def test_bad_metric(self):
gin.bind_parameter("discretizer.discretizer_fn", _identity_discretizer)
gin.bind_parameter("discretizer.num_bins", 10)
ground_truth_data = dummy_data.IdentityObservationsData()
representation_function = lambda x: np.zeros_like(x, dtype=np.float64)
random_state = np.random.RandomState(0)
scores = irs.compute_irs(ground_truth_data, representation_function,
random_state, 0.99, 3000, 3000)
self.assertBetween(scores["IRS"], 0.0, 0.1)
np.random.RandomState(0), 64, 10000,
5000, 10000)
print(' factor_vae: %.6f' % scores['eval_accuracy'])
scores = dci.compute_dci(ground_truth_data, representation_fn,
np.random.RandomState(0), 10000, 5000)
print(' dci: %.6f' % scores['disentanglement'])
scores = sap_score.compute_sap(ground_truth_data,
representation_fn,
np.random.RandomState(0),
10000,
5000,
continuous_factors=False)
print(' sap_score: %.6f' % scores['SAP_score'])
import gin.tf
gin.bind_parameter("discretizer.discretizer_fn", utils._histogram_discretize)
gin.bind_parameter("discretizer.num_bins", 20)
scores = mig.compute_mig(ground_truth_data, representation_fn,
np.random.RandomState(0), 10000)
print(' mig: %.6f' % scores['discrete_mig'])
gin.bind_parameter("irs.batch_size", 16)
scores = irs.compute_irs(ground_truth_data,
representation_fn,
np.random.RandomState(0),
num_train=10000)
print(' irs: %.6f' % scores['IRS'])