This version of the operator has been available since version 1.
axis: The first normalization dimension: normalization will be performed along dimensions axis : rank(inputs). Default value is
epsilon: The epsilon value to use to avoid division by zero. Default value is
stash_type: type used for stash mean/inv_std_var Default value is
X (heterogeneous) - T: Input data tensor from the previous layer.
scale (heterogeneous) - V: Scale tensor.
Between 1 and 2 outputs.
Y (heterogeneous) - V: Output data tensor.
inv_std_var (optional, heterogeneous) - U: Saved inverse standard variance used during training to speed up gradient computation.