InnerQ delivers 1.3x average speedup over prior KV cache quantization and 2.7x over baseline by inner-dimension grouping, hybrid symmetric/asymmetric quantization, high-precision windows for recent and sink tokens, and prefold per-channel key normalization.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
InnerQ: Hardware-Aware Tuning-Free Quantization of KV Cache for Large Language Models
InnerQ delivers 1.3x average speedup over prior KV cache quantization and 2.7x over baseline by inner-dimension grouping, hybrid symmetric/asymmetric quantization, high-precision windows for recent and sink tokens, and prefold per-channel key normalization.