inference/research
verifiedhigh · 0.88updated 2026-10-03

ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inference (Li et al., ByteDance)

paper

Fields

id
sources/qlab-shadowkv-paper
URL
https://arxiv.org/abs/2410.21465
Publisher
arXiv
Kind
paper
Published
2024-10-28
Accessed
2026-10-03
notes
v3. Low-rank key cache on GPU, value cache offloaded to CPU; on-the-fly sparse KV reconstruction.

Notes

v3. Low-rank key cache on GPU, value cache offloaded to CPU; on-the-fly sparse KV reconstruction.

Cross-references

This record references no other records.