HARD-KV bridges dynamic head-adaptive KV cache compression with static inference engine constraints via Cascade Cache and Logits Calibration, reporting up to 2x throughput gains on long-context math benchmarks.
In- context kv-cache eviction for llms via attention-gate
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
-
HARD-KV: Head-Adaptive Regularization for Decoding-time KV Compression
HARD-KV bridges dynamic head-adaptive KV cache compression with static inference engine constraints via Cascade Cache and Logits Calibration, reporting up to 2x throughput gains on long-context math benchmarks.