def make_placeholder(input_size, sequence_length, batch_size, extra_axes=0):
input_axis = ng.make_axis(name='features')
recurrent_axis = ng.make_axis(name='REC_REP')
batch_axis = ng.make_axis(name='N')
input_axes = ng.make_axes([input_axis, recurrent_axis, batch_axis])
input_axes.set_shape((input_size, sequence_length, batch_size))
input_axes = ng.make_axes([ng.make_axis(length=1, name='features_' + str(i))
for i in range(extra_axes)]) + input_axes
input_placeholder = ng.placeholder(input_axes)
rng = RandomTensorGenerator()
input_value = rng.uniform(-0.01, 0.01, input_axes)
return input_placeholder, input_value
def make_placeholder(input_size, sequence_length, batch_size, extra_axes=0):
input_axis = ng.make_axis(name='features')
recurrent_axis = ng.make_axis(name='REC_REP')
batch_axis = ng.make_axis(name='N')
input_axes = ng.make_axes([input_axis, recurrent_axis, batch_axis])
input_axes.set_shape((input_size, sequence_length, batch_size))
input_axes = ng.make_axes([
ng.make_axis(length=1, name='features_' + str(i))
for i in range(extra_axes)
]) + input_axes
input_placeholder = ng.placeholder(input_axes)
rng = RandomTensorGenerator()
input_value = rng.uniform(-0.01, 0.01, input_axes)
return input_placeholder, input_value
# See the License for the specific language governing permissions and
# limitations under the License.
# ----------------------------------------------------------------------------
'''
Test of the batchnorm layer
'''
import pytest
import numpy as np
import ngraph as ng
from ngraph.frontends.neon import BatchNorm, Recurrent, LSTM, Tanh
from ngraph.testing.random import RandomTensorGenerator
from ngraph.testing.execution import ExecutorFactory
pytestmark = pytest.mark.transformer_dependent
rng = RandomTensorGenerator()
rtol = 0
atol = 1e-6
recurrent_atol = 1e-5
class BatchNormReference(object):
def __init__(self,
x,
init_gamma=1.0,
init_beta=0.0,
gmean=0.0,
gvar=1.0,
rho=0.9,
eps=1e-3,
axis=(1, )):