Model size reduction is incredibly important for getting ML running on people's devices. This paper is an awesome summary of the existing research, and its rela

Model size reduction is incredibly important for getting ML running on people's devices. This paper is an awesome summary of the existing research, and its relative lack of coherence. :)