Generating parameters from fixed weight matrix promotes efficiency
#model-compression #weight-matrix
Teach a model to fish, rather than give a model a fish
#knowledge-distillation #model-compression