Gradients for Reliable Feature Attribution

Manifold-Aligned Guided Integrated Gradients method for improving reliability of feature attribution in neural networks.

Ax Xianli Zhu, Jia Yin 5/7/2026

ZNO: Stable Rational Neural Operators in the Z-Domain for Discrete-Time Dynamics

Z-Domain Neural Operator for discrete-time system identification using stable rational filters parameterized in z-plane.

Ax Eitan Kosman, Gabriele Serussi, Chaim Baskin 5/7/2026

Structured Diffusion Bridges: Inductive Bias for Denoising Diffusion Bridges

Diffusion bridge framework for modality translation with alignment constraints to restrict solution space in cross-modal mapping.

Ax Hahyeon Choi, Nojun Kwak 5/7/2026

Toward Structural Multimodal Representations: Specialization, Selection, and Sparsification via Mixture-of-Experts

S3 framework for multimodal learning using mixture-of-experts with specialization, selection, and sparsification for task-specific routing.

Ax Mahmoud Hanouneh, Radu Timofte, Dmitry Ignatov 5/7/2026

From Code to Prediction: Fine-Tuning LLMs for Neural Network Performance Classification in NNGPT

Fine-tuning LLMs to classify neural network performance for AutoML frameworks, exploring LLM reasoning about architecture quality.

Ax Skye Gunasekaran, T\'ea Wright, Rui-Jie Zhu, Jason Eshraghian 5/7/2026

Transformers with Selective Access to Early Representations

Transformer architecture exposing later layers to early representations via selective access mechanisms improving feature recovery.

Ax Yuyang Zhou, Guang Cheng, Zongyao Chen, Shui Yu 5/7/2026

MalPurifier: Enhancing Android Malware Detection with Adversarial Purification against Evasion Attacks

Adversarial purification defense method enhancing Android malware detection robustness against evasion attacks.

Ax Xinwei Shen, Nicolai Meinshausen 5/7/2026

Distributional Principal Autoencoders

Dimension reduction technique learning distributional models matching conditional data distributions for lossless reconstruction.