1
Self-Guided Test-Time Training for Long-Context LLMs
自我引导的测试时训练，专治长上下文LLM的准确率退化难题。
arXiv:2607.09415v1 Announce Type: cross Abstract: Long-context processing has become increasingly important for large language models (LLMs), but simp…
2
Vegas: Self-Speculative Decoding with Verification-Guided Sparse Attention
用验证引导的稀疏注意力加速长上下文LLM推理，自推测解码摆脱KV缓存瓶颈。
arXiv:2602.07223v2 Announce Type: replace Abstract: Long-context large language model (LLM) inference has become the norm for today's AI applications.…
3
Positional Failures in Long-Context LLMs: A Blind Spot in Reasoning Benchmarks
揭示长上下文LLM因位置偏差导致推理失败的盲点，挑战现有基准评估体系。
arXiv:2605.23170v1 Announce Type: cross Abstract: Position-controlled evaluation is standard for retrieval tasks such as Needle-in-a-Haystack and RULE…
4
Runtime-Certified Bounded-Error Quantized Attention
针对长上下文LLM推理中KV缓存量化的近似误差，提出运行时认证的有界误差方法，确保量化后的注意力计算精度可控。
arXiv:2605.20868v1 Announce Type: new Abstract: KV cache quantization reduces the memory cost of long-context LLM inference, but introduces approximat…
5
ProxyKV: Cross-Model Proxy Pruning for Efficient Long-Context LLM Inference
跨模型代理剪枝巧妙兼顾低延迟与高精度，解决长上下文LLM推理中KV缓存内存墙难题
arXiv:2605.16360v1 Announce Type: new Abstract: Efficient long-context inference in Large Language Models (LLMs) is severely constrained by the Key-Va…