Splitting a fixed KV-cache budget into exact main entries and cluster-summary residual entries that share the softmax improves LongBench and RULER scores over eviction baselines.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
ResKV: Reconstructing Omitted Attention Contributions for Fixed-Budget KV Cache Compression
Splitting a fixed KV-cache budget into exact main entries and cluster-summary residual entries that share the softmax improves LongBench and RULER scores over eviction baselines.