train.py

import argparse
import collections
import torch
import numpy as np
import data_loader.CIFAR10_4x_data_loader as module_data
import model.loss as module_loss
import model.metric as module_metric
import model.model as module_arch
import model.transform as module_transform
from parse_config import ConfigParser
from trainer import Trainer
from utils import prepare_device, get_model_storage_size


# fix random seeds for reproducibility
SEED = 16
torch.manual_seed(SEED)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
np.random.seed(SEED)

def main(config: ConfigParser):
    logger = config.get_logger('train')

    # Dynamically build the transformations based on config
    train_transform = module_transform.create_transfrom(config['train_transform'])
    valid_transform = module_transform.create_transfrom(config['valid_transform'])

    # setup data_loader instances
    train_data_loader = config.init_obj('train_data_loader', module_data, transform=train_transform)
    valid_data_loader = config.init_obj('valid_data_loader', module_data, transform=valid_transform)

    # build model architecture, then print to console
    model = module_arch.create_model(config)
    logger.info(model)

    # Get and log model storage size
    model_storage_size = get_model_storage_size(model)
    logger.info(f"Model storage size: {model_storage_size:.2f} MB")

    # prepare for (multi-device) GPU training
    device, device_ids = prepare_device(config['n_gpu'])
    model = model.to(device)
    if len(device_ids) > 1:
        model = torch.nn.DataParallel(model, device_ids=device_ids)

    # get function handles of metrics
    criterion = module_loss.create_criterion(config['loss'])
    metrics = [getattr(module_metric, met) for met in config['metrics']]

    # build optimizer, learning rate scheduler.
    trainable_params = filter(lambda p: p.requires_grad, model.parameters())
    optimizer = config.init_obj('optimizer', torch.optim, trainable_params)
    lr_scheduler = config.init_obj('lr_scheduler', torch.optim.lr_scheduler, optimizer)

    trainer = Trainer(model, criterion, metrics, optimizer,
                      config=config,
                      device=device,
                      train_data_loader=train_data_loader,
                      valid_data_loader=valid_data_loader,
                      lr_scheduler=lr_scheduler)

    trainer.train()


if __name__ == '__main__':
    args = argparse.ArgumentParser(description='PyTorch Template')
    args.add_argument('-c', '--config', default=None, type=str,
                      help='config file path (default: None)')
    args.add_argument('-r', '--resume', default=None, type=str,
                      help='path to latest checkpoint (default: None)')
    args.add_argument('-d', '--device', default=None, type=str,
                      help='indices of GPUs to enable (default: all)')

    # custom cli options to modify configuration from default values given in json file.
    CustomArgs = collections.namedtuple('CustomArgs', 'flags type target')
    options = [
        CustomArgs(['--lr', '--learning_rate'], type=float, target='optimizer;args;lr'),
        CustomArgs(['--bs', '--batch_size'], type=int, target='data_loader;args;batch_size')
    ]
    config = ConfigParser.from_args(args, options)
    main(config)