KV-cache compression

PruneHal

PruneHal: Reducing Hallucinations in Multi-modal Large Language Models through Adaptive KV Cache Pruning

On the frontierfirst seen Oct 22, 2025

Current frontier — recent, not yet superseded in the knowledge base

0 papers critique it · 0 beat it on benchmarks

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.