Knowledge Distillation Explained Teacher to Student
Автор: Zero to Builder: Cloud, AI & Blockchain
Загружено: 2026-11-05
Просмотров: 2
Описание:
Knowledge distillation transfers a large teacher model soft predictions into a small student that runs faster with most of the accuracy.
What you will learn:
What soft targets reveal beyond hard labels.
How temperature softens teacher outputs.
How to balance distillation loss against true-label loss.
Practice task:
Distill a large trained classifier into a smaller student, compare accuracy and inference speed of teacher, student, and a student trained alone.
---
Subscribe to Zero to Builder: / @buildzerotobuilder
Full AI course playlist: • AI Zero to Builder
#ZeroToBuilder
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке: