@misc{indiciaeaa7bafe09289,
  title = {MemDecay: Region-Aware KV Cache Eviction for Efficient LLM Agent Inference},
  author = {Venkatesha Matam and Keon Kim},
  year = {2026},
  url = {https://arxiv.org/abs/2607.10582},
  note = {Source identifier: 2607.10582}
}