Pith. sign in

← back to paper

Review history

arxiv: 2607.01237 · 2 revisions

KARA: Efficient Reasoning LLM Serving via Sliding-Window KV Cache Compression

  1. 2026-07-12 CONDITIONAL MODERATE v1.1.0-grok45 novelty 5.0
    31276 ms 21801 in 3347 out 2026-07-12T17:53:32.046783+00:00
  2. 2026-07-04 UNVERDICTED LOW v0.9.1-grok novelty 6.0
    38106 ms 5766 in 1130 out 2026-07-04T01:21:45.333440+00:00