TY  - RPRT
TI  - H1B-KV: Hybrid One-Bit Caches for Memory-Efficient Large Language Model Inference
AU  - Harshil Vejendla
PY  - 2025
UR  - https://arxiv.org/abs/2510.05529
ID  - 2510.05529
ER  -