def experiment_run(args):
device = 'cuda' if torch.cuda.is_available() else 'cpu'
trainloader, testloader = get_dataloaders(args)
auto_change_dir(args.exp_name)
print("Using device", device) # todo: cambiar a logger
# This will download a model with it's weights. For using another model just instantiate it here.
teacher = load_model(args.teacher, trainable=False, device=device)
student = load_model(args.student)
teacher.eval()
student.train()
best_acc = 0
start_epoch = 0
feat_loss = parse_distillation_loss(args)
kd_loss = last_layer_loss_parser(args.log_dist, string_input=True)
eval_criterion = torch.nn.CrossEntropyLoss()
optimizer = optim.Adam(student.parameters(),
lr=args.lr) # todo: evaluar si mover en exp
flatten = args.student.split("_")[0] == "linear"
layer = args.layer
idxs = [layer]
auto_change_dir(",".join([str(i) for i in idxs]))
writer = SummaryWriter("tb_logs") # todo mover dentro de exp
exp = FeatureExperiment(device=device,
student=student,
teacher=teacher,
optimizer=optimizer,
kd_criterion=kd_loss,
ft_criterion=feat_loss,
eval_criterion=eval_criterion,
linear=flatten,
writer=writer,
testloader=testloader,
trainloader=trainloader,
best_acc=best_acc,
idxs=idxs,
use_regressor=args.feat_dist == "hint",
args=args)
if exp.epoch + 1 < args.epochs:
print("training", exp.epoch, "-", args.epochs)
for epoch in range(exp.epoch, args.epochs):
exp.train_epoch()
exp.test_epoch()
exp.save_model()
else:
print("epochs surpassed")
def main(args):
#global args
device = 'cuda' if torch.cuda.is_available() else 'cpu'
args = parser.parse_args()
trainloader, testloader = get_dataloaders(args.batch_size)
auto_change_dir(args.exp_name)
teacher = load_model(args.teacher, trainable=False, device=device)
student = load_model(args.student)
teacher.eval()
student.train()
#rgs.exp_name is not None:
# os.chdir("/home/jp/Memoria/repo/Cifar10/ResNet101/") #Linux
#os.chdir("test") # Windows
best_acc = 0
start_epoch = 0
criterion = parse_distillation_loss(args) # CD a distillation
eval_criterion = torch.nn.CrossEntropyLoss()
optimizer = optim.Adam(student.parameters(), lr=args.lr)
flatten = args.student.split("_")[0] == "linear"
writer = SummaryWriter("tb_logs")
exp = DistillationExperiment(
device=device, # Todo mover arriba
student=student,
teacher=teacher,
optimizer=optimizer,
criterion=criterion,
eval_criterion=eval_criterion,
linear=flatten,
writer=writer,
testloader=testloader,
trainloader=trainloader,
best_acc=best_acc,
args=args)
for epoch in range(start_epoch, args.epochs):
exp.train_epoch()
exp.test_epoch()
exp.save_model()
def main(args):
device = 'cuda' if torch.cuda.is_available() else 'cpu'
print("using device", device)
#trainloader, testloader, classes = load_cifar10(args)
trainloader, testloader = get_dataloaders(args.batch_size)
auto_change_dir(args.exp_name)
net = load_model(args.model, trainable=False, device=device)
best_acc = 0
start_epoch = 0
writer = SummaryWriter("teacher_trainer")
criterion = nn.CrossEntropyLoss()
#optimizer = optim.SGD(net.parameters(), lr=args.lr, momentum=0.9, weight_decay=5e-4)
optimizer = optim.Adam(net.parameters(), lr=args.lr)
exp = Experiment(device=device,
net=net,
optimizer=optimizer,
criterion=criterion,
linear=args.model.split("_")[0] == "linear",
writer=writer,
testloader=testloader,
trainloader=trainloader,
best_acc=best_acc,
start_epoch=start_epoch)
exp.test_epoch()
exp.save_model(save_checkpoints=False,
overwrite_record=True,
overwrite_epoch=True)
# remove these hooks
#for h in hooks:
# h.remove()
#print(summary["ReLU-39"].mean())
print(summary)
return model,summary
if __name__ == '__main__':
from lib.utils.imagenet.utils import load_model
device = 'cuda' if torch.cuda.is_available() else 'cpu'
model=load_model("MobileNetV2")
model,s=summary(model,(3,224,224),"ReLU-39")
print(s["ReLU-39"].mean())
x = [torch.rand(2,3,224,224)+1]
# print(type(x[0]))
# make a forward pass
# print(x.shape)
model(*x)
print(s["ReLU-39"].mean())