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 298 篇，第 1 / 6 页

1. NeurIPS 2026$\boldsymbol{f}$-OPD: Stabilizing Long-Horizon On-Policy Distillation with Freshness-Aware Control
2. NeurIPS 2026A Self-Evolving Framework for Efficient Terminal Agents via Observational Context Compression
3. NeurIPS 2026A Solvable Model of Chain-of-Thought in In-Context Learning
4. NeurIPS 2026Accelerating Long-Context LLM Prefill via Layer-wise Progressive Token Pruning in Local Deployment
5. NeurIPS 2026AdaCal: Adaptive Calibration for Robust Sparse Attention in Long-Context LLMs
6. NeurIPS 2026Adaptive Mass-Segmented KV Compression for Long-Context Reasoning
7. NeurIPS 2026Align-RAG: Alignment Is All You Need for TSFM In-Context Learning
8. NeurIPS 2026Ask Early, Ask Late, Ask Right: When Does Clarification Timing Matter for Long-Horizon Agents?
9. NeurIPS 2026Attention as In-Context Empirical Bayes: A Two-Stage View via Particle Dynamics
10. NeurIPS 2026Attention Drift: What Auto-Regressive Speculative Decoding Models Learn
11. NeurIPS 2026Attention Itself Could Retrieve. RetrieveVGGT: Training-Free Long Context Streaming 3D Reconstruction via Query-Key Similarity Retrieval
12. NeurIPS 2026Beyond Suspicious Steps: Ontological Trust in Long-Horizon Agents
13. NeurIPS 2026Block Sparse Flash Attention
14. NeurIPS 2026Block-Wise Differentiable Sinkhorn Attention: Tail-Refinement Gradients with a Gap-Aware Dustbin Bridge
15. NeurIPS 2026Breadcrumbing Search Agents: Per-Turn Scheming Over Long-Horizon Trajectories
16. NeurIPS 2026Bridging Risk Approximation Gaps in Model Predictive Task Sampling via In-Context Modeling
17. NeurIPS 2026Can In-Context Learning Support Intrinsic Curiosity?
18. NeurIPS 2026Can Language Models Actually Retrieve In-Context? Drowning in Documents at Million Token Scale
19. NeurIPS 2026CEO-Bench: Can Agents Play the Long Game?
20. NeurIPS 2026Choosing Before Acting: Comparative Value Estimation for Long-Horizon Tool-Use Agents
21. NeurIPS 2026CivBench: A Long-Horizon Benchmark for Tool-Mediated Agents in Civilization VI
22. NeurIPS 2026CLoSeR: Closing the Loop for Long-Context Streaming Reconstruction
23. NeurIPS 2026CoLaX: Context-Aware Local Explanations for Time Series Classification
24. NeurIPS 2026CommunityKV: Efficient Long-Context Decoding via Graph Partitioning
25. NeurIPS 2026CompactAttention: Accelerating Chunked Prefill with Block-Union KV Selection
26. NeurIPS 2026Constrained Modulatory Reservoirs for Context-Dependent Computation
27. NeurIPS 2026Context as Low-Rank Weights: Bounded Parametric Dynamic Memory for Unbounded Context
28. NeurIPS 2026ContextShift: A Controlled Benchmark for Context Dependence in Object Detection
29. NeurIPS 2026Controlling Transient Amplification Improves Long-horizon Rollouts
30. NeurIPS 2026Correspondence Pruning by Iterative Structural Rectification
31. NeurIPS 2026CRUMB: Efficient Prior Fitted Network Inference via Distributionally Matched Context Batching
32. NeurIPS 2026CutAttn: Discovering Cognitive Transition Layers for Efficient Long-Context Prefilling
33. NeurIPS 2026Do Enterprise Systems Need Learned World Models? The Importance of Context to Infer Dynamics
34. NeurIPS 2026DSR-TSF: Spectrum-Driven Dynamic Routing for Efficient Long-Horizon Time Series Forecasting
35. NeurIPS 2026DualKV: Shared-Prompt Flash Attention for Efficient RL Training with Large Rollouts and Long Contexts
36. NeurIPS 2026Dynamic Context Modeling for Longitudinal Mental Health Monitoring under Distribution Shift
37. NeurIPS 2026EAT: Eviction-Aware Training for Long-Context LLM Inference on Edge Devices
38. NeurIPS 2026Effective Context in Transformers: An Analysis of Fragmentation and Tokenization
39. NeurIPS 2026Efficient Scaling of LLM Training with Flexible Context Parallelism
40. NeurIPS 2026Facts Don't Speak Louder than Words: The Behavioral Essence of Long-CoT Distillation
41. NeurIPS 2026FFTSparse: Relative-Position Correlation Guided Sparse Attention for Long-Context Language Models
42. NeurIPS 2026Flow Matching with In-Context Priors for Out-of-Distribution Brain Dynamics
43. NeurIPS 2026Flux Attention: Context-Aware Hybrid Attention for Efficient LLMs Inference
44. NeurIPS 2026Fractional State Space Transition for Long Sequence Modeling
45. NeurIPS 2026FrameScout: Scouting Query-Relevant Frames for Long Video Understanding
46. NeurIPS 2026From History to State: Constant-Context Skill Learning for LLM Agents
47. NeurIPS 2026From Label Priors to Task Evidence: Long-Video Frame Selection via Bayesian GRPO
48. NeurIPS 2026GeneZip: Region-Aware Compression for Long Context DNA Modeling
49. NeurIPS 2026Geometry Matters in Packed 3D Attention
50. NeurIPS 2026Get a GRIP, this will be a long TRIP: A Quantifiable Long-Range Framework for Verifying Over-squashing

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。