def _datasets_loader(r: registry.Registry):
from catalyst.data import dataset as m # noqa: WPS347
r.add_from_module(m)
from catalyst.contrib import datasets as m_contrib # noqa: WPS347
r.add_from_module(m_contrib)
def _engines_loader(r: registry.Registry):
from catalyst.core.engine import IEngine
r.add(IEngine)
from catalyst import engines as m # noqa: WPS347
r.add_from_module(m)
def _runners_loader(r: registry.Registry):
from catalyst.core.runner import IRunner
r.add(IRunner)
r.add(IRunner)
from catalyst import runners as m # noqa: WPS347
r.add_from_module(m)
def _callbacks_loader(r: registry.Registry):
from catalyst.core.callback import Callback, CallbackWrapper
r.add(Callback)
r.add(CallbackWrapper)
from catalyst import callbacks as m # noqa: WPS347
r.add_from_module(m)
def _optimizers_loader(r: registry.Registry):
from catalyst.contrib.nn import optimizers as m
r.add_from_module(m)
if SETTINGS.fairscale_required:
from fairscale import optim as m2
r.add_from_module(m2, prefix=["fairscale."])
def test_add_module():
"""@TODO: Docs. Contribution is welcome."""
r = Registry("")
r.add_from_module(module)
r.get("foo")
with pytest.raises(RegistryException):
r.get_instance("bar")
def _samplers_loader(r: Registry):
from torch.utils.data import sampler as s
factories = {
k: v
for k, v in s.__dict__.items() if "Sampler" in k and k != "Sampler"
}
r.add(**factories)
from catalyst.data import sampler
r.add_from_module(sampler)
def _model_loader(r: Registry):
from catalyst.contrib import models as m
r.add_from_module(m)
try:
import segmentation_models_pytorch as smp
r.add_from_module(smp, prefix="smp.")
except ImportError as ex:
if settings.segmentation_models_required:
logger.warning("segmentation_models_pytorch not available,"
" to install segmentation_models_pytorch,"
" run `pip install segmentation-models-pytorch`.")
raise ex
def _transforms_loader(r: registry.Registry):
from catalyst.data import transforms as t
# add `'transform.'` prefix to avoid nameing conflicts with other catalyst modules
r.add_from_module(t, prefix=["transform."])
if SETTINGS.albu_required:
import albumentations as m
r.add_from_module(m, prefix=["A.", "albu.", "albumentations."])
from albumentations import pytorch as p
r.add_from_module(p, prefix=["A.", "albu.", "albumentations."])
def _transforms_loader(r: Registry):
from torch.jit.frontend import UnsupportedNodeError
try:
import albumentations as m
r.add_from_module(m, prefix=["A.", "albu.", "albumentations."])
from albumentations import pytorch as p
r.add_from_module(p, prefix=["A.", "albu.", "albumentations."])
from catalyst.contrib.data.cv import transforms as t
r.add_from_module(t, prefix=["catalyst.", "C."])
except ImportError as ex:
if settings.albumentations_required:
logger.warning(
"albumentations not available, to install albumentations, "
"run `pip install albumentations`.")
raise ex
try:
from kornia import augmentation as k
r.add_from_module(k, prefix=["kornia."])
except ImportError as ex:
if settings.kornia_required:
logger.warning("kornia not available, to install kornia, "
"run `pip install kornia`.")
raise ex
except UnsupportedNodeError as ex:
logger.warning(
"kornia has requirement torch>=1.5.0, probably you have"
" an old version of torch which is incompatible.\n"
"To update pytorch, run `pip install -U 'torch>=1.5.0'`.")
if settings.kornia_required:
raise ex
def _transforms_loader(r: Registry):
try:
import albumentations as m
r.add_from_module(m, prefix=["A.", "albu.", "albumentations."])
from albumentations import pytorch as p
r.add_from_module(p, prefix=["A.", "albu.", "albumentations."])
from catalyst.contrib.data.cv import transforms as t
r.add_from_module(t, prefix=["catalyst.", "C."])
except ImportError as ex:
if settings.albumentations_required:
logger.warning(
"albumentations not available, to install albumentations, "
"run `pip install albumentations`.")
raise ex
def _modules_loader(r: Registry):
from catalyst.contrib.nn import modules as m
r.add_from_module(m)
def _grad_clip_loader(r: Registry):
from torch.nn.utils import clip_grad as m
r.add_from_module(m)
def _loggers_loader(r: registry.Registry):
from catalyst import loggers as m # noqa: WPS347
r.add_from_module(m)
def _schedulers_loader(r: Registry):
from catalyst.contrib.nn import schedulers as m
r.add_from_module(m)
def _optimizers_loader(r: Registry):
from catalyst.contrib.nn import optimizers as m
r.add_from_module(m)
def _criterion_loader(r: Registry):
from catalyst.contrib.nn import criterion as m
r.add_from_module(m)
def _torch_functional_loader(r: registry.Registry):
import torch.nn.functional as F
r.add_from_module(F, ["F."])
def _torch_loader(r: registry.Registry):
import torch as m
r.add_from_module(m, ["torch."], ignore_all=True)
def _model_loader(r: registry.Registry):
from catalyst.contrib import models as m
r.add_from_module(m)
def _callbacks_loader(r: Registry):
from catalyst.dl import callbacks as m # noqa: WPS347
r.add_from_module(m)
def _callbacks_loader(r: Registry):
from catalyst.core import callbacks as m
r.add_from_module(m)
