WiSP achieves up to 1.95x decode throughput on low-resource MoE serving by dynamically paging reused experts and using MV-WSA to allocate VRAM between experts and KV cache, with the offline policy performing well on both prefill and decode.
DyMoE: Dynamic expert orchestration with mixed- precision quantization for efficient MoE inference on edge.arXiv preprint arXiv:2603.19172,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
WiSP: A Working-Set View of Mixture-of-Experts Serving on Extremely Low-Resource Hardware
WiSP achieves up to 1.95x decode throughput on low-resource MoE serving by dynamically paging reused experts and using MV-WSA to allocate VRAM between experts and KV cache, with the offline policy performing well on both prefill and decode.