原书出处
Designing Large Language Model Applications
One drawback of using adapter-based fine-tuning techniques is that during inference, each batch can support only a single adapter instance, i.e., an adapter fine-tuned for a particular task. Prefix-tuning, in contrast, enables multiple tasks to be run in the same batch.
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/ch07.html
EPUB SHA-256:1b4d2347fec2f949fcddfbaf4fad597a1ae20887a096cfae64c59a619b4781bf