def test_right_shift_rows(self):
shift = m_helpers.right_shift_rows(numpy.array([[1,2,3], [4,5,6]]), 1, -1)
correct = [
[-1, 1, 2],
[-1, 4, 5]
]
helpers.assert_theano_equal(shift, correct)
def test_accuracy(self):
P = ctc.PADDING
activations = numpy.array([
[[0,1,0], [0,1,0]],
[[0,1,0], [0,0,1]],
[[1,0,0], [1,0,0]],
[[0,0,1], [0,0,1]],
[[1,0,0], [1,0,0]],
[[1,0,0], [1,0,0]]
], dtype=theano.config.floatX)
targets = numpy.array([
[1,1],
[2,2],
[P,2],
[P,1]
], dtype=theano.config.floatX)
target_lengths = numpy.array([2, 4], dtype=theano.config.floatX)
helpers.assert_theano_equal(
ctc.accuracy(activations, targets, target_lengths),
0.875
)
def test_skip_allowed(self):
l0 = ctc._LOG_ZERO
l1 = ctc._LOG_ONE
helpers.assert_theano_equal(
ctc._skip_allowed(numpy.array([[0, 1, 0, 2, 0], [0, 1, 0, 1, 0]])),
[[l0, l1, l0, l1, l0], [l0, l0, l0, l1, l0]])
def test_simple(self):
P = -1 # padding
a = numpy.array([[1,2,3]], dtype=theano.config.floatX).T # BAT
b = numpy.array([[4,2,3]], dtype=theano.config.floatX).T # CAT
helpers.assert_theano_equal(
levenshtein.distances(a,b,P),
[1],
)
def test_simple(self):
P = -1 # padding
a = numpy.array([[1, 2, 3]], dtype=theano.config.floatX).T # BAT
b = numpy.array([[4, 2, 3]], dtype=theano.config.floatX).T # CAT
helpers.assert_theano_equal(
levenshtein.distances(a, b, P),
[1],
)
def test_transform_targets(self):
targets = numpy.array([[1, 1], [2, 2], [3, 2], [4, 1]],
dtype=theano.config.floatX)
helpers.assert_theano_equal(
ctc.transform_targets(targets),
[[0., 0.], [1., 1.], [0., 0.], [2., 2.], [0., 0.], [3., 2.],
[0., 0.], [4., 1.], [0., 0.]])
def test_best_path_decode(self):
activations = numpy.array(
[[[0, 1, 0], [0, 1, 0]], [[0, 1, 0], [0, 0, 1]],
[[1, 0, 0], [1, 0, 0]], [[0, 0, 1], [0, 0, 1]],
[[1, 0, 0], [0, 1, 0]], [[1, 0, 0], [1, 0, 0]]],
dtype=theano.config.floatX)
P = ctc.PADDING
helpers.assert_theano_equal(ctc._best_path_decode(activations),
[[1, 1], [2, 2], [P, 2], [P, 1]])
def test_skip_allowed(self):
l0 = ctc._LOG_ZERO
l1 = ctc._LOG_ONE
helpers.assert_theano_equal(
ctc._skip_allowed(
numpy.array([[0, 1, 0, 2, 0],
[0, 1, 0, 1, 0]])
),
[[l0, l1, l0, l1, l0],
[l0, l0, l0, l1, l0]]
)
def test_accuracy(self):
P = ctc.PADDING
activations = numpy.array(
[[[0, 1, 0], [0, 1, 0]], [[0, 1, 0], [0, 0, 1]],
[[1, 0, 0], [1, 0, 0]], [[0, 0, 1], [0, 0, 1]],
[[1, 0, 0], [1, 0, 0]], [[1, 0, 0], [1, 0, 0]]],
dtype=theano.config.floatX)
targets = numpy.array([[1, 1], [2, 2], [P, 2], [P, 1]],
dtype=theano.config.floatX)
target_lengths = numpy.array([2, 4], dtype=theano.config.floatX)
helpers.assert_theano_equal(
ctc.accuracy(activations, targets, target_lengths), 0.875)
def test_best_path_decode(self):
activations = numpy.array([
[[0,1,0], [0,1,0]],
[[0,1,0], [0,0,1]],
[[1,0,0], [1,0,0]],
[[0,0,1], [0,0,1]],
[[1,0,0], [0,1,0]],
[[1,0,0], [1,0,0]]
], dtype=theano.config.floatX)
P = ctc.PADDING
helpers.assert_theano_equal(
ctc._best_path_decode(activations),
[[1, 1],
[2, 2],
[P, 2],
[P, 1]]
)
def test_transform_targets(self):
targets = numpy.array([
[1,1],
[2,2],
[3,2],
[4,1]
], dtype=theano.config.floatX)
helpers.assert_theano_equal(
ctc.transform_targets(targets),
[[ 0., 0.],
[ 1., 1.],
[ 0., 0.],
[ 2., 2.],
[ 0., 0.],
[ 3., 2.],
[ 0., 0.],
[ 4., 1.],
[ 0., 0.]]
)
def test_right_shift_rows(self):
shift = m_helpers.right_shift_rows(numpy.array([[1, 2, 3], [4, 5, 6]]),
1, -1)
correct = [[-1, 1, 2], [-1, 4, 5]]
helpers.assert_theano_equal(shift, correct)
def test_initial_probabilities(self):
helpers.assert_theano_equal(ctc._initial_probabilities(2, 3),
[[0, ctc._LOG_ZERO, ctc._LOG_ZERO],
[0, ctc._LOG_ZERO, ctc._LOG_ZERO]])
def test_initial_probabilities(self):
helpers.assert_theano_equal(
ctc._initial_probabilities(2, 3),
[[0, ctc._LOG_ZERO, ctc._LOG_ZERO],
[0, ctc._LOG_ZERO, ctc._LOG_ZERO]]
)
