train_scratch.py

import argparse
import os
import random
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 registry
import datafree

parser = argparse.ArgumentParser(description='PyTorch ImageNet Training')
# Basic Settings
parser.add_argument('--data_root', default='./data')
parser.add_argument('--model', default='resnet34')
parser.add_argument('--dataset', default='tiny_imagenet')
parser.add_argument('--epochs', default=200, type=int, metavar='N',
                    help='number of total epochs to run')
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('--lr_decay_milestones', default="80,120,160", type=str,
                    help='milestones for learning rate decay')
parser.add_argument('--evaluate_only', action='store_true',
                    help='evaluate model on validation set')
parser.add_argument('--resume', default='', type=str, metavar='PATH',
                    help='path to latest checkpoint (default: none)')
parser.add_argument('--gpu', default=None, type=int,
                    help='GPU id to use.')
parser.add_argument('--ver', default=1, type=int,
                    help='Version')
parser.add_argument('--sche', default="cos", type=str,
                    help='Version')



# Device & FP16
parser.add_argument('--fp16', action='store_true',
                    help='use fp16')
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('--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')

# Misc
parser.add_argument('-j', '--workers', default=16, type=int, metavar='N',
                    help='number of data loading workers (default: 4)')
parser.add_argument('--start-epoch', default=0, type=int, metavar='N',
                    help='manual epoch number (useful on restarts)')
parser.add_argument('--momentum', default=0.9, type=float, metavar='M',
                    help='momentum')
parser.add_argument('--pretrained', dest='pretrained', action='store_true',
                    help='use pre-trained model')
parser.add_argument('--wd', '--weight-decay', default=5e-4, type=float,
                    metavar='W', help='weight decay (default: 1e-4)',
                    dest='weight_decay')
parser.add_argument('-p', '--print-freq', default=0, type=int,
                    metavar='N', help='print frequency (default: 0)')
parser.add_argument('--seed', default=None, type=int,
                    help='seed for initializing training.')

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.')
    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
    args.ngpus_per_node = 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):
    global best_acc1
    args.gpu = gpu

    ############################################
    # GPU and FP16
    ############################################
    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)
    if args.fp16:
        from torch.cuda.amp import autocast, GradScaler
        args.scaler = GradScaler() if args.fp16 else None 
        args.autocast = autocast
    else:
        args.autocast = datafree.utils.dummy_ctx

    ############################################
    # Logger
    ############################################
    log_name = 'R%d-%s-%s'%(args.rank, args.dataset, args.model) if args.multiprocessing_distributed else '%s-%s'%(args.dataset, args.model)
    args.logger = datafree.utils.logger.get_logger(log_name, output='checkpoints/scratch/log-%s-%s.txt'%(args.dataset, args.model))
    if args.rank<=0:
        for k, v in datafree.utils.flatten_dict( vars(args) ).items(): # print args
            args.logger.info( "%s: %s"%(k,v) )
    
    
    ############################################
    # Setup dataset
    ############################################
    num_classes, train_dataset, val_dataset = registry.get_dataset(name=args.dataset, data_root=args.data_root)
    cudnn.benchmark = True
    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(
        val_dataset,
        batch_size=args.batch_size, shuffle=False,
        num_workers=args.workers, pin_memory=True)
    evaluator = datafree.evaluators.classification_evaluator(val_loader)
    args.current_epoch = 0

    ############################################
    # Setup models and datasets
    ############################################
    model = registry.get_model(args.model, num_classes=num_classes, pretrained=args.pretrained)
    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
        model = torch.nn.DataParallel(model).cuda()

    ############################################
    # Setup optimizer
    ############################################
    criterion = nn.CrossEntropyLoss().cuda(args.gpu)
    optimizer = torch.optim.SGD(model.parameters(), args.lr, momentum=args.momentum, weight_decay=args.weight_decay)
    if args.sche == "cos":
        scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(optimizer, args.epochs, eta_min=0)
    else:
        milestones = [ int(ms) for ms in args.lr_decay_milestones.split(',') ]
        scheduler = torch.optim.lr_scheduler.MultiStepLR( optimizer, milestones=milestones, gamma=0.2)

    ############################################
    # Resume
    ############################################
    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)

            if isinstance(model, nn.Module):
                model.load_state_dict(checkpoint['state_dict'])
            else:
                model.module.load_state_dict(checkpoint['state_dict'])
            
            best_acc1 = checkpoint['best_acc1']

            try: 
                args.start_epoch = checkpoint['epoch']
                optimizer.load_state_dict(checkpoint['optimizer'])
                scheduler.load_state_dict(checkpoint['scheduler'])
            except: print("Fails to load additional information")
            print("[!] loaded checkpoint '{}' (epoch {} acc {})"
                  .format(args.resume, checkpoint['epoch'], best_acc1))
        else:
            print("[!] no checkpoint found at '{}'".format(args.resume))

    ############################################
    # Evaluate
    ############################################
    if args.evaluate_only:
        model.eval()
        eval_results = evaluator(model, device=args.gpu)
        (acc1, acc5), val_loss = eval_results['Acc'], eval_results['Loss']
        print('[Eval] Acc@1={acc1:.4f} Acc@5={acc5:.4f} Loss={loss:.4f}'.format(acc1=acc1, acc5=acc5, loss=val_loss))
        return

    ############################################
    # Train Loop
    ############################################
    for epoch in range(args.start_epoch, args.epochs):
        time_start = time.time()
        if args.distributed:
            train_sampler.set_epoch(epoch)
        args.current_epoch=epoch
        train(train_loader, model, criterion, optimizer, args)
        model.eval()
        eval_results = evaluator(model, device=args.gpu)
        (acc1, acc5), val_loss = eval_results['Acc'], eval_results['Loss']
        time_end = time.time() - time_start
        args.logger.info('[Eval] Epoch={current_epoch} Acc@1={acc1:.4f} Acc@5={acc5:.4f} Loss={loss:.4f} Lr={lr:.4f} Time:{t:.4f}'
                .format(current_epoch=args.current_epoch, acc1=acc1, acc5=acc5, loss=val_loss, lr=optimizer.param_groups[0]['lr'], t=time_end))
        scheduler.step()
        is_best = acc1 > best_acc1
        best_acc1 = max(acc1, best_acc1)
        _best_ckpt = 'checkpoints/scratch/%s_%s_%s.pth'%(args.dataset, args.model, args.ver)
        if not args.multiprocessing_distributed or (args.multiprocessing_distributed
                and args.rank % ngpus_per_node == 0):
            if 66.4 < acc1 < 66.5:
                save_checkpoint({
                    'epoch': epoch + 1,
                    'arch': args.model,
                    'state_dict': model.state_dict(),
                    'best_acc1': float(acc1),
                    'optimizer' : optimizer.state_dict(),
                    'scheduler': scheduler.state_dict()
                }, True, _best_ckpt)
                best_acc1 = acc1
            if acc1 == 66.44:
                save_checkpoint({
                    'epoch': epoch + 1,
                    'arch': args.model,
                    'state_dict': model.state_dict(),
                    'best_acc1': float(acc1),
                    'optimizer' : optimizer.state_dict(),
                    'scheduler': scheduler.state_dict()
                }, True, _best_ckpt)
                args.logger.info("Best: %.4f" % acc1)
                return 1

    if args.rank<=0:
        args.logger.info("Best: %.4f"%best_acc1)


def train(train_loader, model, criterion, optimizer, args):
    global best_acc1
    loss_metric = datafree.metrics.RunningLoss(nn.CrossEntropyLoss(reduction='sum'))
    acc_metric = datafree.metrics.TopkAccuracy(topk=(1,5))
    model.train()
    for i, (images, target) in enumerate(train_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)
        with args.autocast(enabled=args.fp16):
            output = model(images)
            loss = criterion(output, target)
        # measure accuracy and record loss
        acc_metric.update(output, target)
        loss_metric.update(output, target)
        optimizer.zero_grad()
        if args.fp16:
            scaler = args.scaler
            scaler.scale(loss).backward()
            scaler.step(optimizer)
            scaler.update()
        else:
            loss.backward()
            optimizer.step()
        if args.print_freq>0 and i % args.print_freq == 0:
            (train_acc1, train_acc5), train_loss = acc_metric.get_results(), loss_metric.get_results()
            args.logger.info('[Train] Epoch={current_epoch} Iter={i}/{total_iters}, train_acc@1={train_acc1:.4f}, train_acc@5={train_acc5:.4f}, train_Loss={train_loss:.4f}, Lr={lr:.4f}'
              .format(current_epoch=args.current_epoch, i=i, total_iters=len(train_loader), train_acc1=train_acc1, train_acc5=train_acc5, train_loss=train_loss, lr=optimizer.param_groups[0]['lr']))
            loss_metric.reset(), acc_metric.reset()
            

def save_checkpoint(state, is_best, filename='checkpoint.pth'):
    if is_best:
        torch.save(state, filename)

if __name__ == '__main__':
    main()