SparKV: Overhead-Aware KV Cache Loading for Efficient On-Device LLM Inference

source_context

{
  "surface": "paper",
  "mode": "paper",
  "query": "SparKV: Overhead-Aware KV Cache Loading for Efficient On-Device LLM Inference",
  "normalized_query": "2604.21231",
  "route": "/paper/sparkv-overhead-aware-kv-cache-loading-for-efficient-on-device-llm-inference",
  "paper_ref": "sparkv-overhead-aware-kv-cache-loading-for-efficient-on-device-llm-inference",
  "topic_slug": null,
  "benchmark_ref": null,
  "dataset_ref": null
}