def testConvOutputKroneckerFactorInit(self):
with tf_ops.Graph().as_default():
random_seed.set_random_seed(200)
tensor = array_ops.ones((2, 3, 4, 5), name='a/b/c')
factor = ff.ConvOutputKroneckerFactor(((tensor, ), ))
factor.instantiate_cov_variables()
self.assertEqual([5, 5], factor.get_cov().get_shape().as_list())
def testMakeCovarianceUpdateOp(self):
with tf_ops.Graph().as_default(), self.test_session() as sess:
random_seed.set_random_seed(200)
tensor = np.arange(1, 17).reshape(2, 2, 2, 2).astype(np.float32)
factor = ff.ConvOutputKroneckerFactor((array_ops.constant(tensor),))
sess.run(tf_variables.global_variables_initializer())
new_cov = sess.run(factor.make_covariance_update_op(.5))
self.assertAllClose([[43, 46.5], [46.5, 51.5]], new_cov)
def testConvOutputKroneckerFactorInitFloat64(self):
with tf_ops.Graph().as_default():
dtype = dtypes.float64_ref
random_seed.set_random_seed(200)
tensor = array_ops.ones((2, 3, 4, 5), dtype=dtype, name='a/b/c')
factor = ff.ConvOutputKroneckerFactor((tensor,))
cov = factor.get_cov()
self.assertEqual(cov.dtype, dtype)
self.assertEqual([5, 5], cov.get_shape().as_list())
def test3DConvolution(self):
with tf_ops.Graph().as_default():
batch_size = 1
width = 3
out_channels = width**3
factor = ff.ConvOutputKroneckerFactor(outputs_grads=([
random_ops.random_uniform(
(batch_size, width, width, width, out_channels), seed=0)
], ))
factor.instantiate_cov_variables()
with self.test_session() as sess:
sess.run(tf_variables.global_variables_initializer())
sess.run(factor.make_covariance_update_op(0.0))
cov = sess.run(factor.get_cov())
# Cov should be rank 3^3, as each spatial position donates a rank-1
# update.
self.assertMatrixRank(width**3, cov)
def testConvOutputKroneckerFactorInitNotEnoughDims(self):
with tf_ops.Graph().as_default():
random_seed.set_random_seed(200)
tensor = array_ops.ones((2, 3), name='a/b/c')
with self.assertRaises(IndexError):
ff.ConvOutputKroneckerFactor((tensor, ))
