原书出处
Designing Large Language Model Applications
Constructing Draft Models How do we ensure that the draft model has a high token acceptance rate? One way is to distill the draft model from the main model. This technique was introduced by Zhou et al. and is called DistillSpec. Zhou et al. introduced self-speculative decoding, where the draft model is a subset of the layers of the main model. In many use cases,
原书定位详情
来源:a19c7933-4eda-4176-aebd-30fe7e1ad292
正文位置:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch09.html
EPUB SHA-256:1b4d2347fec2f949fcddfbaf4fad597a1ae20887a096cfae64c59a619b4781bf