runtime_rewrite_main.py

import argparse
import os
import random
import shutil
import time
import warnings
from enum import Enum

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 tensorboardX import SummaryWriter

from SFPres50 import softpruning, do_Mask
from utils import gensubgradient, find_upper_bound, get_bnparameters, newthres
from reLBFGS import LBFGS
from get_small_model.resnet import resnet50
#from get_small_model.resnet_imagenet import resnet50
from tools import compute_conv_flops, return_output
#runtime
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 dataset')
parser.add_argument('--save', default='./events/', type=str, metavar='PATH',
                    help='path to save model (default: current directory)')
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=248, 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('--thres', default=0.1, type=float, help='threshold to prune')
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('--s', type=float, default=0.00001,
                    help='scale sparse rate (default: 0.00001)')
parser.add_argument('--x', type=float, default=0.01,
                    help='change rate (default: 0.01)')
parser.add_argument('--ub', type=float, default=0.46,
                    help='new_flops/flops_init (default: 0.46)')
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')

best_acc1 = 0


def main():
    args = parser.parse_args()
    global thres
    thres = args.s

    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

    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, thres)


def main_worker(gpu, ngpus_per_node, args, thres):
    global best_acc1

    args.gpu = gpu
    end_flag = False
    chosen_flag = False
    cfg_mask = []
    conv_mask = []
    linear_mask = []
    alpha = 1
    writer = SummaryWriter('runs/exp24', write_to_disk=True)
    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]()
        model = resnet50()
    print(model)
    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])
        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)
    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)
    optimizermu = LBFGS(get_bnparameters(model), lr=args.lr, history_size=5, mu=args.s, batch_size=args.batch_size,
                            line_search_fn="strong_wolfe")

    # 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_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
    flops_init = compute_conv_flops(model)
    print("Flops init =========>{}".format(flops_init))
    ratio = 0.
    flops = flops_init
    flag = True

    for epoch in range(args.start_epoch, args.epochs):
        if args.distributed:
            train_sampler.set_epoch(epoch)
        if epoch < 120:
            adjust_learning_rate(optimizer, epoch, args,0,0)
        # train for one epoch
        thres = train(train_loader, model, criterion, optimizer, optimizermu, epoch, args, alpha, thres, flag, chosen_flag)
        thres_ub = newthres(model, 0.50)
        if thres > thres_ub:
            thres = thres_ub
            end_flag = True
        writer.add_scalar("train/thres", thres, epoch)
        if chosen_flag == False:
            output_size = return_output(model)
            model, flops, ratio, cfg_mask, conv_mask, linear_mask, end_flag = softpruning(args, model, thres, end_flag, output_size)
            print("Pruned! pruned_ratio:{}, flops_drops:{}".format(ratio, 1 - (flops / flops_init)))
        else:
            do_Mask(model, cfg_mask, conv_mask, linear_mask)
            print("Masked! pruned_ratio:{}, flops_drops:{}".format(ratio, 1-(flops/flops_init)))
        writer.add_scalar("train/pruned_ratio", ratio, epoch)
        writer.add_scalar("train/flops_drops", 1-(flops/flops_init), epoch)

        if end_flag == False and (1-(flops/flops_init) >= 0.54):
            print("flops drop has already satisfied!")
            end_flag = True

        # evaluate on validation set
        acc1 = validate(val_loader, model, criterion, args)
        if acc1 < 0.1 and epoch >= 4:
            print("pause because of the acc")
            flag = False
        if end_flag == True and chosen_flag == False and acc1 >= 0.001:
            chosen_flag = True
        if flag == False and acc1 > 0.65:
            thres = newthres(model, ratio)
            flag = True

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

        if not args.multiprocessing_distributed or (args.multiprocessing_distributed
                                                    and args.rank % ngpus_per_node == 0):
            save_checkpoint({
                'epoch': epoch + 1,
                'arch': args.arch,
                'state_dict': model.state_dict(),
                'best_acc1': best_acc1,
                'optimizer': optimizer.state_dict(),
            }, is_best)
        writer.flush()
    writer.close()


def train(train_loader, model, criterion, optimizer, optimizermu, epoch, args, alpha, thres, flag, chosen_flag):
    batch_time = AverageMeter('Time', ':6.3f')
    data_time = AverageMeter('Data', ':6.3f')
    losses = AverageMeter('Loss', ':.4e')
    top1 = AverageMeter('Acc@1', ':6.4f')
    top5 = AverageMeter('Acc@5', ':6.4f')
    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()
    sum_delta = 0

    sum_t = 0
    sum_normal = 0
    for i, (images, target) in enumerate(train_loader):
        if epoch >=120:
            adjust_learning_rate(optimizer, epoch, args, i, len(train_loader))

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

        if (epoch+1) % 2 == 0 and not chosen_flag and flag:
            t_LBFGS = time.time()
            def closure():
                # 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()
                return loss

            loss, update, _ = optimizermu.step(closure)
            sum_t += (time.time() - t_LBFGS)
            optimizer.step()

            sub = gensubgradient(model)
            views = []
            for subs in sub:
                views.extend(subs)
            views = torch.Tensor(views).cuda()
            update = torch.mul(update, views)
            temp = torch.ones(update.numel()).cuda()
            derivative_l = update.dot(temp)
            #delta = args.x * loss / derivative_l  # 0.00000045可以进行更改
            delta = 0.000000001 * loss / derivative_l
            #1e-8
            #print('image.size:{}'.format(images.size(0)))
            #mix = (optimizermu.param_groups[0]['lr'] * alpha * delta) / (images.size(0)*5005)
            mix = (optimizermu.param_groups[0]['lr'] * alpha * delta)/images.size(0)
            #if i % (args.print_freq * 5) == 0:
            #    print('mix:{}'.format(mix))
            thres_before = thres
            thres = thres + mix
            sum_delta_before = sum_delta
            sum_delta += mix

            upper_bound = find_upper_bound(model)
            if thres >= upper_bound or thres < 0:
                thres = thres_before  # 回滚
                sum_delta = sum_delta_before

            #if i % (args.print_freq*5) == 0:
            #    print('After update:  threshold:{}'.format(thres))
        else:
            t_normal = time.time()
            # compute output
            output = model(images)
            loss = criterion(output, target)
            #if i==0:
            #    loss_init = loss
            #elif i==len(train_loader)-1:
            #    loss_end = loss
            #    args.x = (loss_init - loss_end)/(args.batch_size*len(train_loader))
            #    print("args.x = {}".format(args.x))
            # 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()
            optimizer.step()
            sum_normal += (time.time()-t_normal)
        print('')
        # measure elapsed time
        batch_time.update(time.time() - end)
        end = time.time()

        if i % args.print_freq == 0:
            progress.display(i)
    print('normal_runtime:{}, L-BFGS runtime:{}'.format(sum_normal, sum_t))
    return thres


def validate(val_loader, model, criterion, args):
    batch_time = AverageMeter('Time', ':6.3f', Summary.NONE)
    losses = AverageMeter('Loss', ':.4e', Summary.NONE)
    top1 = AverageMeter('Acc@1', ':6.4f', Summary.AVERAGE)
    top5 = AverageMeter('Acc@5', ':6.4f', Summary.AVERAGE)
    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)

        progress.display_summary()

    return top1.avg


def save_checkpoint(state, is_best, filename='exp24_checkpoint.pth.tar'):
    torch.save(state, filename)
    if is_best:
        shutil.copyfile(filename, 'exp24_model_best.pth.tar')


class Summary(Enum):
    NONE = 0
    AVERAGE = 1
    SUM = 2
    COUNT = 3


class AverageMeter(object):
    """Computes and stores the average and current value"""

    def __init__(self, name, fmt=':f', summary_type=Summary.AVERAGE):
        self.name = name
        self.fmt = fmt
        self.summary_type = summary_type
        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__)

    def summary(self):
        fmtstr = ''
        if self.summary_type is Summary.NONE:
            fmtstr = ''
        elif self.summary_type is Summary.AVERAGE:
            fmtstr = '{name} {avg:.3f}'
        elif self.summary_type is Summary.SUM:
            fmtstr = '{name} {sum:.3f}'
        elif self.summary_type is Summary.COUNT:
            fmtstr = '{name} {count:.3f}'
        else:
            raise ValueError('invalid summary type %r' % self.summary_type)

        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 display_summary(self):
        entries = [" *"]
        entries += [meter.summary() for meter in self.meters]
        print(' '.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, args, iteration, train_loader_len):
    """Sets the learning rate to the initial LR decayed by 10 every 40 epochs"""
    lr = args.lr * (0.1 ** (epoch // 40))
    if epoch >=120:
        lr = cos_lr(0.05, epoch=epoch-120, iteration=iteration,
                               num_iter=train_loader_len, num_epoch=args.epochs-120)
    for param_group in optimizer.param_groups:
        param_group['lr'] = lr

def cos_lr(base_lr, epoch, iteration, num_iter, num_epoch, warmup=False):
    from math import cos, pi
    warmup_epoch = 5 if warmup else 0
    warmup_iter = warmup_epoch * num_iter
    current_iter = iteration + epoch * num_iter
    max_iter = num_epoch * num_iter
    lr = base_lr * (1 + cos(pi * (current_iter - warmup_iter) / (max_iter - warmup_iter))) / 2
    if epoch < warmup_epoch:
        lr = base_lr * current_iter / warmup_iter
    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()