def create_model(input_sequence, label_sequence, vocab_dim, hidden_dim):
# Create the rnn that computes the latent representation for the next token.
rnn_with_latent_output = Sequential([
C.Embedding(hidden_dim),
For(
range(num_layers), lambda: Sequential([
Stabilizer(),
Recurrence(LSTM(hidden_dim), go_backwards=False)
])),
])
# Apply it to the input sequence.
latent_vector = rnn_with_latent_output(input_sequence)
# Connect the latent output to (sampled/full) softmax.
if use_sampled_softmax:
weights = load_sampling_weights(token_frequencies_file_path)
smoothed_weights = np.float32(np.power(weights, alpha))
sampling_weights = C.reshape(C.Constant(smoothed_weights),
shape=(1, vocab_dim))
z, ce, errs = cross_entropy_with_sampled_softmax(
latent_vector, label_sequence, vocab_dim, hidden_dim,
softmax_sample_size, sampling_weights)
else:
z, ce, errs = cross_entropy_with_full_softmax(latent_vector,
label_sequence,
vocab_dim, hidden_dim)
return z, ce, errs
def create_model(output_dim):
return Sequential([
LayerStack(
num_layers, lambda: Sequential([
Stabilizer(),
Recurrence(LSTM(hidden_dim), go_backwards=False)
])),
Dense(output_dim)
])
def test_htk_deserializers():
mbsize = 640
epoch_size = 1000 * mbsize
lr = [0.001]
feature_dim = 33
num_classes = 132
context = 2
os.chdir(data_path)
features_file = "glob_0000.scp"
labels_file = "glob_0000.mlf"
label_mapping_file = "state.list"
fd = HTKFeatureDeserializer(
StreamDefs(amazing_features=StreamDef(
shape=feature_dim, context=(context, context), scp=features_file)))
ld = HTKMLFDeserializer(
label_mapping_file,
StreamDefs(
awesome_labels=StreamDef(shape=num_classes, mlf=labels_file)))
reader = MinibatchSource([fd, ld])
features = C.input_variable(((2 * context + 1) * feature_dim))
labels = C.input_variable((num_classes))
model = Sequential(
[For(range(3), lambda: Recurrence(LSTM(256))),
Dense(num_classes)])
z = model(features)
ce = C.cross_entropy_with_softmax(z, labels)
errs = C.classification_error(z, labels)
learner = C.adam_sgd(z.parameters,
lr=C.learning_rate_schedule(lr, C.UnitType.sample,
epoch_size),
momentum=C.momentum_as_time_constant_schedule(1000),
low_memory=True,
gradient_clipping_threshold_per_sample=15,
gradient_clipping_with_truncation=True)
trainer = C.Trainer(z, (ce, errs), learner)
input_map = {
features: reader.streams.amazing_features,
labels: reader.streams.awesome_labels
}
pp = C.ProgressPrinter(freq=0)
# just run and verify it doesn't crash
for i in range(3):
mb_data = reader.next_minibatch(mbsize, input_map=input_map)
trainer.train_minibatch(mb_data)
pp.update_with_trainer(trainer, with_metric=True)
assert True
os.chdir(abs_path)
def create_pooling_neural_network(input_vars, out_dims):
hidden_layer_1 = Dense(2, activation=cntk.ops.relu)
hidden_layer_2 = Dense(16, activation=cntk.ops.relu)
output_layer = Dense(out_dims, activation=None)
model = Sequential([hidden_layer_1, hidden_layer_2,
output_layer])(input_vars)
return model
def _build_model(self):
with default_options(init=he_uniform(), activation=relu, bias=True):
model = Sequential([
Convolution((8, 8), 32, strides=(4, 4)),
Convolution((4, 4), 64, strides=(2, 2)),
Convolution((3, 3), 64, strides=(1, 1)),
Dense(512, init=he_uniform(0.01)),
Dense(self._nb_actions, activation=None, init=he_uniform(0.01))
])
return model
def create_multi_layer_neural_network(input_vars, out_dims, num_hidden_layers):
input_dims = input_vars.shape[0]
num_hidden_neurons = input_dims**3
hidden_layer = lambda: Dense(num_hidden_neurons, activation=cntk.ops.relu)
output_layer = Dense(out_dims, activation=None)
model = Sequential([LayerStack(num_hidden_layers, hidden_layer),
output_layer])(input_vars)
return model
def create_convolutional_neural_network(input_vars, out_dims):
convolutional_layer_1 = Convolution((5, 5),
32,
strides=1,
activation=cntk.ops.relu,
pad=True,
init=glorot_normal(),
init_bias=0.1)
pooling_layer_1 = MaxPooling((2, 2), strides=(2, 2), pad=True)
convolutional_layer_2 = Convolution((5, 5),
64,
strides=1,
activation=cntk.ops.relu,
pad=True,
init=glorot_normal(),
init_bias=0.1)
pooling_layer_2 = MaxPooling((2, 2), strides=(2, 2), pad=True)
convolutional_layer_3 = Convolution((5, 5),
128,
strides=1,
activation=cntk.ops.relu,
pad=True,
init=glorot_normal(),
init_bias=0.1)
pooling_layer_3 = MaxPooling((2, 2), strides=(2, 2), pad=True)
fully_connected_layer = Dense(1024,
activation=cntk.ops.relu,
init=glorot_normal(),
init_bias=0.1)
output_layer = Dense(out_dims,
activation=None,
init=glorot_normal(),
init_bias=0.1)
model = Sequential([
convolutional_layer_1,
pooling_layer_1,
convolutional_layer_2,
pooling_layer_2,
#convolutional_layer_3, pooling_layer_3,
fully_connected_layer,
output_layer
])(input_vars)
return model
def create_vgg9_model(input, num_classes):
with default_options(activation=relu):
model = Sequential([
LayerStack(3, lambda i: [
Convolution((3,3), [64,96,128][i], init=glorot_uniform(), pad=True),
Convolution((3,3), [64,96,128][i], init=glorot_uniform(), pad=True),
MaxPooling((3,3), strides=(2,2))
]),
LayerStack(2, lambda : [
Dense(1024, init=glorot_uniform())
]),
Dense(num_classes, init=glorot_uniform(), activation=None)
])
return model(input)
def ffnet():
inputs = 2
outputs = 2
layers = 2
hidden_dimension = 50
# input variables denoting the features and label data
features = C.input_variable((inputs), np.float32)
label = C.input_variable((outputs), np.float32)
# Instantiate the feedforward classification model
my_model = Sequential(
[Dense(hidden_dimension, activation=C.sigmoid),
Dense(outputs)])
z = my_model(features)
ce = C.cross_entropy_with_softmax(z, label)
pe = C.classification_error(z, label)
# Instantiate the trainer object to drive the model training
lr_per_minibatch = learning_rate_schedule(0.125, UnitType.minibatch)
trainer = C.Trainer(z, ce, pe, [sgd(z.parameters, lr=lr_per_minibatch)])
# Get minibatches of training data and perform model training
minibatch_size = 25
num_minibatches_to_train = 1024
pp = ProgressPrinter(0)
for i in range(num_minibatches_to_train):
train_features, labels = generate_random_data(minibatch_size, inputs,
outputs)
# Specify the mapping of input variables in the model to actual minibatch data to be trained with
trainer.train_minibatch({features: train_features, label: labels})
pp.update_with_trainer(trainer)
last_avg_error = pp.avg_loss_since_start()
test_features, test_labels = generate_random_data(minibatch_size, inputs,
outputs)
avg_error = trainer.test_minibatch({
features: test_features,
label: test_labels
})
print(' error rate on an unseen minibatch: {}'.format(avg_error))
return last_avg_error, avg_error
def create_recurrent_network():
# Input variables denoting the features and label data
features = input_variable(((2 * context + 1) * feature_dim))
labels = input_variable((num_classes))
# create network
model = Sequential(
[For(range(3), lambda: Recurrence(LSTM(256))),
Dense(num_classes)])
z = model(features)
ce = cross_entropy_with_softmax(z, labels)
errs = classification_error(z, labels)
return {
'feature': features,
'label': labels,
'ce': ce,
'errs': errs,
'output': z
}
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 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
def create_alexnet():
# Input variables denoting the features and label data
feature_var = input_variable((num_channels, image_height, image_width))
label_var = input_variable((num_classes))
# apply model to input
# remove mean value
input = minus(feature_var, constant(114), 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)
Convolution((11, 11),
96,
init=normal(0.01),
pad=False,
strides=(4, 4),
name='conv1'),
Activation(activation=relu, name='relu1'),
LocalResponseNormalization(1.0, 2, 0.0001, 0.75, name='norm1'),
MaxPooling((3, 3), (2, 2), name='pool1'),
Convolution((5, 5),
192,
init=normal(0.01),
init_bias=0.1,
name='conv2'),
Activation(activation=relu, name='relu2'),
LocalResponseNormalization(1.0, 2, 0.0001, 0.75, name='norm2'),
MaxPooling((3, 3), (2, 2), name='pool2'),
Convolution((3, 3), 384, init=normal(0.01), name='conv3'),
Activation(activation=relu, name='relu3'),
Convolution((3, 3),
384,
init=normal(0.01),
init_bias=0.1,
name='conv4'),
Activation(activation=relu, name='relu4'),
Convolution((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(num_classes, init=normal(0.01), name='fc8')
])(input)
# loss and metric
ce = cross_entropy_with_softmax(z, label_var)
pe = classification_error(z, label_var)
log_number_of_parameters(z)
print()
return {
'feature': feature_var,
'label': label_var,
'ce': ce,
'pe': pe,
'output': z
}
def conv3d_ucf11(train_reader, test_reader, max_epochs=30):
# Replace 0 with 1 to get detailed log.
set_computation_network_trace_level(0)
# These values must match for both train and test reader.
image_height = train_reader.height
image_width = train_reader.width
num_channels = train_reader.channel_count
sequence_length = train_reader.sequence_length
num_output_classes = train_reader.label_count
# Input variables denoting the features and label data
input_var = input_variable(
(num_channels, sequence_length, image_height, image_width), np.float32)
label_var = input_variable(num_output_classes, np.float32)
# Instantiate simple 3D Convolution network inspired by VGG network
# and http://vlg.cs.dartmouth.edu/c3d/c3d_video.pdf
with default_options(activation=relu):
z = Sequential([
Convolution((3, 3, 3), 64, pad=True),
MaxPooling((1, 2, 2), (1, 2, 2)),
LayerStack(
3, lambda i: [
Convolution((3, 3, 3), [96, 128, 128][i], pad=True),
Convolution((3, 3, 3), [96, 128, 128][i], pad=True),
MaxPooling((2, 2, 2), (2, 2, 2))
]),
LayerStack(2, lambda: [Dense(1024), Dropout(0.5)]),
Dense(num_output_classes, activation=None)
])(input_var)
# loss and classification error.
ce = cross_entropy_with_softmax(z, label_var)
pe = classification_error(z, label_var)
# training config
epoch_size = 1322 # for now we manually specify epoch size
minibatch_size = 4
# Set learning parameters
lr_per_sample = [0.01] * 10 + [0.001] * 10 + [0.0001]
lr_schedule = learning_rate_schedule(lr_per_sample,
epoch_size=epoch_size,
unit=UnitType.sample)
momentum_time_constant = 4096
mm_schedule = momentum_as_time_constant_schedule(momentum_time_constant,
epoch_size=epoch_size)
# Instantiate the trainer object to drive the model training
learner = momentum_sgd(z.parameters, lr_schedule, mm_schedule, True)
trainer = Trainer(z, ce, pe, learner)
log_number_of_parameters(z)
print()
progress_printer = ProgressPrinter(tag='Training')
# Get minibatches of images to train with and perform model training
for epoch in range(max_epochs): # loop over epochs
train_reader.reset()
while train_reader.has_more():
videos, labels, current_minibatch = train_reader.next_minibatch(
minibatch_size)
trainer.train_minibatch({input_var: videos, label_var: labels})
progress_printer.update_with_trainer(
trainer, with_metric=True) # log progress
progress_printer.epoch_summary(with_metric=True)
# Test data for trained model
epoch_size = 332
minibatch_size = 2
# process minibatches and evaluate the model
metric_numer = 0
metric_denom = 0
minibatch_index = 0
test_reader.reset()
while test_reader.has_more():
videos, labels, current_minibatch = test_reader.next_minibatch(
minibatch_size)
# minibatch data to be trained with
metric_numer += trainer.test_minibatch({
input_var: videos,
label_var: labels
}) * current_minibatch
metric_denom += current_minibatch
# Keep track of the number of samples processed so far.
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 create_vgg19():
# Input variables denoting the features and label data
feature_var = input_variable((num_channels, image_height, image_width))
label_var = 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)
LayerStack(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'),
LayerStack(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'),
LayerStack(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'),
LayerStack(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'),
LayerStack(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 = cross_entropy_with_softmax(z, label_var)
pe = classification_error(z, label_var)
pe5 = 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
}