main_l2_vit.py

import argparse
import datetime
import numpy as np
import time
import torch
import torch.backends.cudnn as cudnn
import json

from pathlib import Path

from timm.data import Mixup
from timm.models import create_model
from timm.loss import LabelSmoothingCrossEntropy, SoftTargetCrossEntropy
from timm.scheduler import create_scheduler
from timm.optim import create_optimizer
from timm.utils import NativeScaler, get_state_dict, ModelEma
from torch.nn import parameter

from datasets import build_dataset
from engine_l2 import train_one_epoch, evaluate
from losses_l2 import DistillationLoss, DiffPruningLoss, DistillDiffPruningLoss
from samplers import RASampler
import utils
from utils import SoftTargetCrossEntropy_max
from functools import partial
import torch.nn as nn
from vit_l2 import VisionTransformerDiffPruning, VisionTransformerTeacher, _cfg, checkpoint_filter_fn
from lvvit_l2 import LVViTDiffPruning, LVViT_Teacher
from pit import PoolingTransformer, PoolingTransformerTeacher
import math
import shutil

def get_args_parser():
    parser = argparse.ArgumentParser('DynamicViT training and evaluation script', add_help=False)
    parser.add_argument('--batch-size', default=64, type=int)
    parser.add_argument('--epochs', default=300, type=int)

    # Model parameters
    parser.add_argument('--arch', default='deit_small', type=str, help='Name of model to train')
    parser.add_argument('--input-size', default=224, type=int, help='images input size')
    parser.add_argument('--distillw', type=float, default=0.5, help='distill rate (default: 0.5)')
    parser.add_argument('--ratiow', type=float, default=2.0, metavar='PCT', help='ratio rate (default: 2.0)')
    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=0, 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=0.4, metavar='PCT',
                        help='Color jitter factor (default: 0.4)')
    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)

    # * 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')

    # 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'],
                        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('--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('--distill', action='store_true', default=False, help='Enabling distributed evaluation')
    parser.add_argument('--base_rate', type=float, default=0.7)

    return parser



def get_param_groups(model, weight_decay):
    decay = []
    no_decay = []
    predictor = []
    for name, param in model.named_parameters():
        if 'predictor' in name:
            predictor.append(param)
        elif not param.requires_grad:
            continue  # frozen weights
        elif 'cls_token' in name or 'pos_embed' in name:
            continue  # frozen weights
        elif len(param.shape) == 1 or name.endswith(".bias"):
            no_decay.append(param)
        else:
            decay.append(param)
    return [
        {'params': predictor, 'weight_decay': weight_decay, 'name': 'predictor'},
        {'params': no_decay, 'weight_decay': 0., 'name': 'base_no_decay'},
        {'params': decay, 'weight_decay': weight_decay, 'name': 'base_decay'}
        ]


def adjust_learning_rate(param_groups, init_lr, min_lr, step, max_step, warming_up_step=2, warmup_predictor=False, base_multi=0.1):
    #if step >= 22:
    #    init_lr = 5e-4
    cos_lr = (math.cos(step / max_step * math.pi) + 1) * 0.5
    cos_lr = min_lr + cos_lr * (init_lr - min_lr)
    if warmup_predictor and step < 1:
        cos_lr = init_lr * 0.01
    if step < warming_up_step:
        backbone_lr = 0
    else:
        backbone_lr = min(init_lr * 0.01, cos_lr)
    print('## Using lr  %.7f for BACKBONE, cosine lr = %.7f for PREDICTOR' % (backbone_lr, cos_lr))
    for param_group in param_groups:
        if param_group['name'] == 'predictor':
            param_group['lr'] = cos_lr
        else:
            param_group['lr'] = backbone_lr # init_lr * 0.01 # cos_lr * base_multi

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

    print(args)

    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

    dataset_train, args.nb_classes = build_dataset(is_train=True, args=args)
    dataset_val, _ = build_dataset(is_train=False, args=args)

    if True:  # 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:
        sampler_train = torch.utils.data.RandomSampler(dataset_train)
        sampler_val = torch.utils.data.SequentialSampler(dataset_val)

    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,
    )

    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
    )

    mixup_fn = None
    mixup_active = args.mixup > 0 or args.cutmix > 0. or args.cutmix_minmax is not None
    if mixup_active:
        mixup_fn = Mixup(
            mixup_alpha=args.mixup, cutmix_alpha=args.cutmix, cutmix_minmax=args.cutmix_minmax,
            prob=args.mixup_prob, switch_prob=args.mixup_switch_prob, mode=args.mixup_mode,
            label_smoothing=args.smoothing, num_classes=args.nb_classes)
    else:
        print('Attention: mixup/cutmix are not used')


    base_rate = args.base_rate
    #KEEP_RATE = [base_rate, base_rate ** 2, base_rate ** 3]
    # KEEP_RATE = [1.0, 1.0, 1.0]
    KEEP_RATE = [0.617,0.369,0.137]

    if args.arch == 'deit_small':
        PRUNING_LOC = [3,6,9] 
        print(f"Creating model: {args.arch}")
        print('token_ratio =', KEEP_RATE, 'at layer', PRUNING_LOC)
        model = VisionTransformerDiffPruning(
            patch_size=16, embed_dim=384, depth=12, num_heads=6, mlp_ratio=4, qkv_bias=True, 
            pruning_loc=PRUNING_LOC, token_ratio=KEEP_RATE, distill=args.distill
            )
        model_path = './deit_small_patch16_224-cd65a155.pth'
        checkpoint = torch.load(model_path, map_location="cpu")
        ckpt = checkpoint_filter_fn(checkpoint, model)
        model.default_cfg = _cfg()
        missing_keys, unexpected_keys = model.load_state_dict(ckpt, strict=False)
        print('# missing keys=', missing_keys)
        print('# unexpected keys=', unexpected_keys)
        print('sucessfully loaded from pre-trained weights:', model_path)

        if args.distill:
            print('## Distillation Pruning Mode')
            model_t = VisionTransformerTeacher(
                patch_size=16, embed_dim=384, depth=12, num_heads=6, mlp_ratio=4, qkv_bias=True
            )
            model_t.load_state_dict(ckpt, strict=True)
            model_t.to(device)
            print('sucessfully loaded from pre-trained weights for the teach model')
    elif args.arch == 'lvvit_s':
        PRUNING_LOC = [4,8,12] 
        print(f"Creating model: {args.arch}")
        print('token_ratio =', KEEP_RATE, 'at layer', PRUNING_LOC)
        model = LVViTDiffPruning(
            patch_size=16, embed_dim=384, depth=16, num_heads=6, mlp_ratio=3.,
            p_emb='4_2',skip_lam=2., return_dense=True,mix_token=True,
            pruning_loc=PRUNING_LOC, token_ratio=KEEP_RATE, distill=args.distill
        )
        model_path = './lvvit_s-26M-224-83.3.pth.tar'
        checkpoint = torch.load(model_path, map_location="cpu")
        model.default_cfg = _cfg()
        missing_keys, unexpected_keys = model.load_state_dict(checkpoint, strict=False)
        print('# missing keys=', missing_keys)
        print('# unexpected keys=', unexpected_keys)
        print('sucessfully loaded from pre-trained weights:', model_path)

        if args.distill:
            print('## Distillation Pruning Mode')
            model_t = LVViT_Teacher(
                patch_size=16, embed_dim=384, depth=16, num_heads=6, mlp_ratio=3.,
                p_emb='4_2',skip_lam=2., return_dense=True,mix_token=True
            )
            model_t.load_state_dict(checkpoint, strict=True)
            model_t.to(device)
            print('sucessfully loaded from pre-trained weights for the teach model')
    elif args.arch == 'lvvit_m':
        PRUNING_LOC = [5,10,15] 
        print(f"Creating model: {args.arch}")
        print('token_ratio =', KEEP_RATE, 'at layer', PRUNING_LOC)
        model = LVViTDiffPruning(
            patch_size=16, embed_dim=512, depth=20, num_heads=8, mlp_ratio=3.,
            p_emb='4_2',skip_lam=2., return_dense=True,mix_token=True,
            pruning_loc=PRUNING_LOC, token_ratio=KEEP_RATE, distill=args.distill
        )
        model_path = './lvvit_m-56M-224-84.0.tar'
        checkpoint = torch.load(model_path, map_location="cpu")
        model.default_cfg = _cfg()
        missing_keys, unexpected_keys = model.load_state_dict(checkpoint, strict=False)
        print('# missing keys=', missing_keys)
        print('# unexpected keys=', unexpected_keys)
        print('sucessfully loaded from pre-trained weights:', model_path)

        if args.distill:
            print('## Distillation Pruning Mode')
            model_t = LVViT_Teacher(
                patch_size=16, embed_dim=512, depth=20, num_heads=8, mlp_ratio=3.,
                p_emb='4_2',skip_lam=2., return_dense=True,mix_token=True
            )
            model_t.load_state_dict(checkpoint, strict=True)
            model_t.to(device)
            print('sucessfully loaded from pre-trained weights for the teach model')

    elif args.arch == 'pit_s':
        PRUNING_LOC = [1,3,2]
        print(f"Creating model: {args.arch}")
        print('token_ratio =', KEEP_RATE, 'at layer', PRUNING_LOC)
        model = PoolingTransformer(
            image_size=224,
            patch_size=16,
            stride=8,
            base_dims=[48, 48, 48],
            depth=[2, 6, 4],
            heads=[3, 6, 12],
            mlp_ratio=4,
            pruning_loc=PRUNING_LOC, token_ratio=KEEP_RATE, distill=args.distill
            )
        model_path = './pit_s_809.pth'
        checkpoint = torch.load(model_path, map_location="cpu")
        #ckpt = checkpoint_filter_fn(checkpoint, model)
        model.default_cfg = _cfg()
        missing_keys, unexpected_keys = model.load_state_dict(checkpoint, strict=False)
        print('# missing keys=', missing_keys)
        print('# unexpected keys=', unexpected_keys)
        print('sucessfully loaded from pre-trained weights:', model_path)

        if args.distill:
            print('## Distillation Pruning Mode')
            model_t = PoolingTransformerTeacher(
                image_size=224,
                patch_size=16,
                stride=8,
                base_dims=[48, 48, 48],
                depth=[2, 6, 4],
                heads=[3, 6, 12],
                mlp_ratio=4
            )
            model_t.load_state_dict(checkpoint, strict=False)
            model_t.to(device)
            print('sucessfully loaded from pre-trained weights for the teach model')

    else:
        raise NotImplementedError


    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)

    # model = nn.DataParallel(model)
    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)

    linear_scaled_lr = args.lr * args.batch_size * utils.get_world_size() / 512.0
    args.lr = linear_scaled_lr

    opt_args = dict(lr=args.lr, weight_decay=args.weight_decay)
    if hasattr(args, 'opt_eps') and args.opt_eps is not None:
        opt_args['eps'] = args.opt_eps
    if hasattr(args, 'opt_betas') and args.opt_betas is not None:
        opt_args['betas'] = args.opt_betas
    
    # parameter_group = [
    #     {'params': list(model_without_ddp.blocks.parameters()) + list(model_without_ddp.head.parameters()), 'lr_mult': 0.1},
    #     {'params': model_without_ddp.score_predictor.parameters()},
    # ]

    parameter_group = get_param_groups(model_without_ddp, args.weight_decay)
    
    optimizer = torch.optim.AdamW(parameter_group, **opt_args)
    loss_scaler = NativeScaler()

    # lr_scheduler, _ = create_scheduler(args, optimizer)

    criterion = LabelSmoothingCrossEntropy()

    if args.mixup > 0.:
        # smoothing is handled with mixup label transform
        criterion = SoftTargetCrossEntropy()
    elif args.smoothing:
        criterion = LabelSmoothingCrossEntropy(smoothing=args.smoothing)
    else:
        criterion = torch.nn.CrossEntropyLoss()

    # wrap the criterion in our custom DistillationLoss, which
    # just dispatches to the original criterion if args.distillation_type is 'none'

    if args.distill:
        if 'lvvit' in args.arch: 
            criterion = DistillDiffPruningLoss(
                model_t, criterion, clf_weight=1.0, keep_ratio=KEEP_RATE, ratio_weight=args.ratiow, distill_weight=args.distillw
            )
        else:
            criterion = DistillDiffPruningLoss(
                model_t, criterion, clf_weight=1.0, keep_ratio=KEEP_RATE, mse_token=True, ratio_weight=args.ratiow, distill_weight=args.distillw
            )
    else:
        criterion = DiffPruningLoss(
            criterion, clf_weight=1.0, keep_ratio=KEEP_RATE
        )

    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')
        model_without_ddp.load_state_dict(checkpoint['model'])
        if not args.eval and 'optimizer' 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'])

    if args.eval:
        test_stats = evaluate(data_loader_val, model, device)
        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)

        warmup_step = 5
        adjust_learning_rate(optimizer.param_groups, args.lr, args.min_lr, epoch, args.epochs, warmup_predictor=False, warming_up_step=warmup_step, base_multi=0.1)

        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.finetune == '' # keep in eval mode during finetuning
        )

        # 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)

        test_stats = evaluate(data_loader_val, model, device)
        print(f"Accuracy of the network on the {len(dataset_val)} test images: {test_stats['acc1']:.1f}%")
        max_accuracy = max(max_accuracy, test_stats["acc1"])
        print(f'Max accuracy: {max_accuracy:.2f}%')

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

        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('DynamicViT 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)