Quantization of the KV cache beats rank reduction for matched storage budgets by 4-364 PPL, because dimension removal can flip attention token selection under softmax while bounded quantization noise usually preserves ordering.
H2O: Heavy-Hitter Oracle for Efficient Generative Inference
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Quantization Dominates Rank Reduction for KV-Cache Compression
Quantization of the KV cache beats rank reduction for matched storage budgets by 4-364 PPL, because dimension removal can flip attention token selection under softmax while bounded quantization noise usually preserves ordering.