TY  - RPRT
TI  - HAWK: Heterogeneity-Aware KV Cache Transfer for Efficient GPU-NPU Phase-Disaggregated LLM Inference
AU  - Junhyeok Park
AU  - Jungeun Yoon
AU  - Kyungmin Kim
AU  - Hunsoon Lee
AU  - Shinyoung Ahn
AU  - Kangho Kim
PY  - 2026
UR  - https://arxiv.org/abs/2610.29873
ID  - 2610.29873
ER  -