def output(self, inputs, ops=None, prefix=None, **kwargs):
""" Add output operations to a model graph, like predictions, quality metrics, etc.
Parameters
----------
inputs : tf.Tensor or a sequence of tf.Tensors
input tensors
ops : a sequence of str
operation names::
- 'sigmoid' - add ``sigmoid(inputs)``
- 'proba' - add ``softmax(inputs)``
- 'labels' - add ``argmax(inputs)``
- 'accuracy' - add ``mean(predicted_labels == true_labels)``
prefix : a sequence of str
a prefix for each input if there are multiple inputs
Raises
------
ValueError if the number of outputs does not equal to the number of prefixes
TypeError if inputs is not a Tensor or a sequence of Tensors
"""
kwargs = self.fill_params('output', **kwargs)
predictions_op = self.pop('predictions', kwargs, default=None)
if ops is None:
ops = []
elif not isinstance(ops, (list, tuple)):
ops = [ops]
if not isinstance(inputs, (tuple, list)):
inputs = [inputs]
prefix = prefix or 'output'
prefix = [prefix]
if len(inputs) != len(prefix):
raise ValueError('Each output in multiple output models should have its own prefix')
for i, tensor in enumerate(inputs):
if not isinstance(tensor, tf.Tensor):
raise TypeError("Network output is expected to be a Tensor, but given {}".format(type(inputs)))
current_prefix = prefix[i]
if current_prefix:
ctx = tf.variable_scope(current_prefix)
ctx.__enter__()
else:
ctx = None
attr_prefix = current_prefix + '_' if current_prefix else ''
pred_prefix = '' if len(inputs) == 1 else attr_prefix
self._add_output_op(tensor, predictions_op, 'predictions', pred_prefix, **kwargs)
for oper in ops:
self._add_output_op(tensor, oper, oper, attr_prefix, **kwargs)
if ctx:
ctx.__exit__(None, None, None)
评论列表
文章目录
