长上下文论文趋势 2023–2026

主要结论

• 长上下文论文从 2023 年的 66 篇增长到 2026 年的 341 篇，三年间扩大到约 5.2 倍。
• NeurIPS 是发长上下文论文最多的会议，2023–2026 年共 298 篇，占四个会议总量的 36%。
• 交叉方向里涨得最快的是AI 智能体：同时属于两个主题的论文从 2023 年的 0 篇增至 2026 年的 45 篇。
• 在「长上下文与记忆」的 5 个主题里，它的 2026 年论文数排第 1；同组增长最快的检索记忆是 6.6 倍，长上下文是 5.2 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026Accelerating Long-Context LLM Prefill via Layer-wise Progressive Token Pruning in Local Deployment
• ICML 2026AdaSplash-2: Faster Differentiable Sparse Attention
• ICLR 2026An evolutionary perspective on modes of learning in Transformers
• CVPR 2026AR2-4FV Anchored Referring and Re-identification for Long-Term Grounding in Fixed-View
• NeurIPS 2025A Controllable Examination for Long-Context Language Models
• ICML 2025CateKV: On Sequential Consistency for Long-Context LLM Inference Acceleration
• ICLR 2025$\text{D}_{2}\text{O}$: Dynamic Discriminative Operations for Efficient Long-Context Inference of Large Language Models
• CVPR 2025BIMBA: Selective-Scan Compression for Long-Range Video Question Answering
• NeurIPS 2024[Re] CUDA: Curriculum of Data Augmentation for Long‐tailed Recognition
• ICML 2024Benchmarking and Building Long-Context Retrieval Models with LoCo and M2-BERT
• ICLR 2024CLEX: Continuous Length Extrapolation for Large Language Models
• CVPR 2024GigaTraj: Predicting Long-term Trajectories of Hundreds of Pedestrians in Gigapixel Complex Scenes
• NeurIPS 2023A Single 2D Pose with Context is Worth Hundreds for 3D Human Pose Estimation
• ICML 2023CAB: Comprehensive Attention Benchmarking on Long Sequence Modeling
• ICLR 2023$k$NN Prompting: Beyond-Context Learning with Calibration-Free Nearest Neighbor Inference
• CVPR 2023AutoAD: Movie Description in Context
• NeurIPS 2026AdaCal: Adaptive Calibration for Robust Sparse Attention in Long-Context LLMs
• ICML 2026Artificial Hippocampus Networks for Efficient Long-Context Modeling
• ICLR 2026AutoSP: Unlocking Long-Context LLM Training Via Compiler-Based Sequence Parallelism
• CVPR 2026AstraNav-Memory Contexts Compression for Long Memory

全部长上下文论文

共 230 篇，第 2 / 5 页

51. ICLR 2026MrRoPE: Mixed-radix Rotary Position Embedding
52. ICLR 2026On learning linear dynamical systems in context with attention layers
53. ICLR 2026Out of the Memory Barrier: A Highly Memory-Efficient Training System for LLMs with Million-Token Contexts
54. ICLR 2026PERK: Long-Context Reasoning as Parameter-Efficient Test-Time Learning
55. ICLR 2026PM-KVQ: Progressive Mixed-precision KV Cache Quantization for Long-CoT LLMs
56. ICLR 2026PMDformer: Patch-Mean Decoupling Transformer for Long-term Forecasting
57. ICLR 2026ProtoKV: Long-context Knowledges Are Already Well-Organized Before Your Query
58. ICLR 2026Queue Length Regret Bounds for Contextual Queueing Bandits
59. ICLR 2026RACE Attention: A Strictly Linear-Time Attention for Long-Sequence Training
60. ICLR 2026ReST-KV: Robust KV Cache Eviction with Layer-wise Output Reconstruction and Spatial-Temporal Smoothing
61. ICLR 2026Retrospective Sparse Attention for Efficient Long-Context Generation
62. ICLR 2026Revisiting Long-context Modeling from Context Denoising Perspective
63. ICLR 2026Reward Is Enough: LLMs Are In-Context Reinforcement Learners
64. ICLR 2026RMAAT: Astrocyte-Inspired Memory Compression and Replay for Efficient Long-Context Transformers
65. ICLR 2026SARM: Stage-Aware Reward Modeling for Long Horizon Robot Manipulation
66. ICLR 2026Sculptor: Empowering LLMs with Cognitive Agency via Active Context Management
67. ICLR 2026Short Window Attention Enables Long-Term Memorization
68. ICLR 2026SinkTrack: Attention Sink based Context Anchoring for Large Language Models
69. ICLR 2026Smooth Reading: Bridging the Gap of Recurrent LLM to Self-Attention LLM on Long-Context Understanding
70. ICLR 2026SoLoPO: Unlocking Long-Context Capabilities in LLMs via Short-to-Long Preference Optimization
71. ICLR 2026Stacked from One: Multi-Scale Self-Injection for Context Window Extension
72. ICLR 2026Story-Iter: A Training-free Iterative Paradigm for Long Story Visualization
73. ICLR 2026SWINGARENA: Adversarial Programming Arena for Long-context GitHub Issue Solving
74. ICLR 2026Tactic: Adaptive Sparse Attention with Clustering and Distribution Fitting for Long-Context LLMs
75. ICLR 2026TAPTRv3: Spatial and Temporal Context Foster Robust Tracking of Any Point in Long Video
76. ICLR 2026Test-Time Training Done Right
77. ICLR 2026Transformers Learn Latent Mixture Models In-Context via Mirror Descent
78. ICLR 2026UltraLLaDA: Scaling the Context Length to 128K for Diffusion Large Language Models
79. ICLR 2026UltraMemV2: Memory Networks Scaling to 120B Parameters with Superior Long-Context Learning
80. ICLR 2026Understanding and Improving Length Generalization in Hierarchical Sparse Attention Models
81. ICLR 2026VideoChat-Flash: Hierarchical Compression for Long-Context Video Modeling
82. ICLR 2026When Does Divide and Conquer Work for Long Context LLM? A Noise Decomposition Framework
83. ICLR 2025$\text{D}_{2}\text{O}$: Dynamic Discriminative Operations for Efficient Long-Context Inference of Large Language Models
84. ICLR 2025A Causal Lens for Learning Long-term Fair Policies
85. ICLR 2025A Little Goes a Long Way: Efficient Long Context Training and Inference with Partial Contexts
86. ICLR 2025A Training-Free Sub-quadratic Cost Transformer Model Serving Framework with Hierarchically Pruned Attention
87. ICLR 2025BenTo: Benchmark Reduction with In-Context Transferability
88. ICLR 2025Bridging Context Gaps: Leveraging Coreference Resolution for Long Contextual Understanding
89. ICLR 2025Can Generative AI Solve Your In-Context Learning Problem? A Martingale Perspective
90. ICLR 2025Can In-context Learning Really Generalize to Out-of-distribution Tasks?
91. ICLR 2025Can LLMs Really Learn to Translate a Low-Resource Language from One Grammar Book?
92. ICLR 2025ChatQA 2: Bridging the Gap to Proprietary LLMs in Long Context and RAG Capabilities
93. ICLR 2025Competition Dynamics Shape Algorithmic Phases of In-Context Learning
94. ICLR 2025Context Clues: Evaluating Long Context Models for Clinical Prediction Tasks on EHR Data
95. ICLR 2025Context-Parametric Inversion: Why Instruction Finetuning May Not Actually Improve Context Reliance
96. ICLR 2025CURIE: Evaluating LLMs on Multitask Scientific Long-Context Understanding and Reasoning
97. ICLR 2025DeciMamba: Exploring the Length Extrapolation Potential of Mamba
98. ICLR 2025Density estimation with LLMs: a geometric investigation of in-context learning trajectories
99. ICLR 2025Differential learning kinetics govern the transition from memorization to generalization during in-context learning
100. ICLR 2025Dual Process Learning: Controlling Use of In-Context vs. In-Weights Strategies with Weight Forgetting

常见问题

2023–2026 年四大会议一共有多少篇长上下文论文？

共 838 篇，按年份分别是 2023 年 66 篇、2024 年 176 篇、2025 年 255 篇、2026 年 341 篇。

NeurIPS 2026 有多少篇长上下文论文？

NeurIPS 2026 有 130 篇长上下文论文，上一届是 98 篇。

ICLR 2026 有多少篇长上下文论文？

ICLR 2026 有 82 篇长上下文论文，上一届是 87 篇。

ICML 2026 有多少篇长上下文论文？

ICML 2026 有 98 篇长上下文论文，上一届是 53 篇。

CVPR 2026 有多少篇长上下文论文？

CVPR 2026 有 31 篇长上下文论文，上一届是 17 篇。

哪个会议的长上下文论文最多？

NeurIPS，2023–2026 年共 298 篇。