We describe a procedure for explaining neurons in deep representations by identifying compositional logical concepts that closely approximate ...
確定! 回上一頁