On the Size and Approximation Error of Distilled Datasets | Read Paper on Bytez