Sentinel: Decoding Context Utilization via Attention Probing for Efficient LLM Context Compression

arXiv:2505.23277 · cs.CL, cs.AI · Submitted 2025-05-29 · Read on arXiv

cs.CL, cs.AI

Submitted: 2025-05-29

Updated: 2026-11-07

Code: https://github.com/yzhangchuck/Sentinel

Terminology

Sources

Related papers