def _setup_loss(self):
if self.loss[TYPE] == 'softmax_cross_entropy':
self.train_loss_function = SoftmaxCrossEntropyLoss(
num_classes=self.num_classes,
feature_loss=self.loss,
name='train_loss'
)
elif self.loss[TYPE] == 'sampled_softmax_cross_entropy':
self.train_loss_function = SampledSoftmaxCrossEntropyLoss(
decoder_obj=self.decoder_obj,
num_classes=self.num_classes,
feature_loss=self.loss,
name='train_loss'
)
else:
raise ValueError(
"Loss type {} is not supported. Valid values are "
"'softmax_cross_entropy' or "
"'sampled_softmax_cross_entropy'".format(self.loss[TYPE])
)
self.eval_loss_function = SoftmaxCrossEntropyLoss(
num_classes=self.num_classes,
feature_loss=self.loss,
name='eval_loss')
def __init__(self,
decoder_obj=None,
num_classes=0,
feature_loss=None,
name='sampled_softmax_cross_entropy_metric'):
super(SampledSoftmaxCrossEntropyMetric, self).__init__(name=name)
self.metric_function = SampledSoftmaxCrossEntropyLoss(
decoder_obj=decoder_obj,
num_classes=num_classes,
feature_loss=feature_loss)
