1-late SGD for PyTorch ImageNet example with Horovod

pytorch_imagenet_resnet50_1late.py

from __future__ import print_function

import argparse
import torch.backends.cudnn as cudnn
import torch.nn.functional as F
import torch.optim as optim
import torch.utils.data.distributed
from torchvision import datasets, transforms, models
import horovod.torch as hvd
import tensorboardX
import os
from tqdm import tqdm

# Training settings
parser = argparse.ArgumentParser(description='PyTorch ImageNet Example',
                                 formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('--train-dir', default=os.path.expanduser('~/imagenet/train'),
                    help='path to training data')
parser.add_argument('--val-dir', default=os.path.expanduser('~/imagenet/validation'),
                    help='path to validation data')
parser.add_argument('--log-dir', default='./logs',
                    help='tensorboard log directory')
parser.add_argument('--checkpoint-format', default='./checkpoint-{epoch}.pth.tar',
                    help='checkpoint file format')
parser.add_argument('--batch-size', type=int, default=32,
                    help='input batch size for training')
parser.add_argument('--val-batch-size', type=int, default=32,
                    help='input batch size for validation')
parser.add_argument('--epochs', type=int, default=90,
                    help='number of epochs to train')
parser.add_argument('--base-lr', type=float, default=0.0125,
                    help='learning rate for a single GPU')
parser.add_argument('--warmup-epochs', type=float, default=5,
                    help='number of warmup epochs')
parser.add_argument('--momentum', type=float, default=0.9,
                    help='SGD momentum')
parser.add_argument('--wd', type=float, default=0.00005,
                    help='weight decay')
parser.add_argument('--no-cuda', action='store_true', default=False,
                    help='disables CUDA training')
parser.add_argument('--seed', type=int, default=42,
                    help='random seed')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()

hvd.init()
torch.manual_seed(args.seed)

if args.cuda:
    # Horovod: pin GPU to local rank.
    torch.cuda.set_device(hvd.local_rank())
    torch.cuda.manual_seed(args.seed)

cudnn.benchmark = True

# If set > 0, will resume training from a given checkpoint.
resume_from_epoch = 0
for try_epoch in range(args.epochs, 0, -1):
    if os.path.exists(args.checkpoint_format.format(epoch=try_epoch)):
        resume_from_epoch = try_epoch
        break

# Horovod: broadcast resume_from_epoch from rank 0 (which will have
# checkpoints) to other ranks.
resume_from_epoch = hvd.broadcast(torch.tensor(resume_from_epoch), root_rank=0,
                                  name='resume_from_epoch').item()

# Horovod: print logs on the first worker.
verbose = 1 if hvd.rank() == 0 else 0

# Horovod: write TensorBoard logs on first worker.
log_writer = tensorboardX.SummaryWriter(args.log_dir) if hvd.rank() == 0 else None


kwargs = {'num_workers': 4, 'pin_memory': True} if args.cuda else {}
train_dataset = \
    datasets.ImageFolder(args.train_dir,
                         transform=transforms.Compose([
                             transforms.RandomResizedCrop(224),
                             transforms.RandomHorizontalFlip(),
                             transforms.ToTensor(),
                             transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                  std=[0.229, 0.224, 0.225])
                         ]))
# Horovod: use DistributedSampler to partition data among workers. Manually specify
# `num_replicas=hvd.size()` and `rank=hvd.rank()`.
train_sampler = torch.utils.data.distributed.DistributedSampler(
    train_dataset, num_replicas=hvd.size(), rank=hvd.rank())
train_loader = torch.utils.data.DataLoader(
    train_dataset, batch_size=args.batch_size, sampler=train_sampler, **kwargs)

val_dataset = \
    datasets.ImageFolder(args.val_dir,
                         transform=transforms.Compose([
                             transforms.Resize(256),
                             transforms.CenterCrop(224),
                             transforms.ToTensor(),
                             transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                                  std=[0.229, 0.224, 0.225])
                         ]))
val_sampler = torch.utils.data.distributed.DistributedSampler(
    val_dataset, num_replicas=hvd.size(), rank=hvd.rank())
val_loader = torch.utils.data.DataLoader(val_dataset, batch_size=args.val_batch_size,
                                         sampler=val_sampler, **kwargs)


# Set up standard ResNet-50 model.
model = models.resnet50()

if args.cuda:
    # Move model to GPU.
    model.cuda()

# Horovod: scale learning rate by the number of GPUs.
optimizer = optim.SGD(model.parameters(), lr=args.base_lr * hvd.size(),
                      momentum=args.momentum, weight_decay=args.wd)


# Custom Distributed Optimizer that does 1-late gradient.
class _DistributedOptimizer(torch.optim.Optimizer):
    def __init__(self, params, named_parameters=None):
        super(self.__class__, self).__init__(params)

        if named_parameters is not None:
            named_parameters = list(named_parameters)
        else:
            named_parameters = []

        # make sure that named_parameters are tuples
        if any([not isinstance(p, tuple) for p in named_parameters]):
            raise ValueError('named_parameters should be a sequence of '
                             'tuples (name, parameter), usually produced by '
                             'model.named_parameters().')

        self._parameter_names = {v: k for k, v
                                 in sorted(named_parameters)}
        self._handles = {}
        self._grad_accs = []
        self._last_grad = {}

        if hvd.size() > 1:
            self._register_hooks()

    def _register_hooks(self):
        for param_group in self.param_groups:
            for p in param_group['params']:
                if p.requires_grad:
                    p_tmp = p.expand_as(p)
                    grad_acc = p_tmp.grad_fn.next_functions[0][0]
                    grad_acc.register_hook(self._make_hook(p))
                    self._grad_accs.append(grad_acc)

    def _make_hook(self, p):
        def hook(*ignore):
            assert not p.grad.requires_grad
            name = self._parameter_names.get(p)
            new_grad = p.grad.data.clone()
            if p in self._handles:
                last_grad = hvd.synchronize(self._handles[p])
            else:
                last_grad = torch.zeros_like(new_grad)
            self._last_grad[p] = last_grad
            handle = hvd.allreduce_async_(new_grad, average=True, name=name)
            self._handles[p] = handle
        return hook

    def step(self, closure=None):
        for p, lg in self._last_grad.items():
            p.grad.data.set_(lg)
        return super(self.__class__, self).step(closure)


def DistributedOptimizer(optimizer, named_parameters=None):
    """
    An optimizer that wraps another torch.optim.Optimizer, using an allreduce to
    average gradient values before applying gradients to model weights.

    Allreduce operations are executed after each gradient is computed by `loss.backward()`
    in parallel with each other. The `step()` method ensures that all allreduce operations are
    finished before applying gradients to the model.

    DistributedOptimizer exposes the `synchronize()` method, which forces allreduce operations
    to finish before continuing the execution. It's useful in conjunction with gradient
    clipping, or other operations that modify gradients in place before `step()` is executed.

    Example of gradient clipping:
    ```
    output = model(data)
    loss = F.nll_loss(output, target)
    loss.backward()
    optimizer.synchronize()
    torch.nn.utils.clip_grad_norm(model.parameters(), args.clip)
    optimizer.step()
    ```

    Arguments:
        optimizer: Optimizer to use for computing gradients and applying updates.
        named_parameters: A mapping between parameter names and values. Used for naming of
                          allreduce operations. Typically just `model.named_parameters()`.
    """
    # We dynamically create a new class that inherits from the optimizer that was passed in.
    # The goal is to override the `step()` method with an allreduce implementation.
    cls = type(optimizer.__class__.__name__, (optimizer.__class__,),
               dict(_DistributedOptimizer.__dict__))
    return cls(optimizer.param_groups, named_parameters)


# Horovod: wrap optimizer with DistributedOptimizer.
optimizer = DistributedOptimizer(
    optimizer, named_parameters=model.named_parameters())

# Restore from a previous checkpoint, if initial_epoch is specified.
# Horovod: restore on the first worker which will broadcast weights to other workers.
if resume_from_epoch > 0 and hvd.rank() == 0:
    filepath = args.checkpoint_format.format(epoch=resume_from_epoch)
    checkpoint = torch.load(filepath)
    model.load_state_dict(checkpoint['model'])
    optimizer.load_state_dict(checkpoint['optimizer'])

# Horovod: broadcast parameters & optimizer state.
hvd.broadcast_parameters(model.state_dict(), root_rank=0)
hvd.broadcast_optimizer_state(optimizer, root_rank=0)

def train(epoch):
    model.train()
    train_sampler.set_epoch(epoch)
    train_loss = Metric('train_loss')
    train_accuracy = Metric('train_accuracy')

    with tqdm(total=len(train_loader),
              desc='Train Epoch     #{}'.format(epoch + 1),
              disable=not verbose) as t:
        for batch_idx, (data, target) in enumerate(train_loader):
            adjust_learning_rate(epoch, batch_idx)

            if args.cuda:
                data, target = data.cuda(), target.cuda()
            optimizer.zero_grad()
            output = model(data)
            loss = F.cross_entropy(output, target)
            loss.backward()
            optimizer.step()

            train_loss.update(loss)
            train_accuracy.update(accuracy(output, target))
            t.set_postfix({'loss': train_loss.avg.item(),
                           'accuracy': 100. * train_accuracy.avg.item()})
            t.update(1)

    if log_writer:
        log_writer.add_scalar('train/loss', train_loss.avg, epoch)
        log_writer.add_scalar('train/accuracy', train_accuracy.avg, epoch)


def validate(epoch):
    model.eval()
    val_loss = Metric('val_loss')
    val_accuracy = Metric('val_accuracy')

    with tqdm(total=len(val_loader),
              desc='Validate Epoch  #{}'.format(epoch + 1),
              disable=not verbose) as t:
        with torch.no_grad():
            for data, target in val_loader:
                if args.cuda:
                    data, target = data.cuda(), target.cuda()
                output = model(data)

                val_loss.update(F.cross_entropy(output, target))
                val_accuracy.update(accuracy(output, target))
                t.set_postfix({'loss': val_loss.avg.item(),
                               'accuracy': 100. * val_accuracy.avg.item()})
                t.update(1)

    if log_writer:
        log_writer.add_scalar('val/loss', val_loss.avg, epoch)
        log_writer.add_scalar('val/accuracy', val_accuracy.avg, epoch)


# Horovod: using `lr = base_lr * hvd.size()` from the very beginning leads to worse final
# accuracy. Scale the learning rate `lr = base_lr` ---> `lr = base_lr * hvd.size()` during
# the first five epochs. See https://arxiv.org/abs/1706.02677 for details.
# After the warmup reduce learning rate by 10 on the 30th, 60th and 80th epochs.
def adjust_learning_rate(epoch, batch_idx):
    if epoch < args.warmup_epochs:
        epoch += float(batch_idx + 1) / len(train_loader)
        lr_adj = 1. / hvd.size() * (epoch * (hvd.size() - 1) / args.warmup_epochs + 1)
    elif epoch < 30:
        lr_adj = 1.
    elif epoch < 60:
        lr_adj = 1e-1
    elif epoch < 80:
        lr_adj = 1e-2
    else:
        lr_adj = 1e-3
    for param_group in optimizer.param_groups:
        param_group['lr'] = args.base_lr * hvd.size() * lr_adj


def accuracy(output, target):
    # get the index of the max log-probability
    pred = output.max(1, keepdim=True)[1]
    return pred.eq(target.view_as(pred)).cpu().float().mean()


def save_checkpoint(epoch):
    if hvd.rank() == 0:
        filepath = args.checkpoint_format.format(epoch=epoch + 1)
        state = {
            'model': model.state_dict(),
            'optimizer': optimizer.state_dict(),
        }
        torch.save(state, filepath)


# Horovod: average metrics from distributed training.
class Metric(object):
    def __init__(self, name):
        self.name = name
        self.sum = torch.tensor(0.)
        self.n = torch.tensor(0.)

    def update(self, val):
        self.sum += hvd.allreduce(val.cpu(), name=self.name)
        self.n += 1

    @property
    def avg(self):
        return self.sum / self.n


for epoch in range(resume_from_epoch, args.epochs):
    train(epoch)
    validate(epoch)
    save_checkpoint(epoch)