Specify which shape axis the channel is specified
Do not select CUDNN operator, if available
Epsilon to prevent div 0. Must be no less than CUDNN_BN_MIN_EPSILON defined in cudnn.h when using cudnn (usually 1e-5)
Fix gamma while training
The maximum scalar value in the form of float32 obtained through calibration. If present, it will be used to by quantized batch norm op to calculate primitive scale.Note: this calib_range is to calib bn output.
The minimum scalar value in the form of float32 obtained through calibration. If present, it will be used to by quantized batch norm op to calculate primitive scale.Note: this calib_range is to calib bn output.
Momentum for moving average
Output the mean and inverse std
Whether use global moving statistics instead of local batch-norm. This will force change batch-norm into a scale shift operator.
This Param Object is specifically used for BatchNorm