def test(self):
devices = xm.get_xla_supported_devices()
batch_size = xu.getenv_as('BATCH_SIZE', int, defval=4)
sample_count = xu.getenv_as('SAMPLE_COUNT', int, defval=10)
train_loader = xu.SampleGenerator(
data=(torch.zeros(batch_size, 3, 224,
224), torch.zeros(batch_size, dtype=torch.int64)),
sample_count=sample_count * len(devices))
def loop_fn(model, loader, device, context):
loss_fn = nn.NLLLoss()
optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.5)
for data, target in loader:
with xu.TimedScope(msg='Training loop: ', printfn=None):
optimizer.zero_grad()
output = xu.timed(lambda: model(data), msg='Model: ', printfn=None)
loss = xu.timed(
lambda: loss_fn(output, target), msg='Loss: ', printfn=None)
xu.timed(loss.backward, msg='LossBkw: ', printfn=None)
xu.timed(
lambda: xm.optimizer_step(optimizer), msg='Step: ', printfn=None)
self.assertLess(loss.cpu().item(), 3.0)
model_parallel = dp.DataParallel(
torchvision.models.resnet18, device_ids=devices)
model_parallel(loop_fn, train_loader)
def inference():
model = model_from_name(args.model_name)
devices = (xm.get_xla_supported_devices(
max_devices=args.num_cores) if args.num_cores != 0 else [])
model.load_state_dict(torch.load(args.weight_file))
model_parallel = dp.DataParallel(model, device_ids=devices)
patient = extract_patient(args.csv_file_path, return_label=False)
## For all predictions files must be multiple by num_cores*batch_size
extra = args.num_cores * args.batch_size - len(
patient) % args.num_cores * args.batch_size
patient = np.concatenate([patient, patient[:extra]])
ds = RSNADataset(patient, path=args.path, transform=train_transforms)
loader = D.DataLoader(ds,
batch_size=args.batch_size,
shuffle=False,
num_workers=args.num_workers)
result = model_parallel(infer_loop, loader)
result = np.array(result)
score, patientid = [result[:, i] for i in range(result.shape[-1])]
score = np.concatenate(score)
patientid = np.concatenate(patientid)
prediction_to_df(score, patientid, args.subm_file)
def train():
set_seeds()
logging.info('Loading masks...')
with open(args.json_file, 'r') as f:
masks = json.load(f)
# for example use only 200 images
filename = list(masks.keys())[:200]
global devices, num_steps_per_epoch
devices = (xm.get_xla_supported_devices(
max_devices=args.num_cores) if args.num_cores != 0 else [])
logging.info('Start training model')
if args.model_name == 'deeplabv3_resnet50':
m = torchvision.models.segmentation.deeplabv3_resnet50(False)
else:
m = torchvision.models.segmentation.fcn_resnet50(False)
m.classifier[-1] = torch.nn.Conv2d(m.classifier[-1].in_channels, 46, 1)
# wrapped for parallel training
model = dp.DataParallel(m, device_ids=devices)
ds = FashionDataset(filename,
masks,
path=args.data_path,
transform=train_transform,
size=(256, 256))
loader = D.DataLoader(ds,
batch_size=args.batch_size,
shuffle=True,
num_workers=args.num_worker)
num_steps_per_epoch = len(loader) // len(devices)
for epoch in range(1, args.epochs + 1):
train_loss = model(train_loop_fn, loader)
train_loss = np.array(train_loss).mean()
logging.info('[Epoch {:3d}] Train loss: {:.3f}'.format(
epoch, train_loss))
# Save weights
state_dict = model.models[0].to('cpu').state_dict()
torch.save(state_dict, args.save_file)
logging.info('')
logging.info('Model saved\n')
def train():
set_seeds()
global num_steps_per_epoch
## Create and wrap model ##
model = model_from_name(args.model_name)
devices = (
xm.get_xla_supported_devices(max_devices=args.num_cores) if args.num_cores != 0 else [])
model_parallel = dp.DataParallel(model, device_ids=devices)
logging.info('Model {} loaded and wrapped'.format(args.model_name))
logging.info('')
## Create dataset and loader
patient, label = extract_patient(args.csv_file_path, return_label=True)
if args.use_image:
ds = RSNAImages(patient, label, path=args.path, transform=train_transforms)
else:
ds = RSNADataset(patient, label, path=args.path, transform=train_transforms)
loader = D.DataLoader(ds, batch_size=args.batch_size,
shuffle=True, num_workers=args.num_workers)
logging.info('Dataset created\n')
logging.info('Start training model\n')
num_steps_per_epoch = len(loader) // len(devices)
for epoch in range(1, args.num_epochs + 1):
start_time = time.time()
model_parallel(train_loop_fn, loader)
logging.info('')
logging.info('Epoch training time: {:.2f} minutes\n'.format((time.time() - start_time)/60**1))
# Save weights
state_dict = model_parallel.models[0].to('cpu').state_dict()
torch.save(state_dict, args.save_pht)
logging.info('')
logging.info('Model saved')
logging.info('')
def main(model, dataset, train_pairs, qrels, valid_run, qrelf, model_out_dir):
params = [(k, v) for k, v in model.named_parameters() if v.requires_grad]
non_bert_params = {
'params': [v for k, v in params if not k.startswith('bert.')]
}
bert_params = {
'params': [v for k, v in params if k.startswith('bert.')],
'lr': BERT_LR
}
optimizer = torch.optim.Adam([non_bert_params, bert_params], lr=LR)
# optimizer = torch.optim.SGD([non_bert_params, bert_params], lr=LR, momentum=0.9)
# model.to(device)
model_parallel = dp.DataParallel(model, device_ids=devices)
epoch = 0
top_valid_score = None
for epoch in range(MAX_EPOCH):
# loss = train_iteration(model, optimizer, dataset, train_pairs, qrels)
# print(f'train epoch={epoch} loss={loss}')
# # return
train_set = TrainDataset(it=data.iter_train_pairs(
model, dataset, train_pairs, qrels, 1),
length=BATCH_SIZE * BATCHES_PER_EPOCH)
train_loader = torch.utils.data.DataLoader(
train_set,
batch_size=GRAD_ACC_SIZE,
)
# for i, tr in enumerate(train_loader):
# for tt in tr:
# print(tt, tr[tt].size())
# break
# print('finished')
# return
model_parallel(train_iteration_multi, train_loader)
'''
def train_mnist(FLAGS):
DTYPE = torch.float32
torch.manual_seed(1)
dims = (
FLAGS.batch_size,
1,
784,
)
train_dataset_len = FLAGS.steps_per_epoch if FLAGS.steps_per_epoch else 60000
train_loader = xu.SampleGenerator(
data=(
torch.ones(dims, dtype=DTYPE,),
torch.ones(
FLAGS.batch_size,
dtype=torch.int64 if not _MSE_LOSS else DTYPE,
),
),
sample_count=train_dataset_len
// FLAGS.batch_size
// xm.xrt_world_size(),
)
test_loader = xu.SampleGenerator(
data=(
torch.ones(dims, dtype=DTYPE,),
torch.ones(
FLAGS.batch_size,
dtype=torch.int64 if not _MSE_LOSS else DTYPE,
),
),
sample_count=10000 // FLAGS.batch_size // xm.xrt_world_size(),
)
devices = (
xm.get_xla_supported_devices(max_devices=FLAGS.num_cores)
if FLAGS.num_cores != 0
else []
)
"""
Non multi-processing
"""
# Scale learning rate to num cores
lr = FLAGS.lr * max(len(devices), 1)
model = MNIST(FLAGS)
model_parallel = dp.DataParallel(
model,
device_ids=devices,
)
writer = None
if xm.is_master_ordinal():
writer = test_utils.get_summary_writer(FLAGS.logdir)
# Just some step closure output
def train_output_fn(outputs, ctx, args, tracker):
if ctx.step > 0 and args.log_steps and ctx.step % args.log_steps == 0:
now_time = time.time()
if hasattr(ctx, 'start_time') and ctx.start_time:
per_step_time = (now_time - ctx.start_time) / (
ctx.step - ctx.last_step_timed
)
steps_per_second = 1 / per_step_time
print(
f'[{xm.get_ordinal()}] Round-trip step time: '
f'{per_step_time} seconds, steps per second: {steps_per_second}'
)
if tracker:
_train_update(
device=device,
step=ctx.step,
loss=outputs[0],
tracker=tracker,
epoch=epoch,
writer=writer,
)
print(f'BEGIN Train step {ctx.step}')
ctx.start_time = time.time()
ctx.last_step_timed = ctx.step
else:
ctx.start_time = time.time()
ctx.last_step_timed = ctx.step
ctx.step += 1
def train_loop_fn(model, loader, device=None, context=None):
lr_adder = 0.0
if _MSE_LOSS:
loss_fn = nn.MSELoss()
else:
loss_fn = nn.NLLLoss()
optimizer = context.getattr_or(
'optimizer',
lambda: optim.SGD(
model.parameters(),
lr=lr + lr_adder,
momentum=FLAGS.momentum,
),
)
tracker = xm.RateTracker()
model.train()
def train_inner_loop_fn(batch, ctx):
step = ctx.step
print(f'Step {step}')
data = batch[0]
target = batch[1]
optimizer.zero_grad()
output = model(data)
loss = loss_fn(output, target)
loss.backward()
xm.optimizer_step(
optimizer,
barrier=False,
)
if (
FLAGS.log_steps != 0
and (
FLAGS.log_steps == 1
or (step > 0 and step % FLAGS.log_steps == 0)
)
):
xm.add_step_closure(
_train_update,
args=(device, step, loss, tracker, epoch, writer),
)
if step == 0:
xm.master_print(f"End TRAIN step {step}")
ctx.step += 1
return [loss]
step = 0
# Train
print('Starting new epoch train loop... (epoch={epoch})')
for step, (data, target) in enumerate(loader):
if step % FLAGS.step_print_interval == 0:
xm.master_print(f"Begin TRAIN Step: {step}")
context.step = step
if not FLAGS.use_autograph:
outputs = train_inner_loop_fn((data, target), context)
else:
outputs = ptwse.flow.runner.maybe_run_converted(
train_inner_loop_fn,
(data, target),
context
)
xm.master_print(f"Saving model...")
_save_checkpoint(FLAGS, device, None, model, is_epoch=True)
xm.master_print(f"Model saved")
def test_loop_fn(model, loader, device, context):
print("***********************")
print("ENTERING TEST FUNCTION")
print("***********************")
print('Evaluating...')
total_samples = 0
correct = 0
model.eval()
for step, (data, target) in enumerate(loader):
if step >= FLAGS.test_max_step:
break
output = model(data)
pred = output.max(1, keepdim=True)[1]
if FLAGS.mp:
correct += pred.eq(target.view_as(pred)).sum()
else:
correct += pred.eq(target.view_as(pred)).sum().item()
total_samples += data.size()[0]
if FLAGS.mp:
this_accuracy = 100.0 * correct.item() / total_samples
print("CALLING: mesh_reduce('test_accuracy')")
this_accuracy = xm.mesh_reduce(
'test_accuracy', this_accuracy, np.mean
)
print("BACK FROM: mesh_reduce('test_accuracy')")
else:
this_accuracy = 100.0 * correct / total_samples
test_utils.print_test_update(device, this_accuracy)
print("***********************")
print("LEAVING TEST FUNCTION")
print("***********************")
return this_accuracy
#
# Set up for
#
accuracy = 0.0
num_devices = (
len(xm.xla_replication_devices(devices)) if len(devices) > 1 else 1
)
if not FLAGS.steps_per_epoch:
num_training_steps_per_epoch = train_dataset_len // (
FLAGS.batch_size * num_devices
)
else:
num_training_steps_per_epoch = FLAGS.steps_per_epoch
max_accuracy = 0.0
#
# Epoch loop
#
for epoch in range(1, FLAGS.num_epochs + 1):
#
# Train
#
device = xm.xla_device()
ctx = dp.Context(device=device)
ctx.tracker = xm.RateTracker()
ctx.step = 0
train_loop_fn(model, train_loader, device, ctx)
#
# Test
#
if FLAGS.run_test:
with ptwse.scope.proxy_disabled(disabled=FLAGS.test_off_proxy):
accuracies = model_parallel(test_loop_fn, test_loader)
accuracy = mean(accuracies)
print(
'Epoch: {}, Mean Accuracy: {:.2f}%'.format(epoch, accuracy)
)
global_step = (epoch - 1) * num_training_steps_per_epoch
max_accuracy = max(accuracy, max_accuracy)
test_utils.write_to_summary(
writer,
global_step,
dict_to_write={'Accuracy/test': accuracy},
write_xla_metrics=True,
)
if FLAGS.metrics_debug:
xm.master_print(met.metrics_report())
test_utils.close_summary_writer(writer)
xm.master_print('Max Accuracy: {:.2f}%'.format(max_accuracy))
return max_accuracy
def train_imagenet():
print('==> Preparing data..')
img_dim = get_model_property('img_dim')
if FLAGS.fake_data:
train_dataset_len = 1200000 # Roughly the size of Imagenet dataset.
train_loader = xu.SampleGenerator(
data=(torch.zeros(FLAGS.batch_size, 3, img_dim, img_dim),
torch.zeros(FLAGS.batch_size, dtype=torch.int64)),
sample_count=train_dataset_len // FLAGS.batch_size //
xm.xrt_world_size())
test_loader = xu.SampleGenerator(
data=(torch.zeros(FLAGS.test_set_batch_size, 3, img_dim, img_dim),
torch.zeros(FLAGS.test_set_batch_size, dtype=torch.int64)),
sample_count=50000 // FLAGS.batch_size // xm.xrt_world_size())
else:
normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
train_dataset = torchvision.datasets.ImageFolder(
os.path.join(FLAGS.datadir, 'train'),
transforms.Compose([
transforms.RandomResizedCrop(img_dim),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
normalize,
]))
train_dataset_len = len(train_dataset.imgs)
resize_dim = max(img_dim, 256)
test_dataset = torchvision.datasets.ImageFolder(
os.path.join(FLAGS.datadir, 'val'),
# Matches Torchvision's eval transforms except Torchvision uses size
# 256 resize for all models both here and in the train loader. Their
# version crashes during training on 299x299 images, e.g. inception.
transforms.Compose([
transforms.Resize(resize_dim),
transforms.CenterCrop(img_dim),
transforms.ToTensor(),
normalize,
]))
train_sampler = None
test_sampler = None
if xm.xrt_world_size() > 1:
train_sampler = torch.utils.data.distributed.DistributedSampler(
train_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=True)
test_sampler = torch.utils.data.distributed.DistributedSampler(
test_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=False)
train_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=FLAGS.batch_size,
sampler=train_sampler,
shuffle=False if train_sampler else True,
num_workers=FLAGS.num_workers)
test_loader = torch.utils.data.DataLoader(
test_dataset,
batch_size=FLAGS.test_set_batch_size,
sampler=test_sampler,
shuffle=False,
num_workers=FLAGS.num_workers)
torch.manual_seed(42)
devices = (xm.get_xla_supported_devices(
max_devices=FLAGS.num_cores) if FLAGS.num_cores != 0 else [])
# Pass [] as device_ids to run using the PyTorch/CPU engine.
torchvision_model = get_model_property('model_fn')
model_parallel = dp.DataParallel(torchvision_model, device_ids=devices)
def train_loop_fn(model, loader, device, context):
loss_fn = nn.CrossEntropyLoss()
optimizer = context.getattr_or(
'optimizer', lambda: optim.SGD(model.parameters(),
lr=FLAGS.lr,
momentum=FLAGS.momentum,
weight_decay=5e-4))
lr_scheduler = context.getattr_or(
'lr_scheduler', lambda: schedulers.wrap_optimizer_with_scheduler(
optimizer,
scheduler_type=getattr(FLAGS, 'lr_scheduler_type', None),
scheduler_divisor=getattr(FLAGS, 'lr_scheduler_divisor', None),
scheduler_divide_every_n_epochs=getattr(
FLAGS, 'lr_scheduler_divide_every_n_epochs', None),
num_steps_per_epoch=num_training_steps_per_epoch,
summary_writer=writer if xm.is_master_ordinal() else None))
tracker = xm.RateTracker()
model.train()
for x, (data, target) in loader:
optimizer.zero_grad()
output = model(data)
loss = loss_fn(output, target)
loss.backward()
xm.optimizer_step(optimizer)
tracker.add(FLAGS.batch_size)
if x % FLAGS.log_steps == 0:
test_utils.print_training_update(device, x, loss.item(),
tracker.rate(),
tracker.global_rate())
if lr_scheduler:
lr_scheduler.step()
def test_loop_fn(model, loader, device, context):
total_samples = 0
correct = 0
model.eval()
for x, (data, target) in loader:
output = model(data)
pred = output.max(1, keepdim=True)[1]
correct += pred.eq(target.view_as(pred)).sum().item()
total_samples += data.size()[0]
accuracy = 100.0 * correct / total_samples
test_utils.print_test_update(device, accuracy)
return accuracy
accuracy = 0.0
writer = SummaryWriter(log_dir=FLAGS.logdir) if FLAGS.logdir else None
num_devices = len(
xm.xla_replication_devices(devices)) if len(devices) > 1 else 1
num_training_steps_per_epoch = train_dataset_len // (FLAGS.batch_size *
num_devices)
for epoch in range(1, FLAGS.num_epochs + 1):
model_parallel(train_loop_fn, train_loader)
accuracies = model_parallel(test_loop_fn, test_loader)
accuracy = mean(accuracies)
print('Epoch: {}, Mean Accuracy: {:.2f}%'.format(epoch, accuracy))
global_step = (epoch - 1) * num_training_steps_per_epoch
test_utils.add_scalar_to_summary(writer, 'Accuracy/test', accuracy,
global_step)
if FLAGS.metrics_debug:
print(torch_xla._XLAC._xla_metrics_report())
return accuracy
def train_mnist():
torch.manual_seed(1)
if FLAGS.fake_data:
train_dataset_len = 60000 # Number of images in MNIST dataset.
train_loader = xu.SampleGenerator(
data=(torch.zeros(FLAGS.batch_size, 1, 28, 28),
torch.zeros(FLAGS.batch_size, dtype=torch.int64)),
sample_count=train_dataset_len // FLAGS.batch_size //
xm.xrt_world_size())
test_loader = xu.SampleGenerator(
data=(torch.zeros(FLAGS.batch_size, 1, 28, 28),
torch.zeros(FLAGS.batch_size, dtype=torch.int64)),
sample_count=10000 // FLAGS.batch_size // xm.xrt_world_size())
else:
train_dataset = datasets.MNIST(FLAGS.datadir,
train=True,
download=True,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ),
(0.3081, ))
]))
train_dataset_len = len(train_dataset)
test_dataset = datasets.MNIST(FLAGS.datadir,
train=False,
transform=transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.1307, ),
(0.3081, ))
]))
train_sampler = None
test_sampler = None
if xm.xrt_world_size() > 1:
train_sampler = torch.utils.data.distributed.DistributedSampler(
train_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=True)
test_sampler = torch.utils.data.distributed.DistributedSampler(
test_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=False)
train_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=FLAGS.batch_size,
sampler=train_sampler,
drop_last=FLAGS.drop_last,
shuffle=False if train_sampler else True,
num_workers=FLAGS.num_workers)
test_loader = torch.utils.data.DataLoader(
test_dataset,
batch_size=FLAGS.batch_size,
sampler=test_sampler,
drop_last=FLAGS.drop_last,
shuffle=False,
num_workers=FLAGS.num_workers)
devices = (xm.get_xla_supported_devices(
max_devices=FLAGS.num_cores) if FLAGS.num_cores != 0 else [])
# Scale learning rate to num cores
lr = FLAGS.lr * max(len(devices), 1)
# Pass [] as device_ids to run using the PyTorch/CPU engine.
model_parallel = dp.DataParallel(MNIST, device_ids=devices)
def train_loop_fn(model, loader, device, context):
loss_fn = nn.NLLLoss()
optimizer = context.getattr_or(
'optimizer', lambda: optim.SGD(
model.parameters(), lr=lr, momentum=FLAGS.momentum))
tracker = xm.RateTracker()
model.train()
for x, (data, target) in enumerate(loader):
optimizer.zero_grad()
output = model(data)
loss = loss_fn(output, target)
loss.backward()
xm.optimizer_step(optimizer)
tracker.add(FLAGS.batch_size)
if x % FLAGS.log_steps == 0:
test_utils.print_training_update(device, x, loss.item(),
tracker.rate(),
tracker.global_rate())
def test_loop_fn(model, loader, device, context):
total_samples = 0
correct = 0
model.eval()
for data, target in loader:
output = model(data)
pred = output.max(1, keepdim=True)[1]
correct += pred.eq(target.view_as(pred)).sum().item()
total_samples += data.size()[0]
accuracy = 100.0 * correct / total_samples
test_utils.print_test_update(device, accuracy)
return accuracy
accuracy = 0.0
writer = test_utils.get_summary_writer(FLAGS.logdir)
num_devices = len(
xm.xla_replication_devices(devices)) if len(devices) > 1 else 1
num_training_steps_per_epoch = train_dataset_len // (FLAGS.batch_size *
num_devices)
for epoch in range(1, FLAGS.num_epochs + 1):
model_parallel(train_loop_fn, train_loader)
accuracies = model_parallel(test_loop_fn, test_loader)
accuracy = mean(accuracies)
print('Epoch: {}, Mean Accuracy: {:.2f}%'.format(epoch, accuracy))
global_step = (epoch - 1) * num_training_steps_per_epoch
test_utils.add_scalar_to_summary(writer, 'Accuracy/test', accuracy,
global_step)
if FLAGS.metrics_debug:
print(met.metrics_report())
test_utils.close_summary_writer(writer)
return accuracy
def train_cifar():
print('==> Preparing data..')
if FLAGS.fake_data:
train_loader = xu.SampleGenerator(
data=(torch.zeros(FLAGS.batch_size, 3, 32, 32),
torch.zeros(FLAGS.batch_size, dtype=torch.int64)),
sample_count=50000 // FLAGS.batch_size // xm.xrt_world_size())
test_loader = xu.SampleGenerator(
data=(torch.zeros(FLAGS.batch_size, 3, 32, 32),
torch.zeros(FLAGS.batch_size, dtype=torch.int64)),
sample_count=10000 // FLAGS.batch_size // xm.xrt_world_size())
else:
transform_train = transforms.Compose([
transforms.RandomCrop(32, padding=4),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465),
(0.2023, 0.1994, 0.2010)),
])
transform_test = transforms.Compose([
transforms.ToTensor(),
transforms.Normalize((0.4914, 0.4822, 0.4465),
(0.2023, 0.1994, 0.2010)),
])
train_dataset = torchvision.datasets.CIFAR10(root=FLAGS.datadir,
train=True,
download=True,
transform=transform_train)
test_dataset = torchvision.datasets.CIFAR10(root=FLAGS.datadir,
train=False,
download=True,
transform=transform_test)
train_sampler = None
test_sampler = None
if xm.xrt_world_size() > 1:
train_sampler = torch.utils.data.distributed.DistributedSampler(
train_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=True)
test_sampler = torch.utils.data.distributed.DistributedSampler(
test_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=False)
train_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=FLAGS.batch_size,
sampler=train_sampler,
shuffle=False if train_sampler else True,
num_workers=FLAGS.num_workers)
test_loader = torch.utils.data.DataLoader(
test_dataset,
batch_size=FLAGS.batch_size,
sampler=test_sampler,
shuffle=False,
num_workers=FLAGS.num_workers)
torch.manual_seed(42)
devices = (xm.get_xla_supported_devices(
max_devices=FLAGS.num_cores) if FLAGS.num_cores != 0 else [])
# Pass [] as device_ids to run using the PyTorch/CPU engine.
model = torchvision.models.resnet18 if FLAGS.use_torchvision else ResNet18
model_parallel = dp.DataParallel(model, device_ids=devices)
def train_loop_fn(model, loader, device, context):
loss_fn = nn.CrossEntropyLoss()
optimizer = context.getattr_or(
'optimizer', lambda: optim.SGD(model.parameters(),
lr=FLAGS.lr,
momentum=FLAGS.momentum,
weight_decay=5e-4))
tracker = xm.RateTracker()
model.train()
for x, (data, target) in loader:
optimizer.zero_grad()
output = model(data)
loss = loss_fn(output, target)
loss.backward()
xm.optimizer_step(optimizer)
tracker.add(FLAGS.batch_size)
if x % FLAGS.log_steps == 0:
test_utils.print_training_update(device, x, loss.item(),
tracker.rate(),
tracker.global_rate())
def test_loop_fn(model, loader, device, context):
total_samples = 0
correct = 0
model.eval()
for x, (data, target) in loader:
output = model(data)
pred = output.max(1, keepdim=True)[1]
correct += pred.eq(target.view_as(pred)).sum().item()
total_samples += data.size()[0]
accuracy = 100.0 * correct / total_samples
test_utils.print_test_update(device, accuracy)
return accuracy
accuracy = 0.0
for epoch in range(1, FLAGS.num_epochs + 1):
model_parallel(train_loop_fn, train_loader)
accuracies = model_parallel(test_loop_fn, test_loader)
accuracy = sum(accuracies) / len(accuracies)
print("Epoch: {}, Mean Accuracy: {:.2f}%".format(epoch, accuracy))
if FLAGS.metrics_debug:
print(met.metrics_report())
return accuracy
def train_imagenet():
print("==> Preparing data..")
img_dim = get_model_property("img_dim")
if FLAGS.fake_data:
train_dataset_len = 1200000 # Roughly the size of Imagenet dataset.
train_loader = xu.SampleGenerator(
data=(
torch.zeros(FLAGS.batch_size, 3, img_dim, img_dim),
torch.zeros(FLAGS.batch_size, dtype=torch.int64),
),
sample_count=train_dataset_len // FLAGS.batch_size // xm.xrt_world_size(),
)
test_loader = xu.SampleGenerator(
data=(
torch.zeros(FLAGS.test_set_batch_size, 3, img_dim, img_dim),
torch.zeros(FLAGS.test_set_batch_size, dtype=torch.int64),
),
sample_count=50000 // FLAGS.batch_size // xm.xrt_world_size(),
)
else:
normalize = transforms.Normalize(
mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]
)
train_dataset = torchvision.datasets.ImageFolder(
os.path.join(FLAGS.datadir, "train"),
transforms.Compose(
[
transforms.RandomResizedCrop(img_dim),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
normalize,
]
),
)
train_dataset_len = len(train_dataset.imgs)
resize_dim = max(img_dim, 256)
test_dataset = torchvision.datasets.ImageFolder(
os.path.join(FLAGS.datadir, "val"),
# Matches Torchvision's eval transforms except Torchvision uses size
# 256 resize for all models both here and in the train loader. Their
# version crashes during training on 299x299 images, e.g. inception.
transforms.Compose(
[
transforms.Resize(resize_dim),
transforms.CenterCrop(img_dim),
transforms.ToTensor(),
normalize,
]
),
)
train_sampler = None
test_sampler = None
if xm.xrt_world_size() > 1:
train_sampler = torch.utils.data.distributed.DistributedSampler(
train_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=True,
)
test_sampler = torch.utils.data.distributed.DistributedSampler(
test_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=False,
)
train_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=FLAGS.batch_size,
sampler=train_sampler,
drop_last=True,
shuffle=False if train_sampler else True,
num_workers=FLAGS.num_workers,
)
test_loader = torch.utils.data.DataLoader(
test_dataset,
batch_size=FLAGS.test_set_batch_size,
sampler=test_sampler,
drop_last=False,
shuffle=False,
num_workers=FLAGS.num_workers,
)
torch.manual_seed(42)
devices = (
xm.get_xla_supported_devices(max_devices=FLAGS.num_cores)
if FLAGS.num_cores != 0
else ["cpu"]
)
print("use tpu devices", devices)
# Pass [] as device_ids to run using the PyTorch/CPU engine.
torchvision_model = get_model_property("model_fn")
model_parallel = dp.DataParallel(torchvision_model, device_ids=devices)
def train_loop_fn(model, loader, device, context):
loss_fn = nn.CrossEntropyLoss()
optimizer = context.getattr_or(
"optimizer",
lambda: optim.SGD(
model.parameters(),
lr=FLAGS.lr,
momentum=FLAGS.momentum,
weight_decay=1e-4,
),
)
lr_scheduler = context.getattr_or(
"lr_scheduler",
lambda: schedulers.wrap_optimizer_with_scheduler(
optimizer,
scheduler_type=getattr(FLAGS, "lr_scheduler_type", None),
scheduler_divisor=getattr(FLAGS, "lr_scheduler_divisor", None),
scheduler_divide_every_n_epochs=getattr(
FLAGS, "lr_scheduler_divide_every_n_epochs", None
),
num_steps_per_epoch=num_training_steps_per_epoch,
summary_writer=writer if xm.is_master_ordinal() else None,
),
)
tracker = xm.RateTracker()
model.train()
total_samples = 0
correct = 0
top5_accuracys = 0
losses = 0
for x, (data, target) in loader:
optimizer.zero_grad()
output = model(data)
pred = output.max(1, keepdim=True)[1]
correct += pred.eq(target.view_as(pred)).sum()
total_samples += data.size()[0]
top5_accuracys += topk_accuracy(output, target, topk=5)
loss = loss_fn(output, target)
loss.backward()
losses += loss.item()
xm.optimizer_step(optimizer)
tracker.add(FLAGS.batch_size)
if x % FLAGS.log_steps == 0:
print(
"[{}]({}) Loss={:.5f} Top-1 ACC = {:.2f} Rate={:.2f} GlobalRate={:.2f} Time={}".format(
str(device),
x,
loss.item(),
(100.0 * correct / total_samples).item(),
tracker.rate(),
tracker.global_rate(),
time.asctime(),
)
)
if lr_scheduler:
lr_scheduler.step()
return (
losses / (x + 1),
(100.0 * correct / total_samples).item(),
(top5_accuracys / (x + 1)).item(),
)
def test_loop_fn(model, loader, device, context):
total_samples = 0
correct = 0
top5_accuracys = 0
model.eval()
for x, (data, target) in loader:
output = model(data)
pred = output.max(1, keepdim=True)[1]
correct += pred.eq(target.view_as(pred)).sum().item()
total_samples += data.size()[0]
top5_accuracys += topk_accuracy(output, target, topk=5).item()
accuracy = 100.0 * correct / total_samples
test_utils.print_test_update(device, accuracy)
return accuracy, top5_accuracys
accuracy = 0.0
writer = SummaryWriter(FLAGS.logdir) if FLAGS.logdir else None
num_devices = len(xm.xla_replication_devices(devices)) if len(devices) > 1 else 1
num_training_steps_per_epoch = train_dataset_len // (FLAGS.batch_size * num_devices)
max_accuracy = 0.0
print("train_loader_len", len(train_loader), num_training_steps_per_epoch)
print("test_loader_len", len(test_loader))
for epoch in range(1, FLAGS.num_epochs + 1):
global_step = (epoch - 1) * num_training_steps_per_epoch
# Train evaluate
metrics = model_parallel(train_loop_fn, train_loader)
losses, accuracies, top5_accuracys = zip(*metrics)
loss = mean(losses)
accuracy = mean(accuracies)
top5_accuracy = mean(top5_accuracys)
print(
"Epoch: {} (Train), Loss {}, Mean Top-1 Accuracy: {:.2f} Top-5 accuracy: {}".format(
epoch, loss, accuracy, top5_accuracy
)
)
test_utils.add_scalar_to_summary(writer, "Loss/train", loss, global_step)
test_utils.add_scalar_to_summary(
writer, "Top-1 Accuracy/train", accuracy, global_step
)
test_utils.add_scalar_to_summary(
writer, "Top-5 Accuracy/train", top5_accuracy, global_step
)
# Test evaluate
metrics = model_parallel(test_loop_fn, test_loader)
accuracies, top5_accuracys = zip(*metrics)
top5_accuracys = sum(top5_accuracys)
top5_accuracy = top5_accuracys / len(test_loader)
accuracy = mean(accuracies)
print(
"Epoch: {} (Valid), Mean Top-1 Accuracy: {:.2f} Top-5 accuracy: {}".format(
epoch, accuracy, top5_accuracy
)
)
test_utils.add_scalar_to_summary(
writer, "Top-1 Accuracy/test", accuracy, global_step
)
test_utils.add_scalar_to_summary(
writer, "Top-5 Accuracy/test", top5_accuracy, global_step
)
if FLAGS.metrics_debug:
print(met.metrics_report())
if accuracy > max_accuracy:
max_accuracy = max(accuracy, max_accuracy)
torch.save(
model_parallel.models[0].to("cpu").state_dict(),
f"./reports/resnet50_model-{epoch}.pt",
)
model_parallel.models[0].to(devices[0])
test_utils.close_summary_writer(writer)
print("Max Accuracy: {:.2f}%".format(accuracy))
return max_accuracy
settings.STD,
)
Flag = {}
Flag['lr'] = args.lr
Flag['epoch'] = args.e
Flag['warm'] = args.warm
Flag['batch_size'] = args.b
Flag['milestones'] = settings.MILESTONES
Flag['ignore_idx'] = train_data_loader.dataset.ignore_index
len(train_data_loader.dataset)
net = UNet(3, train_data_loader.dataset.class_num)
devices = (xm.get_xla_supported_devices())
print(devices)
net = dp.DataParallel(net, device_ids=devices)
iter_per_epoch = len(train_data_loader) / 8
best_iou = 0
for epoch in range(1, args.e + 1):
print('training epoch {}'.format(epoch))
t1 = time.time()
net(train_loop_fn, train_data_loader)
print(time.time() - t1)
#result = net(test_loop_fn, test_data_loader)
#pred_res = np.array([res[0] for res in result])
#mask_res = np.array([res[1] for res in result])
def train_unet():
print('==> Preparing data..')
img_dim = 1024
normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
train_dataset = satelliteDataSet(
os.path.join(datadir, 'train'),
transforms.Compose([
transforms.RandomResizedCrop(img_dim),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(), normalize
]))
train_dataset_len = len(train_dataset)
resize_dim = max(img_dim, 256)
test_dataset = satelliteDataSet(
os.path.join(datadir, 'train'),
transforms.Compose([
transforms.Resize(resize_dim),
transforms.CenterCrop(img_dim),
transforms.ToTensor(),
normalize,
]))
train_sampler = None
test_sampler = None
if xm.xrt_world_size() > 1:
train_sampler = torch.utils.data.distributed.DistributedSampler(
train_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=True)
test_sampler = torch.utils.data.distributed.DistributedSampler(
test_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=False)
train_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=batch_size,
sampler=train_sampler,
shuffle=False if train_sampler else True,
num_workers=num_workers)
test_loader = torch.utils.data.DataLoader(test_dataset,
batch_size=test_set_batch_size,
sampler=test_sampler,
shuffle=False,
num_workers=num_workers)
torch.manual_seed(42)
devices = (xm.get_xla_supported_devices(max_devices=num_cores))
torchvision_model = UNet()
model_parallel = dp.DataParallel(torchvision_model, device_ids=devices)
def train_loop_fn(model, loader, device, context):
loss_fn = nn.CrossEntropyLoss()
optimizer = context.getattr_or(
'optimizer', lambda: optim.SGD(model.parameters(),
lr=lr,
momentum=momentum,
weight_decay=1e-4))
lr_scheduler = context.getattr_or(
'lr_scheduler', lambda: schedulers.wrap_optimizer_with_scheduler(
optimizer,
scheduler_type=lr_scheduler_type,
scheduler_divisor=lr_scheduler_divisor,
scheduler_divide_every_n_epochs=
lr_scheduler_divide_every_n_epochs,
num_steps_per_epoch=num_training_steps_per_epoch,
summary_writer=None))
tracker = xm.RateTracker()
model.train()
for x, (data, target) in loader:
optimizer.zero_grad()
data = data.permute(0, 3, 1, 2)
output = model(data)
print('passed through model')
loss = loss_fn(output, target)
loss.backward()
xm.optimizer_step(optimizer)
tracker.add(batch_size)
if x % log_steps == 0:
print(
'device: {}, x: {}, loss: {}, tracker: {}, tracker_global: {} '
.format(device, x, loss.item(), tracker.rate(),
tracker.global_rate()))
if lr_scheduler:
lr_scheduler.step()
def test_loop_fn(model, loader, device, context):
total_samples = 0
correct = 0
model.eval()
for x, (data, target) in loader:
output = model(data)
pred = output.max(1, keepdim=True)[1]
correct += pred.eq(target.view_as(pred)).sum().item()
total_samples += data.size()[0]
accuracy = 100.0 * correct / total_samples
#test_utils.print_test_update(device, accuracy)
print('device: {}, accuracy: {}'.format(device, accuracy))
return accuracy
accuracy = 0.0
num_devices = len(
xm.xla_replication_devices(devices)) if len(devices) > 1 else 1
num_training_steps_per_epoch = train_dataset_len // (batch_size *
num_devices)
for epoch in range(1, num_epochs + 1):
model_parallel(train_loop_fn, train_loader)
accuracies = model_parallel(test_loop_fn, test_loader)
accuracy = mean(accuracies)
print('Epoch: {}, Mean Accuracy: {:.2f}%'.format(epoch, accuracy))
global_step = (epoch - 1) * num_training_steps_per_epoch
print('global step: {}'.format(global_step))
return accuracy
def train_unet():
# logging setup
logger_name = 'train_logger'
logger = initializeLogging(os.path.join(logdir,
'train_history.txt'), logger_name)
# checkpointing setup
checkpoint_frequency = log_steps
torch.manual_seed(1)
'''
train_dataset = datasets.MNIST(
datadir,
train=True,
download=True,
transform=transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,))]))
train_dataset_len = len(train_dataset)
test_dataset = datasets.MNIST(
datadir,
train=False,
transform=transforms.Compose(
[transforms.ToTensor(),
transforms.Normalize((0.1307,), (0.3081,))]))
'''
img_dim = 1024
normalize = transforms.Normalize(
mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
train_dataset = satelliteDataSet(
os.path.join(datadir, 'train'),
transforms.Compose([
transforms.RandomResizedCrop(img_dim),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
normalize
]))
train_dataset_len = len(train_dataset)
resize_dim = max(img_dim, 256)
test_dataset = satelliteDataSet(
os.path.join(datadir, 'train'),
transforms.Compose([
transforms.Resize(resize_dim),
transforms.CenterCrop(img_dim),
transforms.ToTensor(),
normalize,
]))
train_sampler = None
test_sampler = None
if xm.xrt_world_size() > 1:
train_sampler = torch.utils.data.distributed.DistributedSampler(
train_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=True)
test_sampler = torch.utils.data.distributed.DistributedSampler(
test_dataset,
num_replicas=xm.xrt_world_size(),
rank=xm.get_ordinal(),
shuffle=False)
train_loader = torch.utils.data.DataLoader(
train_dataset,
batch_size=batch_size,
sampler=train_sampler,
drop_last=drop_last,
shuffle=False if train_sampler else True,
num_workers=num_workers)
test_loader = torch.utils.data.DataLoader(
test_dataset,
batch_size=batch_size,
sampler=test_sampler,
drop_last=drop_last,
shuffle=False,
num_workers=num_workers)
maxItr = num_epochs * len(train_loader) // train_loader.batch_size + 1
devices = (
xm.get_xla_supported_devices(
max_devices=num_cores) if num_cores != 0 else [])
# Scale learning rate to num cores
lr = 0.0001
lr = lr * max(len(devices), 1)
# Pass [] as device_ids to run using the PyTorch/CPU engine.
model_parallel = dp.DataParallel(UNet, device_ids=devices)
def train_loop_fn(model, loader, device, context):
loss_fn = nn.CrossEntropyLoss()
optimizer = context.getattr_or(
'optimizer',
lambda: optim.Adam(model.parameters(), lr=lr))
tracker = xm.RateTracker()
model.train()
print('# of iterations: {}'.format(maxItr))
logger.info('# of iterations: {}'.format(maxItr))
optimizer.zero_grad()
for x, (data, target) in enumerate(loader):
data = target[0].permute(0,3,1,2)
target = target[1]
output = model(data)
loss = loss_fn(output, target.long())
#_, preds = torch.max(output, 1)
loss.backward()
# backprop every log_step iterations
if x % log_steps == 0:
xm.optimizer_step(optimizer)
optimizer.zero_grad()
tracker.add(batch_size)
# compute the confusion matrix and IoU
#print(preds.shape)
#print(target.shape)
#val_conf = np.zeros((num_classes, num_classes))
#val_conf = val_conf + confusion_matrix(
# target[target >= 0].view(-1).cpu().numpy(),
# preds[target >= 0].view(-1).cpu().numpy())
#pos = np.sum(val_conf, 1)
#res = np.sum(val_conf, 0)
#tp = np.diag(val_conf)
#iou = np.mean(tp / np.maximum(1, pos + res - tp))
#logger.info('device: {}, x: {}, loss: {}, tracker_rate: {}, tracker_global_rate: {}'.format(device, x, loss.item(), tracker.rate(), tracker.global_rate()))
print('device: {}, x: {}, loss: {}, tracker_rate: {}, tracker_global_rate: {}'.format(device, x, loss.item(), tracker.rate(), tracker.global_rate()))
if x % log_steps == 0:
logger.info('device: {}, x: {}, loss: {}, tracker_rate: {}, tracker_global_rate: {}'.format(device, x, loss.item(), tracker.rate(), tracker.global_rate()))
def test_loop_fn(model, loader, device, context):
total_samples = 0
correct = 0
model.eval()
for data, target in loader:
data = target[0].permute(0,3,1,2)
target = target[1]
output = model(data)
#pred = output.max(1, keepdim=True)[1].float()
_, preds = torch.max(output, 1)
preds = preds.float()
correct += preds.eq(target.view_as(preds)).sum().item()
total_samples += target.shape[1]**2
print('device: {}, Running Accuracy: {}'.format(device, correct/total_samples))
accuracy = 100.0 * correct / total_samples
test_utils.print_test_update(device, accuracy)
logger.info('TEST: device: {}, accuracy: {}'.format(device, accuracy))
return accuracy
accuracy = 0.0
writer = test_utils.get_summary_writer(logdir)
num_devices = len(
xm.xla_replication_devices(devices)) if len(devices) > 1 else 1
num_training_steps_per_epoch = train_dataset_len // (
batch_size * num_devices)
print('total epochs: {}'.format(num_epochs))
for epoch in range(1, num_epochs + 1):
print(epoch)
print(train_loader)
model_parallel(train_loop_fn, train_loader)
accuracies = model_parallel(test_loop_fn, test_loader)
accuracy = mean(accuracies)
print('Epoch: {}, Mean Accuracy: {:.2f}%'.format(epoch, accuracy))
logger.info('Epoch: {}, Mean Accuracy: {:.2f}%'.format(epoch, accuracy))
global_step = (epoch - 1) * num_training_steps_per_epoch
if metrics_debug:
print(met.metrics_report())
logger.info(met.metrics_report())
logger.info('saving checkpoint. epoch: {}'.format(epoch))
torch.save(model_parallel, os.path.join(logdir,'model_parallel_chkpt.pt'))
logger.info('checkpoint saved. epoch: {}'.format(epoch))
test_utils.close_summary_writer(writer)
return accuracy
def main():
parser = utils.get_args_parser_with_general_args()
parser.add_argument('--one_tpu', action='store_true', help="Run on one tpu core for degugging. Makes it easy to use break points")
parser.add_argument('--tpu_report', action='store_true', help="Print xla metric report")
args = parser.parse_args()
utils.init(args) # set seeds, init logger, prepare output directory
devices = tpu_xm.get_xla_supported_devices()
if args.one_tpu:
devices = [devices[0]]
n_tpu = len(devices)
logging.info(f'Found {n_tpu} TPU cores')
tokenizer = AutoTokenizer.from_pretrained(args.bert_model)
tokenizer.save_pretrained(args.output_dir)
args.start_epoch = utils.prepare_last_checkpoint(args.bert_model)
model = AutoModelWithLMHead.from_pretrained(args.bert_model) # Only Masked Language Modeling
logging.info(f"Saving initial checkpoint to: {args.output_dir}")
#model.save_pretrained(args.output_dir) #TODO: why does this break?
#xm.save(model.state_dict(), args.output_dir)
model = tpu_dp.DataParallel(model, device_ids=devices)
num_data_epochs, num_train_optimization_steps= utils.get_dataset_stats(args, n_tpu)
def tpu_training_loop(model, loader, device, context):
""" Called by torch_xla_py.data_parallel. This function is executed on each core of the TPU once per epoch"""
param_optimizer = list(model.named_parameters())
no_decay = ['bias', 'LayerNorm.bias', 'LayerNorm.weight']
optimizer_grouped_parameters = [
{'params': [p for n, p in param_optimizer if not any(nd in n for nd in no_decay)], 'weight_decay': 0.01},
{'params': [p for n, p in param_optimizer if any(nd in n for nd in no_decay)], 'weight_decay': 0.0}
]
# one optimizer and scheduler per TPU core. Both objects are saved in `context` to be reused the next epoch
optimizer = context.getattr_or(
'optimizer',
AdamW(optimizer_grouped_parameters, lr=args.learning_rate, eps=args.adam_epsilon, betas=tuple(args.betas)))
# derive warmup info
if args.warmup_proportion is not None:
warmup_steps = int(args.warmup_proportion * num_train_optimization_steps + 0.5)
elif args.warmup_steps is not None:
warmup_steps = args.warmup_steps
else:
raise Exception('What is the warmup?? Specify either warmup proportion or steps')
scheduler = context.getattr_or(
'scheduler',
WarmupLinearSchedule(optimizer, warmup_steps=warmup_steps, t_total=num_train_optimization_steps))
tr_loss = None
pbar = None
if str(pbar_device) == str(device): # All threads are in sync. Use progress bar only on one of them
pbar = tqdm(total=int(pbar_steps), desc=f"device {device}", dynamic_ncols=True)
tracker = tpu_xm.RateTracker()
model.train()
for step, batch in enumerate(loader):
input_ids, input_mask, segment_ids, lm_label_ids, _ = batch
outputs = model(input_ids, segment_ids, input_mask, lm_label_ids)
loss = outputs[0]
if args.gradient_accumulation_steps > 1:
loss = loss / args.gradient_accumulation_steps
loss.sum().backward() # for multiple tensors
tracker.add(args.train_batch_size)
tr_loss = loss * args.gradient_accumulation_steps if step == 0 else tr_loss + loss * args.gradient_accumulation_steps
if pbar is not None:
pbar.update(1)
# pbar.set_description(desc=f'LR: {scheduler.get_lr()}')
if (step + 1) % args.gradient_accumulation_steps == 0:
tpu_xm.optimizer_step(optimizer)
prev_lr = scheduler.get_last_lr()[0]
scheduler.step()
curr_lr = scheduler.get_last_lr()[0]
if args.track_learning_rate:
if pbar is not None:
pbar.set_description(f"Prev LR: {prev_lr} Curr LR: {curr_lr}")
optimizer.zero_grad()
return tr_loss.sum().item() / step # `.item()` requires a trip from TPU to CPU, which is very slow. Use it only once per epoch=
for epoch in range(args.start_epoch, args.epochs):
# Load one training file into memory
epoch_dataset = utils.PregeneratedDataset(epoch=epoch, training_path=args.pregenerated_data, tokenizer=tokenizer,
num_data_epochs=num_data_epochs, reduce_memory=args.reduce_memory)
train_sampler = RandomSampler(epoch_dataset)
train_dataloader = DataLoader(epoch_dataset, sampler=train_sampler, batch_size=args.train_batch_size)
pbar_device = devices[0]
pbar_steps = utils.compute_num_steps_in_epoch(num_samples=train_sampler.num_samples,
batch_size=args.train_batch_size,
grad_accum_steps=1, # the pbar steps should not take into account grad accumulation steps
n_tpu=n_tpu)
logging.info(f'start training, epoch {epoch} on {len(devices)} cores for {pbar_steps} steps')
start = time.time()
losses = model(tpu_training_loop, train_dataloader) # calls `tpu_training_loop` multiple times, once per TPU core
logging.info(f'Epoch {epoch} took {round(time.time() - start, 2)} seconds. Average loss: {sum(losses)/len(losses)}')
utils.save_checkpoint(model._models[0], epoch, args.output_dir)
if args.tpu_report:
logging.info(torch_xla._XLAC._xla_metrics_report())
