Searched refs:max_loss (Results 1 – 2 of 2) sorted by relevance
88 max_loss=None, train_steps=200, argument94 if max_loss is None:96 max_loss = 1.098 max_loss = 0.05 / (data_noise_stddev ** 2)140 self.assertLess(test_loss, max_loss)198 max_loss=50., # Just make sure there are no exceptions.204 max_loss=2.0,217 max_loss=1.5,
126 max_loss=DEFAULT_LOSS_THRESHOLD): argument145 if max_loss is not None and self.loss > max_loss: