bi_tuning.py

"""
@author: Junguang Jiang
@contact: JiangJunguang1123@outlook.com
"""
import random
import time
import warnings
import argparse
import shutil

import torch
import torch.nn as nn
import torch.backends.cudnn as cudnn
from torch.optim import SGD
from torch.utils.data import DataLoader

import utils
from tllib.vision.transforms import MultipleApply
from tllib.utils.metric import accuracy
from tllib.utils.meter import AverageMeter, ProgressMeter
from tllib.utils.data import ForeverDataIterator
from tllib.utils.logger import CompleteLogger
from tllib.regularization.bi_tuning import Classifier, BiTuning

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")


def main(args: argparse.Namespace):
    logger = CompleteLogger(args.log, args.phase)
    print(args)

    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.')

    cudnn.benchmark = True

    # Data loading code
    train_augmentation = utils.get_train_transform(args.train_resizing, not args.no_hflip, args.color_jitter)
    val_transform = utils.get_val_transform(args.val_resizing)
    train_transform = MultipleApply([train_augmentation, train_augmentation])
    print("train_transform: ", train_transform)
    print("val_transform: ", val_transform)

    train_dataset, val_dataset, num_classes = utils.get_dataset(args.data, args.root, train_transform,
                                                                val_transform, args.sample_rate,
                                                                args.num_samples_per_classes)
    train_loader = DataLoader(train_dataset, batch_size=args.batch_size, shuffle=True,
                              num_workers=args.workers, drop_last=True)
    train_iter = ForeverDataIterator(train_loader)
    val_loader = DataLoader(val_dataset, batch_size=args.batch_size, shuffle=False, num_workers=args.workers)
    print("training dataset size: {} test dataset size: {}".format(len(train_dataset), len(val_dataset)))

    # create model
    print("=> using pre-trained model '{}'".format(args.arch))
    backbone_q = utils.get_model(args.arch, args.pretrained)
    pool_layer = nn.Identity() if args.no_pool else None
    classifier_q = Classifier(backbone_q, num_classes, pool_layer=pool_layer, projection_dim=args.projection_dim,
                              finetune=args.finetune)
    if args.pretrained_fc:
        print("=> loading pre-trained fc from '{}'".format(args.pretrained_fc))
        pretrained_fc_dict = torch.load(args.pretrained_fc)
        classifier_q.projector.load_state_dict(pretrained_fc_dict, strict=False)
    classifier_q = classifier_q.to(device)
    backbone_k = utils.get_model(args.arch)
    classifier_k = Classifier(backbone_k, num_classes, pool_layer=pool_layer).to(device)

    bituning = BiTuning(classifier_q, classifier_k, num_classes, K=args.K, m=args.m, T=args.T)

    # define optimizer and lr scheduler
    optimizer = SGD(classifier_q.get_parameters(args.lr), lr=args.lr, momentum=args.momentum, weight_decay=args.wd,
                    nesterov=True)
    lr_scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer, args.lr_decay_epochs, gamma=args.lr_gamma)

    # resume from the best checkpoint
    if args.phase == 'test':
        checkpoint = torch.load(logger.get_checkpoint_path('best'), map_location='cpu')
        classifier_q.load_state_dict(checkpoint)
        acc1 = utils.validate(val_loader, classifier_q, args, device)
        print(acc1)
        return

    # start training
    best_acc1 = 0.0
    for epoch in range(args.epochs):
        print(lr_scheduler.get_lr())
        # train for one epoch
        train(train_iter, bituning, optimizer, epoch, args)
        lr_scheduler.step()

        # evaluate on validation set
        acc1 = utils.validate(val_loader, classifier_q, args, device)

        # remember best acc@1 and save checkpoint
        torch.save(classifier_q.state_dict(), logger.get_checkpoint_path('latest'))
        if acc1 > best_acc1:
            shutil.copy(logger.get_checkpoint_path('latest'), logger.get_checkpoint_path('best'))
        best_acc1 = max(acc1, best_acc1)

    print("best_acc1 = {:3.1f}".format(best_acc1))
    logger.close()


def train(train_iter: ForeverDataIterator, bituning: BiTuning, optimizer: SGD, epoch: int, args: argparse.Namespace):
    batch_time = AverageMeter('Time', ':4.2f')
    data_time = AverageMeter('Data', ':3.1f')
    cls_losses = AverageMeter('Cls Loss', ':3.2f')
    contrastive_losses = AverageMeter('Contrastive Loss', ':3.2f')
    losses = AverageMeter('Loss', ':3.2f')
    cls_accs = AverageMeter('Cls Acc', ':3.1f')

    progress = ProgressMeter(
        args.iters_per_epoch,
        [batch_time, data_time, losses, cls_losses, contrastive_losses, cls_accs],
        prefix="Epoch: [{}]".format(epoch))

    classifier_criterion = torch.nn.CrossEntropyLoss().to(device)
    contrastive_criterion = torch.nn.KLDivLoss(reduction='batchmean').to(device)

    # switch to train mode
    bituning.train()

    end = time.time()
    for i in range(args.iters_per_epoch):
        x, labels = next(train_iter)
        img_q, img_k = x[0], x[1]

        img_q = img_q.to(device)
        img_k = img_k.to(device)
        labels = labels.to(device)

        # measure data loading time
        data_time.update(time.time() - end)

        # compute output
        y, logits_z, logits_y, bituning_labels = bituning(img_q, img_k, labels)
        cls_loss = classifier_criterion(y, labels)
        contrastive_loss_z = contrastive_criterion(logits_z, bituning_labels)
        contrastive_loss_y = contrastive_criterion(logits_y, bituning_labels)
        contrastive_loss = (contrastive_loss_z + contrastive_loss_y)
        loss = cls_loss + contrastive_loss * args.trade_off

        # measure accuracy and record loss
        losses.update(loss.item(), x[0].size(0))
        cls_losses.update(cls_loss.item(), x[0].size(0))
        contrastive_losses.update(contrastive_loss.item(), x[0].size(0))

        cls_acc = accuracy(y, labels)[0]
        cls_accs.update(cls_acc.item(), x[0].size(0))

        # compute gradient and do SGD step
        optimizer.zero_grad()
        loss.backward()
        optimizer.step()

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

        if i % args.print_freq == 0:
            progress.display(i)


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='Bi-tuning for Finetuning')
    # dataset parameters
    parser.add_argument('root', metavar='DIR',
                        help='root path of dataset')
    parser.add_argument('-d', '--data', metavar='DATA')
    parser.add_argument('-sr', '--sample-rate', default=100, type=int,
                        metavar='N',
                        help='sample rate of training dataset (default: 100)')
    parser.add_argument('-sc', '--num-samples-per-classes', default=None, type=int,
                        help='number of samples per classes.')
    parser.add_argument('--train-resizing', type=str, default='default', help='resize mode during training')
    parser.add_argument('--val-resizing', type=str, default='default', help='resize mode during validation')
    parser.add_argument('--no-hflip', action='store_true', help='no random horizontal flipping during training')
    parser.add_argument('--color-jitter', action='store_true', help='apply jitter during training')
    # model parameters
    parser.add_argument('-a', '--arch', metavar='ARCH', default='resnet50',
                        choices=utils.get_model_names(),
                        help='backbone architecture: ' +
                             ' | '.join(utils.get_model_names()) +
                             ' (default: resnet50)')
    parser.add_argument('--no-pool', action='store_true',
                        help='no pool layer after the feature extractor. Used in models such as ViT.')
    parser.add_argument('--finetune', action='store_true', help='whether use 10x smaller lr for backbone')
    parser.add_argument('--pretrained', default=None,
                        help="pretrained checkpoint of the backbone. "
                             "(default: None, use the ImageNet supervised pretrained backbone)")
    parser.add_argument('--pretrained-fc', default=None,
                        help="pretrained checkpoint of the fc. "
                             "(default: None)")
    parser.add_argument('--T', default=0.07, type=float, help="temperature. (default: 0.07)")
    parser.add_argument('--K', type=int, default=40, help="queue size. (default: 40)")
    parser.add_argument('--m', type=float, default=0.999, help="momentum coefficient. (default: 0.999)")
    parser.add_argument('--projection-dim', type=int, default=128,
                        help="dimension of the projection head. (default: 128)")
    parser.add_argument('--trade-off', type=float, default=1.0, help="trade-off parameters. (default: 1.0)")
    # training parameters
    parser.add_argument('-b', '--batch-size', default=48, type=int,
                        metavar='N',
                        help='mini-batch size (default: 48)')
    parser.add_argument('--optimizer', type=str, default='SGD', choices=['SGD', 'Adam'])
    parser.add_argument('--lr', '--learning-rate', default=0.01, type=float,
                        metavar='LR', help='initial learning rate', dest='lr')
    parser.add_argument('--lr-gamma', default=0.1, type=float, help='parameter for lr scheduler')
    parser.add_argument('--lr-decay-epochs', type=int, default=(12,), nargs='+', help='epochs to decay lr')
    parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
                        help='momentum')
    parser.add_argument('--wd', '--weight-decay', default=0.0005, type=float,
                        metavar='W', help='weight decay (default: 5e-4)')
    parser.add_argument('-j', '--workers', default=2, type=int, metavar='N',
                        help='number of data loading workers (default: 2)')
    parser.add_argument('--epochs', default=20, type=int, metavar='N',
                        help='number of total epochs to run')
    parser.add_argument('-i', '--iters-per-epoch', default=500, type=int,
                        help='Number of iterations per epoch')
    parser.add_argument('-p', '--print-freq', default=100, type=int,
                        metavar='N', help='print frequency (default: 100)')
    parser.add_argument('--seed', default=None, type=int,
                        help='seed for initializing training. ')
    parser.add_argument("--log", type=str, default='bi_tuning',
                        help="Where to save logs, checkpoints and debugging images.")
    parser.add_argument("--phase", type=str, default='train', choices=['train', 'test'],
                        help="When phase is 'test', only test the model.")
    args = parser.parse_args()
    main(args)