def evaluation_baseline(option):
if option["model"] == "VGG16":
model = models.VGG(layers=16)
elif option["model"] == "MobileNet":
model = models.MobileNet(alpha=option["alpha"])
elif option["model"] == "LeNet5":
model = models.LeNet5()
elif option["model"] == "ConvNet":
model = models.ConvNet()
elif option["model"] == "ConvNet_s":
model = models.ConvNet_s()
if option["model"] == "LeNet5":
test_dl = load_mnist("test", 1, 1, option["batch"])
else:
test_dl = load_cifar10("test", 1, 1, option["batch"])
load_model = torch.load(option["model_path"][0],
map_location=option["dev"])
model._modules = load_model['_modules']
model.load_state_dict(load_model['state_dict'])
model = model.to(option["dev"])
print("Evaluation")
print("\t{}".format(option["model_path"][0]))
result = evaluate(model, test_dl)
def eprune_baseline(option):
_save_name = option["save_name"]
if option["model"] == "VGG16":
model = models.VGG(layers=16)
elif option["model"] == "MobileNet":
model = models.MobileNet(alpha=option["alpha"])
if option["alpha"] == 1.0:
_save_name = _save_name + "x1.0"
else:
_save_name = _save_name + "x{:n}".format(option["alpha"])
elif option["model"] == "LeNet5":
model = models.LeNet5()
elif option["model"] == "ConvNet":
model = models.ConvNet()
elif option["model"] == "ConvNet_s":
model = models.ConvNet_s()
_save_name = _save_name + "_ep_baseline"
if option["model"] == "LeNet5":
train_dl = load_mnist("train", 1, 1,
option["batch"]) # (train_dl, valid_dl)
test_dl = load_mnist("test", 1, 1, option["batch"])
else:
train_dl = load_cifar10("train", 1, 1, option["batch"])
test_dl = load_cifar10("test", 1, 1, option["batch"])
load_model = torch.load(option["model_path"][0],
map_location=option["dev"])
model._modules = load_model['_modules']
model.load_state_dict(load_model['state_dict'])
model = model.to(option["dev"])
print("Element-wise Pruning - Baseline")
print("\t{}".format(option["model_path"][0]))
pruning = ePruning(model, train_dl[0], train_dl[1],
option["threshold_ratio"])
save_name = _save_name + "_{:03d}".format((option["threshold_ratio"]))
history = pruning.iterative_pruning(finetuning=option["retraining"],
epoch=option["epoch"],
lr=option["lr"],
schedule=option["schedule"],
save_name=save_name,
save_mode=option["save_option"])
print("* Evaluating Test Sets")
result = evaluate(model, test_dl)
def train_baseline(option):
_save_name = option["save_name"]
if option["model"] == "VGG16":
model = models.VGG(layers=16)
elif option["model"] == "MobileNet":
model = models.MobileNet(alpha=option["alpha"])
if option["alpha"] == 1.0:
_save_name = _save_name + "x1.0"
else:
_save_name = _save_name + "x{:n}".format(option["alpha"])
elif option["model"] == "LeNet5":
model = models.LeNet5()
lenet_size = 32
elif option["model"] == "LeNet300100":
model = models.LeNet300100()
lenet_size = 28
elif option["model"] == "ConvNet":
model = models.ConvNet()
elif option["model"] == "ConvNet_s":
model = models.ConvNet_s()
elif option["model"] == "ResNet20":
model = models.ResNet20()
save_name = _save_name + "_baseline"
if option["model"] == "LeNet5" or option["model"] == "LeNet300100":
train_dl = load_mnist("train", 1, 1, option["batch"],
lenet_size) # (train_dl, valid_dl)
test_dl = load_mnist("test", 1, 1, option["batch"], lenet_size)
else:
train_dl = load_cifar10("train", 1, 1, option["batch"])
test_dl = load_cifar10("test", 1, 1, option["batch"])
model = model.to(option["dev"])
loss_fn = nn.CrossEntropyLoss()
opt = optim.Adam(model.parameters(), lr=option["lr"])
history = fit(model, train_dl[0], train_dl[1], loss_fn, opt,
option["epoch"], option["schedule"], save_name,
option["save_option"])
result = evaluate(model, test_dl)
def finetuning_baseline(option):
_save_name = option["save_name"]
if option["model"] == "VGG16":
model = models.VGG(layers=16)
elif option["model"] == "MobileNet":
model = models.MobileNet(alpha=option["alpha"])
if option["alpha"] == 1.0:
_save_name = _save_name + "x1.0"
else:
_save_name = _save_name + "x{:n}".format(option["alpha"])
elif option["model"] == "LeNet5":
model = models.LeNet5()
elif option["model"] == "ConvNet":
model = models.ConvNet()
elif option["model"] == "ConvNet_s":
model = models.ConvNet_s()
save_name = _save_name + "_ft_baseline"
if option["model"] == "LeNet5":
train_dl = load_mnist("train", 1, 1,
option["batch"]) # (train_dl, valid_dl)
test_dl = load_mnist("test", 1, 1, option["batch"])
else:
train_dl = load_cifar10("train", 1, 1, option["batch"])
test_dl = load_cifar10("test", 1, 1, option["batch"])
load_model = torch.load(option["model_path"][0],
map_location=option["dev"])
model._modules = load_model['_modules']
model.load_state_dict(load_model['state_dict'])
model = model.to(option["dev"])
loss_fn = nn.CrossEntropyLoss()
opt = optim.Adam(model.parameters(), lr=option["lr"])
history = fit(model, train_dl[0], train_dl[1], loss_fn, opt,
option["epoch"], option["schedule"], save_name,
option["save_option"])
result = evaluate(model, test_dl)
def evaluation_rcm(option):
if option["model"] == "VGG16":
model = [
models.VGG(layers=16, classification=10)
for _ in range(option["rcm"])
]
elif option["model"] == "MobileNet":
model = [
models.MobileNet(alpha=option["alpha"], classification=10)
for _ in range(option["rcm"])
]
elif option["model"] == "LeNet5":
model = [
models.LeNet5(classification=10) for _ in range(option["rcm"])
]
elif option["model"] == "ConvNet":
model = [
models.ConvNet(classification=10) for _ in range(option["rcm"])
]
elif option["model"] == "ConvNet_s":
model = [
models.ConvNet_s(classification=10) for _ in range(option["rcm"])
]
if option["model"] == "LeNet5":
test_dl = load_mnist("test", 1, 1, option["batch"])
else:
test_dl = load_cifar10("test", 1, 1, option["batch"])
i = 0
for m in model:
load_model = torch.load(option["model_path"][i],
map_location=option["dev"])
m._modules = load_model['_modules']
m.load_state_dict(load_model['state_dict'])
m = m.to(option["dev"])
i = i + 1
result = distributed_evaluate(model, test_dl)
def train_rcm(option):
_save_name = option["save_name"]
if option["model"] == "VGG16":
model = [
models.VGG(layers=16, classification=10)
for _ in range(option["rcm"])
]
elif option["model"] == "MobileNet":
model = [
models.MobileNet(alpha=option["alpha"], classification=10)
for _ in range(option["rcm"])
]
if option["alpha"] == 1.0:
_save_name = _save_name + "x1.0"
else:
_save_name = _save_name + "x{:n}".format(option["alpha"])
elif option["model"] == "LeNet5":
model = [
models.LeNet5(classification=10) for _ in range(option["rcm"])
]
lenet_size = 32
elif option["model"] == "LeNet300100":
model = [models.LeNet300100() for _ in range(option["rcm"])]
lenet_size = 28
elif option["model"] == "ConvNet":
model = [
models.ConvNet(classification=10) for _ in range(option["rcm"])
]
elif option["model"] == "ConvNet_s":
model = [
models.ConvNet_s(classification=10) for _ in range(option["rcm"])
]
elif option["model"] == "ResNet20":
model = [
models.ResNet20(classification=10) for _ in range(option["rcm"])
]
_save_name = _save_name + "_rcm{:d}".format(option["rcm"])
if option["model"] == "LeNet5" or option["model"] == "LeNet300100":
train_dl = [
load_mnist("train", option["rcm"], i + 1, option["batch"],
lenet_size) for i in range(option["rcm"])
] # ([train_dl[0], ...], [valid_dl[0], ...])
test_dl = load_mnist("test", 1, 1, option["batch"], lenet_size)
else:
train_dl = [
load_cifar10("train", option["rcm"], i + 1, option["batch"])
for i in range(option["rcm"])
]
test_dl = load_cifar10("test", 1, 1, option["batch"])
load_model = torch.load(option["model_path"][0],
map_location=option["dev"])
for i in range(option["rcm"]):
print("Reduced Classification model #{:d}".format(i + 1))
classification = int(10 / option["rcm"]) + 1
model[i].load_state_dict(load_model["state_dict"])
last_in_features = model[i].classifier[-1].in_features
model[i].classifier[-1] = nn.Linear(in_features=last_in_features,
out_features=classification)
model[i] = model[i].to(option["dev"])
loss_fn = nn.CrossEntropyLoss()
opt = optim.Adam(model[i].parameters(), lr=option["lr"])
save_name = _save_name + "_{:d}".format(i + 1)
history = fit(model[i], train_dl[i][0], train_dl[i][1], loss_fn, opt,
option["epoch"], option["schedule"], save_name,
option["save_option"])
result = distributed_evaluate(model, test_dl)
def finetuning_rcm(option):
_save_name = option["save_name"]
if option["model"] == "VGG16":
model = [
models.VGG(layers=16, classification=10)
for _ in range(option["rcm"])
]
elif option["model"] == "MobileNet":
model = [
models.MobileNet(alpha=option["alpha"], classification=10)
for _ in range(option["rcm"])
]
if option["alpha"] == 1.0:
_save_name = _save_name + "x1.0"
else:
_save_name = _save_name + "x{:n}".format(option["alpha"])
elif option["model"] == "LeNet5":
model = [
models.LeNet5(classification=10) for _ in range(option["rcm"])
]
elif option["model"] == "ConvNet":
model = [
models.ConvNet(classification=10) for _ in range(option["rcm"])
]
elif option["model"] == "ConvNet_s":
model = [
models.ConvNet_s(classification=10) for _ in range(option["rcm"])
]
_save_name = _save_name + "_ft_rcm{:d}".format(option["rcm"])
if option["model"] == "LeNet5":
train_dl = [
load_mnist("train", option["rcm"], i + 1, option["batch"])
for i in range(option["rcm"])
] # ([train_dl[0], ...], [valid_dl[0], ...])
test_dl = load_mnist("test", 1, 1, option["batch"])
else:
train_dl = [
load_cifar10("train", option["rcm"], i + 1, option["batch"])
for i in range(option["rcm"])
]
test_dl = load_cifar10("test", 1, 1, option["batch"])
i = 0
for m in model:
load_model = torch.load(option["model_path"][i],
map_location=option["dev"])
m._modules = load_model['_modules']
m.load_state_dict(load_model['state_dict'])
m = m.to(option["dev"])
i = i + 1
for i in range(option["rcm"]):
print("Reduced Classification model #{:d}".format(i + 1))
loss_fn = nn.CrossEntropyLoss()
opt = optim.Adam(model[i].parameters(), lr=option["lr"])
save_name = _save_name + "_{:d}".format(i + 1)
history = fit(model[i], train_dl[i][0], train_dl[i][1], loss_fn, opt,
option["epoch"], option["schedule"], save_name,
option["save_option"])
result = distributed_evaluate(model, test_dl)
def eprune_rcm(option):
_save_name = option["save_name"]
classification = int(10 / option["rcm"]) + 1
if option["model"] == "VGG16":
model = [
models.VGG(layers=16, classification=classification)
for _ in range(option["rcm"])
]
elif option["model"] == "MobileNet":
model = [
models.MobileNet(alpha=option["alpha"],
classification=classification)
for _ in range(option["rcm"])
]
if option["alpha"] == 1.0:
_save_name = _save_name + "x1.0"
else:
_save_name = _save_name + "x{:n}".format(option["alpha"])
elif option["model"] == "LeNet5":
model = [
models.LeNet5(classification=classification)
for _ in range(option["rcm"])
]
elif option["model"] == "ConvNet":
model = [
models.ConvNet(classification=classification)
for _ in range(option["rcm"])
]
elif option["model"] == "ConvNet_s":
model = [
models.ConvNet_s(classification=classification)
for _ in range(option["rcm"])
]
_save_name = _save_name + "_ep_rcm{:d}".format(option["rcm"])
if option["model"] == "LeNet5":
train_dl = [
load_mnist("train", option["rcm"], i + 1, option["batch"])
for i in range(option["rcm"])
] # ([train_dl[0], ...], [valid_dl[0], ...])
test_dl = load_mnist("test", 1, 1, option["batch"])
else:
train_dl = [
load_cifar10("train", option["rcm"], i + 1, option["batch"])
for i in range(option["rcm"])
]
test_dl = load_cifar10("test", 1, 1, option["batch"])
i = 0
for m in model:
load_model = torch.load(option["model_path"][i],
map_location=option["dev"])
m._modules = load_model['_modules']
m.load_state_dict(load_model['state_dict'])
m = m.to(option["dev"])
i = i + 1
print("Element-wise Pruning - RCM{}".format(option["rcm"]))
for i in range(option["rcm"]):
print("Reduced Classification model #{:d}".format(i + 1))
print("\t{}".format(option["model_path"][i]))
pruning = ePruning(model[i], train_dl[i][0], train_dl[i][1],
option["threshold_ratio"])
save_name = _save_name + "_{:03d}_{:d}".format(
(option["threshold_ratio"]), i + 1)
history = pruning.iterative_pruning(finetuning=option["retraining"],
epoch=option["epoch"],
lr=option["lr"],
schedule=option["schedule"],
save_name=save_name,
save_mode=option["save_option"])
print("* Evaluating Test Sets")
result = distributed_evaluate(model, test_dl)
def prune_rcm(option):
_save_name = option["save_name"]
classification = int(10 / option["rcm"]) + 1
if option["model"] == "VGG16":
model = [
models.VGG(layers=16, classification=classification)
for _ in range(option["rcm"])
]
elif option["model"] == "MobileNet":
model = [
models.MobileNet(alpha=option["alpha"],
classification=classification)
for _ in range(option["rcm"])
]
if option["alpha"] == 1.0:
_save_name = _save_name + "x1.0"
else:
_save_name = _save_name + "x{:n}".format(option["alpha"])
elif option["model"] == "LeNet5":
model = [
models.LeNet5(classification=classification)
for _ in range(option["rcm"])
]
elif option["model"] == "ConvNet":
model = [
models.ConvNet(classification=classification)
for _ in range(option["rcm"])
]
elif option["model"] == "ConvNet_s":
model = [
models.ConvNet_s(classification=classification)
for _ in range(option["rcm"])
]
_save_name = _save_name + "_fp_rcm{:d}".format(option["rcm"])
i = 0
for m in model:
load_model = torch.load(option["model_path"][i],
map_location=option["dev"])
m._modules = load_model['_modules']
m.load_state_dict(load_model['state_dict'])
m = m.to(option["dev"])
i = i + 1
if option["model"] == "LeNet5":
train_dl = [
load_mnist("train", option["rcm"], i + 1, option["batch"])
for i in range(option["rcm"])
] # ([train_dl[0], ...], [valid_dl[0], ...])
test_dl = load_mnist("test", 1, 1, option["batch"])
else:
train_dl = [
load_cifar10("train", option["rcm"], i + 1, option["batch"])
for i in range(option["rcm"])
]
test_dl = load_cifar10("test", 1, 1, option["batch"])
csv_name = _save_name
write_pruning_result(model, test_dl, option["rcm"], 0, csv_name)
for i in range(option["prune_step"]):
print("Filter Pruning #{:d}".format(i + 1))
for j in range(option["rcm"]):
print("Reduced Classification model #{:d}".format(j + 1))
save_name = _save_name + "_{:03d}_{:d}".format(i + 1, j + 1)
pruning = Pruning(model[j], train_dl[j][0], train_dl[j][1])
history = pruning.iterative_pruning(
one_epoch_remove=option["filters_removed"],
finetuning=option["retraining"],
epoch=option["epoch"],
lr=option["lr"],
save_name=save_name,
save_mode=option["save_option"])
if option["retraining"]:
if save_name is not None:
if not (os.path.isdir("analysis")):
os.makedirs(os.path.join("analysis"))
with open('./analysis/{}_{:d}_train_loss.csv'.format(
csv_name, j + 1),
'a',
newline='') as csvfile:
training_file = csv.writer(csvfile)
training_file.writerows([history['train'][0]])
with open('./analysis/{}_{:d}_train_acc.csv'.format(
csv_name, j + 1),
'a',
newline='') as csvfile:
training_file = csv.writer(csvfile)
training_file.writerows([history['train'][1]])
with open('./analysis/{}_{:d}_valid_loss.csv'.format(
csv_name, j + 1),
'a',
newline='') as csvfile:
validation_file = csv.writer(csvfile)
validation_file.writerows([history['valid'][0]])
with open('./analysis/{}_{:d}_valid_acc.csv'.format(
csv_name, j + 1),
'a',
newline='') as csvfile:
validation_file = csv.writer(csvfile)
validation_file.writerows([history['valid'][1]])
write_pruning_result(model, test_dl, option["rcm"], i + 1, csv_name)
def prune_baseline(option):
_save_name = option["save_name"]
if option["model"] == "VGG16":
model = models.VGG(layers=16)
elif option["model"] == "MobileNet":
model = models.MobileNet(alpha=option["alpha"])
_save_name = _save_name + "x{:n}".format(option["alpha"])
elif option["model"] == "LeNet5":
model = models.LeNet5()
elif option["model"] == "ConvNet":
model = models.ConvNet()
elif option["model"] == "ConvNet_s":
model = models.ConvNet_s()
_save_name = _save_name + "_fp_baseline"
if option["model"] == "LeNet5":
train_dl = load_mnist("train", 1, 1,
option["batch"]) # (train_dl, valid_dl)
test_dl = load_mnist("test", 1, 1, option["batch"])
else:
train_dl = load_cifar10("train", 1, 1, option["batch"])
test_dl = load_cifar10("test", 1, 1, option["batch"])
load_model = torch.load(option["model_path"][0],
map_location=option["dev"])
model._modules = load_model['_modules']
model.load_state_dict(load_model['state_dict'])
model = model.to(option["dev"])
csv_name = _save_name
write_pruning_result(model, test_dl, option["rcm"], 0, csv_name)
for i in range(option["prune_step"]):
print("Filter Pruning #{:d}".format(i + 1))
save_name = _save_name + "_{:03d}".format(i + 1)
pruning = Pruning(model, train_dl[0], train_dl[1])
history = pruning.iterative_pruning(
one_epoch_remove=option["filters_removed"],
finetuning=option["retraining"],
epoch=option["epoch"],
lr=option["lr"],
save_name=save_name,
save_mode=option["save_option"])
if option["retraining"]:
if save_name is not None:
if not (os.path.isdir("analysis")):
os.makedirs(os.path.join("analysis"))
with open('./analysis/{}_train_loss.csv'.format(csv_name),
'a',
newline='') as csvfile:
training_file = csv.writer(csvfile)
training_file.writerows([history['train'][0]])
with open('./analysis/{}_train_acc.csv'.format(csv_name),
'a',
newline='') as csvfile:
training_file = csv.writer(csvfile)
training_file.writerows([history['train'][1]])
with open('./analysis/{}_valid_loss.csv'.format(csv_name),
'a',
newline='') as csvfile:
validation_file = csv.writer(csvfile)
validation_file.writerows([history['valid'][0]])
with open('./analysis/{}_valid_acc.csv'.format(csv_name),
'a',
newline='') as csvfile:
validation_file = csv.writer(csvfile)
validation_file.writerows([history['valid'][1]])
write_pruning_result(model, test_dl, option["rcm"], i + 1, csv_name)