NeurIPS 2025 Oral Papers

NeurIPSOral2025

All 77 papers accepted as Oral at NeurIPS 2025 — the top slice of the accepted programme. Every title links to its own page with authors, affiliation and sources.

All 77 papers

1. Generalized Linear Mode Connectivity for TransformersAlexander Theus, Alessandro Cabodi, Sotiris Anagnostidis et al. · ETHZ - ETH Zurich
2. Deep Compositional Phase Diffusion for Long Motion Sequence GenerationHo Yin Au, Jie Chen, Junkun Jiang et al. · Hong Kong Baptist University
3. GnnXemplar: Exemplars to Explanations - Natural Language Rules for Global GNN InterpretabilityBurouj Armgaan, Eshan Jain, Harsh Pandey et al. · Indian Institute of Technology Delhi
4. RAG4GFM: Bridging Knowledge Gaps in Graph Foundation Models through Graph Retrieval Augmented GenerationXingliang Wang, Zemin Liu, Junxiao Han et al. · Zhejiang University
5. Agnostic Active Learning Is Always Better Than Passive LearningSteve Hanneke · Purdue University
6. Learning Linear Attention in Polynomial TimeMorris Yau, Ekin Akyürek, Jiayuan Mao et al. · Massachusetts Institute of Technology
7. Optimal Mistake Bounds for Transductive Online LearningZachary Chase, Steve Hanneke, Shay Moran et al. · University of California, San Diego
8. State Entropy Regularization for Robust Reinforcement LearningYonatan Ashlag, Uri Koren, Mirco Mutti et al. · Technion - Israel Institute of Technology, Technion
9. On the Closed-Form of Flow Matching: Generalization Does Not Arise from Target StochasticityQuentin Bertrand, Anne Gagneux, Mathurin Massias et al. · INRIA
10. Why Diffusion Models Don’t Memorize: The Role of Implicit Dynamical Regularization in TrainingTony Bonnaire, Raphaël Urfin, Giulio Biroli et al. · Université Paris-Saclay
11. Adjoint Schrödinger Bridge SamplerGuan-Horng Liu, Jaemoo Choi, Yongxin Chen et al. · FAIR, Meta AI
12. Breaking the Performance Ceiling in Reinforcement Learning requires Inference StrategiesFelix Chalumeau, Daniel Rajaonarivonivelomanantsoa, Ruan John de Kock et al. · InstaDeep
13. High-Dimensional Calibration from Swap RegretMaxwell Fishelson, Noah Golowich, Mehryar Mohri et al. · Massachusetts Institute of Technology
14. In Search of Adam’s Secret SauceAntonio Orvieto, Robert M. Gower · ELLIS Institute Tübingen, Max Planck Institute for Intelligent Systems, Tübingen AI Center, Tübingen, Germany
15. An Optimized Franz-Parisi Criterion and its Equivalence with SQ Lower BoundsSiyu Chen, Theodor Misiakiewicz, Ilias Zadik et al. · Yale University
16. MaxSup: Overcoming Representation Collapse in Label SmoothingYuxuan Zhou, Heng Li, Zhi-Qi Cheng et al. · Baidu
17. Memory Mosaics at scaleJianyu Zhang, Leon Bottou · New York University
18. The emergence of sparse attention: impact of data distribution and benefits of repetitionNicolas Zucchet, Francesco D'Angelo, Andrew Kyle Lampinen et al. · ETHZ - ETH Zurich
19. ControlFusion: A Controllable Image Fusion Network with Language-Vision Degradation PromptsLinfeng Tang, Yeda Wang, Zhanchuan Cai et al. · Wuhan University
20. Identifiability of Deep Polynomial Neural NetworksKonstantin Usevich, Ricardo Augusto Borsoi, Clara Dérand et al. · CNRS
21. Understanding and Mitigating Numerical Sources of Nondeterminism in LLM InferenceJiayi Yuan, Hao Li, Xinheng Ding et al. · Rice University
22. PRIMT: Preference-based Reinforcement Learning with Multimodal Feedback and Trajectory Synthesis from Foundation ModelsRuiqi Wang, Dezhong Zhao, Ziqin Yuan et al. · Purdue University
23. A is for Absorption: Studying Feature Splitting and Absorption in Sparse AutoencodersDavid Chanin, James Wilken-Smith, Tomáš Dulka et al. · MATS
24. EvoLM: In Search of Lost Training Dynamics for Language Model ReasoningZhenting Qi, Fan Nie, Alexandre Alahi et al. · Harvard University
25. Analog In-memory Training on General Non-ideal Resistive Elements: The Impact of Response FunctionsZhaoxian Wu, Quan Xiao, Tayfun Gokmen et al. · Cornell University
26. Discovering Opinion Intervals from Conflicts in Signed GraphsPeter Blohm, Florian Chen, Aristides Gionis et al. · Technische Universität Wien
27. A Clean Slate for Offline Reinforcement LearningMatthew Thomas Jackson, Uljad Berdica, Jarek Luca Liesen et al. · Google DeepMind
28. Spectral Perturbation Bounds for Low-Rank Approximation with Applications to PrivacyPhuc Tran, Van Vu, Nisheeth K. Vishnoi · VinUniversity
29. Improved Regret Bounds for Gaussian Process Upper Confidence Bound in Bayesian OptimizationShogo Iwazaki · MI-6 Ltd.
30. Auto-Compressing NetworksVaggelis Dorovatas, Georgios Paraskevopoulos, Alexandros Potamianos · National Technical University of Athens
31. MokA: Multimodal Low-Rank Adaptation for MLLMsYake Wei, Yu Miao, Dongzhan Zhou et al. · Renmin University of China
32. Advancing Expert Specialization for Better MoEHongcan Guo, Haolang Lu, Guoshun Nan et al. · ByteDance Inc.
33. From Condensation to Rank Collapse: A Two-Stage Analysis of Transformer Training DynamicsZheng-An Chen, Tao Luo · Shanghai Jiaotong University
34. Large Language Diffusion ModelsShen Nie, Fengqi Zhu, Zebin You et al. · Renmin University of China
35. Boosting Knowledge Utilization in Multimodal Large Language Models via Adaptive Logits Fusion and Attention ReallocationWenbin An, Jiahao Nie, Feng Tian et al. · Xi'an Jiaotong University
36. Interactive Cross-modal Learning for Text-3D Scene RetrievalYanglin Feng, Yongxiang Li, Yuan Sun et al. · Sichuan University
37. Rethinking Joint Maximum Mean Discrepancy for Visual Domain AdaptationWei Wang, Haifeng Xia, Chao Huang et al. · SUN YAT-SEN UNIVERSITY
38. Pan-LUT: Efficient Pan-sharpening via Learnable Look-Up TablesZhongnan Cai, Yingying Wang, Hui Zheng et al. · Xiamen University
39. Dynamical Decoupling of Generalization and Overfitting in Large Two-Layer NetworksAndrea Montanari, Pierfrancesco Urbani · Stanford University
40. 1000 Layer Networks for Self-Supervised RL: Scaling Depth Can Enable New Goal-Reaching CapabilitiesKevin Wang, Ishaan Javali, Michał Bortkiewicz et al. · Princeton University
41. Depth-Bounds for Neural Networks via the Braid ArrangementMoritz Leo Grillo, Christoph Hertrich, Georg Loho · Max-Planck Institute
42. Tighter CMI-Based Generalization Bounds via Stochastic Projection and QuantizationMilad Sefidgaran, Kimia Nadjahi, Abdellatif Zaidi · Huawei Paris Research Center
43. A Snapshot of Influence: A Local Data Attribution Framework for Online Reinforcement LearningYuzheng Hu, Fan Wu, Haotian Ye et al. · University of Illinois at Urbana-Champaign
44. High-dimensional neuronal activity from low-dimensional latent dynamics: a solvable modelValentin Schmutz, Ali Haydaroğlu, Shuqi Wang et al. · University College London, University of London
45. Adaptive Surrogate Gradients for Sequential Reinforcement Learning in Spiking Neural NetworksKorneel Van den Berghe, Stein Stroobants, Vijay Janapa Reddi et al. · Delft University of Technology
46. Class-wise Balancing Data Replay for Federated Class-Incremental LearningZhuang Qi, Ying-Peng Tang, Lei Meng et al. · Shandong University
47. Task-Optimized Convolutional Recurrent Networks Align with Tactile Processing in the Rodent BrainTrinity Chung, Yuchen Shen, Nathan Kong et al.
48. ElasticMM: Efficient Multimodal LLMs Serving with Elastic Multimodal ParallelismZedong Liu, Shenggan Cheng, Guangming Tan et al. · University of Electronic Science and Technology of China
49. Dynamical Low-Rank Compression of Neural Networks with Robustness under Adversarial AttacksSteffen Schotthöfer, H. Lexie Yang, Stefan Schnake · Oak Ridge National Laboratory
50. QoQ-Med: Building Multimodal Clinical Foundation Models with Domain-Aware GRPO TrainingWei Dai, Peilin Chen, Chanakya Ekbote et al. · Massachusetts Institute of Technology
51. Gated Attention for Large Language Models: Non-linearity, Sparsity, and Attention-Sink-FreeZihan Qiu, Zekun Wang, Bo Zheng et al. · Alibaba Group
52. Learning long range dependencies through time reversal symmetry breakingGuillaume Pourcel, Maxence Ernoult · INRIA
53. FuXi-Ocean: A Global Ocean Forecasting System with Sub-Daily ResolutionQiusheng Huang, Yuan Niu, Xiaohui Zhong et al. · Fudan University
54. Superposition Yields Robust Neural ScalingYizhou Liu, Ziming Liu, Jeff Gore · Massachusetts Institute of Technology
55. ImageNet-trained CNNs are not biased towards texture: Revisiting feature reliance through controlled suppressionTom Burgert, Oliver Stoll, Paolo Rota et al. · Technische Universität Berlin
56. On Linear Mode Connectivity of Mixture-of-Experts ArchitecturesViet-Hoang Tran, Van-Hoan Trinh, Khanh Vinh Bui et al. · National University of Singapore
57. OpenHOI: Open-World Hand-Object Interaction Synthesis with Multimodal Large Language ModelZhenhao Zhang, Ye Shi, Lingxiao Yang et al.
58. Representation Entanglement for Generation: Training Diffusion Transformers Is Much Easier Than You ThinkGe Wu, Shen Zhang, Ruijing Shi et al. · Nankai University
59. Dynam3D: Dynamic Layered 3D Tokens Empower VLM for Vision-and-Language NavigationZihan Wang, Seungjun Lee, Gim Hee Lee · National University of Singapore
60. Learning (Approximately) Equivariant Networks via Constrained OptimizationAndrei Manolache, Luiz F. O. Chamon, Mathias Niepert · Universität Stuttgart
61. SAGE: A Unified Framework for Generalizable Object State Recognition with State-Action Graph EmbeddingYuan Zang, Zitian Tang, Junho Cho et al. · Brown University
62. Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?Yang Yue, Zhiqi Chen, Rui Lu et al.
63. Learning to Learn with Contrastive Meta-ObjectiveShiguang Wu, Yaqing Wang, Yatao Bian et al. · Tsinghua University, Tsinghua University
64. KVzip: Query-Agnostic KV Cache Compression with Context ReconstructionJang-Hyun Kim, Jinuk Kim, Sangwoo Kwon et al. · Apple
65. HyperET: Efficient Training in Hyperbolic Space for Multi-modal Large Language ModelsZelin Peng, Zhengqin Xu, Qingyang Liu et al. · Shanghai Jiaotong University
66. SAVVY: Spatial Awareness via Audio-Visual LLMs through Seeing and HearingMingfei Chen, Zijun Cui, Xiulong Liu et al. · University of Washington
67. A multiscale analysis of mean-field transformers in the moderate interaction regimeGiuseppe Bruno, Federico Pasqualotto, Andrea Agazzi · Universität Bern
68. Exploring Diffusion Transformer Designs via GraftingKeshigeyan Chandrasegaran, Michael Poli, Daniel Y Fu et al. · Stanford University
69. Generalized Gradient Norm Clipping & Non-Euclidean $(L_0,L_1)$-SmoothnessThomas Pethick, Wanyun Xie, Mete Erdogan et al. · Swiss Federal Institute of Technology Lausanne
70. Rethinking Multimodal Learning from the Perspective of Mitigating Classification Ability DisproportionQing-Yuan Jiang, Longfei Huang, Yang Yang · Nanjing University of Science and Technology
71. TransferTraj: A Vehicle Trajectory Learning Model for Region and Task TransferabilityTonglong Wei, Yan Lin, Zeyu Zhou et al. · Beijing Jiaotong University
72. PhySense: Sensor Placement Optimization for Accurate Physics SensingYuezhou Ma, Haixu Wu, Hang Zhou et al. · Tsinghua University, Tsinghua University
73. InfinityStar: Unified Spacetime AutoRegressive Modeling for Visual GenerationJinlai Liu, Jian Han, Bin Yan et al. · ByteDance Inc.
74. Does Stochastic Gradient really succeed for bandits?Dorian Baudry, Emmeran Johnson, Simon Vary et al. · INRIA
75. Perception Encoder: The best visual embeddings are not at the output of the networkDaniel Bolya, Po-Yao Huang, Peize Sun et al. · Meta
76. PlayerOne: Egocentric World SimulatorYuanpeng Tu, Hao Luo, Xi Chen et al. · The University of Hong Kong
77. Mean Flows for One-step Generative ModelingZhengyang Geng, Mingyang Deng, Xingjian Bai et al. · Carnegie Mellon University

← Browse the whole archive