def __init__(self,
model,
optimizer,
train_loader,
test_loader,
statspath,
criterion,
writer,
scheduler=None,
batch_scheduler=False,
L1lambda=0):
self.model = model
self.scheduler = scheduler
self.criterion = criterion
self.batch_scheduler = batch_scheduler
self.optimizer = optimizer
self.stats = ModelStats(model, statspath)
self.train = Train(model, train_loader, optimizer, self.stats,
self.scheduler if self.batch_scheduler else None,
L1lambda, criterion)
self.test = Test(model, test_loader, self.stats, writer,
self.scheduler, criterion)
self.misclass = Misclass(model, test_loader, self.stats)
self.test_loader = test_loader
torch.backends.cudnn.benchmark = True
def __init__(self, model, optimizer, train_loader, test_loader, statspath, scheduler=None, batch_scheduler=False, L1lambda = 0): self.model = model self.scheduler = scheduler self.batch_scheduler = batch_scheduler self.optimizer = optimizer self.stats = ModelStats(model, statspath) self.train = Train(model, train_loader, optimizer, self.stats, self.scheduler if self.scheduler and self.batch_scheduler else None, L1lambda) self.test = Test(model, test_loader, self.stats)
def __init__(self,
model,
optimizer,
train_loader,
test_loader,
statspath,
scheduler=None,
batch_scheduler=False,
criterion1=None,
criterion2=None,
L1lambda=0):
self.tb = SummaryWriter()
self.model = model
#x = torch.rand(1,3,128,128)
#self.tb.add_graph(self.model, x.to(self.model.device), x.to(self.model.device))
self.scheduler = scheduler
self.batch_scheduler = batch_scheduler
self.optimizer = optimizer
self.stats = ModelStats(model, statspath)
self.criterion1 = criterion1
self.criterion2 = criterion2
self.train = Train(model,
train_loader,
optimizer,
self.stats,
self.scheduler if self.batch_scheduler else None,
criterion1=criterion1,
criterion2=criterion2,
L1lambda=L1lambda,
tb=self.tb)
self.test = Test(model,
test_loader,
self.stats,
self.scheduler,
criterion1=criterion1,
criterion2=criterion2,
tb=self.tb)