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 87 篇，第 2 / 2 页

51. ICLR 2025miniCTX: Neural Theorem Proving with (Long-)Contexts
52. ICLR 2025Needle Threading: Can LLMs Follow Threads Through Near-Million-Scale Haystacks?
53. ICLR 2025NovelQA: Benchmarking Question Answering on Documents Exceeding 200K Tokens
54. ICLR 2025OmniKV: Dynamic Context Selection for Efficient Long-Context LLMs
55. ICLR 2025On the Learn-to-Optimize Capabilities of Transformers in In-Context Sparse Recovery
56. ICLR 2025Oscillatory State-Space Models
57. ICLR 2025P-SPIKESSM: HARNESSING PROBABILISTIC SPIKING STATE SPACE MODELS FOR LONG-RANGE DEPENDENCY TASKS
58. ICLR 2025PEARL: Towards Permutation-Resilient LLMs
59. ICLR 2025PICASO: Permutation-Invariant Context Composition with State Space Models
60. ICLR 2025Quest: Query-centric Data Synthesis Approach for Long-context Scaling of Large Language Model
61. ICLR 2025Radar: Fast Long-Context Decoding for Any Transformer
62. ICLR 2025ReAttention: Training-Free Infinite Context with Finite Attention Scope
63. ICLR 2025Rethinking Invariance in In-context Learning
64. ICLR 2025Retrieval Head Mechanistically Explains Long-Context Factuality
65. ICLR 2025RetroInText: A Multimodal Large Language Model Enhanced Framework for Retrosynthetic Planning via In-Context Representation Learning
66. ICLR 2025Revisiting In-context Learning Inference Circuit in Large Language Models
67. ICLR 2025Samba: Simple Hybrid State Space Models for Efficient Unlimited Context Language Modeling
68. ICLR 2025Scaling Instruction-tuned LLMs to Million-token Contexts via Hierarchical Synthetic Data Generation
69. ICLR 2025Scaling Long Context Training Data by Long-Distance Referrals
70. ICLR 2025SCBench: A KV Cache-Centric Analysis of Long-Context Methods
71. ICLR 2025Scrutinize What We Ignore: Reining In Task Representation Shift Of Context-Based Offline Meta Reinforcement Learning
72. ICLR 2025Self-Updatable Large Language Models by Integrating Context into Model Parameters
73. ICLR 2025TimeKAN: KAN-based Frequency Decomposition Learning Architecture for Long-term Time Series Forecasting
74. ICLR 2025Timer-XL: Long-Context Transformers for Unified Time Series Forecasting
75. ICLR 2025Towards Learning High-Precision Least Squares Algorithms with Sequence Models
76. ICLR 2025Training Free Exponential Context Extension via Cascading KV Cache
77. ICLR 2025Transformers are Universal In-context Learners
78. ICLR 2025Transformers Handle Endogeneity in In-Context Linear Regression
79. ICLR 2025Understanding and Mitigating Bottlenecks of State Space Models through the Lens of Recency and Over-smoothing
80. ICLR 2025Unsupervised Meta-Learning via In-Context Learning
81. ICLR 2025Vector-ICL: In-context Learning with Continuous Vector Representations
82. ICLR 2025VL-ICL Bench: The Devil in the Details of Multimodal In-Context Learning
83. ICLR 2025Wavelet-based Positional Representation for Long Context
84. ICLR 2025What is Wrong with Perplexity for Long-context Language Modeling?
85. ICLR 2025Why Does the Effective Context Length of LLMs Fall Short?
86. ICLR 2025Why In-Context Learning Models are Good Few-Shot Learners?
87. ICLR 2025ZETA: Leveraging $Z$-order Curves for Efficient Top-$k$ Attention

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。