The per-layer terminology in that paper is slightly ambiguous. They aren't referring to the layer-specific learning rates.
確定! 回上一頁