InvokeAI/main.py

995 lines
33 KiB
Python
Raw Normal View History

2021-12-21 02:23:41 +00:00
import argparse, os, sys, datetime, glob, importlib, csv
import numpy as np
import time
import torch
2022-08-23 22:26:28 +00:00
2021-12-21 02:23:41 +00:00
import torchvision
import pytorch_lightning as pl
from packaging import version
from omegaconf import OmegaConf
from torch.utils.data import random_split, DataLoader, Dataset, Subset
from functools import partial
from PIL import Image
from pytorch_lightning import seed_everything
from pytorch_lightning.trainer import Trainer
from pytorch_lightning.callbacks import (
ModelCheckpoint,
Callback,
LearningRateMonitor,
)
2021-12-21 02:23:41 +00:00
from pytorch_lightning.utilities.distributed import rank_zero_only
from pytorch_lightning.utilities import rank_zero_info
from ldm.data.base import Txt2ImgIterableBaseDataset
from ldm.util import instantiate_from_config
def fix_func(orig):
if hasattr(torch.backends, 'mps') and torch.backends.mps.is_available():
def new_func(*args, **kw):
device = kw.get("device", "mps")
kw["device"]="cpu"
return orig(*args, **kw).to(device)
return new_func
return orig
torch.rand = fix_func(torch.rand)
torch.rand_like = fix_func(torch.rand_like)
torch.randn = fix_func(torch.randn)
torch.randn_like = fix_func(torch.randn_like)
torch.randint = fix_func(torch.randint)
torch.randint_like = fix_func(torch.randint_like)
torch.bernoulli = fix_func(torch.bernoulli)
torch.multinomial = fix_func(torch.multinomial)
2022-08-23 22:26:28 +00:00
def load_model_from_config(config, ckpt, verbose=False):
print(f'Loading model from {ckpt}')
pl_sd = torch.load(ckpt, map_location='cpu')
sd = pl_sd['state_dict']
2022-08-23 22:26:28 +00:00
config.model.params.ckpt_path = ckpt
model = instantiate_from_config(config.model)
m, u = model.load_state_dict(sd, strict=False)
if len(m) > 0 and verbose:
print('missing keys:')
2022-08-23 22:26:28 +00:00
print(m)
if len(u) > 0 and verbose:
print('unexpected keys:')
2022-08-23 22:26:28 +00:00
print(u)
if torch.cuda.is_available():
model.cuda()
2022-08-23 22:26:28 +00:00
return model
2021-12-21 02:23:41 +00:00
2021-12-21 02:23:41 +00:00
def get_parser(**parser_kwargs):
def str2bool(v):
if isinstance(v, bool):
return v
if v.lower() in ('yes', 'true', 't', 'y', '1'):
2021-12-21 02:23:41 +00:00
return True
elif v.lower() in ('no', 'false', 'f', 'n', '0'):
2021-12-21 02:23:41 +00:00
return False
else:
raise argparse.ArgumentTypeError('Boolean value expected.')
2021-12-21 02:23:41 +00:00
parser = argparse.ArgumentParser(**parser_kwargs)
parser.add_argument(
'-n',
'--name',
2021-12-21 02:23:41 +00:00
type=str,
const=True,
default='',
nargs='?',
help='postfix for logdir',
2021-12-21 02:23:41 +00:00
)
parser.add_argument(
'-r',
'--resume',
2021-12-21 02:23:41 +00:00
type=str,
const=True,
default='',
nargs='?',
help='resume from logdir or checkpoint in logdir',
2021-12-21 02:23:41 +00:00
)
parser.add_argument(
'-b',
'--base',
nargs='*',
metavar='base_config.yaml',
help='paths to base configs. Loaded from left-to-right. '
'Parameters can be overwritten or added with command-line options of the form `--key value`.',
2021-12-21 02:23:41 +00:00
default=list(),
)
parser.add_argument(
'-t',
'--train',
2021-12-21 02:23:41 +00:00
type=str2bool,
const=True,
default=False,
nargs='?',
help='train',
2021-12-21 02:23:41 +00:00
)
parser.add_argument(
'--no-test',
2021-12-21 02:23:41 +00:00
type=str2bool,
const=True,
default=False,
nargs='?',
help='disable test',
2021-12-21 02:23:41 +00:00
)
parser.add_argument(
'-p', '--project', help='name of new or path to existing project'
2021-12-21 02:23:41 +00:00
)
parser.add_argument(
'-d',
'--debug',
2021-12-21 02:23:41 +00:00
type=str2bool,
nargs='?',
2021-12-21 02:23:41 +00:00
const=True,
default=False,
help='enable post-mortem debugging',
2021-12-21 02:23:41 +00:00
)
parser.add_argument(
'-s',
'--seed',
2021-12-21 02:23:41 +00:00
type=int,
default=23,
help='seed for seed_everything',
2021-12-21 02:23:41 +00:00
)
parser.add_argument(
'-f',
'--postfix',
2021-12-21 02:23:41 +00:00
type=str,
default='',
help='post-postfix for default name',
2021-12-21 02:23:41 +00:00
)
parser.add_argument(
'-l',
'--logdir',
2021-12-21 02:23:41 +00:00
type=str,
default='logs',
help='directory for logging dat shit',
2021-12-21 02:23:41 +00:00
)
parser.add_argument(
'--scale_lr',
2021-12-21 02:23:41 +00:00
type=str2bool,
nargs='?',
2021-12-21 02:23:41 +00:00
const=True,
default=True,
help='scale base-lr by ngpu * batch_size * n_accumulate',
2021-12-21 02:23:41 +00:00
)
2022-08-23 22:26:28 +00:00
parser.add_argument(
'--datadir_in_name',
type=str2bool,
nargs='?',
const=True,
default=True,
help='Prepend the final directory in the data_root to the output directory name',
)
2022-08-23 22:26:28 +00:00
parser.add_argument(
'--actual_resume',
type=str,
default='',
help='Path to model to actually resume from',
)
parser.add_argument(
'--data_root',
type=str,
required=True,
help='Path to directory with training images',
)
2022-08-23 22:26:28 +00:00
parser.add_argument(
'--embedding_manager_ckpt',
type=str,
default='',
help='Initialize embedding manager from a checkpoint',
)
parser.add_argument(
'--placeholder_tokens', type=str, nargs='+', default=['*'],
help='Placeholder token which will be used to denote the concept in future prompts')
2022-08-23 22:26:28 +00:00
parser.add_argument(
'--init_word',
type=str,
help='Word to use as source for initial token embedding.',
)
2022-08-23 22:26:28 +00:00
2021-12-21 02:23:41 +00:00
return parser
def nondefault_trainer_args(opt):
parser = argparse.ArgumentParser()
parser = Trainer.add_argparse_args(parser)
args = parser.parse_args([])
return sorted(k for k in vars(args) if getattr(opt, k) != getattr(args, k))
class WrappedDataset(Dataset):
"""Wraps an arbitrary object with __len__ and __getitem__ into a pytorch dataset"""
def __init__(self, dataset):
self.data = dataset
def __len__(self):
return len(self.data)
def __getitem__(self, idx):
return self.data[idx]
def worker_init_fn(_):
worker_info = torch.utils.data.get_worker_info()
dataset = worker_info.dataset
worker_id = worker_info.id
if isinstance(dataset, Txt2ImgIterableBaseDataset):
split_size = dataset.num_records // worker_info.num_workers
# reset num_records to the true number to retain reliable length information
dataset.sample_ids = dataset.valid_ids[
worker_id * split_size : (worker_id + 1) * split_size
]
2021-12-21 02:23:41 +00:00
current_id = np.random.choice(len(np.random.get_state()[1]), 1)
return np.random.seed(np.random.get_state()[1][current_id] + worker_id)
else:
return np.random.seed(np.random.get_state()[1][0] + worker_id)
class DataModuleFromConfig(pl.LightningDataModule):
def __init__(
self,
batch_size,
train=None,
validation=None,
test=None,
predict=None,
wrap=False,
num_workers=None,
shuffle_test_loader=False,
use_worker_init_fn=False,
shuffle_val_dataloader=False,
):
2021-12-21 02:23:41 +00:00
super().__init__()
self.batch_size = batch_size
self.dataset_configs = dict()
self.num_workers = (
num_workers if num_workers is not None else batch_size * 2
)
2021-12-21 02:23:41 +00:00
self.use_worker_init_fn = use_worker_init_fn
if train is not None:
self.dataset_configs['train'] = train
2021-12-21 02:23:41 +00:00
self.train_dataloader = self._train_dataloader
if validation is not None:
self.dataset_configs['validation'] = validation
self.val_dataloader = partial(
self._val_dataloader, shuffle=shuffle_val_dataloader
)
2021-12-21 02:23:41 +00:00
if test is not None:
self.dataset_configs['test'] = test
self.test_dataloader = partial(
self._test_dataloader, shuffle=shuffle_test_loader
)
2021-12-21 02:23:41 +00:00
if predict is not None:
self.dataset_configs['predict'] = predict
2021-12-21 02:23:41 +00:00
self.predict_dataloader = self._predict_dataloader
self.wrap = wrap
def prepare_data(self):
for data_cfg in self.dataset_configs.values():
instantiate_from_config(data_cfg)
def setup(self, stage=None):
self.datasets = dict(
(k, instantiate_from_config(self.dataset_configs[k]))
for k in self.dataset_configs
)
2021-12-21 02:23:41 +00:00
if self.wrap:
for k in self.datasets:
self.datasets[k] = WrappedDataset(self.datasets[k])
def _train_dataloader(self):
is_iterable_dataset = isinstance(
self.datasets['train'], Txt2ImgIterableBaseDataset
)
2021-12-21 02:23:41 +00:00
if is_iterable_dataset or self.use_worker_init_fn:
init_fn = worker_init_fn
else:
init_fn = None
return DataLoader(
self.datasets['train'],
batch_size=self.batch_size,
num_workers=self.num_workers,
shuffle=False if is_iterable_dataset else True,
worker_init_fn=init_fn,
)
2021-12-21 02:23:41 +00:00
def _val_dataloader(self, shuffle=False):
if (
isinstance(self.datasets['validation'], Txt2ImgIterableBaseDataset)
or self.use_worker_init_fn
):
2021-12-21 02:23:41 +00:00
init_fn = worker_init_fn
else:
init_fn = None
return DataLoader(
self.datasets['validation'],
batch_size=self.batch_size,
num_workers=self.num_workers,
worker_init_fn=init_fn,
shuffle=shuffle,
)
2021-12-21 02:23:41 +00:00
def _test_dataloader(self, shuffle=False):
is_iterable_dataset = isinstance(
self.datasets['train'], Txt2ImgIterableBaseDataset
)
2021-12-21 02:23:41 +00:00
if is_iterable_dataset or self.use_worker_init_fn:
init_fn = worker_init_fn
else:
init_fn = None
# do not shuffle dataloader for iterable dataset
shuffle = shuffle and (not is_iterable_dataset)
return DataLoader(
self.datasets['test'],
batch_size=self.batch_size,
num_workers=self.num_workers,
worker_init_fn=init_fn,
shuffle=shuffle,
)
2021-12-21 02:23:41 +00:00
def _predict_dataloader(self, shuffle=False):
if (
isinstance(self.datasets['predict'], Txt2ImgIterableBaseDataset)
or self.use_worker_init_fn
):
2021-12-21 02:23:41 +00:00
init_fn = worker_init_fn
else:
init_fn = None
return DataLoader(
self.datasets['predict'],
batch_size=self.batch_size,
num_workers=self.num_workers,
worker_init_fn=init_fn,
)
2021-12-21 02:23:41 +00:00
class SetupCallback(Callback):
def __init__(
self, resume, now, logdir, ckptdir, cfgdir, config, lightning_config
):
2021-12-21 02:23:41 +00:00
super().__init__()
self.resume = resume
self.now = now
self.logdir = logdir
self.ckptdir = ckptdir
self.cfgdir = cfgdir
self.config = config
self.lightning_config = lightning_config
def on_keyboard_interrupt(self, trainer, pl_module):
if trainer.global_rank == 0:
print('Summoning checkpoint.')
ckpt_path = os.path.join(self.ckptdir, 'last.ckpt')
2021-12-21 02:23:41 +00:00
trainer.save_checkpoint(ckpt_path)
def on_pretrain_routine_start(self, trainer, pl_module):
if trainer.global_rank == 0:
# Create logdirs and save configs
os.makedirs(self.logdir, exist_ok=True)
os.makedirs(self.ckptdir, exist_ok=True)
os.makedirs(self.cfgdir, exist_ok=True)
if 'callbacks' in self.lightning_config:
if (
'metrics_over_trainsteps_checkpoint'
in self.lightning_config['callbacks']
):
os.makedirs(
os.path.join(self.ckptdir, 'trainstep_checkpoints'),
exist_ok=True,
)
print('Project config')
2021-12-21 02:23:41 +00:00
print(OmegaConf.to_yaml(self.config))
OmegaConf.save(
self.config,
os.path.join(self.cfgdir, '{}-project.yaml'.format(self.now)),
)
2021-12-21 02:23:41 +00:00
print('Lightning config')
2021-12-21 02:23:41 +00:00
print(OmegaConf.to_yaml(self.lightning_config))
OmegaConf.save(
OmegaConf.create({'lightning': self.lightning_config}),
os.path.join(
self.cfgdir, '{}-lightning.yaml'.format(self.now)
),
)
2021-12-21 02:23:41 +00:00
else:
# ModelCheckpoint callback created log directory --- remove it
if not self.resume and os.path.exists(self.logdir):
dst, name = os.path.split(self.logdir)
dst = os.path.join(dst, 'child_runs', name)
2021-12-21 02:23:41 +00:00
os.makedirs(os.path.split(dst)[0], exist_ok=True)
try:
os.rename(self.logdir, dst)
except FileNotFoundError:
pass
class ImageLogger(Callback):
def __init__(
self,
batch_frequency,
max_images,
clamp=True,
increase_log_steps=True,
rescale=True,
disabled=False,
log_on_batch_idx=False,
log_first_step=False,
log_images_kwargs=None,
):
2021-12-21 02:23:41 +00:00
super().__init__()
self.rescale = rescale
self.batch_freq = batch_frequency
self.max_images = max_images
self.logger_log_images = { pl.loggers.TestTubeLogger: self._testtube, } if torch.cuda.is_available() else { }
self.log_steps = [
2**n for n in range(int(np.log2(self.batch_freq)) + 1)
]
2021-12-21 02:23:41 +00:00
if not increase_log_steps:
self.log_steps = [self.batch_freq]
self.clamp = clamp
self.disabled = disabled
self.log_on_batch_idx = log_on_batch_idx
self.log_images_kwargs = log_images_kwargs if log_images_kwargs else {}
self.log_first_step = log_first_step
@rank_zero_only
def _testtube(self, pl_module, images, batch_idx, split):
for k in images:
grid = torchvision.utils.make_grid(images[k])
grid = (grid + 1.0) / 2.0 # -1,1 -> 0,1; c,h,w
tag = f'{split}/{k}'
2021-12-21 02:23:41 +00:00
pl_module.logger.experiment.add_image(
tag, grid, global_step=pl_module.global_step
)
2021-12-21 02:23:41 +00:00
@rank_zero_only
def log_local(
self, save_dir, split, images, global_step, current_epoch, batch_idx
):
root = os.path.join(save_dir, 'images', split)
2021-12-21 02:23:41 +00:00
for k in images:
grid = torchvision.utils.make_grid(images[k], nrow=4)
if self.rescale:
grid = (grid + 1.0) / 2.0 # -1,1 -> 0,1; c,h,w
grid = grid.transpose(0, 1).transpose(1, 2).squeeze(-1)
grid = grid.numpy()
grid = (grid * 255).astype(np.uint8)
filename = '{}_gs-{:06}_e-{:06}_b-{:06}.png'.format(
k, global_step, current_epoch, batch_idx
)
2021-12-21 02:23:41 +00:00
path = os.path.join(root, filename)
os.makedirs(os.path.split(path)[0], exist_ok=True)
Image.fromarray(grid).save(path)
def log_img(self, pl_module, batch, batch_idx, split='train'):
check_idx = (
batch_idx if self.log_on_batch_idx else pl_module.global_step
)
if (
self.check_frequency(check_idx)
and hasattr( # batch_idx % self.batch_freq == 0
pl_module, 'log_images'
)
and callable(pl_module.log_images)
and self.max_images > 0
):
2021-12-21 02:23:41 +00:00
logger = type(pl_module.logger)
is_train = pl_module.training
if is_train:
pl_module.eval()
with torch.no_grad():
images = pl_module.log_images(
batch, split=split, **self.log_images_kwargs
)
2021-12-21 02:23:41 +00:00
for k in images:
N = min(images[k].shape[0], self.max_images)
images[k] = images[k][:N]
if isinstance(images[k], torch.Tensor):
images[k] = images[k].detach().cpu()
if self.clamp:
images[k] = torch.clamp(images[k], -1.0, 1.0)
self.log_local(
pl_module.logger.save_dir,
split,
images,
pl_module.global_step,
pl_module.current_epoch,
batch_idx,
)
logger_log_images = self.logger_log_images.get(
logger, lambda *args, **kwargs: None
)
2021-12-21 02:23:41 +00:00
logger_log_images(pl_module, images, pl_module.global_step, split)
if is_train:
pl_module.train()
def check_frequency(self, check_idx):
if (
(check_idx % self.batch_freq) == 0 or (check_idx in self.log_steps)
) and (check_idx > 0 or self.log_first_step):
2021-12-21 02:23:41 +00:00
try:
self.log_steps.pop(0)
except IndexError as e:
print(e)
pass
return True
return False
def on_train_batch_end(
self, trainer, pl_module, outputs, batch, batch_idx, dataloader_idx=None
):
if not self.disabled and (
pl_module.global_step > 0 or self.log_first_step
):
self.log_img(pl_module, batch, batch_idx, split='train')
def on_validation_batch_end(
self, trainer, pl_module, outputs, batch, batch_idx, dataloader_idx=None
):
2021-12-21 02:23:41 +00:00
if not self.disabled and pl_module.global_step > 0:
self.log_img(pl_module, batch, batch_idx, split='val')
2021-12-21 02:23:41 +00:00
if hasattr(pl_module, 'calibrate_grad_norm'):
if (
pl_module.calibrate_grad_norm and batch_idx % 25 == 0
) and batch_idx > 0:
2021-12-21 02:23:41 +00:00
self.log_gradients(trainer, pl_module, batch_idx=batch_idx)
class CUDACallback(Callback):
# see https://github.com/SeanNaren/minGPT/blob/master/mingpt/callback.py
def on_train_epoch_start(self, trainer, pl_module):
# Reset the memory use counter
if torch.cuda.is_available():
torch.cuda.reset_peak_memory_stats(trainer.root_gpu)
torch.cuda.synchronize(trainer.root_gpu)
2021-12-21 02:23:41 +00:00
self.start_time = time.time()
def on_train_epoch_end(self, trainer, pl_module, outputs=None):
if torch.cuda.is_available():
torch.cuda.synchronize(trainer.root_gpu)
2021-12-21 02:23:41 +00:00
epoch_time = time.time() - self.start_time
try:
epoch_time = trainer.training_type_plugin.reduce(epoch_time)
rank_zero_info(f'Average Epoch time: {epoch_time:.2f} seconds')
if torch.cuda.is_available():
max_memory = (
torch.cuda.max_memory_allocated(trainer.root_gpu) / 2**20
)
max_memory = trainer.training_type_plugin.reduce(max_memory)
rank_zero_info(f'Average Peak memory {max_memory:.2f}MiB')
2021-12-21 02:23:41 +00:00
except AttributeError:
pass
class ModeSwapCallback(Callback):
def __init__(self, swap_step=2000):
super().__init__()
self.is_frozen = False
self.swap_step = swap_step
def on_train_epoch_start(self, trainer, pl_module):
if trainer.global_step < self.swap_step and not self.is_frozen:
self.is_frozen = True
trainer.optimizers = [pl_module.configure_opt_embedding()]
if trainer.global_step > self.swap_step and self.is_frozen:
self.is_frozen = False
trainer.optimizers = [pl_module.configure_opt_model()]
2021-12-21 02:23:41 +00:00
if __name__ == '__main__':
2021-12-21 02:23:41 +00:00
# custom parser to specify config files, train, test and debug mode,
# postfix, resume.
# `--key value` arguments are interpreted as arguments to the trainer.
# `nested.key=value` arguments are interpreted as config parameters.
# configs are merged from left-to-right followed by command line parameters.
# model:
# base_learning_rate: float
# target: path to lightning module
# params:
# key: value
# data:
# target: main.DataModuleFromConfig
# params:
# batch_size: int
# wrap: bool
# train:
# target: path to train dataset
# params:
# key: value
# validation:
# target: path to validation dataset
# params:
# key: value
# test:
# target: path to test dataset
# params:
# key: value
# lightning: (optional, has sane defaults and can be specified on cmdline)
# trainer:
# additional arguments to trainer
# logger:
# logger to instantiate
# modelcheckpoint:
# modelcheckpoint to instantiate
# callbacks:
# callback1:
# target: importpath
# params:
# key: value
now = datetime.datetime.now().strftime('%Y-%m-%dT%H-%M-%S')
2021-12-21 02:23:41 +00:00
# add cwd for convenience and to make classes in this file available when
# running as `python main.py`
# (in particular `main.DataModuleFromConfig`)
sys.path.append(os.getcwd())
parser = get_parser()
parser = Trainer.add_argparse_args(parser)
opt, unknown = parser.parse_known_args()
if opt.name and opt.resume:
raise ValueError(
'-n/--name and -r/--resume cannot be specified both.'
'If you want to resume training in a new log folder, '
'use -n/--name in combination with --resume_from_checkpoint'
2021-12-21 02:23:41 +00:00
)
if opt.resume:
if not os.path.exists(opt.resume):
raise ValueError('Cannot find {}'.format(opt.resume))
2021-12-21 02:23:41 +00:00
if os.path.isfile(opt.resume):
paths = opt.resume.split('/')
2021-12-21 02:23:41 +00:00
# idx = len(paths)-paths[::-1].index("logs")+1
# logdir = "/".join(paths[:idx])
logdir = '/'.join(paths[:-2])
2021-12-21 02:23:41 +00:00
ckpt = opt.resume
else:
assert os.path.isdir(opt.resume), opt.resume
logdir = opt.resume.rstrip('/')
ckpt = os.path.join(logdir, 'checkpoints', 'last.ckpt')
2021-12-21 02:23:41 +00:00
opt.resume_from_checkpoint = ckpt
base_configs = sorted(
glob.glob(os.path.join(logdir, 'configs/*.yaml'))
)
2021-12-21 02:23:41 +00:00
opt.base = base_configs + opt.base
_tmp = logdir.split('/')
2021-12-21 02:23:41 +00:00
nowname = _tmp[-1]
else:
if opt.name:
name = '_' + opt.name
2021-12-21 02:23:41 +00:00
elif opt.base:
cfg_fname = os.path.split(opt.base[0])[-1]
cfg_name = os.path.splitext(cfg_fname)[0]
name = '_' + cfg_name
2021-12-21 02:23:41 +00:00
else:
name = ''
2022-08-23 22:26:28 +00:00
if opt.datadir_in_name:
now = os.path.basename(os.path.normpath(opt.data_root)) + now
2021-12-21 02:23:41 +00:00
nowname = now + name + opt.postfix
logdir = os.path.join(opt.logdir, nowname)
ckptdir = os.path.join(logdir, 'checkpoints')
cfgdir = os.path.join(logdir, 'configs')
2021-12-21 02:23:41 +00:00
seed_everything(opt.seed)
try:
# init and save configs
configs = [OmegaConf.load(cfg) for cfg in opt.base]
cli = OmegaConf.from_dotlist(unknown)
config = OmegaConf.merge(*configs, cli)
lightning_config = config.pop('lightning', OmegaConf.create())
2021-12-21 02:23:41 +00:00
# merge trainer cli with config
trainer_config = lightning_config.get('trainer', OmegaConf.create())
2021-12-21 02:23:41 +00:00
# default to ddp
trainer_config['accelerator'] = 'ddp'
2021-12-21 02:23:41 +00:00
for k in nondefault_trainer_args(opt):
trainer_config[k] = getattr(opt, k)
if not 'gpus' in trainer_config:
del trainer_config['accelerator']
2021-12-21 02:23:41 +00:00
cpu = True
else:
gpuinfo = trainer_config['gpus']
print(f'Running on GPUs {gpuinfo}')
2021-12-21 02:23:41 +00:00
cpu = False
trainer_opt = argparse.Namespace(**trainer_config)
lightning_config.trainer = trainer_config
# model
2022-08-23 22:26:28 +00:00
# config.model.params.personalization_config.params.init_word = opt.init_word
config.model.params.personalization_config.params.embedding_manager_ckpt = (
opt.embedding_manager_ckpt
)
config.model.params.personalization_config.params.placeholder_tokens = (
opt.placeholder_tokens
)
2022-08-23 22:26:28 +00:00
if opt.init_word:
config.model.params.personalization_config.params.initializer_words[
0
] = opt.init_word
2022-08-23 22:26:28 +00:00
if opt.actual_resume:
model = load_model_from_config(config, opt.actual_resume)
else:
model = instantiate_from_config(config.model)
2021-12-21 02:23:41 +00:00
# trainer and callbacks
trainer_kwargs = dict()
# default logger configs
if torch.cuda.is_available():
def_logger = 'testtube'
def_logger_target = 'TestTubeLogger'
else:
def_logger = 'csv'
def_logger_target = 'CSVLogger'
2021-12-21 02:23:41 +00:00
default_logger_cfgs = {
'wandb': {
'target': 'pytorch_lightning.loggers.WandbLogger',
'params': {
'name': nowname,
'save_dir': logdir,
'offline': opt.debug,
'id': nowname,
},
2021-12-21 02:23:41 +00:00
},
def_logger: {
'target': 'pytorch_lightning.loggers.' + def_logger_target,
'params': {
'name': def_logger,
'save_dir': logdir,
},
2021-12-21 02:23:41 +00:00
},
}
default_logger_cfg = default_logger_cfgs[def_logger]
if 'logger' in lightning_config:
2021-12-21 02:23:41 +00:00
logger_cfg = lightning_config.logger
else:
logger_cfg = OmegaConf.create()
logger_cfg = OmegaConf.merge(default_logger_cfg, logger_cfg)
trainer_kwargs['logger'] = instantiate_from_config(logger_cfg)
2021-12-21 02:23:41 +00:00
# modelcheckpoint - use TrainResult/EvalResult(checkpoint_on=metric) to
# specify which metric is used to determine best models
default_modelckpt_cfg = {
'target': 'pytorch_lightning.callbacks.ModelCheckpoint',
'params': {
'dirpath': ckptdir,
'filename': '{epoch:06}',
'verbose': True,
'save_last': True,
},
2021-12-21 02:23:41 +00:00
}
if hasattr(model, 'monitor'):
print(f'Monitoring {model.monitor} as checkpoint metric.')
default_modelckpt_cfg['params']['monitor'] = model.monitor
default_modelckpt_cfg['params']['save_top_k'] = 1
2021-12-21 02:23:41 +00:00
if 'modelcheckpoint' in lightning_config:
2021-12-21 02:23:41 +00:00
modelckpt_cfg = lightning_config.modelcheckpoint
else:
modelckpt_cfg = OmegaConf.create()
2021-12-21 02:23:41 +00:00
modelckpt_cfg = OmegaConf.merge(default_modelckpt_cfg, modelckpt_cfg)
print(f'Merged modelckpt-cfg: \n{modelckpt_cfg}')
2021-12-21 02:23:41 +00:00
if version.parse(pl.__version__) < version.parse('1.4.0'):
trainer_kwargs['checkpoint_callback'] = instantiate_from_config(
modelckpt_cfg
)
2021-12-21 02:23:41 +00:00
# add callback which sets up log directory
default_callbacks_cfg = {
'setup_callback': {
'target': 'main.SetupCallback',
'params': {
'resume': opt.resume,
'now': now,
'logdir': logdir,
'ckptdir': ckptdir,
'cfgdir': cfgdir,
'config': config,
'lightning_config': lightning_config,
},
2021-12-21 02:23:41 +00:00
},
'image_logger': {
'target': 'main.ImageLogger',
'params': {
'batch_frequency': 750,
'max_images': 4,
'clamp': True,
},
2021-12-21 02:23:41 +00:00
},
'learning_rate_logger': {
'target': 'main.LearningRateMonitor',
'params': {
'logging_interval': 'step',
2021-12-21 02:23:41 +00:00
# "log_momentum": True
},
2021-12-21 02:23:41 +00:00
},
'cuda_callback': {'target': 'main.CUDACallback'},
2021-12-21 02:23:41 +00:00
}
if version.parse(pl.__version__) >= version.parse('1.4.0'):
default_callbacks_cfg.update(
{'checkpoint_callback': modelckpt_cfg}
)
2021-12-21 02:23:41 +00:00
if 'callbacks' in lightning_config:
2021-12-21 02:23:41 +00:00
callbacks_cfg = lightning_config.callbacks
else:
callbacks_cfg = OmegaConf.create()
if 'metrics_over_trainsteps_checkpoint' in callbacks_cfg:
print(
'Caution: Saving checkpoints every n train steps without deleting. This might require some free space.'
)
2021-12-21 02:23:41 +00:00
default_metrics_over_trainsteps_ckpt_dict = {
'metrics_over_trainsteps_checkpoint': {
'target': 'pytorch_lightning.callbacks.ModelCheckpoint',
'params': {
'dirpath': os.path.join(
ckptdir, 'trainstep_checkpoints'
),
'filename': '{epoch:06}-{step:09}',
'verbose': True,
'save_top_k': -1,
'every_n_train_steps': 10000,
'save_weights_only': True,
},
}
2021-12-21 02:23:41 +00:00
}
default_callbacks_cfg.update(
default_metrics_over_trainsteps_ckpt_dict
)
2021-12-21 02:23:41 +00:00
callbacks_cfg = OmegaConf.merge(default_callbacks_cfg, callbacks_cfg)
if 'ignore_keys_callback' in callbacks_cfg and hasattr(
trainer_opt, 'resume_from_checkpoint'
):
callbacks_cfg.ignore_keys_callback.params[
'ckpt_path'
] = trainer_opt.resume_from_checkpoint
2021-12-21 02:23:41 +00:00
elif 'ignore_keys_callback' in callbacks_cfg:
del callbacks_cfg['ignore_keys_callback']
trainer_kwargs['callbacks'] = [
instantiate_from_config(callbacks_cfg[k]) for k in callbacks_cfg
]
trainer_kwargs['max_steps'] = trainer_opt.max_steps
2021-12-21 02:23:41 +00:00
if hasattr(torch.backends, 'mps') and torch.backends.mps.is_available():
trainer_opt.accelerator = 'mps'
trainer_opt.detect_anomaly = False
2021-12-21 02:23:41 +00:00
trainer = Trainer.from_argparse_args(trainer_opt, **trainer_kwargs)
trainer.logdir = logdir ###
# data
2022-08-23 22:26:28 +00:00
config.data.params.train.params.data_root = opt.data_root
config.data.params.validation.params.data_root = opt.data_root
data = instantiate_from_config(config.data)
2021-12-21 02:23:41 +00:00
# NOTE according to https://pytorch-lightning.readthedocs.io/en/latest/datamodules.html
# calling these ourselves should not be necessary but it is.
# lightning still takes care of proper multiprocessing though
data.prepare_data()
data.setup()
print('#### Data #####')
2021-12-21 02:23:41 +00:00
for k in data.datasets:
print(
f'{k}, {data.datasets[k].__class__.__name__}, {len(data.datasets[k])}'
)
2021-12-21 02:23:41 +00:00
# configure learning rate
bs, base_lr = (
config.data.params.batch_size,
config.model.base_learning_rate,
)
2021-12-21 02:23:41 +00:00
if not cpu:
ngpu = len(lightning_config.trainer.gpus.strip(',').split(','))
2021-12-21 02:23:41 +00:00
else:
ngpu = 1
if 'accumulate_grad_batches' in lightning_config.trainer:
accumulate_grad_batches = (
lightning_config.trainer.accumulate_grad_batches
)
else:
accumulate_grad_batches = 1
print(f'accumulate_grad_batches = {accumulate_grad_batches}')
lightning_config.trainer.accumulate_grad_batches = (
accumulate_grad_batches
)
2021-12-21 02:23:41 +00:00
if opt.scale_lr:
model.learning_rate = accumulate_grad_batches * ngpu * bs * base_lr
print(
'Setting learning rate to {:.2e} = {} (accumulate_grad_batches) * {} (num_gpus) * {} (batchsize) * {:.2e} (base_lr)'.format(
model.learning_rate,
accumulate_grad_batches,
ngpu,
bs,
base_lr,
)
)
2021-12-21 02:23:41 +00:00
else:
model.learning_rate = base_lr
print('++++ NOT USING LR SCALING ++++')
print(f'Setting learning rate to {model.learning_rate:.2e}')
2021-12-21 02:23:41 +00:00
# allow checkpointing via USR1
def melk(*args, **kwargs):
# run all checkpoint hooks
if trainer.global_rank == 0:
print('Summoning checkpoint.')
ckpt_path = os.path.join(ckptdir, 'last.ckpt')
2021-12-21 02:23:41 +00:00
trainer.save_checkpoint(ckpt_path)
def divein(*args, **kwargs):
if trainer.global_rank == 0:
import pudb
2021-12-21 02:23:41 +00:00
pudb.set_trace()
2021-12-21 02:23:41 +00:00
import signal
2022-08-23 22:26:28 +00:00
signal.signal(signal.SIGTERM, melk)
signal.signal(signal.SIGTERM, divein)
2021-12-21 02:23:41 +00:00
# run
if opt.train:
try:
trainer.fit(model, data)
except Exception:
melk()
raise
if not opt.no_test and not trainer.interrupted:
trainer.test(model, data)
except Exception:
if opt.debug and trainer.global_rank == 0:
try:
import pudb as debugger
except ImportError:
import pdb as debugger
debugger.post_mortem()
raise
finally:
# move newly created debug project to debug_runs
if opt.debug and not opt.resume and trainer.global_rank == 0:
dst, name = os.path.split(logdir)
dst = os.path.join(dst, 'debug_runs', name)
2021-12-21 02:23:41 +00:00
os.makedirs(os.path.split(dst)[0], exist_ok=True)
os.rename(logdir, dst)
2022-08-23 22:26:28 +00:00
# if trainer.global_rank == 0:
# print(trainer.profiler.summary())