def test_ar1(self):
my_nc_benchmark = self.load_ar1_scenario()
strategy = AR1(train_epochs=1,
train_mb_size=10,
eval_mb_size=10,
rm_sz=200)
StrategyTest.run_strategy(self, my_nc_benchmark, strategy)
def test_ar1(self):
my_nc_scenario = self.load_ar1_scenario(fast_test=self.fast_test)
strategy = AR1(train_epochs=1,
train_mb_size=10,
eval_mb_size=10,
rm_sz=200)
self.run_strategy(my_nc_scenario, strategy)
def main(args):
# Device config
device = torch.device(f"cuda:{args.cuda}" if torch.cuda.is_available()
and args.cuda >= 0 else "cpu")
# ---------
# --- TRANSFORMATIONS
train_transform = transforms.Compose([
Resize(224),
ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])
test_transform = transforms.Compose([
Resize(224),
ToTensor(),
transforms.Normalize((0.1307, ), (0.3081, ))
])
# ---------
# --- SCENARIO CREATION
scenario = SplitCIFAR10(5,
train_transform=train_transform,
eval_transform=test_transform)
# ---------
# CREATE THE STRATEGY INSTANCE
cl_strategy = AR1(criterion=CrossEntropyLoss(), device=device)
# TRAINING LOOP
print('Starting experiment...')
results = []
for experience in scenario.train_stream:
print("Start of experience: ", experience.current_experience)
print("Current Classes: ", experience.classes_in_this_experience)
cl_strategy.train(experience, num_workers=4)
print('Training completed')
print('Computing accuracy on the whole test set')
results.append(cl_strategy.eval(scenario.test_stream, num_workers=4))
cl_strategy = EWC(model,
Adam(model.parameters(), lr=0.001),
CrossEntropyLoss(),
ewc_lambda=0.5,
mode="separate",
train_mb_size=args.batch_size,
train_epochs=args.num_epochs,
eval_mb_size=args.batch_size * 2,
evaluator=eval_plugin,
device=device)
elif (args.cl_strategy == "AR1"):
cl_strategy = AR1(model,
Adam(model.parameters(), lr=0.001),
CrossEntropyLoss(),
ewc_lambda=0.5,
train_mb_size=args.batch_size,
train_epochs=args.num_epochs,
eval_mb_size=args.batch_size * 2,
evaluator=eval_plugin,
device=device)
elif (args.cl_strategy == "GDumb"):
cl_strategy = GDumb(model,
Adam(model.parameters(), lr=0.001),
CrossEntropyLoss(),
mem_size=200,
train_mb_size=args.batch_size,
train_epochs=args.num_epochs,
eval_mb_size=args.batch_size * 2,
evaluator=eval_plugin,
device=device)
else: