def run_model()

in data_augmentation/my_training.py [0:0]

• 22 lines of code
• 5 McCabe index (conditional complexity)

def run_model(params, ckpt_path=None, repo=None):
    args = params

    if args.seed is not None:
        random.seed(args.seed)
        torch.manual_seed(args.seed)

    if args.gpu is not None:
        warnings.warn(
            "You have chosen a specific GPU. This will completely "
            "disable data parallelism."
        )

    ngpus_per_node = torch.cuda.device_count()

    if args.distributed:
        torch.cuda.set_device(args.rank)
        torch.distributed.init_process_group(
            backend=args.dist_backend,
            init_method="tcp://{}:{}".format("localhost", 10001),
            world_size=args.world_size,
            rank=args.rank,
        )
    main_worker(args.gpu, ngpus_per_node, args, ckpt_path, repo)

    # cleanup distributed
    if args.distributed:
        cleanup_distributed()