1 paper
Abhinand Balachandran, Praveen Prashant
Knowledge distillation trains a smaller student to match the outputs of a larger teacher. Feature-based methods also align intermediate representations, but this extra constraint m…