def run(train_batch_size, val_batch_size, epochs, lr, momentum, log_interval):
cuda = torch.cuda.is_available()
train_loader, val_loader = get_data_loaders(train_batch_size, val_batch_size)
model = Net()
if cuda:
model = model.cuda()
optimizer = SGD(model.parameters(), lr=lr, momentum=momentum)
trainer = create_supervised_trainer(model, optimizer, F.nll_loss, cuda=cuda)
evaluator = create_supervised_evaluator(model,
metrics={'accuracy': CategoricalAccuracy(),
'nll': Loss(F.nll_loss)},
cuda=cuda)
@trainer.on(Events.ITERATION_COMPLETED)
def log_training_loss(engine):
iter = (engine.state.iteration - 1) % len(train_loader) + 1
if iter % log_interval == 0:
print("Epoch[{}] Iteration[{}/{}] Loss: {:.2f}"
"".format(engine.state.epoch, iter, len(train_loader), engine.state.output))
@trainer.on(Events.EPOCH_COMPLETED)
def log_validation_results(engine):
evaluator.run(val_loader)
metrics = evaluator.state.metrics
avg_accuracy = metrics['accuracy']
avg_nll = metrics['nll']
print("Validation Results - Epoch: {} Avg accuracy: {:.2f} Avg loss: {:.2f}"
.format(engine.state.epoch, avg_accuracy, avg_nll))
trainer.run(train_loader, max_epochs=epochs)
def run(mode, noise_fraction, train_batch_size, val_batch_size, epochs, lr, momentum, log_interval, log_dir):
seed = 12345
random.seed(seed)
torch.manual_seed(seed)
now = datetime.now()
log_dir = os.path.join(log_dir, "train_{}_{}__{}".format(mode, noise_fraction, now.strftime("%Y%m%d_%H%M")))
os.makedirs(log_dir)
cuda = torch.cuda.is_available()
train_loader, val_loader = get_data_loaders(noise_fraction, train_batch_size, val_batch_size)
model = Net()
writer = create_summary_writer(log_dir)
if cuda:
model = model.cuda()
optimizer = SGD(model.parameters(), lr=lr, momentum=momentum)
if mode == 'xentropy':
criterion = nn.CrossEntropyLoss()
elif mode == 'soft_bootstrap':
criterion = SoftBootstrappingLoss(beta=0.95)
elif mode == 'hard_bootstrap':
criterion = HardBootstrappingLoss(beta=0.8)
else:
raise TypeError("Wrong mode {}, expected: xentropy, soft_bootstrap or hard_bootstrap".format(mode))
trainer = create_supervised_trainer(model, optimizer, criterion, cuda=cuda)
evaluator = create_supervised_evaluator(model,
metrics={'accuracy': CategoricalAccuracy(),
'nll': Loss(nn.CrossEntropyLoss())},
cuda=cuda)
@trainer.on(Events.ITERATION_COMPLETED)
def log_training_loss(engine):
iter = (engine.state.iteration - 1) % len(train_loader) + 1
if iter % log_interval == 0:
print("Epoch[{}] Iteration[{}/{}] Loss: {:.2f}"
"".format(engine.state.epoch, iter, len(train_loader), engine.state.output))
writer.add_scalar("training/loss", engine.state.output, engine.state.iteration)
@trainer.on(Events.EPOCH_COMPLETED)
def log_validation_results(engine):
evaluator.run(val_loader)
metrics = evaluator.state.metrics
avg_accuracy = metrics['accuracy']
avg_nll = metrics['nll']
print("Validation Results - Epoch: {} Avg accuracy: {:.2f} Avg loss: {:.2f}"
.format(engine.state.epoch, avg_accuracy, avg_nll))
writer.add_scalar("valdation/loss", avg_nll, engine.state.epoch)
writer.add_scalar("valdation/accuracy", avg_accuracy, engine.state.epoch)
# kick everything off
trainer.run(train_loader, max_epochs=epochs)
writer.close()
def test_create_supervised_trainer():
model = Linear(1, 1)
model.weight.data.zero_()
model.bias.data.zero_()
optimizer = SGD(model.parameters(), 0.1)
trainer = create_supervised_trainer(model, optimizer, mse_loss)
x = torch.FloatTensor([[1.0], [2.0]])
y = torch.FloatTensor([[3.0], [5.0]])
data = [(x, y)]
assert model.weight.data[0, 0] == approx(0.0)
assert model.bias.data[0] == approx(0.0)
state = trainer.run(data)
assert state.output == approx(17.0)
assert model.weight.data[0, 0] == approx(1.3)
assert model.bias.data[0] == approx(0.8)
def run(train_batch_size, val_batch_size, epochs, lr, momentum, log_interval, log_dir):
cuda = torch.cuda.is_available()
train_loader, val_loader = get_data_loaders(train_batch_size, val_batch_size)
model = Net()
writer = create_summary_writer(model, log_dir)
if cuda:
model = model.cuda()
optimizer = SGD(model.parameters(), lr=lr, momentum=momentum)
trainer = create_supervised_trainer(model, optimizer, F.nll_loss, cuda=cuda)
evaluator = create_supervised_evaluator(model,
metrics={'accuracy': CategoricalAccuracy(),
'nll': Loss(F.nll_loss)},
cuda=cuda)
@trainer.on(Events.ITERATION_COMPLETED)
def log_training_loss(trainer, state):
iter = (state.iteration - 1) % len(train_loader) + 1
if iter % log_interval == 0:
print("Epoch[{}] Iteration[{}/{}] Loss: {:.2f}".format(state.epoch, iter, len(train_loader), state.output))
writer.add_scalar("training/loss", state.output, state.iteration)
@trainer.on(Events.EPOCH_COMPLETED)
def log_validation_results(trainer, state):
metrics = evaluator.run(val_loader).metrics
avg_accuracy = metrics['accuracy']
avg_nll = metrics['nll']
print("Validation Results - Epoch: {} Avg accuracy: {:.2f} Avg loss: {:.2f}"
.format(state.epoch, avg_accuracy, avg_nll))
writer.add_scalar("valdation/loss", avg_nll, state.epoch)
writer.add_scalar("valdation/accuracy", avg_accuracy, state.epoch)
# kick everything off
trainer.run(train_loader, max_epochs=epochs)
writer.close()
def run(train_batch_size, val_batch_size, epochs, lr, momentum, log_interval):
vis = visdom.Visdom()
if not vis.check_connection():
raise RuntimeError(
"Visdom server not running. Please run python -m visdom.server")
cuda = torch.cuda.is_available()
train_loader, val_loader = get_data_loaders(train_batch_size,
val_batch_size)
model = Net()
if cuda:
model = model.cuda()
optimizer = SGD(model.parameters(), lr=lr, momentum=momentum)
trainer = create_supervised_trainer(model,
optimizer,
F.nll_loss,
cuda=cuda)
evaluator = create_supervised_evaluator(model,
metrics={
'accuracy':
CategoricalAccuracy(),
'nll': Loss(F.nll_loss)
},
cuda=cuda)
train_loss_window = create_plot_window(vis, '#Iterations', 'Loss',
'Training Loss')
val_accuracy_window = create_plot_window(vis, '#Epochs', 'Accuracy',
'Validation Accuracy')
val_loss_window = create_plot_window(vis, '#Epochs', 'Loss',
'Validation Loss')
@trainer.on(Events.ITERATION_COMPLETED)
def log_training_loss(trainer, state):
iter = (state.iteration - 1) % len(train_loader) + 1
if iter % log_interval == 0:
print("Epoch[{}] Iteration[{}/{}] Loss: {:.2f}".format(
state.epoch, iter, len(train_loader), state.output))
vis.line(X=np.array([state.iteration]),
Y=np.array([state.output]),
update='append',
win=train_loss_window)
@trainer.on(Events.EPOCH_COMPLETED)
def log_validation_results(trainer, state):
metrics = evaluator.run(val_loader).metrics
avg_accuracy = metrics['accuracy']
avg_nll = metrics['nll']
print(
"Validation Results - Epoch: {} Avg accuracy: {:.2f} Avg loss: {:.2f}"
.format(state.epoch, avg_accuracy, avg_nll))
vis.line(X=np.array([state.epoch]),
Y=np.array([avg_accuracy]),
win=val_accuracy_window,
update='append')
vis.line(X=np.array([state.epoch]),
Y=np.array([avg_nll]),
win=val_loss_window,
update='append')
# kick everything off
trainer.run(train_loader, max_epochs=epochs)
def run(opt):
if opt.log_file is not None:
logging.basicConfig(filename=opt.log_file, level=logging.INFO)
else:
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger()
# logger.addHandler(logging.StreamHandler())
logger = logger.info
writer = SummaryWriter(log_dir=opt.log_dir)
model_timer, data_timer = Timer(average=True), Timer(average=True)
# Training variables
logger('Loading models')
model, parameters, mean, std = generate_model(opt)
optimizer = SGD(parameters,
lr=opt.lr,
momentum=opt.momentum,
weight_decay=opt.weight_decay,
nesterov=opt.nesterov)
scheduler = lr_scheduler.ReduceLROnPlateau(optimizer,
'min',
patience=opt.lr_patience)
# Loading checkpoint
if opt.checkpoint:
logger('loading checkpoint {}'.format(opt.checkpoint))
checkpoint = torch.load(opt.checkpoint)
opt.begin_epoch = checkpoint['epoch']
model.load_state_dict(checkpoint['state_dict'])
optimizer.load_state_dict(checkpoint['optimizer'])
logger('Loading dataset')
train_transform = get_transform(mean, std, opt.face_size, mode='training')
train_data = get_training_set(opt, transform=train_transform)
train_loader = DataLoader(train_data,
batch_size=opt.batch_size,
shuffle=True,
num_workers=opt.n_threads,
pin_memory=True)
val_transform = get_transform(mean, std, opt.face_size, mode='validation')
val_data = get_validation_set(opt, transform=val_transform)
val_loader = DataLoader(val_data,
batch_size=opt.batch_size,
shuffle=False,
num_workers=opt.n_threads,
pin_memory=True)
trainer = create_supervised_trainer(model,
optimizer,
nn.L1Loss().cuda(),
cuda=True)
evaluator = create_supervised_evaluator(model,
metrics={
'distance':
MeanPairwiseDistance(),
'loss': MeanAbsoluteError()
},
cuda=True)
# Training timer handlers
model_timer.attach(trainer,
start=Events.EPOCH_STARTED,
resume=Events.ITERATION_STARTED,
pause=Events.ITERATION_COMPLETED,
step=Events.ITERATION_COMPLETED)
data_timer.attach(trainer,
start=Events.EPOCH_STARTED,
resume=Events.ITERATION_COMPLETED,
pause=Events.ITERATION_STARTED,
step=Events.ITERATION_STARTED)
# Training log/plot handlers
@trainer.on(Events.ITERATION_COMPLETED)
def log_training_loss(engine):
iter = (engine.state.iteration - 1) % len(train_loader) + 1
if iter % opt.log_interval == 0:
logger(
"Epoch[{}] Iteration[{}/{}] Loss: {:.2f} Model Process: {:.3f}s/batch "
"Data Preparation: {:.3f}s/batch".format(
engine.state.epoch,
iter, len(train_loader), engine.state.output,
model_timer.value(), data_timer.value()))
writer.add_scalar("training/loss", engine.state.output,
engine.state.iteration)
# Log/Plot Learning rate
@trainer.on(Events.EPOCH_STARTED)
def log_learning_rate(engine):
lr = optimizer.param_groups[0]['lr']
logger('Epoch[{}] Starts with lr={}'.format(engine.state.epoch, lr))
writer.add_scalar("learning_rate", lr, engine.state.epoch)
# Checkpointing
@trainer.on(Events.EPOCH_COMPLETED)
def save_checkpoint(engine):
if engine.state.epoch % opt.save_interval == 0:
save_file_path = os.path.join(
opt.result_path, 'save_{}.pth'.format(engine.state.epoch))
states = {
'epoch': engine.state.epoch,
'arch': opt.model,
'state_dict': model.state_dict(),
'optimizer': optimizer.state_dict(),
}
torch.save(states, save_file_path)
# val_evaluator event handlers
@trainer.on(Events.EPOCH_COMPLETED)
def log_validation_results(engine):
evaluator.run(val_loader)
metrics = evaluator.state.metrics
# metric_values = [metrics[m] for m in val_metrics]
logger("Validation Results - Epoch: {} ".format(engine.state.epoch) +
' '.join(
['{}: {:.4f}'.format(m, val)
for m, val in metrics.items()]))
for m, val in metrics.items():
writer.add_scalar('validation/{}'.format(m), val,
engine.state.epoch)
#
if engine.state.epoch == 1:
optimizer.param_groups[0]['lr'] = 1e-4
# Update Learning Rate
scheduler.step(metrics['loss'])
# kick everything off
logger('Start training')
trainer.run(train_loader, max_epochs=opt.n_epochs)
writer.close()