run_class_finetuning.py

import argparse
import datetime
import numpy as np
import time
import torch
import torch.backends.cudnn as cudnn
import json
import os
from functools import partial
from pathlib import Path
from collections import OrderedDict

from mixup import Mixup
from timm.models import create_model
from timm.loss import LabelSmoothingCrossEntropy, SoftTargetCrossEntropy
from timm.utils import ModelEma
from optim_factory import create_optimizer, get_parameter_groups, LayerDecayValueAssigner

from datasets import build_dataset
from engine_for_finetuning import train_one_epoch, validation_one_epoch, final_test, merge, merge_mean_per_class
from utils_mae import NativeScalerWithGradNormCount as NativeScaler
from utils_mae import  multiple_samples_collate
import utils_mae as utils
import modeling_finetune


def get_args():
    parser = argparse.ArgumentParser('VideoMAE fine-tuning and evaluation script for video classification', add_help=False)
    parser.add_argument('--batch_size', default=64, type=int)
    parser.add_argument('--epochs', default=30, type=int)
    parser.add_argument('--update_freq', default=1, type=int)
    parser.add_argument('--save_ckpt_freq', default=100, type=int)
    parser.add_argument('--val_freq', default=1, type=int)

    # Model parameters
    parser.add_argument('--model', default='vit_base_patch16_224', type=str, metavar='MODEL',
                        help='Name of model to train')
    parser.add_argument('--tubelet_size', type=int, default= 2)
    parser.add_argument('--input_size', default=224, type=int,
                        help='videos input size')

    parser.add_argument('--fc_drop_rate', type=float, default=0.0, metavar='PCT',
                        help='Dropout rate (default: 0.)')
    parser.add_argument('--drop', type=float, default=0.0, metavar='PCT',
                        help='Dropout rate (default: 0.)')
    parser.add_argument('--attn_drop_rate', type=float, default=0.0, metavar='PCT',
                        help='Attention 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('--disable_eval_during_finetuning', action='store_true', default=False)
    parser.add_argument('--model_ema', action='store_true', default=False)
    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('--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)')
    parser.add_argument('--weight_decay_end', type=float, default=None, help="""Final value of the
        weight decay. We use a cosine schedule for WD and using a larger decay by
        the end of training improves performance for ViTs.""")

    parser.add_argument('--lr', type=float, default=1e-3, metavar='LR',
                        help='learning rate (default: 1e-3)')
    parser.add_argument('--layer_decay', type=float, default=0.75)

    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-6, metavar='LR',
                        help='lower lr bound for cyclic schedulers that hit 0 (1e-5)')

    parser.add_argument('--warmup_epochs', type=int, default=5, metavar='N',
                        help='epochs to warmup LR, if scheduler supports')
    parser.add_argument('--warmup_steps', type=int, default=-1, metavar='N',
                        help='num of steps to warmup LR, will overload warmup_epochs if set > 0')

    # Augmentation parameters
    parser.add_argument('--color_jitter', type=float, default=0.4, metavar='PCT',
                        help='Color jitter factor (default: 0.4)')
    parser.add_argument('--num_sample', type=int, default=2,
                        help='Repeated_aug (default: 2)')
    parser.add_argument('--aa', type=str, default='rand-m7-n4-mstd0.5-inc1', metavar='NAME',
                        help='Use AutoAugment policy. "v0" or "original". " + "(default: rand-m7-n4-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")')

    # Evaluation parameters
    parser.add_argument('--crop_pct', type=float, default=None)
    parser.add_argument('--short_side_size', type=int, default=224)
    parser.add_argument('--test_num_segment', type=int, default=5)
    parser.add_argument('--test_num_crop', type=int, default=3)
    
    # 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.')
    parser.add_argument('--cutmix', type=float, default=1.0,
                        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('--model_key', default='model|module', type=str)
    parser.add_argument('--model_prefix', default='', type=str)
    parser.add_argument('--init_scale', default=0.001, type=float)
    parser.add_argument('--use_checkpoint', action='store_true')
    parser.set_defaults(use_checkpoint=False)
    parser.add_argument('--use_mean_pooling', action='store_true')
    parser.set_defaults(use_mean_pooling=True)
    parser.add_argument('--use_cls', action='store_false', dest='use_mean_pooling')

    # Dataset parameters
    parser.add_argument('--data_path', default='/path/to/list_kinetics-400', type=str,
                        help='dataset path')
    parser.add_argument('--eval_data_path', default=None, type=str,
                        help='dataset path for evaluation')
    parser.add_argument('--nb_classes', default=400, type=int,
                        help='number of the classification types')
    parser.add_argument('--imagenet_default_mean_and_std', default=True, action='store_true')
    parser.add_argument('--num_segments', type=int, default= 1)
    parser.add_argument('--num_frames', type=int, default= 16)
    parser.add_argument('--sampling_rate', type=int, default= 4)
    parser.add_argument('--data_set', default='Kinetics-400', choices=['Kinetics-400', 'SSV2', 'UCF101', 'HMDB51','image_folder','SSV2-Mini', 'Mini-Kinetics'],
                        type=str, help='dataset')
    parser.add_argument('--output_dir', default='',
                        help='path where to save, empty for no saving')
    parser.add_argument('--log_dir', default=None,
                        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('--auto_resume', action='store_true')
    parser.add_argument('--no_auto_resume', action='store_false', dest='auto_resume')
    parser.set_defaults(auto_resume=True)

    parser.add_argument('--save_ckpt', action='store_true')
    parser.add_argument('--no_save_ckpt', action='store_false', dest='save_ckpt')
    parser.set_defaults(save_ckpt=True)

    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')
    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('--enable_deepspeed', action='store_true', default=False)

    # debug mode
    parser.add_argument('--not_dist', action='store_true', default=False)
    parser.add_argument('--num_outputs', default=8, type=int)

    known_args, _ = parser.parse_known_args()

    if known_args.enable_deepspeed:
        try:
            import deepspeed
            from deepspeed import DeepSpeedConfig
            parser = deepspeed.add_config_arguments(parser)
            ds_init = deepspeed.initialize
        except:
            print("Please 'pip install deepspeed'")
            exit(0)
    else:
        ds_init = None

    return parser.parse_args(), ds_init


def main(args, ds_init):
    if args.not_dist:
        args.distributed = False
    else:
        utils.init_distributed_mode(args)

    if ds_init is not None:
        utils.create_ds_config(args)

    print(args)

    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, test_mode=False, args=args)
    if args.disable_eval_during_finetuning:
        dataset_val = None
    else:
        dataset_val, _ = build_dataset(is_train=False, test_mode=False, args=args)
    dataset_test, _ = build_dataset(is_train=False, test_mode=True, args=args)
    

    num_tasks = utils.get_world_size()
    global_rank = utils.get_rank()
    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=False)
        sampler_test = torch.utils.data.DistributedSampler(
            dataset_test, num_replicas=num_tasks, rank=global_rank, shuffle=False)
    else:
        sampler_val = torch.utils.data.SequentialSampler(dataset_val)

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

    if args.num_sample > 1:
        collate_func = partial(multiple_samples_collate, fold=False)
    else:
        collate_func = 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,
        collate_fn=collate_func,
    )

    if dataset_val is not None:
        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
        )
    else:
        data_loader_val = None

    if dataset_test is not None:
        data_loader_test = torch.utils.data.DataLoader(
            dataset_test, sampler=sampler_test,
            batch_size=args.batch_size,
            num_workers=args.num_workers,
            pin_memory=args.pin_mem,
            drop_last=False
        )
    else:
        data_loader_test = None

    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 = create_model(
        args.model,
        pretrained=False,
        num_classes=args.nb_classes,
        all_frames=args.num_frames * args.num_segments,
        tubelet_size=args.tubelet_size,
        fc_drop_rate=args.fc_drop_rate,
        drop_rate=args.drop,
        drop_path_rate=args.drop_path,
        attn_drop_rate=args.attn_drop_rate,
        drop_block_rate=None,
        use_checkpoint=args.use_checkpoint,
        use_mean_pooling=args.use_mean_pooling,
        init_scale=args.init_scale,
    )

    patch_size = model.patch_embed.patch_size
    print("Patch size = %s" % str(patch_size))
    args.window_size = (args.num_frames // 2, args.input_size // patch_size[0], args.input_size // patch_size[1])
    args.patch_size = patch_size

    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')

        print("Load ckpt from %s" % args.finetune)
        checkpoint_model = None
        for model_key in args.model_key.split('|'):
            if model_key in checkpoint:
                checkpoint_model = checkpoint[model_key]
                print("Load state_dict by model_key = %s" % model_key)
                break
        if checkpoint_model is None:
            checkpoint_model = checkpoint
        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]

        all_keys = list(checkpoint_model.keys())
        new_dict = OrderedDict()
        for key in all_keys:
            if key.startswith('backbone.'):
                new_dict[key[9:]] = checkpoint_model[key]
            elif key.startswith('encoder.'):
                new_dict[key[8:]] = checkpoint_model[key]
            else:
                new_dict[key] = checkpoint_model[key]
        checkpoint_model = new_dict

        # interpolate position embedding
        if 'pos_embed' in checkpoint_model:
            pos_embed_checkpoint = checkpoint_model['pos_embed']
            embedding_size = pos_embed_checkpoint.shape[-1] # channel dim
            num_patches = model.patch_embed.num_patches # 
            num_extra_tokens = model.pos_embed.shape[-2] - num_patches # 0/1

            # height (== width) for the checkpoint position embedding 
            orig_size = int(((pos_embed_checkpoint.shape[-2] - num_extra_tokens)//(args.num_frames // model.patch_embed.tubelet_size)) ** 0.5)
            # height (== width) for the new position embedding
            new_size = int((num_patches // (args.num_frames // model.patch_embed.tubelet_size) )** 0.5)
            # class_token and dist_token are kept unchanged
            if orig_size != new_size:
                print("Position interpolate from %dx%d to %dx%d" % (orig_size, orig_size, new_size, new_size))
                extra_tokens = pos_embed_checkpoint[:, :num_extra_tokens]
                # only the position tokens are interpolated
                pos_tokens = pos_embed_checkpoint[:, num_extra_tokens:]
                # B, L, C -> BT, H, W, C -> BT, C, H, W
                pos_tokens = pos_tokens.reshape(-1, args.num_frames // model.patch_embed.tubelet_size, orig_size, orig_size, embedding_size)
                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)
                # BT, C, H, W -> BT, H, W, C ->  B, T, H, W, C
                pos_tokens = pos_tokens.permute(0, 2, 3, 1).reshape(-1, args.num_frames // model.patch_embed.tubelet_size, new_size, new_size, embedding_size) 
                pos_tokens = pos_tokens.flatten(1, 3) # B, L, C
                new_pos_embed = torch.cat((extra_tokens, pos_tokens), dim=1)
                checkpoint_model['pos_embed'] = new_pos_embed

        utils.load_state_dict(model, checkpoint_model, prefix=args.model_prefix)

    model.to(device)

    model_ema = None
    if args.model_ema:
        model_ema = ModelEma(
            model,
            decay=args.model_ema_decay,
            device='cpu' if args.model_ema_force_cpu else '',
            resume='')
        print("Using EMA with decay = %.8f" % args.model_ema_decay)

    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:', n_parameters)

    total_batch_size = args.batch_size * args.update_freq * utils.get_world_size()
    num_training_steps_per_epoch = len(dataset_train) // total_batch_size
    args.lr = args.lr * total_batch_size / 256
    args.min_lr = args.min_lr * total_batch_size / 256
    args.warmup_lr = args.warmup_lr * total_batch_size / 256
    print("LR = %.8f" % args.lr)
    print("Batch size = %d" % total_batch_size)
    print("Update frequent = %d" % args.update_freq)
    print("Number of training examples = %d" % len(dataset_train))
    print("Number of training training per epoch = %d" % num_training_steps_per_epoch)

    num_layers = model_without_ddp.get_num_layers()
    if args.layer_decay < 1.0:
        assigner = LayerDecayValueAssigner(list(args.layer_decay ** (num_layers + 1 - i) for i in range(num_layers + 2)))
    else:
        assigner = None

    if assigner is not None:
        print("Assigned values = %s" % str(assigner.values))

    skip_weight_decay_list = model.no_weight_decay()
    print("Skip weight decay list: ", skip_weight_decay_list)

    if args.enable_deepspeed:
        loss_scaler = None
        optimizer_params = get_parameter_groups(
            model, args.weight_decay, skip_weight_decay_list,
            assigner.get_layer_id if assigner is not None else None,
            assigner.get_scale if assigner is not None else None)
        model, optimizer, _, _ = ds_init(
            args=args, model=model, model_parameters=optimizer_params, dist_init_required=not args.distributed,
        )

        print("model.gradient_accumulation_steps() = %d" % model.gradient_accumulation_steps())
        assert model.gradient_accumulation_steps() == args.update_freq
    else:
        if args.distributed:
            model = torch.nn.parallel.DistributedDataParallel(model, device_ids=[args.gpu], find_unused_parameters=True)
            model_without_ddp = model.module

        optimizer = create_optimizer(
            args, model_without_ddp, skip_list=skip_weight_decay_list,
            get_num_layer=assigner.get_layer_id if assigner is not None else None, 
            get_layer_scale=assigner.get_scale if assigner is not None else None)
        loss_scaler = NativeScaler()

    print("Use step level LR scheduler!")
    lr_schedule_values = utils.cosine_scheduler(
        args.lr, args.min_lr, args.epochs, num_training_steps_per_epoch,
        warmup_epochs=args.warmup_epochs, warmup_steps=args.warmup_steps,
    )
    if args.weight_decay_end is None:
        args.weight_decay_end = args.weight_decay
    wd_schedule_values = utils.cosine_scheduler(
        args.weight_decay, args.weight_decay_end, args.epochs, num_training_steps_per_epoch)
    print("Max WD = %.7f, Min WD = %.7f" % (max(wd_schedule_values), min(wd_schedule_values)))

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

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

    utils.auto_load_model(
        args=args, model=model, model_without_ddp=model_without_ddp,
        optimizer=optimizer, loss_scaler=loss_scaler, model_ema=model_ema)

    if args.eval:
        if not args.not_dist:
            preds_file = os.path.join(args.output_dir, str(global_rank) + '.txt')
            test_stats = final_test(data_loader_test, model, device, preds_file)
            torch.distributed.barrier()
        else:
            num_tasks = args.num_outputs
        
        if global_rank == 0:
            print("Start merging results...")
            final_top1 ,final_top5 = merge(args.output_dir, num_tasks)
            print(f"Accuracy of the network on the {len(dataset_test)} test videos: Top-1: {final_top1:.2f}%, Top-5: {final_top5:.2f}%")
            log_stats = {'Final top-1': final_top1,
                        'Final Top-5': final_top5}
            
            final_top1_per_class ,final_top5_per_class = merge_mean_per_class(args.output_dir, num_tasks,args.nb_classes)
            print(f"Accuracy of the network on the {len(dataset_test)} test videos: Mean-Top-1: {final_top1_per_class:.2f}%, Mean-Top-5: {final_top5_per_class:.2f}%")
            log_stats["Class-Mean-Top-1"] = final_top1_per_class
            log_stats["Class-Mean-Top-5"] = final_top5_per_class

            if args.output_dir and utils.is_main_process():
                with open(os.path.join(args.output_dir, "log.txt"), mode="a", encoding="utf-8") as f:
                    f.write(json.dumps(log_stats) + "\n")
        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.distributed:
            data_loader_train.sampler.set_epoch(epoch)
        if log_writer is not None:
            log_writer.set_step(epoch * num_training_steps_per_epoch * args.update_freq)
        train_stats = train_one_epoch(
            model, criterion, data_loader_train, optimizer,
            device, epoch, loss_scaler, args.clip_grad, model_ema, mixup_fn,
            log_writer=log_writer, start_steps=epoch * num_training_steps_per_epoch,
            lr_schedule_values=lr_schedule_values, wd_schedule_values=wd_schedule_values,
            num_training_steps_per_epoch=num_training_steps_per_epoch, update_freq=args.update_freq,
        )
        if args.output_dir and args.save_ckpt:
            if (epoch + 1) % args.save_ckpt_freq == 0 or epoch + 1 == args.epochs:
                utils.save_model(
                    args=args, model=model, model_without_ddp=model_without_ddp, optimizer=optimizer,
                    loss_scaler=loss_scaler, epoch=epoch, model_ema=model_ema)
        if data_loader_val is not None and (epoch + 1) % args.val_freq == 0:
            test_stats = validation_one_epoch(data_loader_val, model, device)
            print(f"Accuracy of the network on the {len(dataset_val)} val videos: {test_stats['acc1']:.1f}%")
            if max_accuracy < test_stats["acc1"]:
                max_accuracy = test_stats["acc1"]
                if args.output_dir and args.save_ckpt:
                    utils.save_model(
                        args=args, model=model, model_without_ddp=model_without_ddp, optimizer=optimizer,
                        loss_scaler=loss_scaler, epoch="best", model_ema=model_ema)

            print(f'Max accuracy: {max_accuracy:.2f}%')
            if log_writer is not None:
                log_writer.update(val_acc1=test_stats['acc1'], head="perf", step=epoch)
                log_writer.update(val_acc5=test_stats['acc5'], head="perf", step=epoch)
                log_writer.update(val_loss=test_stats['loss'], head="perf", step=epoch)

            log_stats = {**{f'train_{k}': v for k, v in train_stats.items()},
                         **{f'val_{k}': v for k, v in test_stats.items()},
                         'epoch': epoch,
                         'n_parameters': n_parameters}
        else:
            log_stats = {**{f'train_{k}': v for k, v in train_stats.items()},
                         'epoch': epoch,
                         'n_parameters': n_parameters}
        if args.output_dir and utils.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")

    preds_file = os.path.join(args.output_dir, str(global_rank) + '.txt')
    test_stats = final_test(data_loader_test, model, device, preds_file)
    torch.distributed.barrier()
    if global_rank == 0:
        print("Start merging results...")
        final_top1 ,final_top5 = merge(args.output_dir, num_tasks)
        print(f"Accuracy of the network on the {len(dataset_test)} test videos: Top-1: {final_top1:.2f}%, Top-5: {final_top5:.2f}%")
        log_stats = {'Final top-1': final_top1,
                    'Final Top-5': final_top5}
        if args.output_dir and utils.is_main_process():
            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__':
    opts, ds_init = get_args()
    if opts.output_dir:
        Path(opts.output_dir).mkdir(parents=True, exist_ok=True)
    main(opts, ds_init)