main_pretrain_wp.py

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.

# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.
# --------------------------------------------------------
# References:
# DeiT: https://github.com/facebookresearch/deit
# BEiT: https://github.com/microsoft/unilm/tree/master/beit
# --------------------------------------------------------
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
import torchvision.transforms as transforms
import torchvision.datasets as datasets
import torch.multiprocessing as mp
import timm

assert timm.__version__ == "0.3.2"  # version check
import timm.optim.optim_factory as optim_factory

import util.misc as misc
from util.misc import NativeScalerWithGradNormCount as NativeScaler

import models_mae

from engine_pretrain_wp import train_one_epoch

from policy_models import MAE_Policy, get_policy_conf, load_policy_model


def get_args_parser():
	parser = argparse.ArgumentParser('MAE pre-training', 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=400, 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='mae_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('--mask_ratio', default=0.75, type=float,
						help='Masking ratio (percentage of removed patches).')

	parser.add_argument('--norm_pix_loss', action='store_true',
						help='Use (per-patch) normalized pixels as targets for computing loss')
	parser.set_defaults(norm_pix_loss=False)

	# Optimizer parameters
	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('--min_lr', type=float, default=0., metavar='LR',
						help='lower lr bound for cyclic schedulers that hit 0')

	parser.add_argument('--warmup_epochs', type=int, default=40, metavar='N',
						help='epochs to warmup LR')

	# Dataset parameters
	parser.add_argument('--data_path', default='/datasets01/imagenet_full_size/061417/', type=str,
						help='dataset path')

	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('--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')

	# policy
	parser.add_argument('--mode', default='policy', type=str,
						help='mode = random or policy')
	parser.add_argument('--policy_config', default="", type=str, 
						help = "path to config file of policy network")

	return parser


def main(rank, args):
	args.rank = rank
	misc.init_distributed_mode_v2(args)

	print("mode", args.mode)
	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

	# simple augmentation
	transform_train = transforms.Compose([
			transforms.RandomResizedCrop(args.input_size, scale=(0.2, 1.0), interpolation=3),  # 3 is bicubic
			transforms.RandomHorizontalFlip(),
			transforms.ToTensor(),
			transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])])

	dataset_train = datasets.ImageFolder(os.path.join(args.data_path, 'train'), transform=transform_train)
	print(dataset_train)

	if True:  # args.distributed:
		num_tasks = misc.get_world_size()
		global_rank = misc.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))
	else:
		sampler_train = torch.utils.data.RandomSampler(dataset_train)

	if global_rank == 0 and args.log_dir is not None:
		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,
	)
	
	# define the model
	model = models_mae.__dict__[args.model](norm_pix_loss=args.norm_pix_loss, mode=args.mode)

	model.to(device)

	if args.mode == "policy":
		policy_args = get_policy_conf(args.policy_config) #hard code file name for now
		policy_model = MAE_Policy(policy_args)
		policy_model = load_policy_model(policy_model, policy_args=policy_args)
		print("using optimal mae")
		policy_model.to(device)
		policy_model.eval()
	else:
		policy_model = None
		print("using random mae")

	model_without_ddp = model
	print("Model = %s" % str(model_without_ddp))

	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
	
	# following timm: set wd as 0 for bias and norm layers
	param_groups = optim_factory.add_weight_decay(model_without_ddp, args.weight_decay)
	optimizer = torch.optim.AdamW(param_groups, lr=args.lr, betas=(0.9, 0.95))
	print(optimizer)
	loss_scaler = NativeScaler()

	misc.load_model(args=args, model_without_ddp=model_without_ddp, optimizer=optimizer, loss_scaler=loss_scaler)

	print(f"Start training for {args.epochs} epochs")

	start_time = time.time()
	for epoch in range(args.start_epoch, args.epochs):
		if args.distributed:
			data_loader_train.sampler.set_epoch(epoch)
		train_stats = train_one_epoch(
			model, data_loader_train,
			optimizer, device, epoch, loss_scaler,
			log_writer=log_writer,
			policy_model=policy_model,
			args=args
		)
		if args.output_dir and (epoch % 20 == 0 or epoch + 1 == args.epochs):
			misc.save_model(
				args=args, model=model, model_without_ddp=model_without_ddp, optimizer=optimizer,
				loss_scaler=loss_scaler, epoch=epoch)

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

		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)
	print("entering main")

	#main(args)

	if args.dist_url == "env://" and args.world_size == -1:
		args.world_size = int(os.environ["WORLD_SIZE"])

	ngpus_per_node = torch.cuda.device_count()

	print("gpu count", ngpus_per_node)

	#args.ngpus_per_node = ngpus_per_node
	args.world_size = ngpus_per_node * args.world_size
	mp.spawn(main, 
				nprocs=ngpus_per_node, 
				args=(args,))

'''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)
	print("entering main")

	main(args)
'''