AI & tools
Model distillation and model pruning
Distillation and pruning reduce the parameter count and computational demands of neural networks. Distillation trains a student to imitate a teacher’s predictions; pruning removes unnecessary parameters and connections.