main_imagenet_binary.py

"""
Source: https://github.com/pytorch/examples/tree/master/imagenet
License: BSD 3-clause
"""
import argparse
from bisect import bisect
from datetime import datetime
import json
import os
import random
import shutil
import sys
import time
import warnings

import torch
import torch.nn as nn
import torch.nn.parallel
import torch.backends.cudnn as cudnn
import torch.distributed as dist
import torch.optim
import torch.multiprocessing as mp
import torch.utils.data
import torch.utils.data.distributed
import torchvision.transforms as transforms
import torchvision.datasets as datasets
import torchvision.models as models
from torch.utils.tensorboard import SummaryWriter

from models.resnet_imagenet import resnet18
from models.resnet_imagenet_binary import binarize_all, resnet18 as resnet18_binary
from models.resnet_imagenet_binary import BinaryLinear, BinaryConv2d
from models.resnet_imagenet_binary import restore_fp_weight

from mlmt.time_stamp import TimeStamp

#model_names = sorted(name for name in models.__dict__
#    if name.islower() and not name.startswith("__")
#    and callable(models.__dict__[name]))

parser = argparse.ArgumentParser(description='PyTorch ImageNet Training')
parser.add_argument('data', metavar='DIR',
                    help='path to ImageNet dataset')
"""
parser.add_argument('-a', '--arch', metavar='ARCH', default='resnet18',
                    choices=model_names,
                    help='model architecture: ' +
                        ' | '.join(model_names) +
                        ' (default: resnet18)')
"""
parser.add_argument('-j', '--workers', default=4, type=int, metavar='N',
                    help='number of data loading workers (default: 4)')
parser.add_argument('--epochs', default=90, type=int, metavar='N',
                    help='number of total epochs to run')
parser.add_argument('--start-epoch', default=0, type=int, metavar='N',
                    help='manual epoch number (useful on restarts)')
parser.add_argument('-b', '--batch-size', default=256, type=int,
                    metavar='N',
                    help='mini-batch size (default: 256), this is the total '
                         'batch size of all GPUs on the current node when '
                         'using Data Parallel or Distributed Data Parallel')
parser.add_argument('--lr', '--learning-rate', default=0.1, type=float,
                    metavar='LR', help='initial learning rate', dest='lr')
parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
                    help='momentum')
parser.add_argument('--wd', '--weight-decay', default=1e-4, type=float,
                    metavar='W', help='weight decay (default: 1e-4)',
                    dest='weight_decay')
parser.add_argument('-p', '--print-freq', default=10, type=int,
                    metavar='N', help='print frequency (default: 10)')
parser.add_argument('--resume', default='', type=str, metavar='PATH',
                    help='path to latest checkpoint (default: none)')
parser.add_argument('-e', '--evaluate', dest='evaluate', action='store_true',
                    help='evaluate model on validation set')
#parser.add_argument('--pretrained', dest='pretrained', action='store_true',
#                    help='use pre-trained model')
parser.add_argument('--world-size', default=-1, type=int,
                    help='number of nodes for distributed training')
parser.add_argument('--rank', default=-1, type=int,
                    help='node rank for distributed training')
parser.add_argument('--dist-url', default='tcp://224.66.41.62:23456', type=str,
                    help='url used to set up distributed training')
parser.add_argument('--dist-backend', default='nccl', type=str,
                    help='distributed backend')
parser.add_argument('--seed', default=None, type=int,
                    help='seed for initializing training. ')
parser.add_argument('--gpu', default=None, type=int,
                    help='GPU id to use.')
parser.add_argument('--multiprocessing-distributed', action='store_true',
                    help='Use multi-processing distributed training to launch '
                         'N processes per node, which has N GPUs. This is the '
                         'fastest way to use PyTorch for either single node or '
                         'multi node data parallel training')

parser.add_argument('--save-checkpoint', action='store_true',
                    help='Turn on checkpointing (default: off)')
parser.add_argument('--checkpoint-interval', default=8, type=int,
                    help='Checkpoint interval (default: 8 epochs)')
parser.add_argument('--dry-run', action='store_true',
                    help='Dry run (very short) training (default: off)')

parser.add_argument('logdir', metavar='LOGDIR',
                    help='directory to write tensorboard logs (default: ./logs)')
parser.add_argument('--binarize', action='store_true',
                    help='Turn on binarization (default: off)')
parser.add_argument('--layers-json', default='./models/resnet21_layers_1.json', type=str,
                    help='Layers JSON file')
parser.add_argument('--reverse-layer-binarization', action='store_true',
                    help='Reverse layer binarization (default: off)')
parser.add_argument('--all-binary', action='store_true',
                    help='Turn on binarization for all layers (default: off)')
parser.add_argument('--epochs-per-layer', type=int, default=20, metavar='EPL',
                    help='Epochs per layer for layer-by-layer binarization (default: 20)')
parser.add_argument('--lr-milestones', type=str, default='[84]',
                    help='Reduce learning rate by 10 at beginning of each epoch in this list')


best_acc1 = 0


def save_args(args, writer):
    print("argv: {}".format(sys.argv))
    writer.add_text('argv', str(sys.argv), 0)

    print("args:")
    options = vars(args)
    for k, v in options.items():
        #print("  {} {} {} {}".format(k, v, type(k), type(v)))
        print("  {} {}".format(k, v))
        writer.add_text("args/"+k, str(v), 0)

def experiment_name_from(args) -> str:
    name = f'lr{args.lr}-ep{args.epochs}'
    if args.binarize:
        if args.all_binary:
            name += f'-all_binary'
        else:
            name += f'-epl{args.epochs_per_layer}-{os.path.basename(args.layers_json)}'
        if args.reverse_layer_binarization:
            name += '-reversed_layer'
    if args.dry_run:
        name += '-dryrun'
    return name

def experiment_id() -> str:
    return datetime.now().strftime('%Y-%m-%d_%H%M%S_%f')


def inspect_model(model) -> None:
    print("=============================================")
    """
    if hasattr(model, 'binarize_layer'):
        print("model has binarize_layer()")
    else:
        print("model has no binarize_layer()")
    """
    status = '' if hasattr(model, 'binarize_layer') else 'no'
    print(f'model has {status} binarize_layer')

    status = '' if hasattr(model, '_make_layer') else 'no'
    print(f'model has {status} _make_layer')

    print("Has binarize_?")
    found = []
    for idx, layer in enumerate(model.modules()):
        if hasattr(layer, 'binarize_'):
            found.append(idx)
    print(found)

    print("Has binarize_weight_()?")
    found = []
    for idx, layer in enumerate(model.modules()):
        if hasattr(layer, 'binarize_weight_'):
            found.append(idx)
    print(found)


def show_layers(model: nn.Module) -> None:
    print('---------------------------------------------')
    for idx, layer in enumerate(model.modules()):
        print(idx, layer)


def show_binarized(epoch: int, layers: list, nets: list) -> None:
    """Can deprecate"""
    status = ""
    layers_label = ""
    for n in layers:
        layers_label += "{:3}".format(n)
        status += "{:3}".format(nets[n].binarize_)
    print(f'Epoch  {epoch}')
    print(f'Layer  {layers_label}')
    print(f'Binary {status}')


def show_binarization(model: nn.Module) -> None:
    nets = list(model.modules())
    status = ""
    layers_label = ""
    for n, module in enumerate(nets):
        if isinstance(module, (BinaryLinear, BinaryConv2d)):
            layers_label += "{:3}".format(n)
            status += "{:3}".format(module.binarize_)
    if len(status) > 0:
        print(f'Layer  {layers_label}')
        print(f'Binary {status}')
    else:
        print('No binarization')


def binarize_layer(
        epoch: int, 
        layers: list, 
        model: nn.Module, 
        epochs_per_layer: int, 
        reverse: bool,
) -> None:
    """Set layer binarization on a forward-first schedule.
    Call before training an epoch.

    Parameters
    ----------
    epoch : int
        Must start from 0.
    """
    nets = list(model.modules())
    ending = epoch // epochs_per_layer
    n = 0
    while True: # do ... while()
        if n < len(layers):
            if reverse:
                m = nets[layers[-1-n]]
            else:
                m = nets[layers[n]]
            if isinstance(m, (BinaryLinear, BinaryConv2d)):
                m.binarize_ = True
        n = n + 1
        if n > ending: break


def load_layers(json_file: str):
    print(f"Loading layers from {json_file}...")
    with open(json_file, 'r') as fp:
        return json.load(fp)



def main():
    args = parser.parse_args()

    if args.multiprocessing_distributed:
        raise ValueError("--multiprocessing-distributed not supported")

    if args.seed is not None:
        random.seed(args.seed)
        torch.manual_seed(args.seed)
        cudnn.deterministic = True
        warnings.warn('You have chosen to seed training. '
                      'This will turn on the CUDNN deterministic setting, '
                      'which can slow down your training considerably! '
                      'You may see unexpected behavior when restarting '
                      'from checkpoints.')

    if args.gpu is not None:
        warnings.warn('You have chosen a specific GPU. This will completely '
                      'disable data parallelism.')

    if args.dist_url == "env://" and args.world_size == -1:
        args.world_size = int(os.environ["WORLD_SIZE"])

    args.distributed = args.world_size > 1 or args.multiprocessing_distributed
    if args.distributed:
        raise ValueError("Distributed training not supported")


    ngpus_per_node = torch.cuda.device_count()
    if args.multiprocessing_distributed:
        # Since we have ngpus_per_node processes per node, the total world_size
        # needs to be adjusted accordingly
        args.world_size = ngpus_per_node * args.world_size
        # Use torch.multiprocessing.spawn to launch distributed processes: the
        # main_worker process function
        mp.spawn(main_worker, nprocs=ngpus_per_node, args=(ngpus_per_node, args))
    else:
        # Simply call main_worker function
        main_worker(args.gpu, ngpus_per_node, args)


def main_worker(gpu, ngpus_per_node, args):
    print(f'gpu {gpu}')
    print(f'ngpus_per_node {ngpus_per_node}')
    global best_acc1
    args.gpu = gpu

    if (args.distributed and gpu==0) or not args.distributed:
        print("Creating SummaryWriter ...")
        experiment_dir = os.path.join(args.logdir, experiment_name_from(args) + '_' + experiment_id())
        if not os.path.exists(experiment_dir):
            os.makedirs(experiment_dir)
        tb_writer = SummaryWriter(experiment_dir)
        save_args(args, tb_writer)

    if args.gpu is not None:
        print("Use GPU: {} for training".format(args.gpu))

    if args.distributed:
        if args.dist_url == "env://" and args.rank == -1:
            args.rank = int(os.environ["RANK"])
        if args.multiprocessing_distributed:
            # For multiprocessing distributed training, rank needs to be the
            # global rank among all the processes
            args.rank = args.rank * ngpus_per_node + gpu
        dist.init_process_group(backend=args.dist_backend, init_method=args.dist_url,
                                world_size=args.world_size, rank=args.rank)
    # create model
    """
    if args.pretrained:
        print("=> using pre-trained model '{}'".format(args.arch))
        model = models.__dict__[args.arch](pretrained=True)
    else:
        print("=> creating model '{}'".format(args.arch))
        model = models.__dict__[args.arch]()
    """
    if args.binarize:
        print("=> creating model binarized resnet18")
        model = resnet18_binary()
        #inspect_model(model)
        show_layers(model)
    else:
        print("=> creating model resnet18")
        model = resnet18()

    # load layers_json
    layers = load_layers(args.layers_json)


    if not torch.cuda.is_available():
        print('using CPU, this will be slow')
    elif args.distributed:
        # For multiprocessing distributed, DistributedDataParallel constructor
        # should always set the single device scope, otherwise,
        # DistributedDataParallel will use all available devices.
        if args.gpu is not None:
            torch.cuda.set_device(args.gpu)
            model.cuda(args.gpu)
            # When using a single GPU per process and per
            # DistributedDataParallel, we need to divide the batch size
            # ourselves based on the total number of GPUs we have
            args.batch_size = int(args.batch_size / ngpus_per_node)
            args.workers = int((args.workers + ngpus_per_node - 1) / ngpus_per_node)
            model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu])
            print('model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu])')
            inspect_model(model)
            #show_layers(model)
        else:
            model.cuda()
            # DistributedDataParallel will divide and allocate batch_size to all
            # available GPUs if device_ids are not set
            model = torch.nn.parallel.DistributedDataParallel(model)
    elif args.gpu is not None:
        torch.cuda.set_device(args.gpu)
        model = model.cuda(args.gpu)
        print('model = model.cuda(args.gpu)')
    else:
        # DataParallel will divide and allocate batch_size to all available GPUs
        if args.arch.startswith('alexnet') or args.arch.startswith('vgg'):
            model.features = torch.nn.DataParallel(model.features)
            model.cuda()
        else:
            model = torch.nn.DataParallel(model).cuda()

    # define loss function (criterion) and optimizer
    criterion = nn.CrossEntropyLoss().cuda(args.gpu)

    optimizer = torch.optim.SGD(model.parameters(), args.lr,
                                momentum=args.momentum,
                                weight_decay=args.weight_decay)

    # optionally resume from a checkpoint
    if args.resume:
        if os.path.isfile(args.resume):
            print("=> loading checkpoint '{}'".format(args.resume))
            if args.gpu is None:
                checkpoint = torch.load(args.resume)
            else:
                # Map model to be loaded to specified single gpu.
                loc = 'cuda:{}'.format(args.gpu)
                checkpoint = torch.load(args.resume, map_location=loc)
            args.start_epoch = checkpoint['epoch']
            best_acc1 = checkpoint['best_acc1']
            if args.gpu is not None:
                # best_acc1 may be from a checkpoint from a different GPU
                best_acc1 = best_acc1.to(args.gpu)
            model.load_state_dict(checkpoint['state_dict'])
            optimizer.load_state_dict(checkpoint['optimizer'])
            print("=> loaded checkpoint '{}' (epoch {})"
                  .format(args.resume, checkpoint['epoch']))
        else:
            print("=> no checkpoint found at '{}'".format(args.resume))
            raise ValueError

    cudnn.benchmark = True

    # Data loading code
    traindir = os.path.join(args.data, 'train')
    valdir = os.path.join(args.data, 'val')
    normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406],
                                     std=[0.229, 0.224, 0.225])

    train_dataset = datasets.ImageFolder(
        traindir,
        transforms.Compose([
            transforms.RandomResizedCrop(224),
            transforms.RandomHorizontalFlip(),
            transforms.ToTensor(),
            normalize,
        ]))

    if args.distributed:
        train_sampler = torch.utils.data.distributed.DistributedSampler(train_dataset)
    else:
        train_sampler = None

    train_loader = torch.utils.data.DataLoader(
        train_dataset, batch_size=args.batch_size, shuffle=(train_sampler is None),
        num_workers=args.workers, pin_memory=True, sampler=train_sampler)

    val_loader = torch.utils.data.DataLoader(
        datasets.ImageFolder(valdir, transforms.Compose([
            transforms.Resize(256),
            transforms.CenterCrop(224),
            transforms.ToTensor(),
            normalize,
        ])),
        batch_size=args.batch_size, shuffle=False,
        num_workers=args.workers, pin_memory=True)

    if args.evaluate:
        validate(val_loader, model, criterion, args)
        return

    lr_milestones = json.loads(args.lr_milestones)
    for epoch in range(args.start_epoch, args.epochs):
        if args.distributed:
            train_sampler.set_epoch(epoch)

        learning_rate = adjust_learning_rate(
            optimizer = optimizer, 
            epoch = epoch, 
            base_lr = args.lr, 
            lr_milestones = lr_milestones,
        )
        if (args.distributed and gpu==0) or not args.distributed:
            print(f'Epoch {epoch} LR {learning_rate}')
            tb_writer.add_scalar('train/learning_rate', learning_rate, epoch)

        # train for one epoch
        ts = TimeStamp()
        if args.all_binary:
            print("Binarize all")
            model.apply(binarize_all)
        else:
            binarize_layer(
                epoch = epoch, 
                layers = layers,
                model = model,
                epochs_per_layer = args.epochs_per_layer,
                reverse = args.reverse_layer_binarization,
            )
            show_binarization(model)
        train_acc1, train_acc5 = train(train_loader, model, criterion, optimizer, epoch, args)
        if (args.distributed and gpu==0) or not args.distributed:
            tb_writer.add_scalar('train/duration', ts.elapsed()/60.0, epoch)
            tb_writer.add_scalar('train/acc1', train_acc1, epoch)
            tb_writer.add_scalar('train/acc5', train_acc5, epoch)

        # evaluate on validation set
        ts = TimeStamp()
        acc1, acc5 = validate(val_loader, model, criterion, args)

        # remember best acc@1 and save checkpoint
        is_best = acc1 > best_acc1
        best_acc1 = max(acc1, best_acc1)
        if (args.distributed and gpu==0) or not args.distributed:
            tb_writer.add_scalar('val/duration', ts.elapsed()/60.0, epoch)
            tb_writer.add_scalar('val/acc1', acc1, epoch)
            #tb_writer.add_scalar('val/best_acc1', best_acc1, epoch)
            tb_writer.add_scalar('val/acc5', acc5, epoch)

        if not args.multiprocessing_distributed or (args.multiprocessing_distributed
                and args.rank % ngpus_per_node == 0):
            if args.save_checkpoint:
                save_checkpoint(
                    state = {
                        'epoch': epoch + 1,
                        'state_dict': model.state_dict(),
                        'best_acc1': best_acc1,
                        'optimizer' : optimizer.state_dict(),
                    }, 
                    is_best = is_best,
                    experiment_dir = experiment_dir,
                    epoch = epoch,
                    interval = args.checkpoint_interval,
                )

    
    # Flush tensorboard data
    if (args.distributed and gpu==0) or not args.distributed:
        tb_writer.close()



def train(train_loader, model, criterion, optimizer, epoch, args):
    batch_time = AverageMeter('Time', ':6.3f')
    data_time = AverageMeter('Data', ':6.3f')
    losses = AverageMeter('Loss', ':.4e')
    top1 = AverageMeter('Acc@1', ':6.2f')
    top5 = AverageMeter('Acc@5', ':6.2f')
    progress = ProgressMeter(
        len(train_loader),
        [batch_time, data_time, losses, top1, top5],
        prefix="Epoch: [{}]".format(epoch))

    # switch to train mode
    model.train()

    end = time.time()
    for i, (images, target) in enumerate(train_loader):
        # measure data loading time
        data_time.update(time.time() - end)

        if args.gpu is not None:
            images = images.cuda(args.gpu, non_blocking=True)
        if torch.cuda.is_available():
            target = target.cuda(args.gpu, non_blocking=True)

        # compute output
        output = model(images)
        loss = criterion(output, target)

        # measure accuracy and record loss
        acc1, acc5 = accuracy(output, target, topk=(1, 5))
        losses.update(loss.item(), images.size(0))
        top1.update(acc1[0], images.size(0))
        top5.update(acc5[0], images.size(0))

        # compute gradient and do SGD step
        optimizer.zero_grad()
        loss.backward()
        model.apply(restore_fp_weight)
        optimizer.step() # parameter update

        # measure elapsed time
        batch_time.update(time.time() - end)
        end = time.time()

        if i % args.print_freq == 0:
            progress.display(i)
        if args.dry_run:
            if i > 1: break

    return top1.avg, top5.avg


def validate(val_loader, model, criterion, args):
    batch_time = AverageMeter('Time', ':6.3f')
    losses = AverageMeter('Loss', ':.4e')
    top1 = AverageMeter('Acc@1', ':6.2f')
    top5 = AverageMeter('Acc@5', ':6.2f')
    progress = ProgressMeter(
        len(val_loader),
        [batch_time, losses, top1, top5],
        prefix='Test: ')

    # switch to evaluate mode
    model.eval()

    with torch.no_grad():
        end = time.time()
        for i, (images, target) in enumerate(val_loader):
            if args.gpu is not None:
                images = images.cuda(args.gpu, non_blocking=True)
            if torch.cuda.is_available():
                target = target.cuda(args.gpu, non_blocking=True)

            # compute output
            output = model(images)
            loss = criterion(output, target)

            # measure accuracy and record loss
            acc1, acc5 = accuracy(output, target, topk=(1, 5))
            losses.update(loss.item(), images.size(0))
            top1.update(acc1[0], images.size(0))
            top5.update(acc5[0], images.size(0))

            # measure elapsed time
            batch_time.update(time.time() - end)
            end = time.time()

            if i % args.print_freq == 0:
                progress.display(i)
            if args.dry_run:
                if i > 1: break

        # Outside val_loader to avoid duplicate weight binarization.
        model.apply(restore_fp_weight)
        
        # TODO: this should also be done with the ProgressMeter
        print(' * Acc@1 {top1.avg:.3f} Acc@5 {top5.avg:.3f}'
              .format(top1=top1, top5=top5))

    return top1.avg, top5.avg


def save_checkpoint(*, state, is_best, experiment_dir, epoch, interval):
    """Save checkpoints
    Save the running checkpoint.
    Save checkpoints at given interval.
    Save the running best checkpoint.
    """
    filename = os.path.join(experiment_dir, 'checkpoint.pth.tar')
    print(f'Saving current: {filename}')
    torch.save(state, filename)
    if is_best:
        best = os.path.join(experiment_dir, 'model_best.pth.tar')
        print(f'Saving best: {best}')
        shutil.copyfile(filename, best)
    if (interval > 0) and ((epoch+1) % interval == 0):
        current = os.path.join(experiment_dir, f'checkpoint{epoch}.pth.tar')
        print(f'Saving interval: {current}')
        shutil.copyfile(filename, current)


class AverageMeter(object):
    """Computes and stores the average and current value"""
    def __init__(self, name, fmt=':f'):
        self.name = name
        self.fmt = fmt
        self.reset()

    def reset(self):
        self.val = 0
        self.avg = 0
        self.sum = 0
        self.count = 0

    def update(self, val, n=1):
        self.val = val
        self.sum += val * n
        self.count += n
        self.avg = self.sum / self.count

    def __str__(self):
        fmtstr = '{name} {val' + self.fmt + '} ({avg' + self.fmt + '})'
        return fmtstr.format(**self.__dict__)


class ProgressMeter(object):
    def __init__(self, num_batches, meters, prefix=""):
        self.batch_fmtstr = self._get_batch_fmtstr(num_batches)
        self.meters = meters
        self.prefix = prefix

    def display(self, batch):
        entries = [self.prefix + self.batch_fmtstr.format(batch)]
        entries += [str(meter) for meter in self.meters]
        print('\t'.join(entries))

    def _get_batch_fmtstr(self, num_batches):
        num_digits = len(str(num_batches // 1))
        fmt = '{:' + str(num_digits) + 'd}'
        return '[' + fmt + '/' + fmt.format(num_batches) + ']'


def adjust_learning_rate(optimizer, epoch, base_lr, lr_milestones, gamma=0.1) -> float:
    """Sets the learning rate to decay by gamma (multiplicatively) when reaching each milestone"""
    factor = gamma ** bisect(lr_milestones, epoch)
    lr = base_lr * factor
    for param_group in optimizer.param_groups:
        param_group['lr'] = lr
    return lr


def accuracy(output, target, topk=(1,)):
    """Computes the accuracy over the k top predictions for the specified values of k"""
    with torch.no_grad():
        maxk = max(topk)
        batch_size = target.size(0)

        _, pred = output.topk(maxk, 1, True, True)
        pred = pred.t()
        correct = pred.eq(target.view(1, -1).expand_as(pred))

        res = []
        for k in topk:
            correct_k = correct[:k].reshape(-1).float().sum(0, keepdim=True)
            res.append(correct_k.mul_(100.0 / batch_size))
        return res


if __name__ == '__main__':
    main()