def define_and_train(
iteration,
model_name,
load_for_inference,
failout_survival_setting,
data,
batch_size,
classes,
input_shape,
alpha,
strides,
train_datagen,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
):
mux_adds_str = "mux_adds" if ResiliNetPlus else ""
model_file = (
"models/"
+ str(iteration)
+ " "
+ mux_adds_str
+ str(failout_survival_setting)
+ "cifar_failout_rate.h5"
)
model, parallel_model = define_ResiliNet_CNN_MobileNet(
classes=classes,
input_shape=input_shape,
alpha=alpha,
failout_survival_setting=failout_survival_setting,
strides=strides,
num_gpus=num_gpus,
)
get_model_weights_CNN_cifar(
model,
parallel_model,
model_name,
load_for_inference,
model_file,
data,
train_datagen,
batch_size,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
)
return model
def define_and_train(
iteration,
model_name,
load_for_inference,
data,
batch_size,
classes,
input_shape,
alpha,
strides,
train_datagen,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
):
model, parallel_model, model_file = define_model(
iteration,
model_name,
"cifar",
input_shape,
classes,
alpha,
strides,
num_gpus,
weights=None,
)
get_model_weights_CNN_cifar(
model,
parallel_model,
model_name,
load_for_inference,
model_file,
data,
train_datagen,
batch_size,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
)
return model
def define_and_train(
iteration,
model_name,
load_for_inference,
reliability_setting,
weight_scheme,
data,
batch_size,
classes,
input_shape,
alpha,
strides,
train_datagen,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
):
if model_name == "DFG Hyperconnection Weight":
model_file = ("models/" + str(iteration) + "_" +
str(reliability_setting) + "_" + str(weight_scheme) +
"cifar_resnet_hyperconnection_DFG.h5")
model, parallel_model = define_DFG_CNN_ResNet(
input_shape=input_shape,
classes=classes,
block="basic",
residual_unit="v2",
repetitions=[2, 2, 2, 2],
initial_filters=64,
activation="softmax",
include_top=True,
input_tensor=None,
dropout=None,
transition_dilation_rate=(1, 1),
initial_strides=(2, 2),
initial_kernel_size=(7, 7),
initial_pooling="max",
final_pooling=None,
top="classification",
skip_hyperconnection_config=
default_skip_hyperconnection_config, # binary representating if a skip hyperconnection is alive [e1,IoT]
reliability_setting=reliability_setting,
hyperconnection_weights_scheme=weight_scheme,
num_gpus=num_gpus,
)
else: # model_name is "ResiliNet Hyperconnection Weight"
mux_adds_str = "mux_adds" if ResiliNetPlus else ""
model_file = ("models/" + str(iteration) + "_" + mux_adds_str +
str(reliability_setting) + "_" + str(weight_scheme) +
"cifar_resnet_hyperconnection_ResiliNet.h5")
model, parallel_model = define_ResiliNet_CNN_ResNet(
input_shape=input_shape,
classes=classes,
block="basic",
residual_unit="v2",
repetitions=[2, 2, 2, 2],
initial_filters=64,
activation="softmax",
include_top=True,
input_tensor=None,
dropout=None,
transition_dilation_rate=(1, 1),
initial_strides=(2, 2),
initial_kernel_size=(7, 7),
initial_pooling="max",
final_pooling=None,
top="classification",
failout_survival_setting=[0.95, 0.95],
skip_hyperconnection_config=default_skip_hyperconnection_config,
reliability_setting=reliability_setting,
hyperconnection_weights_scheme=weight_scheme,
num_gpus=num_gpus,
)
get_model_weights_CNN_cifar(
model,
parallel_model,
model_name,
load_for_inference,
model_file,
data,
train_datagen,
batch_size,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
)
return model
def define_and_train(
iteration,
model_name,
load_for_inference,
reliability_setting,
skip_hyperconnection_configuration,
data,
batch_size,
classes,
input_shape,
alpha,
strides,
train_datagen,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
):
if model_name == "DFG Hyperconnection Weight Sensitivity":
model_file = ("models/" + str(iteration) + " " +
str(skip_hyperconnection_configuration) + " " +
"cifar_skiphyperconnection_sensitivity_DFG.h5")
model, parallel_model = define_DFG_CNN_MobileNet(
classes=classes,
input_shape=input_shape,
alpha=alpha,
reliability_setting=reliability_setting,
skip_hyperconnection_config=skip_hyperconnection_configuration,
strides=strides,
num_gpus=num_gpus,
)
else: # model_name is "ResiliNet Hyperconnection Weight Sensitivity"
mux_adds_str = "mux_adds" if ResiliNetPlus else ""
model_file = ("models/" + str(iteration) + " " + mux_adds_str +
str(skip_hyperconnection_configuration) + " " +
"cifar_skiphyperconnection_sensitivity_ResiliNet.h5")
model, parallel_model = define_ResiliNet_CNN_MobileNet(
classes=classes,
input_shape=input_shape,
alpha=alpha,
reliability_setting=reliability_setting,
skip_hyperconnection_config=skip_hyperconnection_configuration,
strides=strides,
num_gpus=num_gpus,
)
get_model_weights_CNN_cifar(
model,
parallel_model,
model_name,
load_for_inference,
model_file,
data,
train_datagen,
batch_size,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
)
return model
def define_and_train(
iteration,
model_name,
load_for_inference,
failout_survival_setting,
data,
batch_size,
classes,
input_shape,
alpha,
strides,
train_datagen,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
):
mux_adds_str = "mux_adds" if ResiliNetPlus else ""
model_file = (
"models/"
+ str(iteration)
+ " "
+ mux_adds_str
+ str(failout_survival_setting)
+ "cifar_resnet_failout_rate.h5"
)
model, parallel_model = define_ResiliNet_CNN_ResNet(
input_shape=input_shape,
classes=classes,
block="basic",
residual_unit="v2",
repetitions=[2, 2, 2, 2],
initial_filters=64,
activation="softmax",
include_top=True,
input_tensor=None,
dropout=None,
transition_dilation_rate=(1, 1),
initial_strides=(2, 2),
initial_kernel_size=(7, 7),
initial_pooling="max",
final_pooling=None,
top="classification",
failout_survival_setting=failout_survival_setting,
skip_hyperconnection_config=default_skip_hyperconnection_config,
num_gpus=num_gpus,
)
get_model_weights_CNN_cifar(
model,
parallel_model,
model_name,
load_for_inference,
model_file,
data,
train_datagen,
batch_size,
epochs,
progress_verbose,
checkpoint_verbose,
train_steps_per_epoch,
val_steps_per_epoch,
num_gpus,
)
return model