def bias_add(value, bias, data_format='NCHW', name=None):
"""
Adds `bias` to `value`.
This is (mostly) a special case of `tf.add` where `bias` is restricted to 1-D.
Broadcasting is supported, so `value` may have any number of dimensions.
Unlike `tf.add`, the type of `bias` is allowed to differ from `value` in the
case where both types are quantized.
Args:
value: A `Tensor`.
bias: A 1-D `Tensor` with size matching the last dimension of `value`.
data_format: A string. 'NHWC' and 'NCHW' are supported.
name: A name for the operation (optional).
Returns:
A `Tensor` with the same type as `value`.
"""
return ops.BiasAdd([value, bias], data_format=data_format, name=None)
def call(self, inputs, training=False):
use_stats = 0 if training else 1
if self.fused:
return ops.FusedBatchNorm([
inputs, self.moving_mean, self.moving_variance, self.gamma,
self.beta
],
axis=self.axis,
momentum=self.momentum,
eps=self.epsilon,
use_stats=use_stats,
mode='DEFAULT')
x_norm = ops.BatchNorm(
[inputs, self.moving_mean, self.moving_variance],
axis=self.axis,
momentum=self.momentum,
eps=self.epsilon,
use_stats=use_stats,
mode='DEFAULT')
if self.gamma is not None:
# use scale
if self.beta is not None:
return ops.Scale([x_norm, self.gamma, self.beta],
axis=self.axis,
num_axes=1)
else:
return ops.Scale([x_norm, self.gamma],
axis=self.axis,
num_axes=1)
else:
# do not use scale
if self.beta is not None:
return ops.BiasAdd([x_norm, self.beta],
data_format=self._data_format)
else:
return x_norm
def bias_add(value, bias, data_format='NHWC', name=None):
return ops.BiasAdd([value, bias], data_format=data_format)