Knowledge Graph

Graph explorer

Search resolved entities, walk their edges, and read the sentence behind every one.

Entities25 of 207,987

Entity

key-value cache quantization

technology4 mentionslast seen Nov 6, 2026

A method to reduce memory bandwidth requirements in long-context LLM decoding by storing quantized key-value caches in dense on-chip non-volatile memory.

Relationships 13

key-value cache quantization

Relationship

key-value cache quantizationsupersedesKIVI K2V2

70% confidence1 supporting mentions

Either name opens that entity’s relationships.

Sources 1

preprintarXivSep 4, 2026Interface-Aware KV Cache Quantization for Dense On-Chip NVM in Long-Context LLM Decoding
KIVI attaches per-group metadata, adding about 25% to the stored KV cache; KVQuant keeps sparse full-precision outliers that a dense array cannot hold in place. This paper examines what these structures cost when the KV cache resides in NVM behind fixed-range converters, and designs a quantization scheme matched to that interface.