def forward_pass(self, batch, training=True):
del training
y = tf.nn.softmax(
tf.matmul(batch['x'], self._variables.weights) + self._variables.bias)
return model_lib.BatchOutput(
predictions=y, labels=batch['y'], num_examples=tf.size(batch['y']))
def forward_pass(self, batch, training=True):
del training
y = batch['x'] * self._variables.weights + self._variables.bias
return model_lib.BatchOutput(predictions=y,
labels=batch['y'],
num_examples=tf.size(batch['y']))
def forward_pass(self, batch_input, training=True):
if hasattr(batch_input, '_asdict'):
batch_input = batch_input._asdict()
if isinstance(batch_input, collections.abc.Mapping):
inputs = batch_input.get('x')
else:
inputs = batch_input[0]
if inputs is None:
raise KeyError('Received a batch_input that is missing required key `x`. '
'Instead have keys {}'.format(list(batch_input.keys())))
predictions = self._keras_model(inputs, training=training)
if isinstance(batch_input, collections.abc.Mapping):
y_true = batch_input.get('y')
else:
y_true = batch_input[1]
return model_lib.BatchOutput(
predictions=predictions,
labels=y_true,
num_examples=tf.shape(tf.nest.flatten(inputs)[0])[0])