run_stage2.py

# Copyright (c) 2024 The Johns Hopkins University Applied Physics Laboratory

# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:

# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.

# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.

"""
Code for Stage 2 of UNITE: supervised fine-tuning on source domain data
"""

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
import yaml

from timm.models import create_model
from timm.loss import LabelSmoothingCrossEntropy, SoftTargetCrossEntropy
from timm.utils import ModelEma
import wandb

from src.optim_factory import create_optimizer, get_parameter_groups, LayerDecayValueAssigner
from src.datasets.mixup import Mixup
from src.datasets import build_dataset
from src.datasets.distributed import DistributedSampler
from src.engines.engine_for_finetuning import train_one_epoch, validation_one_epoch, final_test, merge
from src.utils import NativeScalerWithGradNormCount as NativeScaler
from src.utils import multiple_samples_collate
from src import utils
from src.models import *


def get_args(args=None):
    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)

    # 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('--use_learnable_pos_emb', action='store_true')
    parser.set_defaults(use_learnable_pos_emb=False)
    parser.add_argument('--train_head_only', action='store_true', default=False)
    parser.add_argument('--frozen_layers', default='', type=str,
                        help='which transformer layers to freeze, comma separated')
    parser.add_argument('--freeze_patch_embedding', type=utils.str2bool, nargs='?', const=True, default=False)
    parser.add_argument('--head_type', default='linear', type=str, choices=['linear', 'mlp'])
    parser.add_argument('--head_hidden_dim', default=256, type=int)

    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_schedule', type=str, default='cosine', help='Learning rate schedule (default: constant)', choices=['constant', 'cosine', 'step'])
    parser.add_argument('--step_fraction', type=float, default=0.1, help='Fraction by which to decay at each step (default: 0.1)')
    parser.add_argument('--lr_step_epochs', type=int, nargs='+', default=None, help='Epochs at which to decay learning rate (default: [5])')

    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-6)')

    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('--delete_head', action='store_true', help='whether delete head')
    parser.add_argument('--no_delete_head', action='store_false', dest='delete_head')
    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('--checkpoint_num', default=0, type=int,
                        help='number of layers for using checkpoint')
    parser.add_argument('--use_mean_pooling', action='store_true')
    parser.set_defaults(use_mean_pooling=False)
    parser.add_argument('--use_cls', action='store_false', dest='use_mean_pooling')

    # Dataset parameters
    parser.add_argument('--dataset', default='', type=str, help='name of domain shift dataset. ' 
                        'if this is specified, we will automatically override things like ' 
                        'ann file paths and number of classes.')
    parser.add_argument('--prefix', default='', type=str, help='prefix for data')
    parser.add_argument('--split', default=' ', type=str, help='split for metadata')
    parser.add_argument('--data_path', default='you_data_path', type=str,
                        help='dataset path')
    parser.add_argument('--eval_data_path', default=None, type=str,
                        help='dataset path for evaluation')
    parser.add_argument('--train_fraction', default=1.0, type=float)
    parser.add_argument('--train_repetitions', default=1, type=int)
    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('--use_decord', default=False, action='store_true',
                        help='whether use decord to load video, otherwise load image')
    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', choices=[
        'Kinetics', 'Kinetics_sparse', 
        'SSV2', 'UCF101', 'HMDB51', 'image_folder',
        'mitv1_sparse'
        ], type=str, help='dataset')
    parser.add_argument('--ann_file_train', default=None, type=str, help='annotation path')
    parser.add_argument('--ann_file_val', default=None, type=str, help='annotation path')
    parser.add_argument('--ann_file_test', default=None, type=str, help='annotation path')
    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('--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('--reset_train_dataset', action='store_true', help='recreates the train dataset at every epoch. '
                         'this is useful when using a fractional train set, since we dont sacrifice diversity.')
    parser.add_argument('--no_reset_train_dataset', action='store_false', dest='reset_train_data')
    parser.set_defaults(reset_train_data=False)

    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('--test_best', action='store_true',
                        help='Whether test the best model')
    parser.add_argument('--eval', type=utils.str2bool, nargs='?', const=True, default=False,
                        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)
    parser.add_argument('--auto_reload', action='store_true')
    parser.add_argument('--no_auto_reload', action='store_false', dest='auto_reload')
    parser.set_defaults(auto_reload=True)
    parser.add_argument('--eval_freq', default=1, type=int)

    parser.add_argument('--lp_ft_epochs', default=0, type=int)

    # 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('--distributed', action='store_true', default=False)

    parser.add_argument('--enable_deepspeed', action='store_true', default=False)

    parser.add_argument('--disable_wandb', action='store_true', default=False)
    parser.add_argument('--wandb_entity', type=str)
    parser.add_argument('--wandb_project', type=str)
    parser.add_argument('--wandb_group', default=None, type=str)

    # YAML config
    parser.add_argument('--config', default='', type=str, help='yaml config file path')
    if args is not None:
        known_args, _ = parser.parse_known_args(args)
    else:
        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

    # if args is not None:
    #     cmd_args =  parser.parse_args(args)
    # else:
    #     cmd_args =  parser.parse_args(), ds_init
    # first, read the args
    cmd_args = parser.parse_args()

    if cmd_args.config:
        # Get the configs from the yaml file
        yaml_args = argparse.Namespace()
        with open(cmd_args.config, 'r') as f:
            yaml_args.__dict__ = yaml.safe_load(f)
        # Overwrite yaml args with commandline args
        all_args = parser.parse_args(namespace=yaml_args)
    else:
        all_args = cmd_args

    if all_args.dataset:
        all_args = update_dataset_args_from_yaml(all_args)

    return all_args, ds_init

def update_dataset_args_from_yaml(args):
    # See if yaml file path exists (dataset_mappings.yaml)
    dataset_mappings_path = os.path.join(os.path.dirname(__file__), 'dataset_mappings.yaml')
    if os.path.exists(dataset_mappings_path):
        with open(dataset_mappings_path, 'r') as f:
            dataset_mappings = yaml.safe_load(f)
        try:
            dataset_args = dataset_mappings[args.dataset]
            for k, v in dataset_args.items():
                setattr(args, k, v)
                print("Updated %s to %s" % (k, v))
        except KeyError:
            print(f"Dataset <{args.dataset}> not found in dataset_mappings.yaml")
            raise KeyError
    else:
        print("No dataset_mappings.yaml file found, skipping update_dataset_args_from_yaml!")
        raise FileNotFoundError

    return args

def get_model(args):
    print(f"Creating model: {args.model}")
    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,
        use_learnable_pos_emb=args.use_learnable_pos_emb,
        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,
        checkpoint_num=args.checkpoint_num,
        use_mean_pooling=args.use_mean_pooling,
        init_scale=args.init_scale,
        classifier_type=args.head_type,
        classifier_hidden_dim=args.head_hidden_dim,
    )
    return model

def load_from_ckpt(args, model):
    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

    if 'head.weight' in checkpoint_model.keys():
        if args.delete_head:
            print("Removing head from pretrained checkpoint")
            del checkpoint_model['head.weight']
            del checkpoint_model['head.bias']
        elif checkpoint_model['head.weight'].shape[0] == 710:
            if args.nb_classes == 400:
                checkpoint_model['head.weight'] = checkpoint_model['head.weight'][:args.nb_classes]
                checkpoint_model['head.bias'] = checkpoint_model['head.bias'][:args.nb_classes]
            elif args.nb_classes in [600, 700]:
                # download from https://drive.google.com/drive/folders/17cJd2qopv-pEG8NSghPFjZo1UUZ6NLVm
                map_path = f'k710/label_mixto{args.nb_classes}.json'
                print(f'Load label map from {map_path}')
                with open(map_path) as f:
                    label_map = json.load(f)
                checkpoint_model['head.weight'] = checkpoint_model['head.weight'][label_map]
                checkpoint_model['head.bias'] = checkpoint_model['head.bias'][label_map]
                
    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

        # we use 8 frames for pretraining
        orig_t_size = 8 // model.patch_embed.tubelet_size
        new_t_size = args.num_frames // model.patch_embed.tubelet_size
        # height (== width) for the checkpoint position embedding
        orig_size = int(((pos_embed_checkpoint.shape[-2] - num_extra_tokens)//(orig_t_size)) ** 0.5)
        # height (== width) for the new position embedding
        new_size = int((num_patches // (new_t_size) )** 0.5)
        
        if orig_t_size != new_t_size:
            print(f"Temporal interpolate from {orig_t_size} to {new_t_size}")
            tmp_pos_embed = pos_embed_checkpoint.view(1, orig_t_size, -1, embedding_size)
            tmp_pos_embed = tmp_pos_embed.permute(0, 2, 3, 1).reshape(-1, embedding_size, orig_t_size)
            tmp_pos_embed = torch.nn.functional.interpolate(tmp_pos_embed, size=new_t_size, mode='linear')
            tmp_pos_embed = tmp_pos_embed.view(1, -1, embedding_size, new_t_size)
            tmp_pos_embed = tmp_pos_embed.permute(0, 3, 1, 2).reshape(1, -1, embedding_size)
            checkpoint_model['pos_embed'] = tmp_pos_embed
            pos_embed_checkpoint = tmp_pos_embed

        # 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, new_t_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, new_t_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)
    return model

def remake_train_dataloader(num_tasks, global_rank, collate_func, args):
   dataset_train, args.nb_classes = build_dataset(is_train=True, test_mode=False, args=args)
   sampler_train = torch.utils.data.DistributedSampler(
        dataset_train, num_replicas=num_tasks, rank=global_rank, shuffle=True)
   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,
        persistent_workers=True)
   print("Made new train dataloader.")
   return data_loader_train

def main(args, ds_init):
    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

    use_wandb = (utils.is_main_process() and
                 not args.disable_wandb and
                'scrap' not in args.output_dir.lower()
    )

    if use_wandb:
        wandb_run = wandb.init(entity=args.wandb_entity, 
                               project=args.wandb_project, 
                               config=args, 
                               group=args.wandb_group,
                               name=args.output_dir.split('/')[-1])
    else:
        wandb_run = None

    if utils.is_main_process() and args.output_dir:
        Path(args.output_dir).mkdir(parents=True, exist_ok=True)
        # Save args as a yaml file to output directory
        with open(os.path.join(args.output_dir, "config.yaml"), "w") as f:
            yaml.dump(vars(args), f, default_flow_style=False)

    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 = DistributedSampler(
        dataset_train, num_replicas=num_tasks, rank=global_rank, shuffle=True,
        repetitions=args.train_repetitions
    )
    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.output_dir is not None:
        os.makedirs(args.output_dir, exist_ok=True)
        log_writer = utils.TensorboardLogger(log_dir=args.output_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,
        persistent_workers=True
    )

    if dataset_val is not None:
        data_loader_val = torch.utils.data.DataLoader(
            dataset_val, sampler=sampler_val,
            batch_size=int(2 * args.batch_size),
            num_workers=args.num_workers,
            pin_memory=args.pin_mem,
            drop_last=False,
            persistent_workers=True
        )
    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=int(4 * args.batch_size),
            num_workers=args.num_workers,
            pin_memory=args.pin_mem,
            drop_last=False,
            persistent_workers=True
        )
    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 = get_model(args)

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

    if args.finetune:
        model = load_from_ckpt(args, model)

    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
    args.min_lr = args.min_lr
    args.warmup_lr = args.warmup_lr

    print("LR = %.8f" % args.lr)
    print("Batch size = %d" % total_batch_size)
    print("Repeated sample = %d" % args.num_sample)
    print("Update frequent = %d" % args.update_freq)
    print("Number of training examples = %d" % len(dataset_train))
    print("Number of training steps 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=False)
            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()


    if args.lr_schedule == 'cosine':
        lr_schedule_values = utils.cosine_scheduler(
            args.lr, args.min_lr, args.epochs, num_training_steps_per_epoch,
            warmup_epochs=args.warmup_epochs, start_warmup_value=args.warmup_lr, warmup_steps=args.warmup_steps,
        )
    elif args.lr_schedule == 'constant':
        lr_schedule_values = utils.step_scheduler(
            args.lr, args.step_fraction, args.epochs, num_training_steps_per_epoch,
            warmup_epochs=args.warmup_epochs, start_warmup_value=args.warmup_lr, warmup_steps=args.warmup_steps,
        )
    elif args.lr_schedule == 'step':
        assert args.lr_step_epochs is not None
        lr_schedule_values = utils.step_scheduler(
            args.lr, args.step_fraction, args.epochs, num_training_steps_per_epoch,
            warmup_epochs=args.warmup_epochs, start_warmup_value=args.warmup_lr, warmup_steps=args.warmup_steps,
            steps=args.lr_step_epochs,
        )
    
    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))

    if args.eval:
        preds_file = os.path.join(args.output_dir, str(global_rank) + '.txt')
        test_stats, ece = 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")
            if use_wandb:
                wandb_run.log({'test/acc1': final_top1, 'test/acc5': final_top5})
        exit(0)

    if args.auto_reload:
        utils.auto_load_model(
            args=args, model=model, model_without_ddp=model_without_ddp,
            optimizer=optimizer, loss_scaler=loss_scaler, model_ema=model_ema)
        
    ######################################
    ### Configure Trainable Parameters ###
    ######################################

    def freeze_params(model, frozen_layer_substrings):
        frozen_params = []
        trainable_params = []
        for name, param in model.named_parameters():
            if any([n in name for n in frozen_layer_substrings]):
                param.requires_grad = False
                frozen_params.append(name)
            else:
                param.requires_grad = True
                trainable_params.append(name)
        print("Trainable parameters:\n{}".format(trainable_params))
        print("Frozen parameters:\n{}".format(frozen_params))
        return model

    if args.train_head_only:
        for name, param in model.named_parameters():
            if "head" in name or "norm.weight" in name or "norm.bias"in name:
                param.requires_grad = True
                print("Training {}".format(name))
            else:
                param.requires_grad = False
    elif args.frozen_layers:
        frozen_layer_indices = [int(n) for n in args.frozen_layers.split(",")]
        frozen_layer_substrings = ['blocks.'+str(n)+'.' for n in frozen_layer_indices]
        if args.freeze_patch_embedding:
            frozen_layer_substrings.append('patch_embed')
        model = freeze_params(model, frozen_layer_substrings)
    else:
        pass

    if args.lp_ft_epochs > 0:
        # we will freeze all but last few layers for the first few epochs then unfreeze all layers
        frozen_layer_indices = [0,1,2,3,4,5,6,7,8]
        frozen_layer_substrings = ['blocks.'+str(n)+'.' for n in frozen_layer_indices]
        frozen_layer_substrings.append('patch_embed')
        model = freeze_params(model, frozen_layer_substrings)
        

    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.reset_train_dataset:
            data_loader_train = remake_train_dataloader(num_tasks, global_rank, collate_func, args)
        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)

        if args.lp_ft_epochs > 0 and epoch == args.lp_ft_epochs:
            model = freeze_params(model, [])
        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, num_epochs=args.epochs,
            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, train_head_only=args.train_head_only,
            wandb_run=wandb_run if use_wandb else None, args=args
        )
        if use_wandb:
            try:
                train_acc = train_stats['class_acc']
            except KeyError:
                train_acc = None
            wandb_run.log({'train/accuracy': train_acc, 'train/epoch': epoch})
        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)
            utils.save_latest_model(
                args=args, model=model, model_without_ddp=model_without_ddp, optimizer=optimizer,
                loss_scaler=loss_scaler, epoch=epoch, model_name='latest', model_ema=model_ema)
        if data_loader_val is not None and (epoch + 1) % args.eval_freq == 0:
            test_stats, ece = validation_one_epoch(data_loader_val, model, device)
            if use_wandb:
                # update keys of test_stats to have val/ prefix
                val_stats = {f'val/{k}': v for k, v in test_stats.items()}
                val_stats['val/ece'] = ece
                wandb_run.log(val_stats)
            timestep = time.strftime("%Y-%m-%d %H:%M:%S", time.localtime())
            print(f"[{timestep}] 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_latest_model(
                        args=args, model=model, model_without_ddp=model_without_ddp, optimizer=optimizer,
                        loss_scaler=loss_scaler, epoch=epoch, model_name='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")
    
    ###############
    ### Testing ###
    ###############

    preds_file = os.path.join(args.output_dir, str(global_rank) + '.txt')
    if args.test_best:
        time.sleep(10) # wait for the best model to be saved
        utils.auto_load_model(
            args=args, model=model, model_without_ddp=model_without_ddp,
            optimizer=optimizer, loss_scaler=loss_scaler, model_ema=model_ema)
    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 use_wandb:
        wandb.log({'test/acc1': final_top1, 'test/acc5': final_top5})
        wandb_run.finish()

if __name__ == '__main__':
    opts, ds_init = get_args()
    main(opts, ds_init)