knowledge_distillation.py

import argparse
import os
import random
import shutil
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

# needed for Ondrej
import numpy as np
import os.path
from os import path

# Rene
from sklearn.model_selection import train_test_split
import pretrainedmodels as ptm
from my_codes.train_imagenet.main_stratified import AverageMeter, ProgressMeter, accuracy, save_checkpoint

# For half precision
scaler = torch.cuda.amp.GradScaler()

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

model_names += ptm_names

parser = argparse.ArgumentParser(description='PyTorch ImageNet knowledge distillation')
parser.add_argument('data', metavar='DIR',
                    help='path to dataset')
parser.add_argument('-a', '--arch', metavar='ARCH', default='resnet18',
                    choices=model_names,
                    help='student 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('--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('--output-folder', default='', type=str, metavar='output_folder',
                    help='path to a folder in which training outputs will be stored')
parser.add_argument('--from-ptm', action='store_true', dest='from_ptm',
                    help='whether to load the model from library pretrainedmodels')
best_acc1 = 0


def main():
    args = parser.parse_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.')

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

    # Simply call main_worker function
    main_worker(args.gpu, args)


def dataset_with_indices(cls):
    """
    Modifies the given Dataset class to return a tuple data, target, index
    instead of just data, target.
    """

    def __getitem__(self, index):
        data, target = cls.__getitem__(self, index)
        return data, target, index

    return type(cls.__name__, (cls,), {
        '__getitem__': __getitem__,
    })


def main_worker(gpu, args):
    global best_acc1
    args.gpu = gpu

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

    # create model
    if args.pretrained:
        print("=> using pre-trained model '{}'".format(args.arch))
        if not args.from_ptm:
            model = models.__dict__[args.arch](pretrained=True)
        else:
            model = ptm.__dict__[args.arch](num_classes=1000, pretrained='imagenet')
    else:
        print("=> creating model '{}'".format(args.arch))
        if not args.from_ptm:
            model = models.__dict__[args.arch]()
        else:
            model = ptm.__dict__[args.arch](num_classes=1000, pretrained=False)


    if args.gpu is not None:
        torch.cuda.set_device(args.gpu)
        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))

    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_transform = transforms.Compose([
            transforms.RandomResizedCrop(224),
            transforms.RandomHorizontalFlip(),
            transforms.ToTensor(),
            normalize,
        ])

    valid_transform = transforms.Compose([
            transforms.Resize(256),
            transforms.CenterCrop(224),
            transforms.ToTensor(),
            normalize,
        ])

    ImageFolderWithIndices = dataset_with_indices(datasets.ImageFolder)

    train_dataset_train_transf = ImageFolderWithIndices(
        traindir,
        train_transform
        )

    train_sampler = None

    train_loader = torch.utils.data.DataLoader(
        train_dataset_train_transf, 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(
        ImageFolderWithIndices(valdir, valid_transform),
        batch_size=args.batch_size, shuffle=False,
        num_workers=args.workers, pin_memory=True)

    if args.evaluate:
        validate(None, val_loader, model, criterion, args)
        return

    for epoch in range(args.start_epoch, args.epochs):
        adjust_learning_rate(optimizer, epoch, args)

        # train for one epoch
        train(train_loader, model, criterion, optimizer, epoch, args)

        # evaluate on validation set
        acc1 = validate(epoch, val_loader, model, criterion, args)

        # remember best acc@1 and save checkpoint
        is_best = acc1 > best_acc1
        best_acc1 = max(acc1, best_acc1)

        save_checkpoint({
            'epoch': epoch + 1,
            'arch': args.arch,
            'state_dict': model.state_dict(),
            'best_acc1': best_acc1,
            'optimizer' : optimizer.state_dict(),
        }, is_best, args)


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()

    prediction_list = []
    target_list = []
    complete = (epoch % 17 == 16) or (epoch == args.epochs - 1)

    for i, (images, target, indices) in enumerate(train_loader):
        if complete:
            assert(target.is_cuda == False)
            assert(target.requires_grad == False)
            target_py = target.detach().clone().numpy()
            target_list.append(target_py)

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

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

        # compute output
        with torch.cuda.amp.autocast():
            output = model(images)
            loss = criterion(output, target)
        # print(output.requires_grad)
        assert(output.is_cuda == True)
        assert(output.requires_grad == True)
        # print(output.is_cuda)
        # print(output.shape)
        # loss = criterion(output, target)

        if complete:
            # prediction_list.append(output.cpu().detach().numpy())
            prediction_list.append(output.detach().cpu().clone().numpy())

        # 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()
        scaler.scale(loss).backward()
        #optimizer.step()
        scaler.step(optimizer)
        scaler.update()

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

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

    with open(path.join(args.output_folder, "valid_summary.txt"), "a") as myfile:
        myfile.write("{epoch},{losses.avg:.4e},{top1.avg:.3f},{top5.avg:.3f},\t".
               format(epoch = epoch, losses = losses, top1 = top1, top5 = top5))

    if len(prediction_list) > 0:
        prediction_array = np.concatenate(prediction_list)
        target_array = np. concatenate(target_list)
        np.save(path.join(args.output_folder, "train_output_{}.npy".format(epoch)), prediction_array)
        np.save(path.join(args.output_folder, "train_target_{}.npy".format(epoch)), target_array)



if __name__ == '__main__':
    main()