stylegan2_ada_pytorch/train.py

# Copyright (c) Facebook, Inc. and its 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.

# All contributions by NVIDIA CORPORATION:
# Copyright (c) 2021, NVIDIA CORPORATION.  All rights reserved.

# NVIDIA CORPORATION and its licensors retain all intellectual property
# and proprietary rights in and to this software, related documentation
# and any modifications thereto.  Any use, reproduction, disclosure or
# distribution of this software and related documentation without an express
# license agreement from NVIDIA CORPORATION is strictly prohibited.

"""Train a GAN using the techniques described in the paper
"Training Generative Adversarial Networks with Limited Data"."""
import sys
import os

sys.path.insert(1, os.path.join(sys.path[0], ".."))
import click
import re
import json
import tempfile
import torch
import dnnlib

import numpy as np

import parser

from training import training_loop
from metrics import metric_main
from torch_utils import training_stats
from torch_utils import custom_ops


# ----------------------------------------------------------------------------


class UserError(Exception):
    pass


# ----------------------------------------------------------------------------


def setup_training_loop_kwargs(
    # General options (not included in desc).
    exp_name=None,  # Experiment name
    slurm=None,  # Using SLURM or not <bool>
    gpus=None,  # Number of GPUs: <int>, default = 1 gpu
    nodes=None,  # Number of nodes: <int>, default = 1 node
    snap=None,  # Snapshot interval: <int>, default = 50 ticks
    metrics=None,  # List of metric names: [], ['fid50k_full'] (default), ...
    seed=None,  # Random seed: <int>, default = 0
    # Dataset.
    data=None,  # Training dataset (required): <path>
    class_cond=None,  # Conditioning on a class label <bool>
    subset=None,  # Train with only N images: <int>, default = all
    mirror=None,  # Augment dataset with x-flips: <bool>, default = False
    # IC-GAN dataset parameters.
    instance_cond=None,  # Conditioning on instance features <bool>
    feature_augmentation=None,  # Horizontal flips augmentation to extract instance features <bool>
    root_feats=None,  # Path where to find the hdf5 file with the instance features <str>
    root_nns=None,  # Path where to find the pre-computed nearest neighbors for each instance <str>
    label_dim=None,  # Dimensionality of the class embeddings if we use class conditonings <int>.
    # Base config.
    cfg=None,  # Base config: 'auto' (default), 'stylegan2', 'paper256', 'paper512', 'paper1024', 'cifar'
    lrate=None,  # Override learning rate: <float>
    gamma=None,  # Override R1 gamma: <float>
    kimg=None,  # Override training duration: <int>
    batch=None,  # Override batch size: <int>
    num_channel_g=None,  # Override width of generator network: <int>
    num_channel_d=None,  # Override width of discriminator network: <int>
    channel_max_g=None,  # Override max width of generator network: <int>
    channel_max_d=None,  # Override max width of discriminator network: <int>
    hidden_dim_c=None,  # Override embedding dimensionality for class conditioning inside mapping network
    hidden_dim_h=None,  # Override embedding dimensionality for instance conditioning inside mapping network
    es_patience=None,  # Early stopping patience in number of seen images: <int>
    # Discriminator augmentation.
    aug=None,  # Augmentation mode: 'ada' (default), 'noaug', 'fixed'
    p=None,  # Specify p for 'fixed' (required): <float>
    target=None,  # Override ADA target for 'ada': <float>, default = depends on aug
    augpipe=None,  # Augmentation pipeline: 'blit', 'geom', 'color', 'filter', 'noise', 'cutout', 'bg', 'bgc' (default), ..., 'bgcfnc'
    # Transfer learning.
    resume=None,  # Load previous network: 'noresume' (default), 'ffhq256', 'ffhq512', 'ffhq1024', 'celebahq256', 'lsundog256', <file>, <url>
    freezed=None,  # Freeze-D: <int>, default = 0 discriminator layers
    # Performance options (not included in desc).
    fp32=None,  # Disable mixed-precision training: <bool>, default = False
    nhwc=None,  # Use NHWC memory format with FP16: <bool>, default = False
    allow_tf32=None,  # Allow PyTorch to use TF32 for matmul and convolutions: <bool>, default = False
    nobench=None,  # Disable cuDNN benchmarking: <bool>, default = False
    workers=None,  # Override number of DataLoader workers: <int>, default = 3
    **kwargs,
):
    args = dnnlib.EasyDict()

    # ------------------------------------------
    # General options: gpus, snap, metrics, seed
    # ------------------------------------------

    if gpus is None:
        gpus = 1
    assert isinstance(gpus, int)
    if not (gpus >= 1 and gpus & (gpus - 1) == 0):
        raise UserError("--gpus must be a power of two")
    args.num_gpus = gpus * nodes

    if snap is None:
        snap = 50
    assert isinstance(snap, int)
    if snap < 1:
        raise UserError("--snap must be at least 1")
    args.image_snapshot_ticks = snap
    args.network_snapshot_ticks = snap
    args.es_patience = es_patience

    if metrics is None:
        metrics = ["fid50k_full"]
    assert isinstance(metrics, list)
    if not all(metric_main.is_valid_metric(metric) for metric in metrics):
        raise UserError(
            "\n".join(
                ["--metrics can only contain the following values:"]
                + metric_main.list_valid_metrics()
            )
        )
    args.metrics = metrics

    if seed is None:
        seed = 0
    assert isinstance(seed, int)
    args.random_seed = seed

    # -----------------------------------
    # Dataset: data, cond, subset, mirror
    # -----------------------------------

    assert data is not None
    assert isinstance(data, str)

    class_name = "data_utils.datasets_common.ILSVRC_HDF5_feats"
    args.class_cond = class_cond
    args.instance_cond = instance_cond

    if mirror is None:
        mirror = False
    assert isinstance(mirror, bool)

    args.training_set_kwargs = dnnlib.EasyDict(
        class_name=class_name,
        root=data,
        max_size=None,
        xflip=False,
        load_labels=class_cond,
        load_features=instance_cond,
        root_feats=root_feats,
        root_nns=root_nns,
        transform=None,
        label_dim=label_dim,
        feature_dim=2048,
        apply_norm=False,
        label_onehot=True,
        feature_augmentation=feature_augmentation,
    )
    args.data_loader_kwargs = dnnlib.EasyDict(
        pin_memory=True, num_workers=3, prefetch_factor=2
    )
    try:
        training_set = dnnlib.util.construct_class_by_name(
            **args.training_set_kwargs
        )  # subclass of training.dataset.Dataset
        args.training_set_kwargs.resolution = (
            training_set.resolution
        )  # be explicit about resolution
        args.training_set_kwargs.load_labels = class_cond
        args.training_set_kwargs.max_size = len(
            training_set
        )  # be explicit about dataset size
        desc = os.path.splitext(os.path.basename(data))[0]
        del training_set  # conserve memory
    except IOError as err:
        raise UserError(f"--data: {err}")

    if mirror:
        desc += "-mirror"
        args.training_set_kwargs.xflip = True

    # if load_labels:
    #     if not args.training_set_kwargs.load_labels:
    #         raise UserError('--cond=True requires labels specified in dataset.json')
    #     desc += '-cond'
    # else:
    #     args.training_set_kwargs.load_labels = False
    # if load_features and not load_labels:
    #     args.training_set_kwargs.label_dim=2048

    if subset is not None:
        assert isinstance(subset, int)
        if not 1 <= subset <= args.training_set_kwargs.max_size:
            raise UserError(
                f"--subset must be between 1 and {args.training_set_kwargs.max_size}"
            )
        desc += f"-subset{subset}"
        if subset < args.training_set_kwargs.max_size:
            args.training_set_kwargs.max_size = subset
            args.training_set_kwargs.random_seed = args.random_seed

    # ------------------------------------
    # Base config: cfg, gamma, kimg, batch
    # ------------------------------------

    if cfg is None:
        cfg = "auto"
    assert isinstance(cfg, str)
    desc += f"-{cfg}"

    cfg_specs = {
        "auto": dict(
            ref_gpus=-1,
            kimg=25000,
            mb=-1,
            mbstd=-1,
            fmaps=-1,
            lrate=-1,
            gamma=-1,
            ema=-1,
            ramp=0.05,
            map=2,
        ),  # Populated dynamically based on resolution and GPU count.
        "stylegan2": dict(
            ref_gpus=8,
            kimg=25000,
            mb=32,
            mbstd=4,
            fmaps=1,
            lrate=0.002,
            gamma=10,
            ema=10,
            ramp=None,
            map=8,
        ),  # Uses mixed-precision, unlike the original StyleGAN2.
        "paper256": dict(
            ref_gpus=8,
            kimg=25000,
            mb=64,
            mbstd=8,
            fmaps=0.5,
            lrate=0.0025,
            gamma=1,
            ema=20,
            ramp=None,
            map=8,
        ),
        "paper512": dict(
            ref_gpus=8,
            kimg=25000,
            mb=64,
            mbstd=8,
            fmaps=1,
            lrate=0.0025,
            gamma=0.5,
            ema=20,
            ramp=None,
            map=8,
        ),
        "paper1024": dict(
            ref_gpus=8,
            kimg=25000,
            mb=32,
            mbstd=4,
            fmaps=1,
            lrate=0.002,
            gamma=2,
            ema=10,
            ramp=None,
            map=8,
        ),
        "cifar": dict(
            ref_gpus=2,
            kimg=100000,
            mb=64,
            mbstd=32,
            fmaps=1,
            lrate=0.0025,
            gamma=0.01,
            ema=500,
            ramp=0.05,
            map=2,
        ),
    }

    assert cfg in cfg_specs
    spec = dnnlib.EasyDict(cfg_specs[cfg])
    if cfg == "auto":
        desc += f"{gpus:d}"
        spec.ref_gpus = args.num_gpus
        res = args.training_set_kwargs.resolution
        spec.mb = max(
            min(args.num_gpus * min(4096 // res, 32), 64), args.num_gpus
        )  # keep gpu memory consumption at bay
        spec.mbstd = min(
            spec.mb // args.num_gpus, 4
        )  # other hyperparams behave more predictably if mbstd group size remains fixed
        spec.fmaps = 1 if res >= 512 else 0.5
        spec.lrate = 0.002 if res >= 1024 else 0.0025
        spec.gamma = 0.0002 * (res ** 2) / spec.mb  # heuristic formula
        spec.ema = spec.mb * 10 / 32

    args.G_kwargs = dnnlib.EasyDict(
        class_name="training.networks.Generator",
        z_dim=512,
        w_dim=512,
        mapping_kwargs=dnnlib.EasyDict(),
        synthesis_kwargs=dnnlib.EasyDict(),
    )
    args.D_kwargs = dnnlib.EasyDict(
        class_name="training.networks.Discriminator",
        block_kwargs=dnnlib.EasyDict(),
        mapping_kwargs=dnnlib.EasyDict(),
        epilogue_kwargs=dnnlib.EasyDict(),
    )
    args.G_kwargs.synthesis_kwargs.channel_base = args.D_kwargs.channel_base = int(
        spec.fmaps * 32768
    )
    args.G_kwargs.synthesis_kwargs.channel_max = args.D_kwargs.channel_max = 512
    args.G_kwargs.mapping_kwargs.num_layers = spec.map
    if hidden_dim_c is not None:
        args.G_kwargs.mapping_kwargs.embed_features = hidden_dim_c
        args.D_kwargs.mapping_kwargs.embed_features = hidden_dim_c
    if hidden_dim_h is not None:
        args.G_kwargs.mapping_kwargs.embed_features_feat = hidden_dim_h
        args.D_kwargs.mapping_kwargs.embed_features_feat = hidden_dim_h
    args.G_kwargs.synthesis_kwargs.num_fp16_res = (
        args.D_kwargs.num_fp16_res
    ) = 4  # enable mixed-precision training
    args.G_kwargs.synthesis_kwargs.conv_clamp = (
        args.D_kwargs.conv_clamp
    ) = 256  # clamp activations to avoid float16 overflow
    args.D_kwargs.epilogue_kwargs.mbstd_group_size = spec.mbstd

    args.exp_name = exp_name
    if num_channel_d is not None:
        args.D_kwargs.channel_base = num_channel_d
    if channel_max_d is not None:
        args.D_kwargs.channel_max = channel_max_d
    if num_channel_g is not None:
        args.G_kwargs.synthesis_kwargs.channel_base = num_channel_g
    if channel_max_g is not None:
        args.G_kwargs.synthesis_kwargs.channel_max = channel_max_g

    if lrate is not None:
        spec.lrate = lrate

    args.G_opt_kwargs = dnnlib.EasyDict(
        class_name="torch.optim.Adam", lr=spec.lrate, betas=[0, 0.99], eps=1e-8
    )
    args.D_opt_kwargs = dnnlib.EasyDict(
        class_name="torch.optim.Adam", lr=spec.lrate, betas=[0, 0.99], eps=1e-8
    )
    args.loss_kwargs = dnnlib.EasyDict(
        class_name="training.loss.StyleGAN2Loss", r1_gamma=spec.gamma
    )

    args.total_kimg = spec.kimg
    args.batch_size = spec.mb
    args.batch_gpu = spec.mb // spec.ref_gpus
    args.ema_kimg = spec.ema
    args.ema_rampup = spec.ramp

    if cfg == "cifar":
        args.loss_kwargs.pl_weight = 0  # disable path length regularization
        args.loss_kwargs.style_mixing_prob = 0  # disable style mixing
        args.D_kwargs.architecture = "orig"  # disable residual skip connections

    if gamma is not None:
        assert isinstance(gamma, float)
        if not gamma >= 0:
            raise UserError("--gamma must be non-negative")
        desc += f"-gamma{gamma:g}"
        args.loss_kwargs.r1_gamma = gamma

    if kimg is not None:
        assert isinstance(kimg, int)
        if not kimg >= 1:
            raise UserError("--kimg must be at least 1")
        desc += f"-kimg{kimg:d}"
        args.total_kimg = kimg

    if batch is not None:
        assert isinstance(batch, int)
        if not (batch >= 1 and batch % args.num_gpus == 0):
            raise UserError(
                "--batch must be at least 1 and divisible by --gpus and --nodes"
            )
        desc += f"-batch{batch}"
        args.batch_size = batch
        args.batch_gpu = batch // (args.num_gpus)
    args.slurm = slurm

    # ---------------------------------------------------
    # Discriminator augmentation: aug, p, target, augpipe
    # ---------------------------------------------------

    if aug is None:
        aug = "ada"
    else:
        assert isinstance(aug, str)
        desc += f"-{aug}"

    if aug == "ada":
        args.ada_target = 0.6

    elif aug == "noaug":
        pass

    elif aug == "fixed":
        if p is None:
            raise UserError(f"--aug={aug} requires specifying --p")

    else:
        raise UserError(f"--aug={aug} not supported")

    if p is not None:
        assert isinstance(p, float)
        if aug != "fixed":
            raise UserError("--p can only be specified with --aug=fixed")
        if not 0 <= p <= 1:
            raise UserError("--p must be between 0 and 1")
        desc += f"-p{p:g}"
        args.augment_p = p

    if target is not None:
        assert isinstance(target, float)
        if aug != "ada":
            raise UserError("--target can only be specified with --aug=ada")
        if not 0 <= target <= 1:
            raise UserError("--target must be between 0 and 1")
        desc += f"-target{target:g}"
        args.ada_target = target

    assert augpipe is None or isinstance(augpipe, str)
    if augpipe is None:
        augpipe = "bgc"
    else:
        if aug == "noaug":
            raise UserError("--augpipe cannot be specified with --aug=noaug")
        desc += f"-{augpipe}"

    augpipe_specs = {
        "blit": dict(xflip=1, rotate90=1, xint=1),
        "geom": dict(scale=1, rotate=1, aniso=1, xfrac=1),
        "color": dict(brightness=1, contrast=1, lumaflip=1, hue=1, saturation=1),
        "filter": dict(imgfilter=1),
        "noise": dict(noise=1),
        "cutout": dict(cutout=1),
        "bg": dict(xflip=1, rotate90=1, xint=1, scale=1, rotate=1, aniso=1, xfrac=1),
        "bgc": dict(
            xflip=1,
            rotate90=1,
            xint=1,
            scale=1,
            rotate=1,
            aniso=1,
            xfrac=1,
            brightness=1,
            contrast=1,
            lumaflip=1,
            hue=1,
            saturation=1,
        ),
        "bgcf": dict(
            xflip=1,
            rotate90=1,
            xint=1,
            scale=1,
            rotate=1,
            aniso=1,
            xfrac=1,
            brightness=1,
            contrast=1,
            lumaflip=1,
            hue=1,
            saturation=1,
            imgfilter=1,
        ),
        "bgcfn": dict(
            xflip=1,
            rotate90=1,
            xint=1,
            scale=1,
            rotate=1,
            aniso=1,
            xfrac=1,
            brightness=1,
            contrast=1,
            lumaflip=1,
            hue=1,
            saturation=1,
            imgfilter=1,
            noise=1,
        ),
        "bgcfnc": dict(
            xflip=1,
            rotate90=1,
            xint=1,
            scale=1,
            rotate=1,
            aniso=1,
            xfrac=1,
            brightness=1,
            contrast=1,
            lumaflip=1,
            hue=1,
            saturation=1,
            imgfilter=1,
            noise=1,
            cutout=1,
        ),
    }

    assert augpipe in augpipe_specs
    if aug != "noaug":
        args.augment_kwargs = dnnlib.EasyDict(
            class_name="training.augment.AugmentPipe", **augpipe_specs[augpipe]
        )

    # ----------------------------------
    # Transfer learning: resume, freezed
    # ----------------------------------

    resume_specs = {
        "ffhq256": "https://nvlabs-fi-cdn.nvidia.com/stylegan2-ada-pytorch/pretrained/transfer-learning-source-nets/ffhq-res256-mirror-paper256-noaug.pkl",
        "ffhq512": "https://nvlabs-fi-cdn.nvidia.com/stylegan2-ada-pytorch/pretrained/transfer-learning-source-nets/ffhq-res512-mirror-stylegan2-noaug.pkl",
        "ffhq1024": "https://nvlabs-fi-cdn.nvidia.com/stylegan2-ada-pytorch/pretrained/transfer-learning-source-nets/ffhq-res1024-mirror-stylegan2-noaug.pkl",
        "celebahq256": "https://nvlabs-fi-cdn.nvidia.com/stylegan2-ada-pytorch/pretrained/transfer-learning-source-nets/celebahq-res256-mirror-paper256-kimg100000-ada-target0.5.pkl",
        "lsundog256": "https://nvlabs-fi-cdn.nvidia.com/stylegan2-ada-pytorch/pretrained/transfer-learning-source-nets/lsundog-res256-paper256-kimg100000-noaug.pkl",
    }

    assert resume is None or isinstance(resume, str)
    if resume is None:
        resume = "noresume"
    elif resume == "noresume":
        desc += "-noresume"
    elif resume in resume_specs:
        desc += f"-resume{resume}"
        args.resume_pkl = resume_specs[resume]  # predefined url
    else:
        desc += "-resumecustom"
        args.resume_pkl = resume  # custom path or url

    if resume != "noresume":
        args.ada_kimg = 100  # make ADA react faster at the beginning
        args.ema_rampup = None  # disable EMA rampup

    if freezed is not None:
        assert isinstance(freezed, int)
        if not freezed >= 0:
            raise UserError("--freezed must be non-negative")
        desc += f"-freezed{freezed:d}"
        args.D_kwargs.block_kwargs.freeze_layers = freezed

    # -------------------------------------------------
    # Performance options: fp32, nhwc, nobench, workers
    # -------------------------------------------------

    if fp32 is None:
        fp32 = False
    assert isinstance(fp32, bool)
    if fp32:
        args.G_kwargs.synthesis_kwargs.num_fp16_res = args.D_kwargs.num_fp16_res = 0
        args.G_kwargs.synthesis_kwargs.conv_clamp = args.D_kwargs.conv_clamp = None

    if nhwc is None:
        nhwc = False
    assert isinstance(nhwc, bool)
    if nhwc:
        args.G_kwargs.synthesis_kwargs.fp16_channels_last = (
            args.D_kwargs.block_kwargs.fp16_channels_last
        ) = True

    if nobench is None:
        nobench = False
    assert isinstance(nobench, bool)
    if nobench:
        args.cudnn_benchmark = False

    if allow_tf32 is None:
        allow_tf32 = False
    assert isinstance(allow_tf32, bool)
    if allow_tf32:
        args.allow_tf32 = True

    if workers is not None:
        assert isinstance(workers, int)
        if not workers >= 1:
            raise UserError("--workers must be at least 1")
        args.data_loader_kwargs.num_workers = workers

    return desc, args


# ----------------------------------------------------------------------------


def subprocess_fn(rank, args, world_size=1, dist_url="", temp_dir="", slurm=False):
    dnnlib.util.Logger(
        file_name=os.path.join(args.run_dir, "log.txt"),
        file_mode="a",
        should_flush=True,
    )

    # Init torch.distributed.
    if not slurm and args.num_gpus > 1:
        init_file = os.path.abspath(os.path.join(temp_dir, ".torch_distributed_init"))
        if os.name == "nt":
            init_method = "file:///" + init_file.replace("\\", "/")
            torch.distributed.init_process_group(
                backend="gloo",
                init_method=init_method,
                rank=rank,
                world_size=args.num_gpus,
            )
        else:
            init_method = f"file://{init_file}"
            torch.distributed.init_process_group(
                backend="nccl",
                init_method=init_method,
                rank=rank,
                world_size=args.num_gpus,
            )
        # Init torch_utils.
        sync_device = torch.device("cuda", rank) if args.num_gpus > 1 else None
        training_stats.init_multiprocessing(rank=rank, sync_device=sync_device)
        local_rank = rank

    elif slurm:
        rank = int(os.environ.get("SLURM_PROCID"))
        local_rank = int(os.environ.get("SLURM_LOCALID"))
        torch.distributed.init_process_group(
            backend="nccl", init_method=dist_url, rank=rank, world_size=world_size
        )
    else:
        rank = local_rank = 0

    if rank != 0:
        custom_ops.verbosity = "none"

    # Execute training loop.
    training_loop.training_loop(
        rank=rank, local_rank=local_rank, temp_dir=temp_dir, **args
    )


# ----------------------------------------------------------------------------


class CommaSeparatedList(click.ParamType):
    name = "list"

    def convert(self, value, param, ctx):
        _ = param, ctx
        if value is None or value.lower() == "none" or value == "":
            return []
        return value.split(",")


# ----------------------------------------------------------------------------


def main(args, outdir, master_node="", port=40000, dry_run=False, **config_kwargs):
    """Train a GAN using the techniques described in the paper
    "Training Generative Adversarial Networks with Limited Data".

    Examples:

    \b
    # Train with custom dataset using 1 GPU.
    python train.py --outdir=~/training-runs --data=~/mydataset.zip --gpus=1

    \b
    # Train class-conditional CIFAR-10 using 2 GPUs.
    python train.py --outdir=~/training-runs --data=~/datasets/cifar10.zip \\
        --gpus=2 --cfg=cifar --cond=1

    \b
    # Transfer learn MetFaces from FFHQ using 4 GPUs.
    python train.py --outdir=~/training-runs --data=~/datasets/metfaces.zip \\
        --gpus=4 --cfg=paper1024 --mirror=1 --resume=ffhq1024 --snap=10

    \b
    # Reproduce original StyleGAN2 config F.
    python train.py --outdir=~/training-runs --data=~/datasets/ffhq.zip \\
        --gpus=8 --cfg=stylegan2 --mirror=1 --aug=noaug

    \b
    Base configs (--cfg):
      auto       Automatically select reasonable defaults based on resolution
                 and GPU count. Good starting point for new datasets.
      stylegan2  Reproduce results for StyleGAN2 config F at 1024x1024.
      paper256   Reproduce results for FFHQ and LSUN Cat at 256x256.
      paper512   Reproduce results for BreCaHAD and AFHQ at 512x512.
      paper1024  Reproduce results for MetFaces at 1024x1024.
      cifar      Reproduce results for CIFAR-10 at 32x32.

    \b
    Transfer learning source networks (--resume):
      ffhq256        FFHQ trained at 256x256 resolution.
      ffhq512        FFHQ trained at 512x512 resolution.
      ffhq1024       FFHQ trained at 1024x1024 resolution.
      celebahq256    CelebA-HQ trained at 256x256 resolution.
      lsundog256     LSUN Dog trained at 256x256 resolution.
      <PATH or URL>  Custom network pickle.
    """
    dnnlib.util.Logger(should_flush=True)

    # Setup training options.
    config_kwargs = vars(args)
    run_desc, args = setup_training_loop_kwargs(**config_kwargs)
    args.metrics = ["fid50k_full"]

    if args.exp_name is None:
        # Pick output directory.
        prev_run_dirs = []
        if os.path.isdir(outdir):
            prev_run_dirs = [
                x for x in os.listdir(outdir) if os.path.isdir(os.path.join(outdir, x))
            ]
        prev_run_ids = [re.match(r"^\d+", x) for x in prev_run_dirs]
        prev_run_ids = [int(x.group()) for x in prev_run_ids if x is not None]
        cur_run_id = max(prev_run_ids, default=-1) + 1
        args.run_dir = os.path.join(outdir, f"{cur_run_id:05d}-{run_desc}")
        assert not os.path.exists(args.run_dir)
    else:
        args.run_dir = os.path.join(outdir, args.exp_name)

    # Print options.
    print()
    print("Training options:")
    #  print(json.dumps(args, indent=2))
    print()
    print(f"Output directory:   {args.run_dir}")
    print(f"Training data:      {args.training_set_kwargs.root}")
    print(f"Training duration:  {args.total_kimg} kimg")
    print(f"Number of GPUs:     {args.num_gpus}")
    print(f"Number of images:   {args.training_set_kwargs.max_size}")
    print(f"Image resolution:   {args.training_set_kwargs.resolution}")
    print(f"Conditional model:  {args.training_set_kwargs.load_labels}")
    print(f"Dataset x-flips:    {args.training_set_kwargs.xflip}")
    print()

    # Dry run?
    if dry_run:
        print("Dry run; exiting.")
        return

    # Create output directory.
    print("Creating output directory...")
    if not os.path.exists(args.run_dir):
        os.makedirs(args.run_dir, exist_ok=True)
    with open(os.path.join(args.run_dir, "training_options.json"), "wt") as f:
        json.dump(args, f, indent=2)

    ## Multi-gpu or multi-node training ##
    if args.slurm:
        n_nodes = int(os.environ.get("SLURM_JOB_NUM_NODES"))
        n_gpus_per_node = int(os.environ.get("SLURM_TASKS_PER_NODE").split("(")[0])
        world_size = n_gpus_per_node * n_nodes
        dist_url = "tcp://"
        dist_url += master_node
        dist_url += ":" + str(port)
        print("Dist url ", dist_url)
        temp_dir = "/scratch/slurm_tmpdir/" + str(os.environ.get("SLURM_JOB_ID"))
        subprocess_fn(
            rank=-1,
            args=args,
            world_size=world_size,
            dist_url=dist_url,
            temp_dir=temp_dir,
            slurm=args.slurm,
        )
    else:
        # Launch processes.
        print("Launching processes...")
        torch.multiprocessing.set_start_method("spawn")
        with tempfile.TemporaryDirectory() as temp_dir:
            if args.num_gpus == 1:
                subprocess_fn(rank=0, args=args, temp_dir=temp_dir)
            else:
                torch.multiprocessing.spawn(
                    fn=subprocess_fn,
                    args=(args, args.num_gpus, "", temp_dir),
                    nprocs=args.num_gpus,
                )


# ----------------------------------------------------------------------------

if __name__ == "__main__":
    parser_ = parser.get_parser()
    args = parser_.parse_args()
    main(args)  # pylint: disable=no-value-for-parameter

# ----------------------------------------------------------------------------