def initialize(self):
BaseOptions.initialize(self)
self.parser.add_argument('--results_dir',
type=str,
default='./results/',
help='saves results here.')
self.parser.add_argument('--phase',
type=str,
default='test',
help='train, val, test, etc') #todo delete.
self.parser.add_argument(
'--which_epoch',
type=str,
default='latest',
help='which epoch to load? set to latest to use latest cached model'
)
self.parser.add_argument('--num_aug',
type=int,
default=1,
help='# of augmentation files')
self.parser.add_argument('--dropout_p',
type=float,
default=0,
help='dropout layer probability')
self.is_train = False
def initialize(self):
BaseOptions.initialize(self)
self.is_Train = False
self.parser.add_argument('--batchsize',
type=int,
default=3,
help='input batch size')
def initialize(self):
BaseOptions.initialize(self)
self.is_Train = True
self.parser.add_argument('--batchsize', type=int, default=64, help='input batch size')
self.parser.add_argument('--lr_G', type=float, default=0.0002, help='initial learning rate of Generator')
self.parser.add_argument('--lr_D', type=float, default=0.0002, help='initial learning rate of Discriminator')
self.parser.add_argument('--count_epoch', type=int, default=0, help='the starting count epoch count')
self.parser.add_argument('--epochs', type=int, default=10000, help='number of epochs for train')
self.parser.add_argument('--beta1', type=float, default=0.5, help='adam optimizer parameter')
self.parser.add_argument('--beta2', type=float, default=0.999, help='adam optimizer parameter')
self.parser.add_argument('--save_epoch_freq', type=int, default=5, help='frequency of saving the latest results')
self.parser.add_argument('--D_interval', type=int, default=20, help='the interval of each optimization of D')
self.parser.add_argument('--w_L1', type=int, default=1, help='the weight of the L1 loss')
def initialize(self, parser):
parser = BaseOptions.initialize(self, parser)
parser.add_argument('--batch_size',
type=int,
default=4,
help='input batch size')
parser.add_argument('--mode', type=str, default='test')
parser.add_argument('--model_load_path',
type=str,
default='checkpoints',
help='dir for model saving')
parser.add_argument('--save_path',
type=str,
default='',
help='result saving path')
return parser
def initialize(self, parser):
parser = BaseOptions.initialize(self, parser)
parser.add_argument('--batch_size',
type=int,
default=16,
help='input batch size')
parser.add_argument('--lr',
type=float,
default=0.0001,
help='initial learning rate for Adam')
parser.add_argument('--mode', type=str, default='train')
parser.add_argument('--weight-decay',
'--wd',
default=1e-4,
type=float,
metavar='W',
help='weight decay (default: 1e-4) for Adam')
parser.set_defaults(model_save_path='checkpoints')
parser.set_defaults(epochs=21)
parser.set_defaults(batch_size=8)
return parser
def initialize(self):
BaseOptions.initialize(self)
self.parser.add_argument(
'--display_freq',
type=int,
default=100,
help='frequency of showing training results on screen')
self.parser.add_argument(
'--print_freq',
type=int,
default=100,
help='frequency of showing training results on console')
self.parser.add_argument('--save_latest_freq',
type=int,
default=1000,
help='frequency of saving the latest results')
self.parser.add_argument(
'--save_epoch_freq',
type=int,
default=1,
help='frequency of saving checkpoints at the end of epochs')
self.parser.add_argument(
'--continue_train',
action='store_true',
help='continue training: load the latest model')
self.parser.add_argument('--phase',
type=str,
default='train',
help='train, val, test, etc')
self.parser.add_argument(
'--which_epoch',
type=str,
default='latest',
help='which epoch to load? set to latest to use latest cached model'
)
self.parser.add_argument('--niter',
type=int,
default=10,
help='# of iter at starting learning rate')
self.parser.add_argument(
'--niter_decay',
type=int,
default=10,
help='# of iter to linearly decay learning rate to zero')
self.parser.add_argument('--beta1',
type=float,
default=0.5,
help='momentum term of adam')
self.parser.add_argument('--lr',
type=float,
default=0.0002,
help='initial learning rate for adam')
self.parser.add_argument('--TTUR',
action='store_true',
help='Use TTUR training scheme')
self.parser.add_argument('--gan_mode',
type=str,
default='ls',
help='(ls|original|hinge)')
self.parser.add_argument(
'--pool_size',
type=int,
default=1,
help=
'the size of image buffer that stores previously generated images')
self.parser.add_argument(
'--no_html',
action='store_true',
help=
'do not save intermediate training results to [opt.checkpoints_dir]/[opt.name]/web/'
)
# for discriminators
self.parser.add_argument(
'--num_D',
type=int,
default=2,
help='number of patch scales in each discriminator')
self.parser.add_argument('--n_layers_D',
type=int,
default=3,
help='number of layers in discriminator')
self.parser.add_argument('--no_vgg',
action='store_true',
help='do not use VGG feature matching loss')
self.parser.add_argument('--no_ganFeat',
action='store_true',
help='do not match discriminator features')
self.parser.add_argument('--lambda_feat',
type=float,
default=10.0,
help='weight for feature matching')
self.parser.add_argument(
'--sparse_D',
action='store_true',
help='use sparse temporal discriminators to save memory')
# for temporal
self.parser.add_argument('--lambda_T',
type=float,
default=10.0,
help='weight for temporal loss')
self.parser.add_argument('--lambda_F',
type=float,
default=10.0,
help='weight for flow loss')
self.parser.add_argument(
'--n_frames_D',
type=int,
default=3,
help='number of frames to feed into temporal discriminator')
self.parser.add_argument(
'--n_scales_temporal',
type=int,
default=2,
help='number of temporal scales in the temporal discriminator')
self.parser.add_argument(
'--max_frames_per_gpu',
type=int,
default=1,
help='max number of frames to load into one GPU at a time')
self.parser.add_argument('--max_frames_backpropagate',
type=int,
default=1,
help='max number of frames to backpropagate')
self.parser.add_argument(
'--max_t_step',
type=int,
default=1,
help=
'max spacing between neighboring sampled frames. If greater than 1, the network may randomly skip frames during training.'
)
self.parser.add_argument(
'--n_frames_total',
type=int,
default=30,
help='the overall number of frames in a sequence to train with')
self.parser.add_argument(
'--niter_step',
type=int,
default=5,
help='how many epochs do we change training batch size again')
self.parser.add_argument(
'--niter_fix_global',
type=int,
default=0,
help=
'if specified, only train the finest spatial layer for the given iterations'
)
self.isTrain = True
def initialize(self):
BaseOptions.initialize(self)
self.parser.add_argument(
'--print_freq',
type=int,
default=10,
help='frequency of showing training results on console')
self.parser.add_argument('--save_latest_freq',
type=int,
default=250,
help='frequency of saving the latest results')
self.parser.add_argument(
'--save_epoch_freq',
type=int,
default=1,
help='frequency of saving checkpoints at the end of epochs')
self.parser.add_argument(
'--run_test_freq',
type=int,
default=1,
help='frequency of running test in training script')
self.parser.add_argument(
'--continue_train',
action='store_true',
help='continue training: load the latest model')
self.parser.add_argument(
'--epoch_count',
type=int,
default=1,
help=
'the starting epoch count, we save the model by <epoch_count>, <epoch_count>+<save_latest_freq>, ...'
)
self.parser.add_argument('--phase',
type=str,
default='train',
help='train, val, test, etc')
self.parser.add_argument(
'--which_epoch',
type=str,
default='latest',
help='which epoch to load? set to latest to use latest cached model'
)
self.parser.add_argument('--niter',
type=int,
default=150,
help='# of iter at starting learning rate')
self.parser.add_argument(
'--niter_decay',
type=int,
default=150,
help='# of iter to linearly decay learning rate to zero')
self.parser.add_argument('--beta1',
type=float,
default=0.9,
help='momentum term of adam')
self.parser.add_argument('--lr',
type=float,
default=0.0005,
help='initial learning rate for adam')
self.parser.add_argument(
'--lr_policy',
type=str,
default='lambda',
help='learning rate policy: lambda|step|plateau')
self.parser.add_argument(
'--lr_decay_iters',
type=int,
default=50,
help='multiply by a gamma every lr_decay_iters iterations')
self.parser.add_argument('--reg_weight',
type=float,
default=0,
help='l1 regularization weight')
# data augmentation stuff
self.parser.add_argument('--num_aug',
type=int,
default=5,
help='# of augmentation files')
self.parser.add_argument(
'--scale_verts',
action='store_true',
help='non-uniformly scale the mesh e.g., in x, y or z')
self.parser.add_argument(
'--slide_verts',
type=float,
default=0.1,
help='percent vertices which will be shifted along the mesh surface'
)
self.parser.add_argument('--flip_edges',
type=float,
default=0.1,
help='percent of edges to randomly flip')
# tensorboard visualization
self.parser.add_argument('--no_vis',
action='store_true',
help='will not use tensorboard')
self.parser.add_argument('--verbose_plot',
action='store_true',
help='plots network weights, etc.')
self.parser.add_argument('--optim',
type=str,
default='Adam',
help='plots network weights, etc.')
self.parser.add_argument('--dropout_p',
type=float,
default=0,
help='dropout layer probability')
self.is_train = True
import sys
import os
import torch.optim as optim
import torch
import collections
import numpy as np
import math
from multiprocessing import cpu_count
from base_options import BaseOptions
from my_face_dataset import FaceDataSet
SymbolWithOptimizer = collections.namedtuple('SymbolWithOptimizer',
['symbol', 'optimizer'])
if __name__ == "__main__":
opt = BaseOptions().parse()
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
num_workers = cpu_count() if torch.cuda.is_available() else 0
criterion = torch.nn.MSELoss()
symbol_list = []
from importlib import import_module
for symbol_file_name in opt.symbols:
symbol_module = import_module('symbols.' + symbol_file_name)
symbol = symbol_module.get_symbol()
if opt.load_epoch:
mode_ends = '_latest.pth' if opt.load_epoch == 'latest' else '%05d.pth' % opt.load_epoch
checkpoint_path = os.path.join(opt.checkpoints_dir,
def initialize(self):
BaseOptions.initialize(self)
self.parser.add_argument('--phase', type=str, default='test', help='train, val, test, etc')
self.isTrain = False
def initialize(self):
BaseOptions.initialize(self)
# # visdom and HTML visualization parameters
self.parser.add_argument(
'--display_freq',
type=int,
default=400,
help='frequency of showing training results on screen')
# self.parser.add_argument('--display_ncols', type=int, default=4, help='if positive, display all images in a single visdom web panel with certain number of images per row.')
# self.parser.add_argument('--display_id', type=int, default=1, help='window id of the web display')
# self.parser.add_argument('--display_server', type=str, default="http://localhost", help='visdom server of the web display')
# self.parser.add_argument('--display_env', type=str, default='main', help='visdom display environment name (default is "main")')
# self.parser.add_argument('--display_port', type=int, default=8097, help='visdom port of the web display')
self.parser.add_argument(
'--update_html_freq',
type=int,
default=1000,
help='frequency of saving training results to html')
self.parser.add_argument(
'--print_freq',
type=int,
default=100,
help='frequency of showing training results on console')
self.parser.add_argument('--eval_freq',
type=int,
default=100,
help='frequency of evaluating results')
# self.parser.add_argument('--no_html', action='store_true', help='do not save intermediate training results to [opt.checkpoints_dir]/[opt.name]/web/')
# # network saving and loading parameters
# self.parser.add_argument('--save_latest_freq', type=int, default=5000, help='frequency of saving the latest results')
# self.parser.add_argument('--save_epoch_freq', type=int, default=5, help='frequency of saving checkpoints at the end of epochs')
# self.parser.add_argument('--save_by_iter', action='store_true', help='whether saves model by iteration')
self.parser.add_argument(
'--continue_train',
action='store_true',
help='continue training: load the latest model')
self.parser.add_argument(
'--start_epoch',
type=int,
default=0,
help=
'the starting epoch count, we save the model by <epoch_count>, <epoch_count>+<save_latest_freq>, ...'
)
# self.parser.add_argument('--epoch_count', type=int, default=1, help='the starting epoch count, we save the model by <epoch_count>, <epoch_count>+<save_latest_freq>, ...')
# self.parser.add_argument('--phase', type=str, default='train', help='train, val, test, etc')
# # training parameters
self.parser.add_argument(
'--n_epochs',
type=int,
default=100,
help='number of epochs with the initial learning rate')
self.parser.add_argument(
'--n_epochs_decay',
type=int,
default=100,
help='number of epochs to linearly decay learning rate to zero')
self.parser.add_argument('--beta1',
type=float,
default=0.5,
help='momentum term of adam')
self.parser.add_argument('--lr',
type=float,
default=0.0002,
help='initial learning rate for adam')
# self.parser.add_argument('--gan_mode', type=str, default='lsgan', help='the type of GAN objective. [vanilla| lsgan | wgangp]. vanilla GAN loss is the cross-entropy objective used in the original GAN paper.')
# self.parser.add_argument('--pool_size', type=int, default=50, help='the size of image buffer that stores previously generated images')
self.parser.add_argument(
'--lr_policy',
type=str,
default='linear',
help='learning rate policy. [linear | step | plateau | cosine]')
self.parser.add_argument(
'--lr_decay_iters',
type=int,
default=50,
help='multiply by a gamma every lr_decay_iters iterations')
self.parser.add_argument('--resume',
type=bool,
default=True,
help='load pretrained model or not')
self.initialized = True
self.isTrain = True