提出Hyperflux，一种基于L0的新型网络剪枝方法，将剪枝建模为连续演化系统，揭示剪枝过程中神经元重要性变化。
arXiv:2504.05349v4 Announce Type: replace-cross Abstract: Network pruning is used to reduce inference latency and power consumption in large neural ne…
长文本生成中的KV缓存压缩新方法，利用动量策略在解码时高效压缩，大幅降低内存占用。
arXiv:2605.29873v1 Announce Type: new Abstract: Key-Value (KV) cache remains a major bottleneck for deploying Large Language Models (LLMs) in long-gen…
专为有状态LLM应用设计的记忆基础管理系统，VLDB26接收的高质量学术论文
arXiv:2605.29640v1 Announce Type: new Abstract: Large Language Models have revolutionized interactive applications; however, their finite context wind…
用可组合的元标记压缩KV缓存，高效保留上下文信息，大模型推理再提速。
arXiv:2605.22337v1 Announce Type: new Abstract: The KV cache used in large language models has linearly growing time complexity, so LLMs face memory b…