Download
https://hgpu.org/?p=23851
Mixed-Precision Embedding Using a Cache
Share