def get_projection_map(out_dim, in_dim):
if in_dim > out_dim:
raise ValueError("Can only project from lower to higher dimensionality")
projection_map_values = np.zeros(in_dim * out_dim, dtype=np.float32)
for i in range(0, in_dim):
projection_map_values[(i * out_dim) + i] = 1.0
shape = (in_dim, 1, 1, out_dim)
return constant(value=projection_map_values.reshape(shape))
def get_projection_map(out_dim, in_dim):
if in_dim > out_dim:
raise ValueError(
"Can only project from lower to higher dimensionality")
projection_map_values = np.zeros(in_dim * out_dim, dtype=np.float32)
for i in range(0, in_dim):
projection_map_values[(i * out_dim) + i] = 1.0
shape = (in_dim, 1, 1, out_dim)
return constant(value=projection_map_values.reshape(shape))
def create_model(self):
mean_removed_features = minus(self.input,
constant(114),
name='mean_removed_input')
with default_options(activation=None, pad=True, bias=True):
self.model = Sequential([
Convolution2D((11, 11),
96,
init=normal(0.01),
pad=False,
name='conv1'),
Activation(activation=relu, name='relu1'),
self.__local_response_normalization(1.0,
2,
0.0001,
0.75,
name='norm1'),
MaxPooling((3, 3), (2, 2), name='pool1'),
Convolution2D((5, 5),
192,
init=normal(0.01),
init_bias=0.1,
name='conv2'),
Activation(activation=relu, name='relu2'),
self.__local_response_normalization(1.0,
2,
0.0001,
0.75,
name='norm2'),
MaxPooling((3, 3), (2, 2), name='pool2'),
Convolution2D((3, 3), 384, init=normal(0.01), name='conv3'),
Activation(activation=relu, name='relu3'),
Convolution2D((3, 3),
384,
init=normal(0.01),
init_bias=0.1,
name='conv4'),
Activation(activation=relu, name='relu4'),
Convolution2D((3, 3),
256,
init=normal(0.01),
init_bias=0.1,
name='conv5'),
Activation(activation=relu, name='relu5'),
MaxPooling((3, 3), (2, 2), name='pool5'),
Dense(4096, init=normal(0.005), init_bias=0.1, name='fc6'),
Activation(activation=relu, name='relu6'),
Dropout(0.5, name='drop6'),
Dense(4096, init=normal(0.005), init_bias=0.1, name='fc7'),
Activation(activation=relu, name='relu7'),
Dropout(0.5, name='drop7'),
Dense(self.number_labels, init=normal(0.01), name='fc8')
])(mean_removed_features)
def eval(node, clean_up=True):
"""
It evaluates a node that has taken a numpy array as input. Note that sequences
are not supported yet by this method
Examples:
Plus with two matrices
>>> print (cntk.eval(cntk.ops.plus([[-30.,40.], [1.,2.]], [[-30.,40.], [1.,2.]])))
# [array([[[-60., 80.], [2., 4.]]])]
Times with broadcast of a scalar over a matrix
>>> print (cntk.eval(cntk.ops.element_times([[-30.,40.], [1.,2.]], 5)))
# [array([[[-150., 200.], [5., 10.]]])]
Args:
node (:class:`cntk.graph.ComputationNode`): the node to evaluate
clean_up (bool): whether the temporary directory should be removed when the context is left
Returns:
NumPy array containing the result
"""
from cntk.context import get_new_context
from cntk.ops import input_numpy, constant
from cntk.graph import ComputationNode, _InputComputationNodeBase
import numpy as np
# call a helper method to get a context
with get_new_context() as ctx:
ctx.clean_up = clean_up
first = True
# The params are passed as arryas, e.g. plus([1,2], [3,4]), and we need to
# wrap them with input and parameter nodes.
if node.params:
for p in node.params:
if p in node.inputs:
val = getattr(node, p)
if not isinstance(val, ComputationNode):
# One param needs to be an Input() node. This will be fixed in
# CNTK soon, so that we can remove this workaround and evaluate a
# network with no inputs.
if first:
ir = input_numpy([val], alias=p, name=p)
setattr(node, p, ir)
first = False
else:
setattr(node, p, constant(getattr(node, p),
name=p))
else:
if isinstance(val,
_InputComputationNodeBase) and first:
first = False
return ctx.eval(node)
def create_network(num_convolution_layers):
""" Create network
"""
# Input variables denoting the features and label data
input_var = cntk.input_variable(
(_NUM_CHANNELS, _IMAGE_HEIGHT, _IMAGE_WIDTH))
label_var = cntk.input_variable((_NUM_CLASSES))
# create model, and configure learning parameters
# Instantiate the feedforward classification model
input_removemean = minus(input_var, constant(128))
scaled_input = element_times(constant(0.00390625), input_removemean)
print('Creating NN model')
with layers.default_options(activation=relu, pad=True):
model = layers.Sequential([
layers.For(
range(num_convolution_layers), lambda: [
layers.Convolution2D((3, 3), 64),
layers.Convolution2D((3, 3), 64),
layers.MaxPooling((3, 3), (2, 2))
]),
layers.For(
range(2),
lambda i: [layers.Dense([256, 128][i]),
layers.Dropout(0.5)]),
layers.Dense(_NUM_CLASSES, activation=None)
])(scaled_input)
# loss and metric
ce = cross_entropy_with_softmax(model, label_var)
pe = classification_error(model, label_var)
return {
'name': 'convnet',
'feature': input_var,
'label': label_var,
'ce': ce,
'pe': pe,
'output': model
}
def test_simple_mnist():
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
epoch_size = sys.maxsize
minibatch_size = 32
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 3
num_minibatches_to_train = (num_samples_per_sweep * num_sweeps_to_train_with) / minibatch_size
lr = cntk_py.learning_rates_per_sample(0.003125)
input = variable((input_dim,), np.float32, needs_gradient=False, name="features")
scaled_input = element_times(constant((), 0.00390625), input)
label = variable((num_output_classes,), np.float32, needs_gradient=False, name="labels")
dev = cntk_py.DeviceDescriptor.cpudevice()
netout = fully_connected_classifier_net(scaled_input.output(), num_output_classes, hidden_layers_dim, num_hidden_layers, dev, sigmoid)
ce = cross_entropy_with_softmax(netout.output(), label)
pe = classification_error(netout.output(), label)
ffnet = combine([ce, pe, netout], "classifier_model")
cm = create_mb_source(input_dim, num_output_classes, epoch_size)
stream_infos = cm.stream_infos()
for si in stream_infos:
if si.m_name == 'features':
features_si = si
elif si.m_name == 'labels':
labels_si = si
minibatch_size_limits = dict()
minibatch_size_limits[features_si] = (0,minibatch_size)
minibatch_size_limits[labels_si] = (0,minibatch_size)
trainer = cntk_py.Trainer(ffnet, ce.output(), [cntk_py.sgdlearner(ffnet.parameters(), lr)])
for i in range(0,int(num_minibatches_to_train)):
mb=cm.get_next_minibatch(minibatch_size_limits, dev)
arguments = dict()
arguments[input] = mb[features_si].m_data
arguments[label] = mb[labels_si].m_data
trainer.train_minibatch(arguments, dev)
freq = 20
if i % freq == 0:
training_loss = get_train_loss(trainer)
print(str(i+freq) + ": " + str(training_loss))
#TODO: move the testing code into a separate test module ?
assert np.allclose(training_loss, 0.6142425537109375, atol=TOLERANCE_ABSOLUTE)
def eval(node):
"""
It evaluates a node that has taken a numpy array as input. Note that sequences
are not supported yet by this method
Examples:
Plus with two matrices
>>> print (cntk.eval(cntk.ops.plus([[-30.,40.], [1.,2.]], [[-30.,40.], [1.,2.]])))
# [array([[[-60., 80.], [2., 4.]]])]
Times with broadcast of a scalar over a matrix
>>> print (cntk.eval(cntk.ops.element_times([[-30.,40.], [1.,2.]], 5)))
# [array([[[-150., 200.], [5., 10.]]])]
Args:
node (:class:`cntk.graph.ComputationNode`): the node to evaluate
Returns:
NumPy array containing the result
"""
from cntk.context import get_context
from cntk.ops import input_numpy, constant
from cntk.graph import ComputationNode
# call a helper method to get a context
ctx = get_context()
first = True
# The params are passed as arryas, e.g. plus([1,2], [3,4]), and we need to
# wrap them with input and parameter nodes.
if node.params:
for p in node.params:
if p in node.inputs:
val = getattr(node, p)
if not isinstance(val, ComputationNode):
# One param needs to be an Input() node. This will being fixed in
# CNTK soon, so that we can remove this workaround and evaluate a
# network with no inputs.
if first:
if not isinstance(val, list):
# inputs have the outmost dimension for sequences
val = [val]
ir = input_numpy([val], alias=p, name=p)
setattr(node, p, ir)
first = False
else:
setattr(node, p, constant(getattr(node, p), name=p))
else:
if val.op_name == 'CNTK2.Input' and first:
first = False
return ctx.eval(node)
def eval(node):
"""
It evaluates a node that has taken a numpy array as input. Note that sequences
are not supported yet by this method
Examples:
Plus with two matrices
>>> print (cntk.eval(cntk.ops.plus([[-30.,40.], [1.,2.]], [[-30.,40.], [1.,2.]])))
# [array([[[-60., 80.], [2., 4.]]])]
Times with broadcast of a scalar over a matrix
>>> print (cntk.eval(cntk.ops.element_times([[-30.,40.], [1.,2.]], 5)))
# [array([[[-150., 200.], [5., 10.]]])]
Args:
node (:class:`cntk.graph.ComputationNode`): the node to evaluate
Returns:
NumPy array containing the result
"""
from cntk.context import get_context
from cntk.ops import input_numpy, constant
from cntk.graph import ComputationNode
# call a helper method to get a context
ctx = get_context()
first = True
# The params are passed as arryas, e.g. plus([1,2], [3,4]), and we need to
# wrap them with input and parameter nodes.
if node.params:
for p in node.params:
if p in node.inputs:
val = getattr(node, p)
if not isinstance(val, ComputationNode):
# One param needs to be an Input() node. This will being fixed in
# CNTK soon, so that we can remove this workaround and evaluate a
# network with no inputs.
if first:
if not isinstance(val, list):
# inputs have the outmost dimension for sequences
val = [val]
ir = input_numpy([val], alias=p, name=p)
setattr(node, p, ir)
first = False
else:
setattr(node, p, constant(getattr(node, p), name=p))
else:
if val.op_name == 'CNTK2.Input' and first:
first = False
return ctx.eval(node)
def simple_mnist():
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
input = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant((), 0.00390625), input)
netout = fully_connected_classifier_net(scaled_input, num_output_classes,
hidden_layers_dim,
num_hidden_layers, sigmoid)
ce = cross_entropy_with_softmax(netout, label)
pe = classification_error(netout, label)
rel_path = r"../../../../Examples/Image/MNIST/Data/Train-28x28_cntk_text.txt"
path = os.path.join(os.path.dirname(os.path.abspath(__file__)), rel_path)
feature_stream_name = 'features'
labels_stream_name = 'labels'
mb_source = text_format_minibatch_source(path, [
StreamConfiguration(feature_stream_name, input_dim),
StreamConfiguration(labels_stream_name, num_output_classes)
])
features_si = mb_source.stream_info(feature_stream_name)
labels_si = mb_source.stream_info(labels_stream_name)
# Instantiate the trainer object to drive the model training
lr = learning_rates_per_sample(0.003125)
trainer = Trainer(netout, ce, pe,
[sgd_learner(netout.owner.parameters(), lr)])
# Get minibatches of images to train with and perform model training
minibatch_size = 32
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 1
num_minibatches_to_train = (num_samples_per_sweep *
num_sweeps_to_train_with) / minibatch_size
training_progress_output_freq = 20
for i in range(0, int(num_minibatches_to_train)):
mb = mb_source.get_next_minibatch(minibatch_size)
# Specify the mapping of input variables in the model to actual minibatch data to be trained with
arguments = {
input: mb[features_si].m_data,
label: mb[labels_si].m_data
}
trainer.train_minibatch(arguments)
print_training_progress(trainer, i, training_progress_output_freq)
def eval(node, clean_up=True):
"""
It evaluates a node that has taken a numpy array as input. Note that sequences
are not supported yet by this method
Examples:
Plus with two matrices
>>> print (cntk.eval(cntk.ops.plus([[-30.,40.], [1.,2.]], [[-30.,40.], [1.,2.]])))
# [array([[[-60., 80.], [2., 4.]]])]
Times with broadcast of a scalar over a matrix
>>> print (cntk.eval(cntk.ops.element_times([[-30.,40.], [1.,2.]], 5)))
# [array([[[-150., 200.], [5., 10.]]])]
Args:
node (:class:`cntk.graph.ComputationNode`): the node to evaluate
clean_up (bool): whether the temporary directory should be removed when the context is left
Returns:
NumPy array containing the result
"""
from cntk.context import get_new_context
from cntk.ops import input_numpy, constant
from cntk.graph import ComputationNode, _InputComputationNodeBase
import numpy as np
# call a helper method to get a context
with get_new_context() as ctx:
ctx.clean_up = clean_up
first = True
# The params are passed as arryas, e.g. plus([1,2], [3,4]), and we need to
# wrap them with input and parameter nodes.
if node.params:
for p in node.params:
if p in node.inputs:
val = getattr(node, p)
if not isinstance(val, ComputationNode):
# One param needs to be an Input() node. This will be fixed in
# CNTK soon, so that we can remove this workaround and evaluate a
# network with no inputs.
if first:
ir = input_numpy([val], alias=p, name=p)
setattr(node, p, ir)
first = False
else:
setattr(node, p, constant(getattr(node, p), name=p))
else:
if isinstance(val, _InputComputationNodeBase) and first:
first = False
return ctx.eval(node)
def sanitize_input(arg, fallback_dtype=np.float32, reshape=None):
"""sanitize_input(arg, fallback_dtype=np.float32, reshape=None)
Convert to :class:`~cntk.variables.Variable` so that it can be passed
as Variable to the CNTK operators.
* If ``arg`` is a NumPy array and its type is neither `np.float32` nor
`np.float64`, it sets it to `np.float32`.
* If ``arg`` is an op, it is assumed that it has only one output, which
will be returned.
Args:
arg (number, NumPy array, :class:`~cntk.variables.Variable`, or :class:`~cntk.ops.functions.Function`): input
fallback_dtype (NumPy dtype): fallback dtype in case ``arg`` is a list
Returns:
Leaves Constant, Parameter, and Variable as is. Returns Constant, if
``arg`` is a number or NumPy array. Variable otherwise.
"""
from cntk.ops.functions import UserFunction
from cntk.variables import Constant, Variable, Parameter
from cntk.ops.functions import Function
from cntk.ops import constant
from ..core import asarray
# is it a Variable or a Function?
if isinstance(arg,
(Constant, cntk_py.Constant,
Variable, cntk_py.Variable,
Parameter, cntk_py.Parameter,
Function, cntk_py.Function)):
return arg
if isinstance(arg, Variable._Type):
raise ValueError("Input is a type object (" + str(arg) + "). Did you mean to pass 'input(" + str(arg) + ")'?")
# maybe a Python list that we can interpret as a NumPy array?
if isinstance(arg, list) and not arg:
raise ValueError('input is empty')
if not isinstance(arg, np.ndarray) or arg.dtype != fallback_dtype:
# TODO: check whether Values can be ingested directly
arg = asarray(arg, fallback_dtype)
if arg.shape == ():
arg.shape = (1,)
if reshape:
arg = np.reshape(arg, reshape)
return constant(value=arg)
def simple_mnist():
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
input = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant((), 0.00390625), input)
netout = fully_connected_classifier_net(scaled_input, num_output_classes, hidden_layers_dim, num_hidden_layers, sigmoid)
ce = cross_entropy_with_softmax(netout, label)
pe = classification_error(netout, label)
rel_path = os.path.join(*"../../../../Examples/Image/MNIST/Data/Train-28x28_cntk_text.txt".split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
if not os.path.exists(path):
readme_file = os.path.normpath(os.path.join(os.path.dirname(path), "..", "README.md"))
raise RuntimeError("File '%s' does not exist. Please follow the instructions at %s to download and prepare it."%(path, readme_file))
feature_stream_name = 'features'
labels_stream_name = 'labels'
mb_source = text_format_minibatch_source(path, [
StreamConfiguration( feature_stream_name, input_dim ),
StreamConfiguration( labels_stream_name, num_output_classes) ])
features_si = mb_source.stream_info(feature_stream_name)
labels_si = mb_source.stream_info(labels_stream_name)
# Instantiate the trainer object to drive the model training
lr = learning_rates_per_sample(0.003125)
trainer = Trainer(netout, ce, pe, [sgd_learner(netout.owner.parameters(), lr)])
# Get minibatches of images to train with and perform model training
minibatch_size = 32
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 1
num_minibatches_to_train = (num_samples_per_sweep * num_sweeps_to_train_with) / minibatch_size
training_progress_output_freq = 20
for i in range(0, int(num_minibatches_to_train)):
mb = mb_source.get_next_minibatch(minibatch_size)
# Specify the mapping of input variables in the model to actual minibatch data to be trained with
arguments = {input : mb[features_si].m_data, label : mb[labels_si].m_data}
trainer.train_minibatch(arguments)
print_training_progress(trainer, i, training_progress_output_freq)
def sanitize_input(arg, fallback_dtype=np.float32, reshape=None):
"""sanitize_input(arg, fallback_dtype=np.float32, reshape=None)
Convert to :class:`~cntk.variables.Variable` so that it can be passed
as Variable to the CNTK operators.
* If ``arg`` is a NumPy array and its type is not among (`np.float32`,
`np.float64`, `np.float16`), it sets it to `np.float32`.
* If ``arg`` is an op, it is assumed that it has only one output, which
will be returned.
Args:
arg (number, NumPy array, :class:`~cntk.variables.Variable`, or :class:`~cntk.ops.functions.Function`): input
fallback_dtype (NumPy dtype): fallback dtype in case ``arg`` is a list
Returns:
Leaves Constant, Parameter, and Variable as is. Returns Constant, if
``arg`` is a number or NumPy array. Variable otherwise.
"""
from cntk.ops.functions import UserFunction
from cntk.variables import Constant, Variable, Parameter
from cntk.ops.functions import Function
from cntk.ops import constant
from ..core import asarray
# is it a Variable or a Function?
if isinstance(arg,
(Constant, cntk_py.Constant,
Variable, cntk_py.Variable,
Parameter, cntk_py.Parameter,
Function, cntk_py.Function)):
return arg
if isinstance(arg, Variable._Type):
raise ValueError("Input is a type object (" + str(arg) + "). Did you mean to pass 'input_variable(**" + str(arg) + ")'?")
# maybe a Python list that we can interpret as a NumPy array?
if isinstance(arg, list) and not arg:
raise ValueError('input is empty')
if not isinstance(arg, np.ndarray) or arg.dtype != fallback_dtype:
# TODO: check whether Values can be ingested directly
arg = asarray(arg, fallback_dtype)
if reshape:
arg = np.reshape(arg, reshape)
return constant(value=arg)
def sanitize_input(arg, fallback_dtype=np.float32, reshape=None):
"""
Convert to :class:`~cntk.ops.variables.Variable` so that it can be passed as Variable to the
CNTK operators.
* If ``arg`` is a NumPy array and its type is neither `np.float32` nor `np.float64`, it sets it to `np.float32`.
* If ``arg`` is an op, it is assumed that it has only one output, which will be returned.
Args:
arg (number, NumPy array, :class:`~cntk.ops.variables.Variable`, or :class:`~cntk.ops.functions.Function`): input
fallback_dtype (NumPy dtype): fallback dtype in case ``arg`` is a list
Returns:
Leaves Constant, Parameter, and Variable as is. Returns Constant, if
``arg`` is a number or NumPy array. Variable otherwise.
"""
from cntk.ops.variables import Constant, Variable, Parameter
from cntk.ops import constant
# is it a Variable?
if isinstance(arg,
(Constant, cntk_py.Constant,
Variable, cntk_py.Variable,
Parameter, cntk_py.Parameter)):
return arg
# or a Function?
if isinstance(arg, cntk_py.Function):
try:
return arg.output
except RuntimeError:
raise ValueError(
'the argument has more than one output, please provide the one you want')
# maybe a Python list that we can interpret as a NumPy array?
if isinstance(arg, list) and not arg:
raise ValueError('input is empty')
if not isinstance(arg, np.ndarray) or arg.dtype!=fallback_dtype:
arg = np.asarray(arg, dtype=fallback_dtype)
if arg.shape == ():
arg.shape = (1,)
if reshape:
arg = np.reshape(arg, reshape)
return constant(value=arg)
def sanitize_input(arg, fallback_dtype=np.float32, reshape=None):
"""
Convert to :class:`~cntk.ops.variables.Variable` so that it can be passed as Variable to the
CNTK operators.
* If ``arg`` is a NumPy array and its type is neither `np.float32` nor `np.float64`, it sets it to `np.float32`.
* If ``arg`` is an op, it is assumed that it has only one output, which will be returned.
Args:
arg (number, NumPy array, :class:`~cntk.ops.variables.Variable`, or :class:`~cntk.ops.functions.Function`): input
fallback_dtype (NumPy dtype): fallback dtype in case ``arg`` is a list
Returns:
Leaves Constant, Parameter, and Variable as is. Returns Constant, if
``arg`` is a number or NumPy array. Variable otherwise.
"""
from cntk.ops.variables import Constant, Variable, Parameter
from cntk.ops import constant
# is it a Variable?
if isinstance(arg,
(Constant, cntk_py.Constant,
Variable, cntk_py.Variable,
Parameter, cntk_py.Parameter)):
return arg
# or a Function?
if isinstance(arg, cntk_py.Function):
try:
return arg.output
except RuntimeError:
raise ValueError(
'the argument has more than one output, please provide the one you want')
# maybe a Python list that we can interpret as a NumPy array?
if isinstance(arg, list) and not arg:
raise ValueError('input is empty')
if not isinstance(arg, np.ndarray) or arg.dtype!=fallback_dtype:
arg = np.asarray(arg, dtype=fallback_dtype)
if arg.shape == ():
arg.shape = (1,)
if reshape:
arg = np.reshape(arg, reshape)
return constant(value=arg)
def batch_norm(cntk_layer, inputs):
'''
Setup batch normalization op with given parameters
Args:
cntk_layer (:class:`~cntk.contrib.crosstalkcaffe.unimodel.cntkmodel.CntkLayersDefinition`):
the layer definition of batch normalization op
inputs (list): a list contains all :class:`~cntk.ops.functions.Function` or
:class:`~cntk.input`
Return:
:func:`~cntk.ops.functions.Function`: instaced cntk batch normalization op
'''
sanitize_input = internal.sanitize_input(inputs[0])
parameter_tensor = (sanitize_input.shape[0], )
scale_init = 1
bias_init = 0
mean_init = 1
var_init = 0
if cntk_layer.parameter_tensor:
if len(cntk_layer.parameter_tensor) < 3:
raise AssertionError('At least three tensors (saved_mean, saved_variance and scale) are needed')
mean_tensor = cntk_layer.parameter_tensor[0]
variance_tensor = cntk_layer.parameter_tensor[1]
global_scale = cntk_layer.parameter_tensor[2].data[0]
moving_average_factor = 1 / global_scale if global_scale != 0 else 0
mean_init = np.asarray(mean_tensor.data, dtype=np.float32) * moving_average_factor
var_init = np.asarray(variance_tensor.data, dtype=np.float32) * moving_average_factor
if len(cntk_layer.parameter_tensor) == 5:
scale_tensor = cntk_layer.parameter_tensor[3]
bias_tensor = cntk_layer.parameter_tensor[4]
scale_init = np.asarray(scale_tensor.data, dtype=np.float32)
bias_init = np.asarray(bias_tensor.data, dtype=np.float32)
scale_parameters = ops.parameter(parameter_tensor, init=scale_init, name='.'.join((cntk_layer.op_name, 'scale')))
bias_parameters = ops.parameter(parameter_tensor, init=bias_init, name='.'.join((cntk_layer.op_name, 'bias')))
mean_parameters = ops.parameter(parameter_tensor, init=mean_init, name='.'.join((cntk_layer.op_name, 'mean')))
var_parameters = ops.parameter(parameter_tensor, init=var_init, name='.'.join((cntk_layer.op_name, 'var')))
epsilon = cntk_layer.parameters.epsilon
return ops.batch_normalization(sanitize_input, scale_parameters, bias_parameters, mean_parameters,
var_parameters, True, use_cudnn_engine=False, epsilon=epsilon,
running_count=ops.constant(0),
name=cntk_layer.op_name)
def sanitize_input(arg, fallback_dtype=np.float32):
"""
Convert to Variable or Constant so that it can be passed as Variable to the CNTK
operators.
* If `arg` is a NumPy array and its type is neither `np.float32` nor
`np.float64`, it sets it to `np.float32`.
* If `arg` is an op, it is assumed that it has only one output, which will be returned.
Args:
arg (number, NumPy array, `Variable`, or `Function`): input
fallback_dtype (numpy dtype): fallback dtype in case `arg` is a list
Returns:
Constant, if `arg` was a number or NumPy array. Variable otherwise.
"""
from cntk.ops.variables import Constant, Variable
from cntk.ops import constant
# is it a Variable?
if isinstance(arg,
(Constant, Variable, cntk_py.Constant, cntk_py.Variable)):
return arg
# or a Function?
# FIXME soon to be replaced by Function
#if isinstance(arg, (Function, cntk_py.Function)):
if isinstance(arg, cntk_py.Function):
try:
return arg.output()
except RuntimeError:
raise ValueError(
'the argument has more than one output, please provide the one you want'
)
# maybe a Python list that we can interpret as a NumPy array?
if isinstance(arg, list) and not arg:
raise ValueError('input is empty')
if not isinstance(arg, np.ndarray):
arg = np.asarray(arg, dtype=fallback_dtype)
return constant(value=arg)
def sanitize_input(arg, fallback_dtype=np.float32):
"""
Convert to Variable or Constant so that it can be passed as Variable to the
CNTK operators.
* If `arg` is a NumPy array and its type is neither `np.float32` nor
`np.float64`, it sets it to `np.float32`.
* If `arg` is an op, it is assumed that it has only one output, which will
be returned.
Args:
arg (number, NumPy array, `Variable`, or `Function`): input
fallback_dtype (numpy dtype): fallback dtype in case `arg` is a list
Returns:
Constant, if `arg` was a number or NumPy array. Variable otherwise.
"""
from cntk.ops.variables import Constant, Variable
from cntk.ops import constant
# is it a Variable?
if isinstance(arg,
(Constant, Variable, cntk_py.Constant, cntk_py.Variable)):
return arg
# or a Function?
if isinstance(arg, cntk_py.Function):
try:
return arg.output()
except RuntimeError:
raise ValueError(
'the argument has more than one output, please provide the one you want')
# maybe a Python list that we can interpret as a NumPy array?
if isinstance(arg, list) and not arg:
raise ValueError('input is empty')
if not isinstance(arg, np.ndarray):
arg = np.asarray(arg, dtype=fallback_dtype)
return constant(value=arg)
def sanitize_input(arg, fallback_dtype=np.float32):
"""
Convert to Variable or Constant so that it can be passed as Variable to the CNTK
operators.
* If `arg` is a NumPy array and its type is neither `np.float32` nor
`np.float64`, it sets it to `np.float32`.
* If `arg` is an op, it is assumed that it has only one output, which will be returned.
Args:
arg (number, NumPy array, `Variable`, or `Function`): input
fallback_dtype (numpy dtype): fallback dtype in case `arg` is a list
Returns:
Constant, if `arg` was a number or NumPy array. Variable otherwise.
"""
from cntk.ops.variables import Constant, Variable
from cntk.ops import constant
if isinstance(arg,
(Constant, Variable, cntk_py.Constant, cntk_py.Variable)):
return arg
try:
var_output = arg.output()
if isinstance(var_output, (Variable, cntk_py.Variable)):
return var_output
else:
raise ValueError(
'Cannot convert argument of type "%s" to Variable' % type(arg))
except AttributeError:
# no function or function with more then one output
pass
if isinstance(arg, list) and not arg:
raise ValueError('input is empty')
if not isinstance(arg, np.ndarray):
arg = np.asarray(arg, dtype=fallback_dtype)
return constant(value=arg)
def sanitize_input(arg, fallback_dtype=np.float32):
"""
Convert to Variable or Constant so that it can be passed as Variable to the CNTK
operators.
* If `arg` is a NumPy array and its type is neither `np.float32` nor
`np.float64`, it sets it to `np.float32`.
* If `arg` is an op, it is assumed that it has only one output, which will be returned.
Args:
arg (number, NumPy array, `Variable`, or `Function`): input
fallback_dtype (numpy dtype): fallback dtype in case `arg` is a list
Returns:
Constant, if `arg` was a number or NumPy array. Variable otherwise.
"""
from cntk.ops.variables import Constant, Variable
from cntk.ops import constant
if isinstance(arg, (Constant, Variable, cntk_py.Constant, cntk_py.Variable)):
return arg
try:
var_output = arg.output()
if isinstance(var_output, (Variable, cntk_py.Variable)):
return var_output
else:
raise ValueError('Cannot convert argument of type "%s" to Variable'%type(arg))
except AttributeError:
# no function or function with more then one output
pass
if isinstance(arg, list) and not arg:
raise ValueError('input is empty')
if not isinstance(arg, np.ndarray):
arg = np.asarray(arg, dtype=fallback_dtype)
return constant(value=arg)
def simple_mnist():
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
features = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), features)
netout = fully_connected_classifier_net(
scaled_input, num_output_classes, hidden_layers_dim, num_hidden_layers, relu)
ce = cross_entropy_with_softmax(netout, label)
pe = classification_error(netout, label)
try:
rel_path = os.path.join(os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Train-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(*"../Image/DataSets/MNIST/Train-28x28_cntk_text.txt".split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
reader_train = create_reader(path, True, input_dim, num_output_classes)
input_map = {
features: reader_train.streams.features,
label: reader_train.streams.labels
}
# Instantiate progress writers.
logdir = os.path.join(os.path.dirname(os.path.abspath(__file__)), "mnist_log")
tensorboard_writer = TensorBoardProgressWriter(freq=1, log_dir=logdir, model=netout)
progress_printer = ProgressPrinter(freq=10, tag='Training')
# Instantiate the trainer object to drive the model training
lr_per_minibatch = learning_rate_schedule(0.2, UnitType.minibatch)
learner = sgd(netout.parameters, lr=lr_per_minibatch)
trainer = Trainer(netout, (ce, pe), learner, [tensorboard_writer, progress_printer])
# Get minibatches of images to train with and perform model training
minibatch_size = 64
num_samples_per_sweep = 6000
num_sweeps_to_train_with = 2
num_minibatches_to_train = (num_samples_per_sweep * num_sweeps_to_train_with) / minibatch_size
for minibatch_idx in range(0, int(num_minibatches_to_train)):
trainer.train_minibatch(reader_train.next_minibatch(minibatch_size, input_map=input_map))
# Log max/min/mean of each parameter tensor, so that we can confirm that the parameters change indeed.
# Don't want to do that very often though, otherwise will spend too much time computing min/max/mean.
if minibatch_idx % 10 == 9:
for p in netout.parameters:
tensorboard_writer.write_value(p.uid + "/max", reduce_max(p).eval(), minibatch_idx)
tensorboard_writer.write_value(p.uid + "/min", reduce_min(p).eval(), minibatch_idx)
tensorboard_writer.write_value(p.uid + "/mean", reduce_mean(p).eval(), minibatch_idx)
trainer.summarize_training_progress()
# Load test data
try:
rel_path = os.path.join(os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Test-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(*"../Image/DataSets/MNIST/Test-28x28_cntk_text.txt".split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
reader_test = create_reader(path, False, input_dim, num_output_classes)
input_map = {
features: reader_test.streams.features,
label: reader_test.streams.labels
}
# Test data for trained model
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size, input_map=input_map)
test_result += trainer.test_minibatch(mb)
# Average of evaluation errors of all test minibatches
trainer.summarize_test_progress()
return test_result / num_minibatches_to_test
def convnet_cifar10_dataaug(reader_train, reader_test, max_epochs = 80):
set_computation_network_trace_level(0)
# Input variables denoting the features and label data
input_var = input_variable((num_channels, image_height, image_width))
label_var = input_variable((num_classes))
# apply model to input
scaled_input = element_times(constant(0.00390625), input_var)
with default_options (activation=relu, pad=True):
z = Sequential([
LayerStack(2, lambda : [
Convolution((3,3), 64),
Convolution((3,3), 64),
MaxPooling((3,3), (2,2))
]),
LayerStack(2, lambda i: [
Dense([256,128][i]),
Dropout(0.5)
]),
Dense(num_classes, activation=None)
])(scaled_input)
# loss and metric
ce = cross_entropy_with_softmax(z, label_var)
pe = classification_error(z, label_var)
# training config
epoch_size = 50000 # for now we manually specify epoch size
minibatch_size = 64
# Set learning parameters
lr_per_sample = [0.0015625]*20+[0.00046875]*20+[0.00015625]*20+[0.000046875]*10+[0.000015625]
lr_schedule = learning_rate_schedule(lr_per_sample, unit=UnitType.sample, epoch_size=epoch_size)
mm_time_constant = [0]*20+[600]*20+[1200]
mm_schedule = momentum_as_time_constant_schedule(mm_time_constant, epoch_size=epoch_size)
l2_reg_weight = 0.002
# trainer object
learner = momentum_sgd(z.parameters, lr_schedule, mm_schedule,
l2_regularization_weight = l2_reg_weight)
trainer = Trainer(z, ce, pe, learner)
# define mapping from reader streams to network inputs
input_map = {
input_var: reader_train.streams.features,
label_var: reader_train.streams.labels
}
log_number_of_parameters(z) ; print()
progress_printer = ProgressPrinter(tag='Training')
# perform model training
for epoch in range(max_epochs): # loop over epochs
sample_count = 0
while sample_count < epoch_size: # loop over minibatches in the epoch
data = reader_train.next_minibatch(min(minibatch_size, epoch_size-sample_count), input_map=input_map) # fetch minibatch.
trainer.train_minibatch(data) # update model with it
sample_count += trainer.previous_minibatch_sample_count # count samples processed so far
progress_printer.update_with_trainer(trainer, with_metric=True) # log progress
progress_printer.epoch_summary(with_metric=True)
persist.save_model(z, os.path.join(model_path, "ConvNet_CIFAR10_DataAug_{}.dnn".format(epoch)))
### Evaluation action
epoch_size = 10000
minibatch_size = 16
# process minibatches and evaluate the model
metric_numer = 0
metric_denom = 0
sample_count = 0
minibatch_index = 0
while sample_count < epoch_size:
current_minibatch = min(minibatch_size, epoch_size - sample_count)
# Fetch next test min batch.
data = reader_test.next_minibatch(current_minibatch, input_map=input_map)
# minibatch data to be trained with
metric_numer += trainer.test_minibatch(data) * current_minibatch
metric_denom += current_minibatch
# Keep track of the number of samples processed so far.
sample_count += data[label_var].num_samples
minibatch_index += 1
print("")
print("Final Results: Minibatch[1-{}]: errs = {:0.2f}% * {}".format(minibatch_index+1, (metric_numer*100.0)/metric_denom, metric_denom))
print("")
return metric_numer/metric_denom
def simple_mnist(tensorboard_logdir=None):
input_dim = 784
num_output_classes = 10
num_hidden_layers = 2
hidden_layers_dim = 200
# Input variables denoting the features and label data
feature = C.input_variable(input_dim, np.float32)
label = C.input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), feature)
z = Sequential([
For(range(num_hidden_layers),
lambda i: Dense(hidden_layers_dim, activation=relu)),
Dense(num_output_classes)
])(scaled_input)
ce = cross_entropy_with_softmax(z, label)
pe = classification_error(z, label)
data_dir = os.path.dirname(os.path.abspath(__file__))
path = os.path.join(data_dir, 'Train-28x28_cntk_text.txt')
reader_train = create_reader(path, True, input_dim, num_output_classes)
input_map = {
feature: reader_train.streams.features,
label: reader_train.streams.labels
}
# Training config
minibatch_size = 64
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 10
# Instantiate progress writers.
# training_progress_output_freq = 100
progress_writers = [
ProgressPrinter(
# freq=training_progress_output_freq,
tag='Training',
num_epochs=num_sweeps_to_train_with)
]
if tensorboard_logdir is not None:
progress_writers.append(
TensorBoardProgressWriter(freq=10,
log_dir=tensorboard_logdir,
model=z))
# Instantiate the trainer object to drive the model training
lr = 0.001
trainer = Trainer(z, (ce, pe), sgd(z.parameters, lr), progress_writers)
training_session(trainer=trainer,
mb_source=reader_train,
mb_size=minibatch_size,
model_inputs_to_streams=input_map,
max_samples=num_samples_per_sweep *
num_sweeps_to_train_with,
progress_frequency=num_samples_per_sweep).train()
# Load test data
path = os.path.normpath(os.path.join(data_dir, "Test-28x28_cntk_text.txt"))
check_path(path)
reader_test = create_reader(path, False, input_dim, num_output_classes)
input_map = {
feature: reader_test.streams.features,
label: reader_test.streams.labels
}
# Test data for trained model
C.debugging.start_profiler()
C.debugging.enable_profiler()
C.debugging.set_node_timing(True)
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size,
input_map=input_map)
eval_error = trainer.test_minibatch(mb)
test_result = test_result + eval_error
C.debugging.stop_profiler()
trainer.print_node_timing()
# Average of evaluation errors of all test minibatches
return test_result * 100 / num_minibatches_to_test
def simple_mnist():
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
epoch_size = sys.maxsize
minibatch_size = 32
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 1
num_minibatches_to_train = (num_samples_per_sweep *
num_sweeps_to_train_with) / minibatch_size
lr = learning_rates_per_sample(0.003125)
input = variable(input_dim,
np.float32,
needs_gradient=False,
name="features")
scaled_input = element_times(constant((), 0.00390625), input)
label = variable(num_output_classes,
np.float32,
needs_gradient=False,
name="labels")
dev = -1
cntk_dev = cntk_device(dev)
netout = fully_connected_classifier_net(scaled_input, num_output_classes,
hidden_layers_dim,
num_hidden_layers, dev, sigmoid)
ce = cross_entropy_with_softmax(netout, label)
pe = classification_error(netout, label)
#TODO: add save and load module code
ffnet = combine([ce, pe, netout], "classifier_model")
rel_path = r"../../../../Examples/Image/MNIST/Data/Train-28x28_cntk_text.txt"
path = os.path.join(os.path.dirname(os.path.abspath(__file__)), rel_path)
cm = create_text_mb_source(path, input_dim, num_output_classes, epoch_size)
stream_infos = cm.stream_infos()
for si in stream_infos:
if si.m_name == 'features':
features_si = si
elif si.m_name == 'labels':
labels_si = si
trainer = Trainer(netout, ce, pe,
[sgdlearner(netout.owner.parameters(), lr)])
for i in range(0, int(num_minibatches_to_train)):
mb = cm.get_next_minibatch(minibatch_size, cntk_dev)
arguments = dict()
arguments[input] = mb[features_si].m_data
arguments[label] = mb[labels_si].m_data
trainer.train_minibatch(arguments, cntk_dev)
freq = 20
if i % freq == 0:
training_loss = get_train_loss(trainer)
eval_crit = get_train_eval_criterion(trainer)
print(
"Minibatch: {}, Train Loss: {}, Train Evaluation Criterion: {}"
.format(i, training_loss, eval_crit))
def Constant(init, shape=None, name=''):
p = constant(init, shape, name=name)
return _name_node(
p, 'constant') # these are factory methods for things with state
def convnet_cifar10(debug_output=False):
set_computation_network_trace_level(0)
image_height = 32
image_width = 32
num_channels = 3
input_dim = image_height * image_width * num_channels
num_output_classes = 10
# Input variables denoting the features and label data
input_var = input_variable((num_channels, image_height, image_width), np.float32)
label_var = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
input_removemean = minus(input_var, constant(128))
scaled_input = element_times(constant(0.00390625), input_removemean)
with default_options (activation=relu, pad=True):
z = Sequential([
LayerStack(2, lambda : [
Convolution((3,3), 64),
Convolution((3,3), 64),
MaxPooling((3,3), (2,2))
]),
LayerStack(2, lambda i: [
Dense([256,128][i]),
Dropout(0.5)
]),
Dense(num_output_classes, activation=None)
])(scaled_input)
ce = cross_entropy_with_softmax(z, label_var)
pe = classification_error(z, label_var)
reader_train = create_reader(os.path.join(data_path, 'Train_cntk_text.txt'), True, input_dim, num_output_classes)
# training config
epoch_size = 50000 # for now we manually specify epoch size
minibatch_size = 64
# Set learning parameters
lr_per_sample = [0.0015625]*10+[0.00046875]*10+[0.00015625]
lr_schedule = learning_rate_schedule(lr_per_sample, epoch_size=epoch_size)
momentum_time_constant = [0]*20+[-minibatch_size/np.log(0.9)]
mm_schedule = momentum_as_time_constant_schedule(momentum_time_constant, epoch_size=epoch_size)
l2_reg_weight = 0.002
# Instantiate the trainer object to drive the model training
learner = momentum_sgd(z.parameters, lr_schedule, mm_schedule, l2_regularization_weight = l2_reg_weight)
trainer = Trainer(z, ce, pe, learner)
# define mapping from reader streams to network inputs
input_map = {
input_var : reader_train.streams.features,
label_var : reader_train.streams.labels
}
log_number_of_parameters(z) ; print()
progress_printer = ProgressPrinter(tag='Training')
# Get minibatches of images to train with and perform model training
max_epochs = 30
for epoch in range(max_epochs): # loop over epochs
sample_count = 0
while sample_count < epoch_size: # loop over minibatches in the epoch
data = reader_train.next_minibatch(min(minibatch_size, epoch_size - sample_count), input_map=input_map) # fetch minibatch.
trainer.train_minibatch(data) # update model with it
sample_count += data[label_var].num_samples # count samples processed so far
progress_printer.update_with_trainer(trainer, with_metric=True) # log progress
progress_printer.epoch_summary(with_metric=True)
persist.save_model(z, os.path.join(model_path, "ConvNet_CIFAR10_{}.dnn".format(epoch)))
# Load test data
reader_test = create_reader(os.path.join(data_path, 'Test_cntk_text.txt'), False, input_dim, num_output_classes)
input_map = {
input_var : reader_test.streams.features,
label_var : reader_test.streams.labels
}
# Test data for trained model
epoch_size = 10000
minibatch_size = 16
# process minibatches and evaluate the model
metric_numer = 0
metric_denom = 0
sample_count = 0
minibatch_index = 0
while sample_count < epoch_size:
current_minibatch = min(minibatch_size, epoch_size - sample_count)
# Fetch next test min batch.
data = reader_test.next_minibatch(current_minibatch, input_map=input_map)
# minibatch data to be trained with
metric_numer += trainer.test_minibatch(data) * current_minibatch
metric_denom += current_minibatch
# Keep track of the number of samples processed so far.
sample_count += data[label_var].num_samples
minibatch_index += 1
print("")
print("Final Results: Minibatch[1-{}]: errs = {:0.2f}% * {}".format(minibatch_index+1, (metric_numer*100.0)/metric_denom, metric_denom))
print("")
return metric_numer/metric_denom
def simple_mnist(tensorboard_logdir=None):
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
feature = C.input_variable(input_dim, np.float32)
label = C.input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), feature)
z = Sequential([For(range(num_hidden_layers), lambda i: Dense(hidden_layers_dim, activation=relu)),
Dense(num_output_classes)])(scaled_input)
ce = cross_entropy_with_softmax(z, label)
pe = classification_error(z, label)
data_dir = os.path.join(abs_path, "..", "..", "..", "DataSets", "MNIST")
path = os.path.normpath(os.path.join(data_dir, "Train-28x28_cntk_text.txt"))
check_path(path)
reader_train = create_reader(path, True, input_dim, num_output_classes)
input_map = {
feature : reader_train.streams.features,
label : reader_train.streams.labels
}
# Training config
minibatch_size = 64
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 10
# Instantiate progress writers.
#training_progress_output_freq = 100
progress_writers = [ProgressPrinter(
#freq=training_progress_output_freq,
tag='Training',
num_epochs=num_sweeps_to_train_with)]
if tensorboard_logdir is not None:
progress_writers.append(TensorBoardProgressWriter(freq=10, log_dir=tensorboard_logdir, model=z))
# Instantiate the trainer object to drive the model training
lr = learning_parameter_schedule_per_sample(1)
trainer = Trainer(z, (ce, pe), adadelta(z.parameters, lr), progress_writers)
training_session(
trainer=trainer,
mb_source = reader_train,
mb_size = minibatch_size,
model_inputs_to_streams = input_map,
max_samples = num_samples_per_sweep * num_sweeps_to_train_with,
progress_frequency=num_samples_per_sweep
).train()
# Load test data
path = os.path.normpath(os.path.join(data_dir, "Test-28x28_cntk_text.txt"))
check_path(path)
reader_test = create_reader(path, False, input_dim, num_output_classes)
input_map = {
feature : reader_test.streams.features,
label : reader_test.streams.labels
}
# Test data for trained model
C.debugging.start_profiler()
C.debugging.enable_profiler()
C.debugging.set_node_timing(True)
#C.cntk_py.disable_cpueval_optimization() # uncomment this to check CPU eval perf without optimization
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size, input_map=input_map)
eval_error = trainer.test_minibatch(mb)
test_result = test_result + eval_error
C.debugging.stop_profiler()
trainer.print_node_timing()
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
def simple_mnist():
input_dim = 784
num_output_classes = 10
num_hidden_layers = 2
hidden_layers_dim = 200
# Input variables denoting the features and label data
feature = C.input_variable(input_dim)
label = C.input_variable(num_output_classes)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), feature)
# z = Sequential([
# Dense(hidden_layers_dim, activation=relu),
# Dense(hidden_layers_dim, activation=relu),
# Dense(num_output_classes)])(scaled_input)
with default_options(activation=relu, init=C.glorot_uniform()):
z = Sequential([For(range(num_hidden_layers),
lambda i: Dense(hidden_layers_dim)),
Dense(num_output_classes, activation=None)])(scaled_input)
ce = cross_entropy_with_softmax(z, label)
pe = classification_error(z, label)
# setup the data
path = abs_path + "\Train-28x28_cntk_text.txt"
reader_train = MinibatchSource(CTFDeserializer(path, StreamDefs(
features=StreamDef(field='features', shape=input_dim),
labels=StreamDef(field='labels', shape=num_output_classes))))
input_map = {
feature: reader_train.streams.features,
label: reader_train.streams.labels
}
# Training config
minibatch_size = 64
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 10
# Instantiate progress writers.
progress_writers = [ProgressPrinter(
tag='Training',
num_epochs=num_sweeps_to_train_with)]
# Instantiate the trainer object to drive the model training
lr = learning_rate_schedule(1, UnitType.sample)
trainer = Trainer(z, (ce, pe), [adadelta(z.parameters, lr)], progress_writers)
training_session(
trainer=trainer,
mb_source=reader_train,
mb_size=minibatch_size,
model_inputs_to_streams=input_map,
max_samples=num_samples_per_sweep * num_sweeps_to_train_with,
progress_frequency=num_samples_per_sweep
).train()
# Load test data
path = abs_path + "\Test-28x28_cntk_text.txt"
reader_test = MinibatchSource(CTFDeserializer(path, StreamDefs(
features=StreamDef(field='features', shape=input_dim),
labels=StreamDef(field='labels', shape=num_output_classes))))
input_map = {
feature: reader_test.streams.features,
label: reader_test.streams.labels
}
# Test data for trained model
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size, input_map=input_map)
eval_error = trainer.test_minibatch(mb)
test_result = test_result + eval_error
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
def convnet_mnist(debug_output=False):
image_height = 28
image_width = 28
num_channels = 1
input_dim = image_height * image_width * num_channels
num_output_classes = 10
# Input variables denoting the features and label data
input_var = input_variable((num_channels, image_height, image_width), np.float32)
label_var = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), input_var)
with default_options (activation=relu, pad=False):
conv1 = Convolution((5,5), 32, pad=True)(scaled_input)
pool1 = MaxPooling((3,3), (2,2))(conv1)
conv2 = Convolution((3,3), 48)(pool1)
pool2 = MaxPooling((3,3), (2,2))(conv2)
conv3 = Convolution((3,3), 64)(pool2)
f4 = Dense(96)(conv3)
drop4 = Dropout(0.5)(f4)
z = Dense(num_output_classes, activation=None)(drop4)
ce = cross_entropy_with_softmax(z, label_var)
pe = classification_error(z, label_var)
reader_train = create_reader(os.path.join(data_path, 'Train-28x28_cntk_text.txt'), True, input_dim, num_output_classes)
# training config
epoch_size = 60000 # for now we manually specify epoch size
minibatch_size = 128
# Set learning parameters
lr_per_sample = [0.001]*10+[0.0005]*10+[0.0001]
lr_schedule = learning_rate_schedule(lr_per_sample, epoch_size)
momentum_time_constant = [0]*5+[1024]
# Instantiate the trainer object to drive the model training
learner = momentum_sgd(z.parameters, lr_schedule, momentum_time_constant)
trainer = Trainer(z, ce, pe, learner)
# define mapping from reader streams to network inputs
input_map = {
input_var : reader_train.streams.features,
label_var : reader_train.streams.labels
}
log_number_of_parameters(z) ; print()
progress_printer = ProgressPrinter(tag='Training')
# Get minibatches of images to train with and perform model training
max_epochs = 40
for epoch in range(max_epochs): # loop over epochs
sample_count = 0
while sample_count < epoch_size: # loop over minibatches in the epoch
data = reader_train.next_minibatch(min(minibatch_size, epoch_size - sample_count), input_map=input_map) # fetch minibatch.
trainer.train_minibatch(data) # update model with it
sample_count += data[label_var].num_samples # count samples processed so far
progress_printer.update_with_trainer(trainer, with_metric=True) # log progress
progress_printer.epoch_summary(with_metric=True)
persist.save_model(z, os.path.join(model_path, "ConvNet_MNIST_{}.dnn".format(epoch)))
# Load test data
reader_test = create_reader(os.path.join(data_path, 'Test-28x28_cntk_text.txt'), False, input_dim, num_output_classes)
input_map = {
input_var : reader_test.streams.features,
label_var : reader_test.streams.labels
}
# Test data for trained model
epoch_size = 10000
minibatch_size = 1024
# process minibatches and evaluate the model
metric_numer = 0
metric_denom = 0
sample_count = 0
minibatch_index = 0
while sample_count < epoch_size:
current_minibatch = min(minibatch_size, epoch_size - sample_count)
# Fetch next test min batch.
data = reader_test.next_minibatch(current_minibatch, input_map=input_map)
# minibatch data to be trained with
metric_numer += trainer.test_minibatch(data) * current_minibatch
metric_denom += current_minibatch
# Keep track of the number of samples processed so far.
sample_count += data[label_var].num_samples
minibatch_index += 1
print("")
print("Final Results: Minibatch[1-{}]: errs = {:0.2f}% * {}".format(minibatch_index+1, (metric_numer*100.0)/metric_denom, metric_denom))
print("")
return metric_numer/metric_denom
def simple_mnist():
input_dim = 784
num_output_classes = 10
num_hidden_layers = 2
hidden_layers_dim = 200
# Input variables denoting the features and label data
feature = C.input_variable(input_dim)
label = C.input_variable(num_output_classes)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), feature)
# z = Sequential([
# Dense(hidden_layers_dim, activation=relu),
# Dense(hidden_layers_dim, activation=relu),
# Dense(num_output_classes)])(scaled_input)
with default_options(activation=relu, init=C.glorot_uniform()):
z = Sequential([
For(range(num_hidden_layers), lambda i: Dense(hidden_layers_dim)),
Dense(num_output_classes, activation=None)
])(scaled_input)
ce = cross_entropy_with_softmax(z, label)
pe = classification_error(z, label)
# setup the data
path = abs_path + "\Train-28x28_cntk_text.txt"
reader_train = MinibatchSource(
CTFDeserializer(
path,
StreamDefs(features=StreamDef(field='features', shape=input_dim),
labels=StreamDef(field='labels',
shape=num_output_classes))))
input_map = {
feature: reader_train.streams.features,
label: reader_train.streams.labels
}
# Training config
minibatch_size = 64
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 10
# Instantiate progress writers.
progress_writers = [
ProgressPrinter(tag='Training', num_epochs=num_sweeps_to_train_with)
]
# Instantiate the trainer object to drive the model training
lr = learning_rate_schedule(1, UnitType.sample)
trainer = Trainer(z, (ce, pe), [adadelta(z.parameters, lr)],
progress_writers)
training_session(trainer=trainer,
mb_source=reader_train,
mb_size=minibatch_size,
model_inputs_to_streams=input_map,
max_samples=num_samples_per_sweep *
num_sweeps_to_train_with,
progress_frequency=num_samples_per_sweep).train()
# Load test data
path = abs_path + "\Test-28x28_cntk_text.txt"
reader_test = MinibatchSource(
CTFDeserializer(
path,
StreamDefs(features=StreamDef(field='features', shape=input_dim),
labels=StreamDef(field='labels',
shape=num_output_classes))))
input_map = {
feature: reader_test.streams.features,
label: reader_test.streams.labels
}
# Test data for trained model
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size,
input_map=input_map)
eval_error = trainer.test_minibatch(mb)
test_result = test_result + eval_error
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
def simple_mnist(debug_output=False):
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
input = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), input)
z = fully_connected_classifier_net(
scaled_input, num_output_classes, hidden_layers_dim, num_hidden_layers, relu)
ce = cross_entropy_with_softmax(z, label)
pe = classification_error(z, label)
try:
rel_path = os.path.join(os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Train-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(abs_path, "..", "..", "..", "..", "..", "Examples", "Image", "DataSets", "MNIST", "Train-28x28_cntk_text.txt")
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
reader_train = create_reader(path, True, input_dim, num_output_classes)
input_map = {
input : reader_train.streams.features,
label : reader_train.streams.labels
}
lr_per_minibatch=learning_rate_schedule(0.2, UnitType.minibatch)
# Instantiate the trainer object to drive the model training
trainer = Trainer(z, ce, pe, sgd(z.parameters, lr=lr_per_minibatch))
# Get minibatches of images to train with and perform model training
minibatch_size = 64
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 10
num_minibatches_to_train = (num_samples_per_sweep * num_sweeps_to_train_with) / minibatch_size
training_progress_output_freq = 500
if debug_output:
training_progress_output_freq = training_progress_output_freq/4
for i in range(0, int(num_minibatches_to_train)):
mb = reader_train.next_minibatch(minibatch_size, input_map=input_map)
trainer.train_minibatch(mb)
print_training_progress(trainer, i, training_progress_output_freq)
# Load test data
try:
rel_path = os.path.join(os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Test-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(abs_path, "..", "..", "..", "..", "..", "Examples", "Image", "DataSets", "MNIST", "Test-28x28_cntk_text.txt")
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
reader_test = create_reader(path, False, input_dim, num_output_classes)
input_map = {
input : reader_test.streams.features,
label : reader_test.streams.labels
}
# Test data for trained model
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size, input_map=input_map)
eval_error = trainer.test_minibatch(mb)
test_result = test_result + eval_error
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
def create_vgg19():
# Input variables denoting the features and label data
feature_var = C.input_variable((num_channels, image_height, image_width))
label_var = C.input_variable((num_classes))
# apply model to input
# remove mean value
input = minus(feature_var,
constant([[[104]], [[117]], [[124]]]),
name='mean_removed_input')
with default_options(activation=None, pad=True, bias=True):
z = Sequential([
# we separate Convolution and ReLU to name the output for feature extraction (usually before ReLU)
For(
range(2), lambda i: [
Convolution2D((3, 3), 64, name='conv1_{}'.format(i)),
Activation(activation=relu, name='relu1_{}'.format(i)),
]),
MaxPooling((2, 2), (2, 2), name='pool1'),
For(
range(2), lambda i: [
Convolution2D((3, 3), 128, name='conv2_{}'.format(i)),
Activation(activation=relu, name='relu2_{}'.format(i)),
]),
MaxPooling((2, 2), (2, 2), name='pool2'),
For(
range(4), lambda i: [
Convolution2D((3, 3), 256, name='conv3_{}'.format(i)),
Activation(activation=relu, name='relu3_{}'.format(i)),
]),
MaxPooling((2, 2), (2, 2), name='pool3'),
For(
range(4), lambda i: [
Convolution2D((3, 3), 512, name='conv4_{}'.format(i)),
Activation(activation=relu, name='relu4_{}'.format(i)),
]),
MaxPooling((2, 2), (2, 2), name='pool4'),
For(
range(4), lambda i: [
Convolution2D((3, 3), 512, name='conv5_{}'.format(i)),
Activation(activation=relu, name='relu5_{}'.format(i)),
]),
MaxPooling((2, 2), (2, 2), name='pool5'),
Dense(4096, name='fc6'),
Activation(activation=relu, name='relu6'),
Dropout(0.5, name='drop6'),
Dense(4096, name='fc7'),
Activation(activation=relu, name='relu7'),
Dropout(0.5, name='drop7'),
Dense(num_classes, name='fc8')
])(input)
# loss and metric
ce = C.cross_entropy_with_softmax(z, label_var)
pe = C.classification_error(z, label_var)
pe5 = C.classification_error(z, label_var, topN=5)
log_number_of_parameters(z)
print()
return {
'feature': feature_var,
'label': label_var,
'ce': ce,
'pe': pe,
'pe5': pe5,
'output': z
}
def simple_mnist(debug_output=False):
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
input = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), input)
netout = fully_connected_classifier_net(scaled_input, num_output_classes,
hidden_layers_dim,
num_hidden_layers, relu)
ce = cross_entropy_with_softmax(netout, label)
pe = classification_error(netout, label)
try:
rel_path = os.path.join(
os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Train-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(
*
"../../../../Examples/Image/DataSets/MNIST/Train-28x28_cntk_text.txt"
.split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
reader_train = create_reader(path, True, input_dim, num_output_classes)
input_map = {
input: reader_train.streams.features,
label: reader_train.streams.labels
}
# Instantiate the trainer object to drive the model training
trainer = Trainer(netout, ce, pe, sgd(netout.parameters, lr=0.003125))
# Get minibatches of images to train with and perform model training
minibatch_size = 64
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 10
num_minibatches_to_train = (num_samples_per_sweep *
num_sweeps_to_train_with) / minibatch_size
training_progress_output_freq = 500
if debug_output:
training_progress_output_freq = training_progress_output_freq / 4
for i in range(0, int(num_minibatches_to_train)):
mb = reader_train.next_minibatch(minibatch_size, input_map=input_map)
trainer.train_minibatch(mb)
print_training_progress(trainer, i, training_progress_output_freq)
# Load test data
try:
rel_path = os.path.join(
os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Test-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(
*
"../../../../Examples/Image/DataSets/MNIST/Test-28x28_cntk_text.txt"
.split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
reader_test = create_reader(path, False, input_dim, num_output_classes)
input_map = {
input: reader_test.streams.features,
label: reader_test.streams.labels
}
# Test data for trained model
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size,
input_map=input_map)
eval_error = trainer.test_minibatch(mb)
test_result = test_result + eval_error
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
def simple_mnist(debug_output=False):
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
input = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant((), 0.00390625), input)
netout = fully_connected_classifier_net(scaled_input, num_output_classes,
hidden_layers_dim,
num_hidden_layers, sigmoid)
ce = cross_entropy_with_softmax(netout, label)
pe = classification_error(netout, label)
try:
rel_path = os.path.join(
os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Train-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(
*"../../../../Examples/Image/MNIST/Data/Train-28x28_cntk_text.txt".
split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
feature_stream_name = 'features'
labels_stream_name = 'labels'
mb_source = text_format_minibatch_source(path, [
StreamConfiguration(feature_stream_name, input_dim),
StreamConfiguration(labels_stream_name, num_output_classes)
])
features_si = mb_source.stream_info(feature_stream_name)
labels_si = mb_source.stream_info(labels_stream_name)
# Instantiate the trainer object to drive the model training
trainer = Trainer(netout, ce, pe, [sgd(netout.parameters(), lr=0.003125)])
# Get minibatches of images to train with and perform model training
minibatch_size = 32
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 1
num_minibatches_to_train = (num_samples_per_sweep *
num_sweeps_to_train_with) / minibatch_size
training_progress_output_freq = 20
for i in range(0, int(num_minibatches_to_train)):
mb = mb_source.get_next_minibatch(minibatch_size)
# Specify the mapping of input variables in the model to actual
# minibatch data to be trained with
arguments = {
input: mb[features_si].m_data,
label: mb[labels_si].m_data
}
trainer.train_minibatch(arguments)
if debug_output:
print_training_progress(trainer, i, training_progress_output_freq)
# Load test data
try:
rel_path = os.path.join(
os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Test-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(
*"../../../../Examples/Image/MNIST/Data/Test-28x28_cntk_text.txt".
split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
test_mb_source = text_format_minibatch_source(path, [
StreamConfiguration(feature_stream_name, input_dim),
StreamConfiguration(labels_stream_name, num_output_classes)
])
features_si = test_mb_source.stream_info(feature_stream_name)
labels_si = test_mb_source.stream_info(labels_stream_name)
# Test data for trained model
test_minibatch_size = 512
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = test_mb_source.get_next_minibatch(test_minibatch_size)
# Specify the mapping of input variables in the model to actual
# minibatch data to be tested with
arguments = {
input: mb[features_si].m_data,
label: mb[labels_si].m_data
}
eval_error = trainer.test_minibatch(arguments)
test_result = test_result + eval_error
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
def convnet_cifar10(debug_output=False):
set_computation_network_trace_level(0)
image_height = 32
image_width = 32
num_channels = 3
input_dim = image_height * image_width * num_channels
num_output_classes = 10
# Input variables denoting the features and label data
input_var = input_variable((num_channels, image_height, image_width),
np.float32)
label_var = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
input_removemean = minus(input_var, constant(128))
scaled_input = element_times(constant(0.00390625), input_removemean)
with default_options(activation=relu, pad=True):
z = Sequential([
LayerStack(
2, lambda: [
Convolution((3, 3), 64),
Convolution((3, 3), 64),
MaxPooling((3, 3), (2, 2))
]),
LayerStack(2, lambda i: [Dense([256, 128][i]),
Dropout(0.5)]),
Dense(num_output_classes, activation=None)
])(scaled_input)
ce = cross_entropy_with_softmax(z, label_var)
pe = classification_error(z, label_var)
reader_train = create_reader(
os.path.join(data_path, 'Train_cntk_text.txt'), True, input_dim,
num_output_classes)
# training config
epoch_size = 50000 # for now we manually specify epoch size
minibatch_size = 64
# Set learning parameters
lr_per_sample = [0.0015625] * 10 + [0.00046875] * 10 + [0.00015625]
lr_schedule = learning_rate_schedule(lr_per_sample,
epoch_size=epoch_size,
unit=UnitType.sample)
momentum_time_constant = [0] * 20 + [-minibatch_size / np.log(0.9)]
mm_schedule = momentum_as_time_constant_schedule(momentum_time_constant,
epoch_size=epoch_size)
l2_reg_weight = 0.002
# Instantiate the trainer object to drive the model training
learner = momentum_sgd(z.parameters,
lr_schedule,
mm_schedule,
l2_regularization_weight=l2_reg_weight)
trainer = Trainer(z, ce, pe, learner)
# define mapping from reader streams to network inputs
input_map = {
input_var: reader_train.streams.features,
label_var: reader_train.streams.labels
}
log_number_of_parameters(z)
print()
progress_printer = ProgressPrinter(tag='Training')
# Get minibatches of images to train with and perform model training
max_epochs = 30
for epoch in range(max_epochs): # loop over epochs
sample_count = 0
while sample_count < epoch_size: # loop over minibatches in the epoch
data = reader_train.next_minibatch(
min(minibatch_size, epoch_size - sample_count),
input_map=input_map) # fetch minibatch.
trainer.train_minibatch(data) # update model with it
sample_count += trainer.previous_minibatch_sample_count # count samples processed so far
progress_printer.update_with_trainer(
trainer, with_metric=True) # log progress
progress_printer.epoch_summary(with_metric=True)
persist.save_model(
z, os.path.join(model_path,
"ConvNet_CIFAR10_{}.dnn".format(epoch)))
# Load test data
reader_test = create_reader(os.path.join(data_path, 'Test_cntk_text.txt'),
False, input_dim, num_output_classes)
input_map = {
input_var: reader_test.streams.features,
label_var: reader_test.streams.labels
}
# Test data for trained model
epoch_size = 10000
minibatch_size = 16
# process minibatches and evaluate the model
metric_numer = 0
metric_denom = 0
sample_count = 0
minibatch_index = 0
while sample_count < epoch_size:
current_minibatch = min(minibatch_size, epoch_size - sample_count)
# Fetch next test min batch.
data = reader_test.next_minibatch(current_minibatch,
input_map=input_map)
# minibatch data to be trained with
metric_numer += trainer.test_minibatch(data) * current_minibatch
metric_denom += current_minibatch
# Keep track of the number of samples processed so far.
sample_count += data[label_var].num_samples
minibatch_index += 1
print("")
print("Final Results: Minibatch[1-{}]: errs = {:0.2f}% * {}".format(
minibatch_index + 1, (metric_numer * 100.0) / metric_denom,
metric_denom))
print("")
return metric_numer / metric_denom
from cntk.io import MinibatchSource, CTFDeserializer, StreamDef, StreamDefs from cntk.ops import relu, element_times, constant from cntk.layers import Dense from cntk.learners import learning_rate_schedule, UnitType, adadelta from cntk import cross_entropy_with_softmax, classification_error, Trainer from cntk.logging import ProgressPrinter #define network input_dim = 784 num_output_classes = 10 hidden_layers_dim = 200 feature = C.input_variable(input_dim) label = C.input_variable(num_output_classes) scaled_input = element_times(constant(0.00390625), feature) #define network topology h1 = Dense(hidden_layers_dim, activation=relu)(scaled_input) h2 = Dense(hidden_layers_dim, activation=relu)(h1) z = Dense(num_output_classes, activation=None)(h2) #define loss and error functions ce = cross_entropy_with_softmax(z, label) pe = classification_error(z, label) #Data source for training and testing path_train = "D:/MachineLearning/CNTK-2.0/Examples/Image/DataSets/MNIST/Train-28x28_cntk_text.txt" path_test = "D:/MachineLearning/CNTK-2.0/Examples/Image/DataSets/MNIST/Test-28x28_cntk_text.txt"
def test_op_times(left_operand, right_operand, device_id, precision,
left_matrix_type, right_matrix_type):
if right_matrix_type == 'sparse':
pytest.skip('second operator of times() has to be dense')
dt = PRECISION_TO_TYPE[precision]
# Forward pass test
#==================
# we compute the expected output for the forward pass
# we need two surrounding brackets
# the first for sequences (length=1, since we have dynamic_axis='')
# the second for batch of one sample
expected = [[np.dot(AA(left_operand, dtype=dt), AA(right_operand, dtype=dt))]]
if left_matrix_type == 'sparse':
a = SI(*batch_dense_to_sparse([left_operand]))
else:
a = I([left_operand])
b = I([right_operand])
from cntk.ops import times, constant
left_as_input = times(a, constant(right_operand))
right_as_input = times(constant(left_operand), b)
unittest_helper(left_as_input, None, expected, device_id=device_id,
precision=precision, clean_up=True, backward_pass=False)
unittest_helper(right_as_input, None, expected, device_id=device_id,
precision=precision, clean_up=True, backward_pass=False)
unittest_helper(times(a, b), None, expected, device_id=device_id,
precision=precision, clean_up=True, backward_pass=False)
# Backward pass test
#==================
def op_grad(A, B):
'''
Compute derivative of A with respect to B. For simplicity, assume A
and B to be matrices.
Let A be 2x2 and B be 2x1, then we have
[a11 a12] [b11] = [ a11 b11 + a12 b21 ]
[a21 a22] [b21] [ a21 b11 + a22 b21 ]
The derivative for A with respect to B is
[b11 b21]
[b11 b21]
The derivative for B with respect to A:
[a11 + a12]
[a21 + a22]
'''
assert len(A.shape) == len(B.shape) == 2
D = np.zeros_like(A)
D[:,:] = B.sum(axis=1)
return D
if 'sparse' not in [left_matrix_type, right_matrix_type]:
# FIXME: disabling until the Pass node supports sparse
expected_left = [[op_grad(AA(left_operand, dtype=dt), AA(right_operand, dtype=dt))]]
expected_right = [[op_grad(AA(right_operand, dtype=dt).T, AA(left_operand, dtype=dt).T).T]]
unittest_helper(left_as_input, None, expected_left, device_id=device_id,
precision=precision, clean_up=True, backward_pass=True, input_node=a)
# BUG: Fails because of Pass node?
unittest_helper(right_as_input, None, expected_right, device_id=device_id,
precision=precision, clean_up=True, backward_pass=True, input_node=b)
def simple_mnist(tensorboard_logdir=None):
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
feature = input(input_dim, np.float32)
label = input(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), feature)
z = fully_connected_classifier_net(scaled_input, num_output_classes,
hidden_layers_dim, num_hidden_layers,
relu)
ce = cross_entropy_with_softmax(z, label)
pe = classification_error(z, label)
data_dir = os.path.join(abs_path, "..", "..", "..", "DataSets", "MNIST")
path = os.path.normpath(os.path.join(data_dir,
"Train-28x28_cntk_text.txt"))
check_path(path)
reader_train = create_reader(path, True, input_dim, num_output_classes)
input_map = {
feature: reader_train.streams.features,
label: reader_train.streams.labels
}
# Training config
minibatch_size = 64
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 10
# Instantiate progress writers.
#training_progress_output_freq = 100
progress_writers = [
ProgressPrinter(
#freq=training_progress_output_freq,
tag='Training',
num_epochs=num_sweeps_to_train_with)
]
if tensorboard_logdir is not None:
progress_writers.append(
TensorBoardProgressWriter(freq=10,
log_dir=tensorboard_logdir,
model=z))
# Instantiate the trainer object to drive the model training
trainer = Trainer(z, (ce, pe), adadelta(z.parameters), progress_writers)
training_session(trainer=trainer,
mb_source=reader_train,
mb_size=minibatch_size,
var_to_stream=input_map,
max_samples=num_samples_per_sweep *
num_sweeps_to_train_with,
progress_frequency=num_samples_per_sweep).train()
# Load test data
path = os.path.normpath(os.path.join(data_dir, "Test-28x28_cntk_text.txt"))
check_path(path)
reader_test = create_reader(path, False, input_dim, num_output_classes)
input_map = {
feature: reader_test.streams.features,
label: reader_test.streams.labels
}
# Test data for trained model
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size,
input_map=input_map)
eval_error = trainer.test_minibatch(mb)
test_result = test_result + eval_error
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
def convnet_cifar10_dataaug(reader_train,
reader_test,
distributed_trainer,
max_epochs=80):
set_computation_network_trace_level(0)
# Input variables denoting the features and label data
input_var = input_variable((num_channels, image_height, image_width))
label_var = input_variable((num_classes))
# apply model to input
scaled_input = element_times(constant(0.00390625), input_var)
with default_options(activation=relu, pad=True):
z = Sequential([
LayerStack(
2, lambda: [
Convolution((3, 3), 64),
Convolution((3, 3), 64),
MaxPooling((3, 3), (2, 2))
]),
LayerStack(2, lambda i: [Dense([256, 128][i]),
Dropout(0.5)]),
Dense(num_classes, activation=None)
])(scaled_input)
# loss and metric
ce = cross_entropy_with_softmax(z, label_var)
pe = classification_error(z, label_var)
# training config
epoch_size = 50000 # for now we manually specify epoch size
minibatch_size = 64
# Set learning parameters
lr_per_sample = [0.0015625] * 20 + [0.00046875] * 20 + [
0.00015625
] * 20 + [0.000046875] * 10 + [0.000015625]
lr_schedule = learning_rate_schedule(lr_per_sample,
unit=UnitType.sample,
epoch_size=epoch_size)
momentum_time_constant = [0] * 20 + [600] * 20 + [1200]
mm_schedule = momentum_as_time_constant_schedule(momentum_time_constant,
epoch_size=epoch_size)
l2_reg_weight = 0.002
# trainer object
learner = momentum_sgd(z.parameters,
lr_schedule,
mm_schedule,
l2_regularization_weight=l2_reg_weight)
trainer = Trainer(z, ce, pe, learner, distributed_trainer)
# define mapping from reader streams to network inputs
input_map = {
input_var: reader_train.streams.features,
label_var: reader_train.streams.labels
}
log_number_of_parameters(z)
print()
progress_printer = ProgressPrinter(tag='Training')
# perform model training
for epoch in range(max_epochs): # loop over epochs
sample_count = 0
while sample_count < epoch_size: # loop over minibatches in the epoch
data = reader_train.next_minibatch(
min(minibatch_size, epoch_size - sample_count),
input_map=input_map) # fetch minibatch.
trainer.train_minibatch(data) # update model with it
sample_count += trainer.previous_minibatch_sample_count # count samples processed so far
progress_printer.update_with_trainer(
trainer, with_metric=True) # log progress
progress_printer.epoch_summary(with_metric=True)
if distributed_trainer.communicator().current_worker(
).global_rank == 0:
persist.save_model(
z,
os.path.join(model_path,
"ConvNet_CIFAR10_DataAug_{}.dnn".format(epoch)))
### Evaluation action
epoch_size = 10000
minibatch_size = 16
# process minibatches and evaluate the model
metric_numer = 0
metric_denom = 0
sample_count = 0
minibatch_index = 0
while sample_count < epoch_size:
current_minibatch = min(minibatch_size, epoch_size - sample_count)
# Fetch next test min batch.
data = reader_test.next_minibatch(current_minibatch,
input_map=input_map)
# minibatch data to be trained with
metric_numer += trainer.test_minibatch(data) * current_minibatch
metric_denom += current_minibatch
# Keep track of the number of samples processed so far.
sample_count += trainer.previous_minibatch_sample_count
minibatch_index += 1
print("")
print("Final Results: Minibatch[1-{}]: errs = {:0.2f}% * {}".format(
minibatch_index + 1, (metric_numer * 100.0) / metric_denom,
metric_denom))
print("")
return metric_numer / metric_denom
def simple_mnist(tensorboard_logdir=None):
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
input = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), input)
z = fully_connected_classifier_net(
scaled_input, num_output_classes, hidden_layers_dim, num_hidden_layers, relu)
ce = cross_entropy_with_softmax(z, label)
pe = classification_error(z, label)
data_dir = os.path.join(abs_path, "..", "..", "..", "DataSets", "MNIST")
path = os.path.normpath(os.path.join(data_dir, "Train-28x28_cntk_text.txt"))
check_path(path)
reader_train = create_reader(path, True, input_dim, num_output_classes)
input_map = {
input : reader_train.streams.features,
label : reader_train.streams.labels
}
# Training config
minibatch_size = 64
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 10
# Instantiate progress writers.
#training_progress_output_freq = 100
progress_writers = [ProgressPrinter(
#freq=training_progress_output_freq,
tag='Training',
num_epochs=num_sweeps_to_train_with)]
if tensorboard_logdir is not None:
progress_writers.append(TensorBoardProgressWriter(freq=10, log_dir=tensorboard_logdir, model=z))
# Instantiate the trainer object to drive the model training
lr_per_minibatch = learning_rate_schedule(0.2, UnitType.minibatch)
trainer = Trainer(z, (ce, pe), sgd(z.parameters, lr=lr_per_minibatch), progress_writers)
training_session(
trainer=trainer,
mb_source = reader_train,
mb_size = minibatch_size,
var_to_stream = input_map,
max_samples = num_samples_per_sweep * num_sweeps_to_train_with,
progress_frequency=num_samples_per_sweep
).train()
# Load test data
path = os.path.normpath(os.path.join(data_dir, "Test-28x28_cntk_text.txt"))
check_path(path)
reader_test = create_reader(path, False, input_dim, num_output_classes)
input_map = {
input : reader_test.streams.features,
label : reader_test.streams.labels
}
# Test data for trained model
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size, input_map=input_map)
eval_error = trainer.test_minibatch(mb)
test_result = test_result + eval_error
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
def simple_mnist(debug_output=False):
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
input = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant((), 0.00390625), input)
netout = fully_connected_classifier_net(
scaled_input, num_output_classes, hidden_layers_dim, num_hidden_layers, sigmoid
)
ce = cross_entropy_with_softmax(netout, label)
pe = classification_error(netout, label)
try:
rel_path = os.path.join(
os.environ["CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY"],
*"Image/MNIST/v0/Train-28x28_cntk_text.txt".split("/")
)
except KeyError:
rel_path = os.path.join(*"../../../../Examples/Image/Datasets/MNIST/Train-28x28_cntk_text.txt".split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
feature_stream_name = "features"
labels_stream_name = "labels"
mb_source = text_format_minibatch_source(
path,
[
StreamConfiguration(feature_stream_name, input_dim),
StreamConfiguration(labels_stream_name, num_output_classes),
],
)
features_si = mb_source[feature_stream_name]
labels_si = mb_source[labels_stream_name]
# Instantiate the trainer object to drive the model training
trainer = Trainer(netout, ce, pe, [sgd(netout.parameters(), lr=0.003125)])
# Get minibatches of images to train with and perform model training
minibatch_size = 32
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 1
num_minibatches_to_train = (num_samples_per_sweep * num_sweeps_to_train_with) / minibatch_size
training_progress_output_freq = 80
if debug_output:
training_progress_output_freq = training_progress_output_freq / 4
for i in range(0, int(num_minibatches_to_train)):
mb = mb_source.get_next_minibatch(minibatch_size)
# Specify the mapping of input variables in the model to actual
# minibatch data to be trained with
arguments = {input: mb[features_si], label: mb[labels_si]}
trainer.train_minibatch(arguments)
print_training_progress(trainer, i, training_progress_output_freq)
# Load test data
try:
rel_path = os.path.join(
os.environ["CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY"], *"Image/MNIST/v0/Test-28x28_cntk_text.txt".split("/")
)
except KeyError:
rel_path = os.path.join(*"../../../../Examples/Image/Datasets/MNIST/Test-28x28_cntk_text.txt".split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
test_mb_source = text_format_minibatch_source(
path,
[
StreamConfiguration(feature_stream_name, input_dim),
StreamConfiguration(labels_stream_name, num_output_classes),
],
randomize=False,
)
features_si = test_mb_source[feature_stream_name]
labels_si = test_mb_source[labels_stream_name]
# Test data for trained model
test_minibatch_size = 512
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = test_mb_source.get_next_minibatch(test_minibatch_size)
# Specify the mapping of input variables in the model to actual
# minibatch data to be tested with
arguments = {input: mb[features_si], label: mb[labels_si]}
eval_error = trainer.test_minibatch(arguments)
test_result = test_result + eval_error
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
def create_vgg19():
# Input variables denoting the features and label data
feature_var = C.input_variable((num_channels, image_height, image_width))
label_var = C.input_variable((num_classes))
# apply model to input
# remove mean value
input = minus(feature_var, constant([[[104]], [[117]], [[124]]]), name='mean_removed_input')
with default_options(activation=None, pad=True, bias=True):
z = Sequential([
# we separate Convolution and ReLU to name the output for feature extraction (usually before ReLU)
For(range(2), lambda i: [
Convolution2D((3,3), 64, name='conv1_{}'.format(i)),
Activation(activation=relu, name='relu1_{}'.format(i)),
]),
MaxPooling((2,2), (2,2), name='pool1'),
For(range(2), lambda i: [
Convolution2D((3,3), 128, name='conv2_{}'.format(i)),
Activation(activation=relu, name='relu2_{}'.format(i)),
]),
MaxPooling((2,2), (2,2), name='pool2'),
For(range(4), lambda i: [
Convolution2D((3,3), 256, name='conv3_{}'.format(i)),
Activation(activation=relu, name='relu3_{}'.format(i)),
]),
MaxPooling((2,2), (2,2), name='pool3'),
For(range(4), lambda i: [
Convolution2D((3,3), 512, name='conv4_{}'.format(i)),
Activation(activation=relu, name='relu4_{}'.format(i)),
]),
MaxPooling((2,2), (2,2), name='pool4'),
For(range(4), lambda i: [
Convolution2D((3,3), 512, name='conv5_{}'.format(i)),
Activation(activation=relu, name='relu5_{}'.format(i)),
]),
MaxPooling((2,2), (2,2), name='pool5'),
Dense(4096, name='fc6'),
Activation(activation=relu, name='relu6'),
Dropout(0.5, name='drop6'),
Dense(4096, name='fc7'),
Activation(activation=relu, name='relu7'),
Dropout(0.5, name='drop7'),
Dense(num_classes, name='fc8')
])(input)
# loss and metric
ce = C.cross_entropy_with_softmax(z, label_var)
pe = C.classification_error(z, label_var)
pe5 = C.classification_error(z, label_var, topN=5)
log_number_of_parameters(z) ; print()
return {
'feature': feature_var,
'label': label_var,
'ce' : ce,
'pe' : pe,
'pe5': pe5,
'output': z
}
def simple_mnist():
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
features = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), features)
netout = fully_connected_classifier_net(scaled_input, num_output_classes,
hidden_layers_dim,
num_hidden_layers, relu)
ce = cross_entropy_with_softmax(netout, label)
pe = classification_error(netout, label)
try:
rel_path = os.path.join(
os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Train-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(
*"../Image/DataSets/MNIST/Train-28x28_cntk_text.txt".split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
reader_train = create_reader(path, True, input_dim, num_output_classes)
input_map = {
features: reader_train.streams.features,
label: reader_train.streams.labels
}
# Instantiate progress writers.
logdir = os.path.join(os.path.dirname(os.path.abspath(__file__)),
"mnist_log")
tensorboard_writer = TensorBoardProgressWriter(freq=1,
log_dir=logdir,
model=netout)
progress_printer = ProgressPrinter(freq=10, tag='Training')
# Instantiate the trainer object to drive the model training
lr_per_minibatch = learning_rate_schedule(0.2, UnitType.minibatch)
learner = sgd(netout.parameters, lr=lr_per_minibatch)
trainer = Trainer(netout, (ce, pe), learner,
[tensorboard_writer, progress_printer])
# Get minibatches of images to train with and perform model training
minibatch_size = 64
num_samples_per_sweep = 6000
num_sweeps_to_train_with = 2
num_minibatches_to_train = (num_samples_per_sweep *
num_sweeps_to_train_with) / minibatch_size
for minibatch_idx in range(0, int(num_minibatches_to_train)):
trainer.train_minibatch(
reader_train.next_minibatch(minibatch_size, input_map=input_map))
# Log max/min/mean of each parameter tensor, so that we can confirm that the parameters change indeed.
# Don't want to do that very often though, otherwise will spend too much time computing min/max/mean.
if minibatch_idx % 10 == 9:
for p in netout.parameters:
tensorboard_writer.write_value(p.uid + "/max",
reduce_max(p).eval(),
minibatch_idx)
tensorboard_writer.write_value(p.uid + "/min",
reduce_min(p).eval(),
minibatch_idx)
tensorboard_writer.write_value(p.uid + "/mean",
reduce_mean(p).eval(),
minibatch_idx)
trainer.summarize_training_progress()
# Load test data
try:
rel_path = os.path.join(
os.environ['CNTK_EXTERNAL_TESTDATA_SOURCE_DIRECTORY'],
*"Image/MNIST/v0/Test-28x28_cntk_text.txt".split("/"))
except KeyError:
rel_path = os.path.join(
*"../Image/DataSets/MNIST/Test-28x28_cntk_text.txt".split("/"))
path = os.path.normpath(os.path.join(abs_path, rel_path))
check_path(path)
reader_test = create_reader(path, False, input_dim, num_output_classes)
input_map = {
features: reader_test.streams.features,
label: reader_test.streams.labels
}
# Test data for trained model
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size,
input_map=input_map)
test_result += trainer.test_minibatch(mb)
# Average of evaluation errors of all test minibatches
trainer.summarize_test_progress()
return test_result / num_minibatches_to_test
def batch_norm(cntk_layer, inputs):
'''
Setup batch normalization op with given parameters
Args:
cntk_layer (:class:`~cntk.contrib.crosstalkcaffe.unimodel.cntkmodel.CntkLayersDefinition`):
the layer definition of batch normalization op
inputs (list): a list contains all :class:`~cntk.ops.functions.Function` or
:class:`~cntk.input`
Return:
:func:`~cntk.ops.functions.Function`: instaced cntk batch normalization op
'''
sanitize_input = internal.sanitize_input(inputs[0])
parameter_tensor = (sanitize_input.shape[0], )
scale_init = 1
bias_init = 0
mean_init = 1
var_init = 0
if cntk_layer.parameter_tensor:
if len(cntk_layer.parameter_tensor) < 3:
raise AssertionError(
'At least three tensors (saved_mean, saved_variance and scale) are needed'
)
mean_tensor = cntk_layer.parameter_tensor[0]
variance_tensor = cntk_layer.parameter_tensor[1]
global_scale = cntk_layer.parameter_tensor[2].data[0]
moving_average_factor = 1 / global_scale if global_scale != 0 else 0
mean_init = np.asarray(mean_tensor.data,
dtype=np.float32) * moving_average_factor
var_init = np.asarray(variance_tensor.data,
dtype=np.float32) * moving_average_factor
if len(cntk_layer.parameter_tensor) == 5:
scale_tensor = cntk_layer.parameter_tensor[3]
bias_tensor = cntk_layer.parameter_tensor[4]
scale_init = np.asarray(scale_tensor.data, dtype=np.float32)
bias_init = np.asarray(bias_tensor.data, dtype=np.float32)
scale_parameters = ops.parameter(parameter_tensor,
init=scale_init,
name='.'.join(
(cntk_layer.op_name, 'scale')))
bias_parameters = ops.parameter(parameter_tensor,
init=bias_init,
name='.'.join(
(cntk_layer.op_name, 'bias')))
mean_parameters = ops.parameter(parameter_tensor,
init=mean_init,
name='.'.join(
(cntk_layer.op_name, 'mean')))
var_parameters = ops.parameter(parameter_tensor,
init=var_init,
name='.'.join(
(cntk_layer.op_name, 'var')))
epsilon = cntk_layer.parameters.epsilon
return ops.batch_normalization(sanitize_input,
scale_parameters,
bias_parameters,
mean_parameters,
var_parameters,
True,
use_cudnn_engine=False,
epsilon=epsilon,
running_count=ops.constant(0),
name=cntk_layer.op_name)
def convnet_mnist(debug_output=False):
image_height = 28
image_width = 28
num_channels = 1
input_dim = image_height * image_width * num_channels
num_output_classes = 10
# Input variables denoting the features and label data
input_var = input_variable((num_channels, image_height, image_width),
np.float32)
label_var = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), input_var)
with default_options(activation=relu, pad=False):
conv1 = Convolution((5, 5), 32, pad=True)(scaled_input)
pool1 = MaxPooling((3, 3), (2, 2))(conv1)
conv2 = Convolution((3, 3), 48)(pool1)
pool2 = MaxPooling((3, 3), (2, 2))(conv2)
conv3 = Convolution((3, 3), 64)(pool2)
f4 = Dense(96)(conv3)
drop4 = Dropout(0.5)(f4)
z = Dense(num_output_classes, activation=None)(drop4)
ce = cross_entropy_with_softmax(z, label_var)
pe = classification_error(z, label_var)
reader_train = create_reader(
os.path.join(data_path, 'Train-28x28_cntk_text.txt'), True, input_dim,
num_output_classes)
# training config
epoch_size = 60000 # for now we manually specify epoch size
minibatch_size = 128
# Set learning parameters
lr_per_sample = [0.001] * 10 + [0.0005] * 10 + [0.0001]
lr_schedule = learning_rate_schedule(lr_per_sample, UnitType.sample,
epoch_size)
mm_time_constant = [0] * 5 + [1024]
mm_schedule = momentum_as_time_constant_schedule(mm_time_constant,
epoch_size)
# Instantiate the trainer object to drive the model training
learner = momentum_sgd(z.parameters, lr_schedule, mm_schedule)
trainer = Trainer(z, ce, pe, learner)
# define mapping from reader streams to network inputs
input_map = {
input_var: reader_train.streams.features,
label_var: reader_train.streams.labels
}
log_number_of_parameters(z)
print()
progress_printer = ProgressPrinter(tag='Training')
# Get minibatches of images to train with and perform model training
max_epochs = 40
for epoch in range(max_epochs): # loop over epochs
sample_count = 0
while sample_count < epoch_size: # loop over minibatches in the epoch
data = reader_train.next_minibatch(
min(minibatch_size, epoch_size - sample_count),
input_map=input_map) # fetch minibatch.
trainer.train_minibatch(data) # update model with it
sample_count += data[
label_var].num_samples # count samples processed so far
progress_printer.update_with_trainer(
trainer, with_metric=True) # log progress
progress_printer.epoch_summary(with_metric=True)
z.save_model(
os.path.join(model_path, "ConvNet_MNIST_{}.dnn".format(epoch)))
# Load test data
reader_test = create_reader(
os.path.join(data_path, 'Test-28x28_cntk_text.txt'), False, input_dim,
num_output_classes)
input_map = {
input_var: reader_test.streams.features,
label_var: reader_test.streams.labels
}
# Test data for trained model
epoch_size = 10000
minibatch_size = 1024
# process minibatches and evaluate the model
metric_numer = 0
metric_denom = 0
sample_count = 0
minibatch_index = 0
while sample_count < epoch_size:
current_minibatch = min(minibatch_size, epoch_size - sample_count)
# Fetch next test min batch.
data = reader_test.next_minibatch(current_minibatch,
input_map=input_map)
# minibatch data to be trained with
metric_numer += trainer.test_minibatch(data) * current_minibatch
metric_denom += current_minibatch
# Keep track of the number of samples processed so far.
sample_count += trainer.previous_minibatch_sample_count
minibatch_index += 1
print("")
print("Final Results: Minibatch[1-{}]: errs = {:0.2f}% * {}".format(
minibatch_index + 1, (metric_numer * 100.0) / metric_denom,
metric_denom))
print("")
return metric_numer / metric_denom
def simple_mnist(debug_output=False):
input_dim = 784
num_output_classes = 10
num_hidden_layers = 1
hidden_layers_dim = 200
# Input variables denoting the features and label data
input = input_variable(input_dim, np.float32)
label = input_variable(num_output_classes, np.float32)
# Instantiate the feedforward classification model
scaled_input = element_times(constant(0.00390625), input)
z = fully_connected_classifier_net(
scaled_input, num_output_classes, hidden_layers_dim, num_hidden_layers, relu)
ce = cross_entropy_with_softmax(z, label)
pe = classification_error(z, label)
data_dir = os.path.join(abs_path, "..", "..", "..", "DataSets", "MNIST")
path = os.path.normpath(os.path.join(data_dir, "Train-28x28_cntk_text.txt"))
check_path(path)
reader_train = create_reader(path, True, input_dim, num_output_classes)
input_map = {
input : reader_train.streams.features,
label : reader_train.streams.labels
}
lr_per_minibatch=learning_rate_schedule(0.2, UnitType.minibatch)
# Instantiate the trainer object to drive the model training
trainer = Trainer(z, ce, pe, sgd(z.parameters, lr=lr_per_minibatch))
# Get minibatches of images to train with and perform model training
minibatch_size = 64
num_samples_per_sweep = 60000
num_sweeps_to_train_with = 10
num_minibatches_to_train = (num_samples_per_sweep * num_sweeps_to_train_with) / minibatch_size
training_progress_output_freq = 500
if debug_output:
training_progress_output_freq = training_progress_output_freq/4
for i in range(0, int(num_minibatches_to_train)):
mb = reader_train.next_minibatch(minibatch_size, input_map=input_map)
trainer.train_minibatch(mb)
print_training_progress(trainer, i, training_progress_output_freq)
# Load test data
path = os.path.normpath(os.path.join(data_dir, "Test-28x28_cntk_text.txt"))
check_path(path)
reader_test = create_reader(path, False, input_dim, num_output_classes)
input_map = {
input : reader_test.streams.features,
label : reader_test.streams.labels
}
# Test data for trained model
test_minibatch_size = 1024
num_samples = 10000
num_minibatches_to_test = num_samples / test_minibatch_size
test_result = 0.0
for i in range(0, int(num_minibatches_to_test)):
mb = reader_test.next_minibatch(test_minibatch_size, input_map=input_map)
eval_error = trainer.test_minibatch(mb)
test_result = test_result + eval_error
# Average of evaluation errors of all test minibatches
return test_result / num_minibatches_to_test
