Meta-Soft: Leveraging Composable Meta-Tokens for Context-Preserving KV Cache Compression 事件

Meta-Soft: Leveraging Composable Meta-Tokens for Context-Preserving KV Cache Compression arXiv:2605.22337v2 Announce Type: replace Abstract: The KV cache used in large language models has linearly growing time complexity, so LLMs face memory blow-up and reduced decoding efficiency when they process long contexts. Current KV Cache eviction has become an important research direction; however, existing methods based on fixed Soft Tokens (e.g., Judge Q) rely on a static parameter set as the query t