|
""" |
|
Train a diffusion model on images. |
|
""" |
|
import random |
|
import json |
|
import sys |
|
import os |
|
|
|
sys.path.append('.') |
|
import torch.distributed as dist |
|
|
|
import traceback |
|
|
|
import torch as th |
|
|
|
import torch.multiprocessing as mp |
|
import numpy as np |
|
|
|
import argparse |
|
import dnnlib |
|
from guided_diffusion import dist_util, logger |
|
from guided_diffusion.script_util import ( |
|
args_to_dict, |
|
add_dict_to_argparser, |
|
) |
|
|
|
from nsr.train_nv_util import TrainLoop3DRecNV, TrainLoop3DRec, TrainLoop3DRecNVPatch, TrainLoop3DRecNVPatchSingleForward, TrainLoop3DRecNVPatchSingleForwardMV, TrainLoop3DRecNVPatchSingleForwardMVAdvLoss |
|
|
|
from nsr.script_util import create_3DAE_model, encoder_and_nsr_defaults, loss_defaults, rendering_options_defaults, eg3d_options_default, dataset_defaults |
|
from nsr.losses.builder import E3DGELossClass, E3DGE_with_AdvLoss |
|
|
|
from pdb import set_trace as st |
|
|
|
|
|
|
|
|
|
|
|
|
|
enable_tf32 = th.backends.cuda.matmul.allow_tf32 |
|
|
|
th.backends.cuda.matmul.allow_tf32 = enable_tf32 |
|
th.backends.cudnn.allow_tf32 = enable_tf32 |
|
th.backends.cudnn.enabled = True |
|
|
|
|
|
def training_loop(args): |
|
|
|
dist_util.setup_dist(args) |
|
|
|
th.autograd.set_detect_anomaly(False) |
|
|
|
|
|
SEED = args.seed |
|
|
|
|
|
logger.log(f"{args.local_rank=} init complete, seed={SEED}") |
|
th.cuda.set_device(args.local_rank) |
|
th.cuda.empty_cache() |
|
|
|
|
|
th.cuda.manual_seed_all(SEED) |
|
np.random.seed(SEED) |
|
random.seed(SEED) |
|
|
|
|
|
logger.configure(dir=args.logdir) |
|
|
|
logger.log("creating encoder and NSR decoder...") |
|
|
|
device = th.device("cuda", args.local_rank) |
|
|
|
|
|
opts = eg3d_options_default() |
|
|
|
if args.sr_training: |
|
args.sr_kwargs = dnnlib.EasyDict( |
|
channel_base=opts.cbase, |
|
channel_max=opts.cmax, |
|
fused_modconv_default='inference_only', |
|
use_noise=True |
|
) |
|
|
|
auto_encoder = create_3DAE_model( |
|
**args_to_dict(args, |
|
encoder_and_nsr_defaults().keys())) |
|
auto_encoder.to(device) |
|
auto_encoder.train() |
|
|
|
logger.log("creating data loader...") |
|
|
|
|
|
if args.objv_dataset: |
|
from datasets.g_buffer_objaverse import load_data, load_eval_data, load_memory_data, load_wds_data |
|
else: |
|
from datasets.shapenet import load_data, load_eval_data, load_memory_data |
|
|
|
if args.overfitting: |
|
data = load_memory_data( |
|
file_path=args.data_dir, |
|
batch_size=args.batch_size, |
|
reso=args.image_size, |
|
reso_encoder=args.image_size_encoder, |
|
num_workers=args.num_workers, |
|
|
|
|
|
**args_to_dict(args, |
|
dataset_defaults().keys())) |
|
eval_data = None |
|
else: |
|
if args.use_wds: |
|
|
|
if args.data_dir == 'NONE': |
|
with open(args.shards_lst) as f: |
|
shards_lst = [url.strip() for url in f.readlines()] |
|
data = load_wds_data( |
|
shards_lst, |
|
args.image_size, |
|
args.image_size_encoder, |
|
args.batch_size, |
|
args.num_workers, |
|
|
|
|
|
|
|
**args_to_dict(args, |
|
dataset_defaults().keys())) |
|
|
|
elif not args.inference: |
|
data = load_wds_data(args.data_dir, |
|
args.image_size, |
|
args.image_size_encoder, |
|
args.batch_size, |
|
args.num_workers, |
|
plucker_embedding=args.plucker_embedding, |
|
mv_input=args.mv_input, |
|
split_chunk_input=args.split_chunk_input) |
|
else: |
|
data = None |
|
|
|
|
|
if args.eval_data_dir == 'NONE': |
|
with open(args.eval_shards_lst) as f: |
|
eval_shards_lst = [url.strip() for url in f.readlines()] |
|
else: |
|
eval_shards_lst = args.eval_data_dir |
|
|
|
eval_data = load_wds_data( |
|
eval_shards_lst, |
|
args.image_size, |
|
args.image_size_encoder, |
|
args.eval_batch_size, |
|
args.num_workers, |
|
|
|
|
|
|
|
|
|
|
|
**args_to_dict(args, |
|
dataset_defaults().keys())) |
|
|
|
|
|
else: |
|
|
|
if args.inference: |
|
data = None |
|
else: |
|
data = load_data( |
|
file_path=args.data_dir, |
|
batch_size=args.batch_size, |
|
reso=args.image_size, |
|
reso_encoder=args.image_size_encoder, |
|
num_workers=args.num_workers, |
|
load_depth=True, |
|
preprocess=auto_encoder.preprocess, |
|
dataset_size=args.dataset_size, |
|
trainer_name=args.trainer_name, |
|
use_lmdb=args.use_lmdb, |
|
use_wds=args.use_wds, |
|
use_lmdb_compressed=args.use_lmdb_compressed, |
|
plucker_embedding=args.plucker_embedding |
|
|
|
) |
|
|
|
if args.pose_warm_up_iter > 0: |
|
overfitting_dataset = load_memory_data( |
|
file_path=args.data_dir, |
|
batch_size=args.batch_size, |
|
reso=args.image_size, |
|
reso_encoder=args.image_size_encoder, |
|
num_workers=args.num_workers, |
|
|
|
|
|
**args_to_dict(args, |
|
dataset_defaults().keys())) |
|
data = [data, overfitting_dataset, args.pose_warm_up_iter] |
|
|
|
eval_data = load_eval_data( |
|
file_path=args.eval_data_dir, |
|
batch_size=args.eval_batch_size, |
|
reso=args.image_size, |
|
reso_encoder=args.image_size_encoder, |
|
num_workers=args.num_workers, |
|
load_depth=True, |
|
preprocess=auto_encoder.preprocess, |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
**args_to_dict(args, |
|
dataset_defaults().keys())) |
|
|
|
logger.log("creating data loader done...") |
|
|
|
args.img_size = [args.image_size_encoder] |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
dist_util.synchronize() |
|
|
|
|
|
|
|
opt = dnnlib.EasyDict(args_to_dict(args, loss_defaults().keys())) |
|
|
|
if 'disc' in args.trainer_name: |
|
loss_class = E3DGE_with_AdvLoss( |
|
device, |
|
opt, |
|
|
|
disc_factor=args.patchgan_disc_factor, |
|
disc_weight=args.patchgan_disc_g_weight).to(device) |
|
else: |
|
loss_class = E3DGELossClass(device, opt).to(device) |
|
|
|
|
|
|
|
logger.log("training...") |
|
|
|
TrainLoop = { |
|
'input_rec': TrainLoop3DRec, |
|
'nv_rec': TrainLoop3DRecNV, |
|
|
|
'nv_rec_patch': TrainLoop3DRecNVPatchSingleForward, |
|
'nv_rec_patch_mvE': TrainLoop3DRecNVPatchSingleForwardMV, |
|
'nv_rec_patch_mvE_disc': TrainLoop3DRecNVPatchSingleForwardMVAdvLoss, |
|
}[args.trainer_name] |
|
|
|
logger.log("creating TrainLoop done...") |
|
|
|
|
|
|
|
auto_encoder.decoder.rendering_kwargs = args.rendering_kwargs |
|
train_loop = TrainLoop( |
|
rec_model=auto_encoder, |
|
loss_class=loss_class, |
|
data=data, |
|
eval_data=eval_data, |
|
|
|
**vars(args)) |
|
|
|
if args.inference: |
|
camera = th.load('assets/objv_eval_pose.pt', map_location=dist_util.dev()) |
|
train_loop.eval_novelview_loop(camera=camera, |
|
save_latent=args.save_latent) |
|
else: |
|
train_loop.run_loop() |
|
|
|
|
|
def create_argparser(**kwargs): |
|
|
|
|
|
defaults = dict( |
|
seed=0, |
|
dataset_size=-1, |
|
trainer_name='input_rec', |
|
use_amp=False, |
|
overfitting=False, |
|
num_workers=4, |
|
image_size=128, |
|
image_size_encoder=224, |
|
iterations=150000, |
|
anneal_lr=False, |
|
lr=5e-5, |
|
weight_decay=0.0, |
|
lr_anneal_steps=0, |
|
batch_size=1, |
|
eval_batch_size=12, |
|
microbatch=-1, |
|
ema_rate="0.9999", |
|
log_interval=50, |
|
eval_interval=2500, |
|
save_interval=10000, |
|
resume_checkpoint="", |
|
use_fp16=False, |
|
fp16_scale_growth=1e-3, |
|
data_dir="", |
|
eval_data_dir="", |
|
|
|
logdir="/mnt/lustre/yslan/logs/nips23/", |
|
|
|
pose_warm_up_iter=-1, |
|
inference=False, |
|
export_latent=False, |
|
save_latent=False, |
|
) |
|
|
|
defaults.update(dataset_defaults()) |
|
defaults.update(encoder_and_nsr_defaults()) |
|
defaults.update(loss_defaults()) |
|
|
|
parser = argparse.ArgumentParser() |
|
add_dict_to_argparser(parser, defaults) |
|
|
|
return parser |
|
|
|
|
|
if __name__ == "__main__": |
|
|
|
|
|
|
|
|
|
|
|
args = create_argparser().parse_args() |
|
args.local_rank = int(os.environ["LOCAL_RANK"]) |
|
|
|
|
|
args.gpus = th.cuda.device_count() |
|
|
|
opts = args |
|
|
|
args.rendering_kwargs = rendering_options_defaults(opts) |
|
|
|
|
|
with open(os.path.join(args.logdir, 'args.json'), 'w') as f: |
|
json.dump(vars(args), f, indent=2) |
|
|
|
|
|
print('Launching processes...') |
|
|
|
try: |
|
training_loop(args) |
|
|
|
except Exception as e: |
|
|
|
traceback.print_exc() |
|
dist_util.cleanup() |
|
|