main.py

# Copyright (c) 2015-present, Facebook, Inc.
# All rights reserved.
import argparse
import datetime
from typing import Optional, Callable, Dict
import numpy as np
import time
import torch
import torch.backends.cudnn as cudnn
import json
from models_v2 import PatchEmbedHybrid

from augment import AugCollatePatchSampleMixup

from pathlib import Path

from timm.models import create_model
from timm.loss import SoftTargetCrossEntropy
from timm.scheduler import create_scheduler
from timm.optim import create_optimizer
from timm.utils import NativeScaler, get_state_dict, ModelEma

from datasets_builders import build_dataset
from engine import train_one_epoch, evaluate
from losses import DistillationLoss
from prepare_imagenet_memfs import prepare_imagenet
from data_samplers import RASampler
from augment import new_data_aug_generator

# noinspection PyUnresolvedReferences
import models_v2

import utils


def get_args_parser():
    parser = argparse.ArgumentParser('DeiT training and evaluation script', add_help=False)
    parser.add_argument('--batch-size', default=64, type=int)
    parser.add_argument('--epochs', default=300, type=int)
    parser.add_argument('--bce-loss', action='store_true')
    parser.add_argument('--unscale-lr', action='store_true')

    # Model parameters
    parser.add_argument('--model', default='deit_base_patch16_224', type=str, metavar='MODEL',
                        help='Name of model to train')
    parser.add_argument('--input-size', default=224, type=int, help='images input size')

    parser.add_argument('--drop', type=float, default=0.0, metavar='PCT',
                        help='Dropout rate (default: 0.)')
    parser.add_argument('--drop-path', type=float, default=0.1, metavar='PCT',
                        help='Drop path rate (default: 0.1)')

    parser.add_argument('--model-ema', action='store_true')
    parser.add_argument('--no-model-ema', action='store_false', dest='model_ema')
    parser.set_defaults(model_ema=True)
    parser.add_argument('--model-ema-decay', type=float, default=0.99996, help='')
    parser.add_argument('--model-ema-force-cpu', action='store_true', default=False, help='')

    # Optimizer parameters
    parser.add_argument('--opt', default='adamw', type=str, metavar='OPTIMIZER',
                        help='Optimizer (default: "adamw"')
    parser.add_argument('--opt-eps', default=1e-8, type=float, metavar='EPSILON',
                        help='Optimizer Epsilon (default: 1e-8)')
    parser.add_argument('--opt-betas', default=None, type=float, nargs='+', metavar='BETA',
                        help='Optimizer Betas (default: None, use opt default)')
    parser.add_argument('--clip-grad', type=float, default=None, metavar='NORM',
                        help='Clip gradient norm (default: None, no clipping)')
    parser.add_argument('--momentum', type=float, default=0.9, metavar='M',
                        help='SGD momentum (default: 0.9)')
    parser.add_argument('--weight-decay', type=float, default=0.05,
                        help='weight decay (default: 0.05)')
    # Learning rate schedule parameters
    parser.add_argument('--sched', default='cosine', type=str, metavar='SCHEDULER',
                        help='LR scheduler (default: "cosine"')
    parser.add_argument('--lr', type=float, default=5e-4, metavar='LR',
                        help='learning rate (default: 5e-4)')
    parser.add_argument('--lr-noise', type=float, nargs='+', default=None, metavar='pct, pct',
                        help='learning rate noise on/off epoch percentages')
    parser.add_argument('--lr-noise-pct', type=float, default=0.67, metavar='PERCENT',
                        help='learning rate noise limit percent (default: 0.67)')
    parser.add_argument('--lr-noise-std', type=float, default=1.0, metavar='STDDEV',
                        help='learning rate noise std-dev (default: 1.0)')
    parser.add_argument('--warmup-lr', type=float, default=1e-6, metavar='LR',
                        help='warmup learning rate (default: 1e-6)')
    parser.add_argument('--min-lr', type=float, default=1e-5, metavar='LR',
                        help='lower lr bound for cyclic schedulers that hit 0 (1e-5)')

    parser.add_argument('--decay-epochs', type=float, default=30, metavar='N',
                        help='epoch interval to decay LR')
    parser.add_argument('--warmup-epochs', type=int, default=5, metavar='N',
                        help='epochs to warmup LR, if scheduler supports')
    parser.add_argument('--cooldown-epochs', type=int, default=10, metavar='N',
                        help='epochs to cooldown LR at min_lr, after cyclic schedule ends')
    parser.add_argument('--patience-epochs', type=int, default=10, metavar='N',
                        help='patience epochs for Plateau LR scheduler (default: 10')
    parser.add_argument('--decay-rate', '--dr', type=float, default=0.1, metavar='RATE',
                        help='LR decay rate (default: 0.1)')

    # Augmentation parameters
    parser.add_argument('--color-jitter', type=float, default=0.3, metavar='PCT',
                        help='Color jitter factor (default: 0.3)')
    parser.add_argument('--aa', type=str, default='rand-m9-mstd0.5-inc1', metavar='NAME',
                        help='Use AutoAugment policy. "v0" or "original". " + \
                             "(default: rand-m9-mstd0.5-inc1)'),
    parser.add_argument('--smoothing', type=float, default=0.1, help='Label smoothing (default: 0.1)')
    parser.add_argument('--train-interpolation', type=str, default='bicubic',
                        help='Training interpolation (random, bilinear, bicubic default: "bicubic")')

    parser.add_argument('--repeated-aug', action='store_true')
    parser.add_argument('--no-repeated-aug', action='store_false', dest='repeated_aug')
    parser.set_defaults(repeated_aug=True)

    parser.add_argument('--train-mode', action='store_true')
    parser.add_argument('--no-train-mode', action='store_false', dest='train_mode')
    parser.set_defaults(train_mode=True)

    parser.add_argument('--ThreeAugment', action='store_true')  # 3augment

    parser.add_argument('--src', action='store_true')  # simple random crop

    # * Random Erase params
    parser.add_argument('--reprob', type=float, default=0.25, metavar='PCT',
                        help='Random erase prob (default: 0.25)')
    parser.add_argument('--remode', type=str, default='pixel',
                        help='Random erase mode (default: "pixel")')
    parser.add_argument('--recount', type=int, default=1,
                        help='Random erase count (default: 1)')
    parser.add_argument('--resplit', action='store_true', default=False,
                        help='Do not random erase first (clean) augmentation split')

    # * Mixup params
    parser.add_argument('--mixup', type=float, default=0.8,
                        help='mixup alpha, mixup enabled if > 0. (default: 0.8)')
    parser.add_argument('--cutmix', type=float, default=1.0,
                        help='cutmix alpha, cutmix enabled if > 0. (default: 1.0)')
    parser.add_argument('--cutmix-minmax', type=float, nargs='+', default=None,
                        help='cutmix min/max ratio, overrides alpha and enables cutmix if set (default: None)')
    parser.add_argument('--mixup-prob', type=float, default=1.0,
                        help='Probability of performing mixup or cutmix when either/both is enabled')
    parser.add_argument('--mixup-switch-prob', type=float, default=0.5,
                        help='Probability of switching to cutmix when both mixup and cutmix enabled')
    parser.add_argument('--mixup-mode', type=str, default='batch',
                        help='How to apply mixup/cutmix params. Per "batch", "pair", or "elem"')

    # Distillation parameters
    parser.add_argument('--teacher-model', default='regnety_160', type=str, metavar='MODEL',
                        help='Name of teacher model to train (default: "regnety_160"')
    parser.add_argument('--teacher-path', type=str, default='')
    parser.add_argument('--distillation-type', default='none', choices=['none', 'soft', 'hard'], type=str, help="")
    parser.add_argument('--distillation-alpha', default=0.5, type=float, help="")
    parser.add_argument('--distillation-tau', default=1.0, type=float, help="")

    # * Finetuning params
    parser.add_argument('--finetune', default='', help='finetune from checkpoint')
    parser.add_argument('--attn-only', action='store_true')

    # Dataset parameters
    parser.add_argument('--data-path', default='/datasets01/imagenet_full_size/061417/', type=str,
                        help='dataset path')
    parser.add_argument('--data-set', default='IMNET', choices=['CIFAR', 'IMNET', 'INAT', 'INAT19', 'HUGGINGFACE'],
                        type=str, help='Image Net dataset path')
    parser.add_argument('--inat-category', default='name',
                        choices=['kingdom', 'phylum', 'class', 'order', 'supercategory', 'family', 'genus', 'name'],
                        type=str, help='semantic granularity')

    parser.add_argument('--output_dir', default='',
                        help='path where to save, empty for no saving')
    parser.add_argument('--device', default='cuda',
                        help='device to use for training / testing')
    parser.add_argument('--seed', default=0, type=int)
    parser.add_argument('--resume', default='', help='resume from checkpoint')
    parser.add_argument('--start_epoch', default=0, type=int, metavar='N',
                        help='start epoch')
    parser.add_argument('--eval', action='store_true', help='Perform evaluation only')
    parser.add_argument('--eval-crop-ratio', default=0.875, type=float, help="Crop ratio for evaluation")
    parser.add_argument('--dist-eval', action='store_true', default=False, help='Enabling distributed evaluation')
    parser.add_argument('--num_workers', default=10, type=int)
    parser.add_argument('--pin-mem', action='store_true',
                        help='Pin CPU memory in DataLoader for more efficient (sometimes) transfer to GPU.')
    parser.add_argument('--no-pin-mem', action='store_false', dest='pin_mem',
                        help='')
    parser.set_defaults(pin_mem=True)

    # distributed training parameters
    parser.add_argument('--world_size', default=1, type=int,
                        help='number of distributed processes')
    parser.add_argument('--dist_url', default='env://', help='url used to set up distributed training')
    parser.add_argument('--memfs-imnet', default=False, type=bool, action=argparse.BooleanOptionalAction,
                        help='unpack imagenet in ram. for use with the plgrid athena cluster.')


    #sampling strategies
    parser.add_argument('--multiscale-patches', default=None, type=str,
                        help='sample patches using the given sequence of patch sizes. This argument is evaluated,'
                             ' can be any python code')
    parser.add_argument('--random-patches', default=None, type=int,
                        help='use random patch scale sampling with a given number of patches')
    parser.add_argument('--random-patch-min-size', default=16, type=int, help='min size of a randomly sampled patch')
    parser.add_argument('--random-patch-max-size', default=48, type=int, help='max size of a randomly sampled patch')
    parser.add_argument('--grid-patch-size', default=None, type=int, help='use grid sampling with a given patch size')
    parser.add_argument('--central_patch_sizes', default=None, nargs="+", help='use central multiscale sampling with a given patch scales')
    parser.add_argument('--log-to-wandb', default=True, type=bool, action=argparse.BooleanOptionalAction,
                        help='log to weights and biases dashboard')
    parser.add_argument('--wandb_run_name', default=None, type=str,
                        help='log to weights and biases dashboard')
    parser.add_argument('--wandb_model_name', default=None, type=str,
                        help='log to weights and biases dashboard')

    #patch sampler modification parameters
    parser.add_argument('--modify_patches', default=False, action="store_true", help="Perform patch modifications when sampling patches")
    parser.add_argument('--merge_patches', default=0, type=int, help="Merge patches back into image")
    parser.add_argument('--use_learned_pos_embed', default=0, type=int, help="Merge patches back into image")
    parser.add_argument('--quantize_pos_embed', default=0, type=int, help="Merge patches back into image")
    parser.add_argument('--patch_zoom', default=0, type=int, help="number of pixels to zoom-in or zoom-out")
    parser.add_argument('--patch_shake', default=0, type=int, help="number of pixels to displace patch randomly (up, left, right, top)")
    parser.add_argument('--patch_dropout', default=0, type=int, help="number of patches to dropout / mask per image")
    parser.add_argument('--patch_select', default=0, type=int, help="select only one patch from block of N patches")
    parser.add_argument('--patch_select_mode', default="drop", type=str, choices=["drop", "resize"], help="Either drop rest of patches from selected blocks or resize the block to 16x16 px")
    parser.add_argument('--patch_dropout_blocks', default=1, type=int, help="number of patches in a block to dropout / mask per image")
    parser.add_argument('--hard_dropout', default=1, type=int, help="weather to drop patches or just mask them")
    
    parser.add_argument('--random-patch-min-scale', default=None, type=float,
                        help='(sampler v2) min scale of a randomly sampled patch')
    parser.add_argument('--random-patch-max-scale', default=0.22, type=float,
                        help='(sampler v2) max scale of a randomly sampled patch')
    parser.add_argument('--smart-patches', default=None, type=int, help='smart sampling')
    parser.add_argument('--scale-aware-embed', default=True, type=bool, action=argparse.BooleanOptionalAction,
                        help='use scale-aware embeds')
    parser.add_argument('--grid-to-random-ratio', default=0.7, type=float, help='hybrid sampler grid to random ratio')

    return parser


def main(args):
    utils.init_distributed_mode(args)

    print(args)

    ext_logger: Optional[Callable[[Dict, int], None]] = None
    if args.log_to_wandb and utils.is_main_process():
        import wandb
        wandb.login()
        wandb.init(
            # Set the project where this run will be logged
            project="elasticvit",
            # Track hyperparameters and run metadata
            config=vars(args),
            name=args.wandb_run_name
        )

        def log_to_wandb(log_dict, step):
            if utils.is_main_process():
                wandb.log(log_dict, step=step)

        ext_logger = log_to_wandb

    if args.distillation_type != 'none' and args.finetune and not args.eval:
        raise NotImplementedError("Finetuning with distillation not yet supported")

    device = torch.device(args.device)

    # fix the seed for reproducibility
    seed = args.seed + utils.get_rank()
    torch.manual_seed(seed)
    np.random.seed(seed)
    # random.seed(seed)

    cudnn.benchmark = True

    if args.memfs_imnet:
        data_path = prepare_imagenet(args.data_path)
    else:
        data_path = args.data_path

    if not args.eval:
        dataset_train, args.nb_classes = build_dataset(is_train=True, data_path=data_path, args=args)
    dataset_val, args.nb_classes = build_dataset(is_train=False, data_path=data_path, args=args)

    if args.distributed:
        num_tasks = utils.get_world_size()
        global_rank = utils.get_rank()
        if args.repeated_aug:
            sampler_train = RASampler(
                dataset_train, num_replicas=num_tasks, rank=global_rank, shuffle=True
            )
        else:
            sampler_train = torch.utils.data.DistributedSampler(
                dataset_train, num_replicas=num_tasks, rank=global_rank, shuffle=True
            )
        if args.dist_eval:
            if len(dataset_val) % num_tasks != 0:
                print('Warning: Enabling distributed evaluation with an eval dataset not divisible by process number. '
                      'This will slightly alter validation results as extra duplicate entries are added to achieve '
                      'equal num of samples per-process.')
            sampler_val = torch.utils.data.DistributedSampler(
                dataset_val, num_replicas=num_tasks, rank=global_rank, shuffle=False)
        else:
            sampler_val = torch.utils.data.SequentialSampler(dataset_val)
    else:
        if not args.eval:
            sampler_train = torch.utils.data.RandomSampler(dataset_train)
        sampler_val = torch.utils.data.SequentialSampler(dataset_val)

    patch_sampler_args = {
        'crop_sizes': args.multiscale_patches,
        'random_patches': args.random_patches,
        'random_min_size': args.random_patch_min_size,
        'random_max_size': args.random_patch_max_size,
        'grid_patch_size': args.grid_patch_size,
        'central_patch_sizes': [int(x) for x in args.central_patch_sizes] if args.central_patch_sizes is not None else [],
        'merge_patches': args.merge_patches,
        'hard_dropout': args.hard_dropout,
        'random_patch_min_scale': args.random_patch_min_scale,
        'random_patch_max_scale': args.random_patch_max_scale,
        'smart_patches': args.smart_patches,
        'grid_to_random_ratio': args.grid_to_random_ratio
    }
    if args.modify_patches:
        patch_sampler_args['patch_modifications'] = {
            'patch_zoom': args.patch_zoom,
            'patch_shake': args.patch_shake,
            'patch_dropout': args.patch_dropout,
            'patch_dropout_blocks': args.patch_dropout_blocks,
            'patch_select': args.patch_select,
            'patch_select_mode': args.patch_select_mode
        }

    train_collate_fn = AugCollatePatchSampleMixup(**patch_sampler_args,
                                                  mixup_alpha=args.mixup, cutmix_alpha=args.cutmix,
                                                  prob=args.mixup_prob, switch_prob=args.mixup_switch_prob,
                                                  label_smoothing=args.smoothing, num_classes=args.nb_classes)

    if not args.eval:
        data_loader_train = torch.utils.data.DataLoader(
            dataset_train, sampler=sampler_train,
            batch_size=args.batch_size,
            num_workers=args.num_workers,
            pin_memory=args.pin_mem,
            drop_last=True,
            collate_fn=train_collate_fn
        )
        if args.ThreeAugment:
            data_loader_train.dataset.transform = new_data_aug_generator(args)

    val_collate_fn = AugCollatePatchSampleMixup(**patch_sampler_args, augment=False)

    data_loader_val = torch.utils.data.DataLoader(
        dataset_val, sampler=sampler_val,
        batch_size=int(1.5 * args.batch_size),
        num_workers=args.num_workers,
        pin_memory=args.pin_mem,
        drop_last=False,
        collate_fn=val_collate_fn
    )

    mixup_fn = None

    print(f"Creating model: {args.model}")
    model = create_model(
        args.model,
        pretrained=False,
        num_classes=args.nb_classes,
        drop_rate=args.drop,
        drop_path_rate=args.drop_path,
        drop_block_rate=None,
        img_size=args.input_size,
        
        Patch_layer=PatchEmbedHybrid,
        use_learned_pos_embed = args.use_learned_pos_embed,
        quantize_pos_embed = args.quantize_pos_embed
    )

    if args.finetune:
        if args.finetune.startswith('https'):
            checkpoint = torch.hub.load_state_dict_from_url(
                args.finetune, map_location='cpu', check_hash=True)
        else:
            checkpoint = torch.load(args.finetune, map_location='cpu')

        checkpoint_model = checkpoint['model']
        state_dict = model.state_dict()
        for k in ['head.weight', 'head.bias', 'head_dist.weight', 'head_dist.bias']:
            if k in checkpoint_model and checkpoint_model[k].shape != state_dict[k].shape:
                print(f"Removing key {k} from pretrained checkpoint")
                del checkpoint_model[k]

        # interpolate position embedding
        pos_embed_checkpoint = checkpoint_model['pos_embed']
        embedding_size = pos_embed_checkpoint.shape[-1]
        num_patches = model.patch_embed.num_patches
        num_extra_tokens = model.pos_embed.shape[-2] - num_patches
        # height (== width) for the checkpoint position embedding
        orig_size = int((pos_embed_checkpoint.shape[-2] - num_extra_tokens) ** 0.5)
        # height (== width) for the new position embedding
        new_size = int(num_patches ** 0.5)
        # class_token and dist_token are kept unchanged
        extra_tokens = pos_embed_checkpoint[:, :num_extra_tokens]
        # only the position tokens are interpolated
        pos_tokens = pos_embed_checkpoint[:, num_extra_tokens:]
        pos_tokens = pos_tokens.reshape(-1, orig_size, orig_size, embedding_size).permute(0, 3, 1, 2)
        pos_tokens = torch.nn.functional.interpolate(
            pos_tokens, size=(new_size, new_size), mode='bicubic', align_corners=False)
        pos_tokens = pos_tokens.permute(0, 2, 3, 1).flatten(1, 2)
        new_pos_embed = torch.cat((extra_tokens, pos_tokens), dim=1)
        checkpoint_model['pos_embed'] = new_pos_embed

        model.load_state_dict(checkpoint_model, strict=False)

    if args.attn_only:
        for name_p, p in model.named_parameters():
            if '.attn.' in name_p:
                p.requires_grad = True
            else:
                p.requires_grad = False
        try:
            model.head.weight.requires_grad = True
            model.head.bias.requires_grad = True
        except:
            model.fc.weight.requires_grad = True
            model.fc.bias.requires_grad = True
        try:
            model.pos_embed.requires_grad = True
        except:
            print('no position encoding')
        try:
            for p in model.patch_embed.parameters():
                p.requires_grad = False
        except:
            print('no patch embed')

    model.to(device)

    model_ema = None
    if args.model_ema:
        # Important to create EMA model after cuda(), DP wrapper, and AMP but before SyncBN and DDP wrapper
        model_ema = ModelEma(
            model,
            decay=args.model_ema_decay,
            device='cpu' if args.model_ema_force_cpu else '',
            resume='')

    model_without_ddp = model
    if args.distributed:
        model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu])
        model_without_ddp = model.module
    n_parameters = sum(p.numel() for p in model.parameters() if p.requires_grad)
    print('number of params:', n_parameters)
    if not args.unscale_lr:
        linear_scaled_lr = args.lr * args.batch_size * utils.get_world_size() / 512.0
        args.lr = linear_scaled_lr
    optimizer = create_optimizer(args, model_without_ddp)
    loss_scaler = NativeScaler()

    lr_scheduler, _ = create_scheduler(args, optimizer)

    criterion = SoftTargetCrossEntropy()
    if args.bce_loss:
        criterion = torch.nn.BCEWithLogitsLoss()

    teacher_model = None
    if args.distillation_type != 'none':
        assert args.teacher_path, 'need to specify teacher-path when using distillation'
        print(f"Creating teacher model: {args.teacher_model}")
        teacher_model = create_model(
            args.teacher_model,
            pretrained=False,
            num_classes=args.nb_classes,
            global_pool='avg',
        )
        if args.teacher_path.startswith('https'):
            checkpoint = torch.hub.load_state_dict_from_url(
                args.teacher_path, map_location='cpu', check_hash=True)
        else:
            checkpoint = torch.load(args.teacher_path, map_location='cpu')
        teacher_model.load_state_dict(checkpoint['model'])
        teacher_model.to(device)
        teacher_model.eval()

    # wrap the criterion in our custom DistillationLoss, which
    # just dispatches to the original criterion if args.distillation_type is 'none'
    criterion = DistillationLoss(
        criterion, teacher_model, args.distillation_type, args.distillation_alpha, args.distillation_tau
    )

    output_dir = Path(args.output_dir)
    if args.resume:
        if args.resume.startswith('https'):
            checkpoint = torch.hub.load_state_dict_from_url(
                args.resume, map_location='cpu', check_hash=True)
        else:
            checkpoint = torch.load(args.resume, map_location='cpu')
        #del checkpoint['model']['pos_embed']
        print(model_without_ddp.load_state_dict(checkpoint['model'], strict=False))
        if not args.eval and 'optimizer' in checkpoint and 'lr_scheduler' in checkpoint and 'epoch' in checkpoint:
            optimizer.load_state_dict(checkpoint['optimizer'])
            lr_scheduler.load_state_dict(checkpoint['lr_scheduler'])
            args.start_epoch = checkpoint['epoch'] + 1
            if args.model_ema:
                utils._load_checkpoint_for_ema(model_ema, checkpoint['model_ema'])
            if 'scaler' in checkpoint:
                loss_scaler.load_state_dict(checkpoint['scaler'])
        lr_scheduler.step(args.start_epoch)
    if args.eval:
        test_stats = evaluate(data_loader_val, model, device, ext_logger=ext_logger)
        print(f"Accuracy of the network on the {len(dataset_val)} test images: {test_stats['acc1']:.1f}%")
        return

    print(f"Start training for {args.epochs} epochs")
    start_time = time.time()
    max_accuracy = 0.0
    for epoch in range(args.start_epoch, args.epochs):
        if args.distributed:
            data_loader_train.sampler.set_epoch(epoch)

        train_stats = train_one_epoch(
            model, criterion, data_loader_train,
            optimizer, device, epoch, loss_scaler,
            args.clip_grad, model_ema, mixup_fn,
            set_training_mode=args.train_mode,
            # keep in eval mode for deit finetuning / train mode for training and deit III finetuning
            args=args, ext_logger=ext_logger
        )

        lr_scheduler.step(epoch)
        if args.output_dir:
            checkpoint_paths = [output_dir / 'checkpoint.pth']
            for checkpoint_path in checkpoint_paths:
                utils.save_on_master({
                    'model': model_without_ddp.state_dict(),
                    'optimizer': optimizer.state_dict(),
                    'lr_scheduler': lr_scheduler.state_dict(),
                    'epoch': epoch,
                    'model_ema': get_state_dict(model_ema),
                    'scaler': loss_scaler.state_dict(),
                    'args': args,
                }, checkpoint_path)

        data_loader_val.collate_fn.on_epoch()
        test_stats = evaluate(data_loader_val, model, device, epoch=epoch, ext_logger=ext_logger)
        print(f"Accuracy of the network on the {len(dataset_val)} test images: {test_stats['acc1']:.1f}%")

        if max_accuracy < test_stats["acc1"]:
            max_accuracy = test_stats["acc1"]
            if args.output_dir:
                checkpoint_paths = [output_dir / 'best_checkpoint.pth']
                for checkpoint_path in checkpoint_paths:
                    utils.save_on_master({
                        'model': model_without_ddp.state_dict(),
                        'optimizer': optimizer.state_dict(),
                        'lr_scheduler': lr_scheduler.state_dict(),
                        'epoch': epoch,
                        'model_ema': get_state_dict(model_ema),
                        'scaler': loss_scaler.state_dict(),
                        'args': args,
                    }, checkpoint_path)

        print(f'Max accuracy: {max_accuracy:.2f}%')

        log_stats = {**{f'train_{k}': v for k, v in train_stats.items()},
                     **{f'test_{k}': v for k, v in test_stats.items()},
                     'epoch': epoch,
                     'n_parameters': n_parameters}

        if args.output_dir and utils.is_main_process():
            with (output_dir / "log.txt").open("a") as f:
                f.write(json.dumps(log_stats) + "\n")

    total_time = time.time() - start_time
    total_time_str = str(datetime.timedelta(seconds=int(total_time)))
    print('Training time {}'.format(total_time_str))


if __name__ == '__main__':
    parser = argparse.ArgumentParser('DeiT training and evaluation script', parents=[get_args_parser()])
    args = parser.parse_args()
    if args.output_dir:
        Path(args.output_dir).mkdir(parents=True, exist_ok=True)
    main(args)