def wflw_data_iterator(data_dir=None,
dataset_mode="encoder_ref",
mode="train",
use_reference=False,
batch_size=1,
shuffle=True,
rng=None,
with_memory_cache=False,
with_file_cache=False,
transform=None):
if use_reference:
logger.info(
'WFLW Dataset for Encoder using reference .npz file is created.')
return data_iterator(
WFLWDataEncoderRefSource(data_dir,
shuffle=shuffle,
rng=rng,
transform=transform,
mode=mode), batch_size, rng,
with_memory_cache, with_file_cache)
else:
logger.info('WFLW Dataset for Encoder is created.')
return data_iterator(
WFLWDataEncoderSource(data_dir,
shuffle=shuffle,
rng=rng,
transform=transform,
mode=mode), batch_size, rng,
with_memory_cache, with_file_cache)
def celebv_data_iterator(dataset_mode=None, celeb_name=None, data_dir=None, ref_dir=None,
mode="all", batch_size=1, shuffle=False, rng=None,
with_memory_cache=False, with_file_cache=False,
resize_size=(64, 64), line_thickness=3, gaussian_kernel=(5, 5), gaussian_sigma=3
):
if dataset_mode == 'transformer':
if ref_dir:
assert os.path.exists(ref_dir), f'{ref_dir} not found.'
logger.info(
'CelebV Dataiterator using reference .npz file for Transformer is created.')
return data_iterator(CelebVDataRefSource(
celeb_name=celeb_name, data_dir=data_dir, ref_dir=ref_dir,
need_image=False, need_heatmap=True, need_resized_heatmap=False,
mode=mode, shuffle=shuffle, rng=rng),
batch_size, rng, with_memory_cache, with_file_cache)
else:
logger.info('CelebV Dataiterator for Transformer is created.')
return data_iterator(CelebVDataSource(
celeb_name=celeb_name, data_dir=data_dir,
need_image=False, need_heatmap=True, need_resized_heatmap=False,
mode=mode, shuffle=shuffle, rng=rng,
resize_size=resize_size, line_thickness=line_thickness,
gaussian_kernel=gaussian_kernel, gaussian_sigma=gaussian_sigma),
batch_size, rng, with_memory_cache, with_file_cache)
elif dataset_mode == 'decoder':
if ref_dir:
assert os.path.exists(ref_dir), f'{ref_dir} not found.'
logger.info(
'CelebV Dataiterator using reference .npz file for Decoder is created.')
return data_iterator(CelebVDataRefSource(
celeb_name=celeb_name, data_dir=data_dir, ref_dir=ref_dir,
need_image=True, need_heatmap=True, need_resized_heatmap=True,
mode=mode, shuffle=shuffle, rng=rng),
batch_size, rng, with_memory_cache, with_file_cache)
else:
logger.info('CelebV Dataiterator for Decoder is created.')
return data_iterator(CelebVDataSource(
celeb_name=celeb_name, data_dir=data_dir,
need_image=True, need_heatmap=True, need_resized_heatmap=True,
mode=mode, shuffle=shuffle, rng=rng,
resize_size=resize_size, line_thickness=line_thickness,
gaussian_kernel=gaussian_kernel, gaussian_sigma=gaussian_sigma),
batch_size, rng, with_memory_cache, with_file_cache)
else:
logger.error(
'Specified Dataitaretor is wrong? given: {}'.format(dataset_mode))
import sys
sys.exit()
def create_data_iterator(batch_size,
data_list,
load_shape,
crop_shape,
comm=None,
shuffle=True,
rng=None,
with_memory_cache=False,
with_parallel=False,
with_file_cache=False,
flip=True):
ds = Ade20kIterator(data_list,
load_shape,
crop_shape,
shuffle=shuffle,
rng=rng,
flip=flip)
# ds.slice turns withMemoryCache flag on forcibly.
# For data augmentation, this is not desirable and ds.slice is not used here.
ds = _get_sliced_data_source(ds, comm, shuffle)
return data_iterator(ds, batch_size, with_memory_cache, with_parallel,
with_file_cache)
def data_iterator_mnist(batch_size,
train=True,
rng=None,
shuffle=True,
with_memory_cache=False,
with_file_cache=False):
'''
Provide DataIterator with :py:class:`MnistDataSource`
with_memory_cache and with_file_cache option's default value is all False,
because :py:class:`MnistDataSource` is able to store all data into memory.
For example,
.. code-block:: python
with data_iterator_mnist(True, batch_size) as di:
for data in di:
SOME CODE TO USE data.
'''
return data_iterator(MnistDataSource(train=train, shuffle=shuffle, rng=rng),
batch_size,
rng,
with_memory_cache,
with_file_cache)
def data_iterator_stl10(batch_size,
train=True,
rng=None,
shuffle=True,
with_memory_cache=False,
with_file_cache=False,
output_dir=None):
'''
Provide DataIterator with :py:class:`STL10DataSource`
with_memory_cache and with_file_cache option's default value is all False,
because :py:class:`STL10DataSource` is able to store all data into memory.
'''
"""
_data_iterator = data_iterator(
STL10DataSource(train=train, shuffle=shuffle, rng=rng),
batch_size,
rng,
with_memory_cache,
with_file_cache
)
return _data_iterator
"""
with STL10DataSource(train=train, shuffle=shuffle, rng=rng, output_dir=output_dir) as ds, \
data_iterator(ds,
batch_size,
rng=rng,
with_memory_cache=with_memory_cache,
with_file_cache=with_file_cache) as di:
yield di
def data_iterator_modelnet40_normal_resampled(
data_dir: str,
batch_size: int,
train: bool,
shuffle: bool,
num_points: int,
normalize: bool,
stop_exhausted: bool = True,
with_memory_cache: bool = True,
with_file_cache: bool = False,
rng: Optional[int] = None,
) -> DataIterator:
dataset = ModelNet40NormalResampledDataset(
data_dir,
batch_size,
train,
shuffle,
num_points,
normalize,
)
return data_iterator(
dataset,
batch_size,
rng=rng,
with_memory_cache=with_memory_cache,
with_file_cache=with_file_cache,
stop_exhausted=stop_exhausted,
)
def SimpleDataIterator(batch_size,
root_dir,
image_size,
comm=None,
shuffle=True,
rng=None,
on_memory=True,
fix_aspect_ratio=True):
# get all files
paths = [
os.path.join(root_dir, x) for x in os.listdir(root_dir)
if os.path.splitext(x)[-1] in SUPPORT_IMG_EXTS
]
if len(paths) == 0:
raise ValueError(
f"[SimpleDataIterator] '{root_dir}' is not found. "
"Please make sure that you specify the correct directory path.")
ds = SimpleDatasource(img_paths=paths,
img_size=image_size,
rng=rng,
on_memory=on_memory,
fix_aspect_ratio=fix_aspect_ratio)
logger.info(f"Initialized data iterator. {ds.size} images are found.")
ds = _get_sliced_data_source(ds, comm, shuffle)
return data_iterator(ds,
batch_size,
with_memory_cache=False,
use_thread=True,
with_file_cache=False)
def data_iterator_yolo(root,
args,
batch_size,
shuffle=True,
train=False,
image_sizes=None,
image_size_change_freq=640,
on_memory_data=None,
use_cv2=True,
shape=None):
# "dataItertor for YoloDataSource"
assert image_size_change_freq % batch_size == 0, 'image_size_change_freq should be divisible by batch_size'
return data_iterator(YoloDataSource(
root,
args,
shuffle=shuffle,
train=train,
image_sizes=image_sizes,
image_size_change_freq=image_size_change_freq,
on_memory_data=on_memory_data,
use_cv2=use_cv2,
shape=shape),
batch_size=batch_size)
def data_iterator_cifar100(batch_size,
train=True,
rng=None,
shuffle=True,
with_memory_cache=False,
with_parallel=False,
with_file_cache=False):
'''
Provide DataIterator with :py:class:`Cifar100DataSource`
with_memory_cache, with_parallel and with_file_cache option's default value is all False,
because :py:class:`Cifar100DataSource` is able to store all data into memory.
For example,
.. code-block:: python
with data_iterator_cifar100(True, batch_size) as di:
for data in di:
SOME CODE TO USE data.
'''
with Cifar100DataSource(train=train, shuffle=shuffle, rng=rng) as ds, \
data_iterator(ds,
batch_size,
with_memory_cache,
with_parallel,
with_file_cache) as di:
yield di
def data_iterator_cifar10(batch_size,
train=True,
rng=None,
shuffle=True,
with_memory_cache=False,
with_parallel=False,
with_file_cache=False):
'''
Provide DataIterator with :py:class:`Cifar10DataSource`
with_memory_cache, with_parallel and with_file_cache option's default value is all False,
because :py:class:`Cifar10DataSource` is able to store all data into memory.
For example,
.. code-block:: python
with data_iterator_cifar10(True, batch_size) as di:
for data in di:
SOME CODE TO USE data.
'''
with Cifar10DataSource(train=train, shuffle=shuffle, rng=rng) as ds, \
data_iterator(ds,
batch_size,
with_memory_cache,
with_parallel,
with_file_cache) as di:
yield di
def create_data_iterator(batch_size, data_list, image_shape, shuffle=True, rng=None,
with_memory_cache=False, with_parallel=False, with_file_cache=False, flip=True):
return data_iterator(CityScapesIterator(data_list, image_shape, shuffle=shuffle, rng=rng, flip=flip),
batch_size,
with_memory_cache,
with_parallel,
with_file_cache)
def run(args):
"""Runs the algorithm."""
Path(hp.output_path).mkdir(parents=True, exist_ok=True)
# setup nnabla context
ctx = get_extension_context(args.context, device_id='0')
nn.set_default_context(ctx)
hp.comm = CommunicatorWrapper(ctx)
hp.event = StreamEventHandler(int(hp.comm.ctx.device_id))
if hp.comm.n_procs > 1 and hp.comm.rank == 0:
n_procs = hp.comm.n_procs
logger.info(f'Distributed training with {n_procs} processes.')
rng = np.random.RandomState(hp.seed)
# train data
train_loader = data_iterator(LJSpeechDataSource('metadata_train.csv',
hp,
shuffle=True,
rng=rng),
batch_size=hp.batch_size,
with_memory_cache=False)
# valid data
valid_loader = data_iterator(LJSpeechDataSource('metadata_valid.csv',
hp,
shuffle=False,
rng=rng),
batch_size=hp.batch_size,
with_memory_cache=False)
dataloader = dict(train=train_loader, valid=valid_loader)
model = Tacotron2(hp)
# setup optimizer
anneal_steps = [
x * (train_loader.size // hp.batch_size) for x in hp.anneal_steps
]
lr_scheduler = AnnealingScheduler(hp.alpha,
warmup=hp.warmup,
anneal_steps=anneal_steps,
anneal_factor=hp.anneal_factor)
optimizer = Optimizer(weight_decay=hp.weight_decay,
max_norm=hp.max_norm,
lr_scheduler=lr_scheduler,
name='Adam',
alpha=hp.alpha)
Tacotron2Trainer(model, dataloader, optimizer, hp).run()
def jsi_iterator(batch_size,
conf,
train,
with_memory_cache=False,
with_file_cache=False,
rng=None):
return data_iterator(JSIData(conf, train, shuffle=True, rng=None),
batch_size, rng, with_memory_cache, with_file_cache)
def data_iterator_librispeech(batch_size,
data_dir,
shuffle=True,
rng=None,
with_memory_cache=False,
with_file_cache=False):
return data_iterator(
LibriSpeechDataSource(data_dir, shuffle=shuffle, rng=rng), batch_size,
rng, with_memory_cache, with_file_cache)
def load_data(bs_train, bs_valid):
x_train = np.load(os.path.join(args.input, "x_train.npy"))
y_train = np.load(os.path.join(args.input, "y_shuffle_train.npy"))
x_val = np.load(os.path.join(args.input, "x_val.npy"))
y_val = np.load(os.path.join(args.input, "y_val.npy"))
data_source_train = Cifar10NumpySource(x_train, y_train, shuffle=False)
data_source_val = Cifar10NumpySource(x_val, y_val)
train_samples, val_samples = len(data_source_train.labels), len(
data_source_val.labels
)
train_loader = data_iterator(
data_source_train, bs_train, None, False, False)
val_loader = data_iterator(data_source_val, bs_valid, None, False, False)
return train_loader, val_loader, train_samples, val_samples
def get_photo_tourism_dataiterator(config, split, comm):
print(
f'Loading {split} images downscaled by a factor of {config.data.downscale}...')
data_source = PhototourismDataSource(config.data.root, img_downscale=int(config.data.downscale),
use_cache=config.data.use_cache, split=split)
if split == 'train':
di = data_iterator(data_source, batch_size=config.train.ray_batch_size)
if comm is not None:
di_ = di.slice(
rng=None, num_of_slices=comm.n_procs, slice_pos=comm.rank)
else:
di_ = di
elif split == 'val':
di_ = data_iterator(data_source, batch_size=1)
elif split == 'test':
return data_source
return di_
def simple_data_iterator(_data,
batch_size,
train=True,
rng=None,
shuffle=True,
with_memory_cache=False,
with_parallel=False,
with_file_cache=False):
return data_iterator(
SimpleDataSource(_data, train=train, shuffle=shuffle, rng=rng),
batch_size, with_memory_cache, with_parallel, with_file_cache)
def get_data_iterator_ffhq(data_config, batch_size, img_size, comm):
data_source = FFHQData(data_config, img_size)
data_iterator_ffhq = data_iterator(data_source, batch_size=batch_size)
if comm is not None:
if comm.n_procs > 1:
data_iterator_ffhq = data_iterator_ffhq.slice(
rng=None, num_of_slices=comm.n_procs, slice_pos=comm.rank)
return data_iterator_ffhq
def data_iterator_dtumvs(data_source,
batch_size,
rng=None,
with_memory_cache=False,
with_file_cache=False):
'''
Provide DataIterator with :py:class:`DTUMVSDataSource`
with_memory_cache and with_file_cache option's default value is all False,
because :py:class:`DTUMVSDataSource` is able to store all data into memory.
'''
return data_iterator(data_source, batch_size, rng, with_memory_cache,
with_file_cache)
def create_data_iterator(batch_size, data_list, image_shape, comm=None, shuffle=True, rng=None,
with_memory_cache=False, with_parallel=False, with_file_cache=False, flip=True):
ds = CityScapesIterator(data_list, image_shape,
shuffle=shuffle, rng=rng, flip=flip)
ds = _get_sliced_data_source(ds, comm, shuffle=shuffle)
return data_iterator(ds,
batch_size,
with_memory_cache,
with_parallel,
with_file_cache)
def facade_data_iterator(
images_root_path,
batch_size,
random_crop=True,
shuffle=True,
rng=None,
with_memory_cache=True,
with_parallel=False,
with_file_cache=False):
return data_iterator(FacadeDataSource(images_root_path, random_crop=random_crop, shuffle=shuffle, rng=rng),
batch_size,
with_memory_cache,
with_file_cache)
def data_iterator_cifar100(batch_size,
train=True,
rng=None,
shuffle=True,
with_memory_cache=False,
with_file_cache=False):
'''
Provide DataIterator with :py:class:`Cifar100DataSource`
with_memory_cache and with_file_cache option's default value is all False,
because :py:class:`Cifar100DataSource` is able to store all data into memory.
'''
data_source = Cifar100DataSource(train=train, shuffle=shuffle, rng=rng)
return data_source, data_iterator(data_source, batch_size, rng,
with_memory_cache, with_file_cache)
def Cifar10DataIterator(batch_size,
image_size=(32, 32),
comm=None,
shuffle=True,
rng=None,
train=True):
ds = Cifar10DataSource(train=train, shuffle=shuffle, rng=rng)
ds = _get_sliced_data_source(ds, comm, shuffle)
return data_iterator(ds,
batch_size,
with_memory_cache=False,
use_thread=True,
with_file_cache=False)
def get_data_iterator_mix(data_root,
comm,
batch_size,
image_size,
img_exts=['content.png', 'style.png', 'mix.png']):
data_source = MixingFacesData(data_root, image_size, img_exts)
data_iterator_ = data_iterator(data_source, batch_size=batch_size)
if comm is not None:
if comm.n_procs > 1:
data_iterator_ = data_iterator_.slice(rng=None,
num_of_slices=comm.n_procs,
slice_pos=comm.rank)
return data_iterator_
def get_data_iterator_attribute(data_root,
comm,
batch_size,
image_size,
img_exts=['o.png', 'y.png']):
data_source = AttributeFacesData(data_root, image_size, img_exts)
data_iterator_ = data_iterator(data_source, batch_size=batch_size)
if comm is not None:
if comm.n_procs > 1:
data_iterator_ = data_iterator_.slice(rng=None,
num_of_slices=comm.n_procs,
slice_pos=comm.rank)
return data_iterator_
def data_iterator_timeseries(dataset_path,
batch_size,
x_input_length=32,
x_output_length=16,
x_split_step=32,
rng=None,
shuffle=True,
with_memory_cache=False,
with_parallel=False,
with_file_cache=False):
return data_iterator(
TimeseriesDataSource(dataset_path=dataset_path,
x_input_length=x_input_length,
x_output_length=x_output_length,
x_split_step=x_split_step,
shuffle=shuffle,
rng=rng), batch_size, with_memory_cache,
with_parallel, with_file_cache)
def data_iterator_caltech101(batch_size,
train=True,
rng=None,
shuffle=True,
width=128,
height=128,
padding=True,
with_memory_cache=False,
with_file_cache=False):
'''
Provide DataIterator with :py:class:`Caltech101DataSource`
with_memory_cache and with_file_cache option's default value is all False,
because :py:class:`Caltech101DataSource` is able to store all data into memory.
'''
return data_iterator(Caltech101DataSource(width=width, height=height, padding=padding, train=train, shuffle=shuffle, rng=rng),
batch_size,
rng,
with_memory_cache,
with_file_cache)
def run(args):
# create output path
Path(hp.output_path).mkdir(parents=True, exist_ok=True)
# setup nnabla context
ctx = get_extension_context(args.context, device_id='0')
nn.set_default_context(ctx)
hp.comm = CommunicatorWrapper(ctx)
hp.event = StreamEventHandler(int(hp.comm.ctx.device_id))
with open(hp.speaker_dir) as f:
hp.n_speakers = len(f.read().split('\n'))
logger.info(f'Training data with {hp.n_speakers} speakers.')
if hp.comm.n_procs > 1 and hp.comm.rank == 0:
n_procs = hp.comm.n_procs
logger.info(f'Distributed training with {n_procs} processes.')
rng = np.random.RandomState(hp.seed)
train_loader = data_iterator(VCTKDataSource('metadata_train.csv',
hp,
shuffle=True,
rng=rng),
batch_size=hp.batch_size,
with_memory_cache=False,
rng=rng)
dataloader = dict(train=train_loader, valid=None)
gen = NVCNet(hp)
gen_optim = Optimizer(weight_decay=hp.weight_decay,
name='Adam',
alpha=hp.g_lr,
beta1=hp.beta1,
beta2=hp.beta2)
dis = Discriminator(hp)
dis_optim = Optimizer(weight_decay=hp.weight_decay,
name='Adam',
alpha=hp.d_lr,
beta1=hp.beta1,
beta2=hp.beta2)
Trainer(gen, gen_optim, dis, dis_optim, dataloader, rng, hp).run()
def frame_data_iterator(root_dir,
frame_shape=(256, 256, 3),
id_sampling=False,
is_train=True,
random_seed=0,
augmentation_params=None,
batch_size=1,
shuffle=True,
with_memory_cache=False,
with_file_cache=False):
return data_iterator(FramesDataSource(
root_dir=root_dir,
frame_shape=frame_shape,
id_sampling=id_sampling,
is_train=is_train,
random_seed=random_seed,
augmentation_params=augmentation_params,
shuffle=shuffle),
batch_size=batch_size,
rng=random_seed,
with_memory_cache=with_memory_cache,
with_file_cache=with_file_cache)
def train():
# Check NNabla version
if utils.get_nnabla_version_integer() < 11900:
raise ValueError(
'Please update the nnabla version to v1.19.0 or latest version since memory efficiency of core engine is improved in v1.19.0'
)
parser, args = get_train_args()
# Get context.
ctx = get_extension_context(args.context, device_id=args.device_id)
comm = CommunicatorWrapper(ctx)
nn.set_default_context(comm.ctx)
ext = import_extension_module(args.context)
# Monitors
# setting up monitors for logging
monitor_path = args.output
monitor = Monitor(monitor_path)
monitor_best_epoch = MonitorSeries('Best epoch', monitor, interval=1)
monitor_traing_loss = MonitorSeries('Training loss', monitor, interval=1)
monitor_validation_loss = MonitorSeries('Validation loss',
monitor,
interval=1)
monitor_lr = MonitorSeries('learning rate', monitor, interval=1)
monitor_time = MonitorTimeElapsed("training time per iteration",
monitor,
interval=1)
if comm.rank == 0:
print("Mixing coef. is {}, i.e., MDL = {}*TD-Loss + FD-Loss".format(
args.mcoef, args.mcoef))
if not os.path.isdir(args.output):
os.makedirs(args.output)
# Initialize DataIterator for MUSDB.
train_source, valid_source, args = load_datasources(parser, args)
train_iter = data_iterator(train_source,
args.batch_size,
RandomState(args.seed),
with_memory_cache=False,
with_file_cache=False)
valid_iter = data_iterator(valid_source,
1,
RandomState(args.seed),
with_memory_cache=False,
with_file_cache=False)
if comm.n_procs > 1:
train_iter = train_iter.slice(rng=None,
num_of_slices=comm.n_procs,
slice_pos=comm.rank)
valid_iter = valid_iter.slice(rng=None,
num_of_slices=comm.n_procs,
slice_pos=comm.rank)
# Calculate maxiter per GPU device.
max_iter = int((train_source._size // args.batch_size) // comm.n_procs)
weight_decay = args.weight_decay * comm.n_procs
print("max_iter", max_iter)
# Calculate the statistics (mean and variance) of the dataset
scaler_mean, scaler_std = utils.get_statistics(args, train_source)
max_bin = utils.bandwidth_to_max_bin(train_source.sample_rate, args.nfft,
args.bandwidth)
unmix = OpenUnmix_CrossNet(input_mean=scaler_mean,
input_scale=scaler_std,
nb_channels=args.nb_channels,
hidden_size=args.hidden_size,
n_fft=args.nfft,
n_hop=args.nhop,
max_bin=max_bin)
# Create input variables.
mixture_audio = nn.Variable([args.batch_size] +
list(train_source._get_data(0)[0].shape))
target_audio = nn.Variable([args.batch_size] +
list(train_source._get_data(0)[1].shape))
vmixture_audio = nn.Variable(
[1] + [2, valid_source.sample_rate * args.valid_dur])
vtarget_audio = nn.Variable([1] +
[8, valid_source.sample_rate * args.valid_dur])
# create training graph
mix_spec, M_hat, pred = unmix(mixture_audio)
Y = Spectrogram(*STFT(target_audio, n_fft=unmix.n_fft, n_hop=unmix.n_hop),
mono=(unmix.nb_channels == 1))
loss_f = mse_loss(mix_spec, M_hat, Y)
loss_t = sdr_loss(mixture_audio, pred, target_audio)
loss = args.mcoef * loss_t + loss_f
loss.persistent = True
# Create Solver and set parameters.
solver = S.Adam(args.lr)
solver.set_parameters(nn.get_parameters())
# create validation graph
vmix_spec, vM_hat, vpred = unmix(vmixture_audio, test=True)
vY = Spectrogram(*STFT(vtarget_audio, n_fft=unmix.n_fft,
n_hop=unmix.n_hop),
mono=(unmix.nb_channels == 1))
vloss_f = mse_loss(vmix_spec, vM_hat, vY)
vloss_t = sdr_loss(vmixture_audio, vpred, vtarget_audio)
vloss = args.mcoef * vloss_t + vloss_f
vloss.persistent = True
# Initialize Early Stopping
es = utils.EarlyStopping(patience=args.patience)
# Initialize LR Scheduler (ReduceLROnPlateau)
lr_scheduler = ReduceLROnPlateau(lr=args.lr,
factor=args.lr_decay_gamma,
patience=args.lr_decay_patience)
best_epoch = 0
# Training loop.
for epoch in trange(args.epochs):
# TRAINING
losses = utils.AverageMeter()
for batch in range(max_iter):
mixture_audio.d, target_audio.d = train_iter.next()
solver.zero_grad()
loss.forward(clear_no_need_grad=True)
if comm.n_procs > 1:
all_reduce_callback = comm.get_all_reduce_callback()
loss.backward(clear_buffer=True,
communicator_callbacks=all_reduce_callback)
else:
loss.backward(clear_buffer=True)
solver.weight_decay(weight_decay)
solver.update()
losses.update(loss.d.copy(), args.batch_size)
training_loss = losses.avg
# clear cache memory
ext.clear_memory_cache()
# VALIDATION
vlosses = utils.AverageMeter()
for batch in range(int(valid_source._size // comm.n_procs)):
x, y = valid_iter.next()
dur = int(valid_source.sample_rate * args.valid_dur)
sp, cnt = 0, 0
loss_tmp = nn.NdArray()
loss_tmp.zero()
while 1:
vmixture_audio.d = x[Ellipsis, sp:sp + dur]
vtarget_audio.d = y[Ellipsis, sp:sp + dur]
vloss.forward(clear_no_need_grad=True)
cnt += 1
sp += dur
loss_tmp += vloss.data
if x[Ellipsis,
sp:sp + dur].shape[-1] < dur or x.shape[-1] == cnt * dur:
break
loss_tmp = loss_tmp / cnt
if comm.n_procs > 1:
comm.all_reduce(loss_tmp, division=True, inplace=True)
vlosses.update(loss_tmp.data.copy(), 1)
validation_loss = vlosses.avg
# clear cache memory
ext.clear_memory_cache()
lr = lr_scheduler.update_lr(validation_loss, epoch=epoch)
solver.set_learning_rate(lr)
stop = es.step(validation_loss)
if comm.rank == 0:
monitor_best_epoch.add(epoch, best_epoch)
monitor_traing_loss.add(epoch, training_loss)
monitor_validation_loss.add(epoch, validation_loss)
monitor_lr.add(epoch, lr)
monitor_time.add(epoch)
if validation_loss == es.best:
# save best model
nn.save_parameters(os.path.join(args.output, 'best_xumx.h5'))
best_epoch = epoch
if stop:
print("Apply Early Stopping")
break
def cycle_gan_data_iterator(data_source, batch_size):
return data_iterator(data_source,
batch_size=batch_size,
with_memory_cache=False,
with_file_cache=False)