def testCustomTransform(self):
# Test applying sigmoid to the 2nd and 4th metrics.
sigmoid_metric_mask = [False, True, False, True]
def sigmoid_metric_transform(metrics: tf.Tensor):
batch_size = tf.shape(tf.nest.flatten(metrics)[0])[0]
sigmoid_batch_mask = tf.reshape(
tf.tile(sigmoid_metric_mask, [batch_size]),
[batch_size, len(sigmoid_metric_mask)])
return tf.where(sigmoid_batch_mask, tf.sigmoid(metrics), metrics)
sigmoid_scalarizer = multi_objective_scalarizer.LinearScalarizer(
[1, 2, 3, -1], sigmoid_metric_transform)
self.assertAllClose(sigmoid_scalarizer(self._batch_multi_objectives),
[10.77958, 26.995390, -9.7795804])
def testInvalidWeights(self):
with self.assertRaisesRegex(ValueError, 'at least two objectives'):
multi_objective_scalarizer.LinearScalarizer([])
with self.assertRaisesRegex(ValueError, 'at least two objectives'):
multi_objective_scalarizer.LinearScalarizer([1])
def setUp(self):
super(LinearScalarizerTest, self).setUp()
self._scalarizer = multi_objective_scalarizer.LinearScalarizer(
[1, 2, 3, -1])
self._batch_multi_objectives = tf.constant(
[[1, 2, 3, 4], [5, 6, 7, 8], [-1, -2, -3, -4]], dtype=tf.float32)
def setUp(self):
self._scalarizer = multi_objective_scalarizer.LinearScalarizer(
[1, 2, 3, -1])
super(LinearScalarizerTest, self).setUp()
