main_finetune.py

import argparse
import datetime
import json
import numpy as np
import os
import time
from pathlib import Path

import torch
import torch.backends.cudnn as cudnn
from torch.utils.tensorboard import SummaryWriter


from timm.models.layers import trunc_normal_
from timm.data.mixup import Mixup
from timm.loss import LabelSmoothingCrossEntropy, SoftTargetCrossEntropy
from timm.utils import ModelEma

import util.lr_decay as lrd
import util.misc as misc
from util.datasets import build_dataset
from util.pos_embed import interpolate_pos_embed
from util.misc import NativeScalerWithGradNormCount as NativeScaler

import modeling.encoders.models_vit as models_vit

from engine_finetune import train_one_epoch, evaluate
from samplers import RASampler


def get_args_parser():
    parser = argparse.ArgumentParser('ADDP fine-tuning for image classification', add_help=False)
    parser.add_argument('--batch_size', default=64, type=int,
                        help='Batch size per GPU (effective batch size is batch_size * accum_iter * # gpus')
    parser.add_argument('--epochs', default=50, type=int)
    parser.add_argument('--accum_iter', default=1, type=int,
                        help='Accumulate gradient iterations (for increasing the effective batch size under memory constraints)')

    # Model parameters
    parser.add_argument('--model', default='vit_large_patch16', 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_path', type=float, default=0.1, metavar='PCT',
                        help='Drop path rate (default: 0.1)')
    
    parser.add_argument('--model-ema', default=False, action='store_true')
    parser.add_argument('--no-model-ema', action='store_false', dest='model_ema')
    parser.add_argument('--model-ema-decay', type=float, default=0.9999, help='')
    parser.add_argument('--model-ema-force-cpu', action='store_true', default=False, help='')

    # Optimizer parameters
    parser.add_argument('--clip_grad', type=float, default=None, metavar='NORM',
                        help='Clip gradient norm (default: None, no clipping)')
    parser.add_argument('--weight_decay', type=float, default=0.05,
                        help='weight decay (default: 0.05)')

    parser.add_argument('--lr', type=float, default=None, metavar='LR',
                        help='learning rate (absolute lr)')
    parser.add_argument('--blr', type=float, default=1e-3, metavar='LR',
                        help='base learning rate: absolute_lr = base_lr * total_batch_size / 256')
    parser.add_argument('--layer_decay', type=float, default=0.75,
                        help='layer-wise lr decay from ELECTRA/BEiT')

    parser.add_argument('--min_lr', type=float, default=1e-6, metavar='LR',
                        help='lower lr bound for cyclic schedulers that hit 0')

    parser.add_argument('--warmup_epochs', type=int, default=5, metavar='N',
                        help='epochs to warmup LR')
    
    parser.add_argument('--beta2', type=float, default=0.999)

    # Augmentation parameters
    parser.add_argument('--color_jitter', type=float, default=None, metavar='PCT',
                        help='Color jitter factor (enabled only when not using Auto/RandAug)')
    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('--crop_min', type=float, default=0.08,)

    parser.add_argument('--repeated-aug', action='store_true', default=False)
    # 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,
                        help='mixup alpha, mixup enabled if > 0.')
    parser.add_argument('--cutmix', type=float, default=0,
                        help='cutmix alpha, cutmix enabled if > 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"')

    # * Finetuning params
    parser.add_argument('--finetune', default='',
                        help='finetune from checkpoint')
    parser.add_argument('--global_pool', default='avg', choices=['avg', 'cls', 'pixel+vq', 'pixel'])
    # parser.set_defaults(global_pool=True)
    parser.add_argument('--cls_token', action='store_false', dest='global_pool',
                        help='Use class token instead of global pool for classification')

    # Dataset parameters
    parser.add_argument('--data_path', default='/datasets01/imagenet_full_size/061417/', type=str,
                        help='dataset path')
    parser.add_argument('--nb_classes', default=1000, type=int,
                        help='number of the classification types')

    parser.add_argument('--output_dir', default='./output_dir',
                        help='path where to save, empty for no saving')
    parser.add_argument('--log_dir', default='./output_dir',
                        help='path where to tensorboard log')
    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 (recommended during training for faster monitor')
    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')
    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('--local_rank', default=-1, type=int)
    parser.add_argument('--dist_on_itp', action='store_true')
    parser.add_argument('--dist_url', default='env://',
                        help='url used to set up distributed training')
    parser.add_argument('--port', default=28507, type=int)
    
    parser.add_argument('--auto_resume', action='store_true', default=True)
    parser.add_argument('--init_values', default=1.0, type=float)

    parser.add_argument('--early_stop_epoch', type=int, help='early stop at which epoch')

    parser.add_argument('--loss_type', default='ce', choices=['ce', 'bce'])
    parser.add_argument('--resume_model_only', default=False, action='store_true')

    return parser


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

    print('job dir: {}'.format(os.path.dirname(os.path.realpath(__file__))))
    print("{}".format(args).replace(', ', ',\n'))

    device = torch.device(args.device)

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

    cudnn.benchmark = True
    torch.backends.cuda.matmul.allow_tf32 = False
    torch.backends.cudnn.allow_tf32 = False

    dataset_train = build_dataset(is_train=True, args=args)
    dataset_val = build_dataset(is_train=False, args=args)

    if True:  # args.distributed:
        num_tasks = misc.get_world_size()
        global_rank = misc.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
            )
        print("Sampler_train = %s" % str(sampler_train))
        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=True)  # shuffle=True to reduce monitor bias
        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)

    if global_rank == 0 and args.log_dir is not None and not args.eval:
        os.makedirs(args.log_dir, exist_ok=True)
        log_writer = SummaryWriter(log_dir=args.log_dir)
    else:
        log_writer = None

    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=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:
        print("Mixup is activated!")
        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)

    model = models_vit.__dict__[args.model](
        num_classes=args.nb_classes,
        drop_path_rate=args.drop_path,
        global_pool=args.global_pool,
        init_values=args.init_values if args.init_values != 1.0 else None,
        args=args,
        img_size=args.input_size,
    )

    if args.finetune and not args.eval:
        checkpoint = torch.load(args.finetune, map_location='cpu')
        print("Load pre-trained checkpoint from: %s" % args.finetune)
        checkpoint_model = checkpoint['model']
        if 'resnet' in args.model:
            for k in list(checkpoint_model.keys()):
                if k.startswith('resnet'):
                    new_k = k.replace('resnet.', '')
                    if not new_k.startswith('fc'):
                        checkpoint_model[new_k] = checkpoint_model[k]
                    del checkpoint_model[k]

        state_dict = model.state_dict()
        for k in ['head.weight', 'head.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]

        if 'resnet' not in args.model:
            # interpolate position embedding
            interpolate_pos_embed(model, checkpoint_model)

        # load pre-trained model
        msg = model.load_state_dict(checkpoint_model, strict=False)
        print(msg)

        # if args.global_pool:
        #     assert set(msg.missing_keys) == {'head.weight', 'head.bias', 'fc_norm.weight', 'fc_norm.bias'}
        # else:
        #     assert set(msg.missing_keys) == {'head.weight', 'head.bias'}

        # manually initialize fc layer
        if hasattr(model, 'head'):
            trunc_normal_(model.head.weight, std=2e-5)

    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
    n_parameters = sum(p.numel() for p in model.parameters() if p.requires_grad)

    print("Model = %s" % str(model_without_ddp))
    print('number of params (M): %.2f' % (n_parameters / 1.e6))

    eff_batch_size = args.batch_size * args.accum_iter * misc.get_world_size()
    
    if args.lr is None:  # only base_lr is specified
        args.lr = args.blr * eff_batch_size / 256

    print("base lr: %.2e" % (args.lr * 256 / eff_batch_size))
    print("actual lr: %.2e" % args.lr)

    print("accumulate grad iterations: %d" % args.accum_iter)
    print("effective batch size: %d" % eff_batch_size)

    if args.distributed:
        model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu], find_unused_parameters=True)
        model_without_ddp = model.module

    if 'resnet' in args.model:
        param_groups = model_without_ddp.parameters()
    else:
        # build optimizer with layer-wise lr decay (lrd)
        param_groups = lrd.param_groups_lrd(model_without_ddp, args.weight_decay,
            no_weight_decay_list=model_without_ddp.no_weight_decay(),
            layer_decay=args.layer_decay
        )
    optimizer = torch.optim.AdamW(param_groups, lr=args.lr, betas=(0.9, args.beta2))
    loss_scaler = NativeScaler()

    if mixup_fn is not None:
        # smoothing is handled with mixup label transform
        if args.loss_type == 'bce':
            criterion = torch.nn.BCEWithLogitsLoss()
        else:
            criterion = SoftTargetCrossEntropy()
    elif args.smoothing > 0.:
        criterion = LabelSmoothingCrossEntropy(smoothing=args.smoothing)
    else:
        criterion = torch.nn.CrossEntropyLoss()

    print("criterion = %s" % str(criterion))

    misc.auto_load_model(
        args=args, model_without_ddp=model_without_ddp, optimizer=optimizer, loss_scaler=loss_scaler, model_ema=model_ema)

    if args.eval:
        test_stats = evaluate(data_loader_val, model, device, args)
        print(f"Accuracy of the network on the {len(dataset_val)} test images: {test_stats['acc1']:.1f}%")
        exit(0)

    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.early_stop_epoch is None else args.early_stop_epoch):
        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, mixup_fn,
            log_writer=log_writer,
            args=args,
            model_ema=model_ema,
        )
        if args.output_dir:
            misc.save_model(
                args=args, model=model, model_without_ddp=model_without_ddp, optimizer=optimizer,
                loss_scaler=loss_scaler, epoch=epoch, latest=True, model_ema=model_ema)

        if (epoch+1)%1 == 0:
            test_stats = evaluate(data_loader_val, model, device, args)
            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 log_writer is not None:
                log_writer.add_scalar('perf/test_acc1', test_stats['acc1'], epoch)
                log_writer.add_scalar('perf/test_acc5', test_stats['acc5'], epoch)
                log_writer.add_scalar('perf/test_loss', test_stats['loss'], epoch)

            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 misc.is_main_process():
                if log_writer is not None:
                    log_writer.flush()
                with open(os.path.join(args.output_dir, "log.txt"), mode="a", encoding="utf-8") 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__':
    args = get_args_parser()
    args = args.parse_args()
    if args.output_dir:
        Path(args.output_dir).mkdir(parents=True, exist_ok=True)
    main(args)