def test_update_config():
config = dict()
updates = {"a/b": 1.0, "a": 2}
update_config(config, updates)
ref = {"a": {"b": 1.0}}
assert config == ref
config = {"a": {"x": 0}}
updates = {"a/y": 1}
update_config(config, updates)
ref = {"a": {"x": 0, "y": 1}}
assert config == ref
config = {"a": {"x": 0}}
updates = {"a/x": 1}
update_config(config, updates)
ref = {"a": {"x": 1}}
assert config == ref
config = {"a": {"x": 0}}
updates = {"a/0": 1}
update_config(config, updates)
ref = {"a": {"x": 0, 0: 1}}
assert config == ref
config = {"a": {"x": 0}}
updates = {"a/x/1": 1}
update_config(config, updates)
ref = {"a": {"x": [None, 1]}}
assert config == ref
def test_config_format():
config = dict()
updates = {"a/b": 1.0, "x": "{a/b}"}
update_config(config, updates)
print(config)
ref = {"a": {"b": 1.0}, "x": 1.0}
assert config == ref
def standalone_eval_csv_file(path_to_csv, callbacks, additional_kwargs={}):
"""Runs all given callbacks on the data in the :class:`EvalDataFolder`
constructed from the given csv.abs
Parameters
----------
path_to_csv : str
Path to the csv file.
callbacks : list(str or Callable)
Import commands used to construct the functions applied to the Data
extracted from :attr:`path_to_csv`.
additional_kwargs : dict
Keypath-value pairs added to the config, which is extracted from
the ``model_outputs.csv``.
Returns
-------
The collected outputs of the callbacks.
"""
import importlib
from edflow.main import get_implementations_from_config
from edflow.config import update_config
import sys
sys.path.append(os.getcwd()) # convenience: load implementations from cwd
out_data = EvalDataFolder(path_to_csv)
config = read_meta_data(path_to_csv)
update_config(config, additional_kwargs)
dataset_str = config["dataset"]
impl = get_implementations_from_config(config, ["dataset"])
in_data = impl["dataset"](config)
if not isinstance(callbacks, list):
callbacks = [callbacks]
outputs = []
for cb in callbacks:
if isinstance(cb, str):
module = ".".join(cb.split(".")[:-1])
module = importlib.import_module(module)
cb = getattr(module, cb.split(".")[-1])
outputs += [
cb(os.path.dirname(path_to_csv), in_data, out_data, config)
]
return outputs
parser.add_argument(
"-d",
"--disable_cache",
action="store_true",
help="Disable caching dataset instantiation.",
)
parser.add_argument(
"-b",
"--base",
nargs="*",
metavar="config.yaml",
help="Paths to base configs.",
default=list(),
)
try:
sys.path.append(os.getcwd())
opt, unknown = parser.parse_known_args()
additional_kwargs = parse_unknown_args(unknown)
config = dict()
for base_config in opt.base:
with open(base_config) as f:
config.update(yaml.full_load(f))
update_config(config, additional_kwargs)
except SystemExit as e:
# This exception will be raised if --help or invalid command line arguments
# are used. Currently streamlit prevents the program from exiting normally
# so we have to do a hard exit.
os._exit(e.code)
explore(config, disable_cache=opt.disable_cache)
def standalone_eval_csv_file(path_to_csv,
callbacks,
additional_kwargs={},
other_config=None):
"""Runs all given callbacks on the data in the :class:`EvalDataFolder`
constructed from the given csv.abs
Parameters
----------
path_to_csv : str
Path to the csv file.
callbacks : dict(name: str or Callable)
Import commands used to construct the functions applied to the Data
extracted from :attr:`path_to_csv`.
additional_kwargs : dict
Keypath-value pairs added to the config, which is extracted from
the ``model_outputs.csv``. These will overwrite parameters in the
original config extracted from the csv.
other_config : str
Path to additional config used to update the existing one as taken from
the ``model_outputs.csv`` . Cannot overwrite the dataset. Only used for
callbacks. Parameters in this other config will overwrite the
parameters in the original config and those of the commandline
arguments.
Returns
-------
outputs: dict
The collected outputs of the callbacks.
"""
from edflow.main import get_implementations_from_config
from edflow.config import update_config
import yaml
if other_config is not None:
with open(other_config, "r") as f:
other_config = yaml.full_load(f)
else:
other_config = {}
out_data = EvalDataFolder(path_to_csv)
config = read_meta_data(path_to_csv)
dataset_str = config["dataset"]
impl = get_implementations_from_config(config, ["dataset"])
in_data = impl["dataset"](config)
update_config(config, additional_kwargs)
config.update(other_config)
config_callbacks, callback_kwargs = config2cbdict(config)
callbacks.update(config_callbacks)
callbacks = load_callbacks(callbacks)
root = os.path.dirname(path_to_csv)
outputs = apply_callbacks(callbacks, root, in_data, out_data, config,
callback_kwargs)
return outputs
def standalone_eval_meta_dset(path_to_meta_dir,
callbacks,
additional_kwargs={},
other_config=None):
"""Runs all given callbacks on the data in the :class:`EvalDataFolder`
constructed from the given csv.abs
Parameters
----------
path_to_csv : str
Path to the csv file.
callbacks : dict(name: str or Callable)
Import commands used to construct the functions applied to the Data
extracted from :attr:`path_to_csv`.
additional_kwargs : dict
Keypath-value pairs added to the config, which is extracted from
the ``model_outputs.csv``. These will overwrite parameters in the
original config extracted from the csv.
other_config : str
Path to additional config used to update the existing one as taken from
the ``model_outputs.csv`` . Cannot overwrite the dataset. Only used for
callbacks. Parameters in this other config will overwrite the
parameters in the original config and those of the commandline
arguments.
Returns
-------
outputs: dict
The collected outputs of the callbacks.
"""
from edflow.util import get_obj_from_str
from edflow.config import update_config
import yaml
if other_config is not None:
with open(other_config, "r") as f:
other_config = yaml.full_load(f)
else:
other_config = {}
out_data = MetaDataset(path_to_meta_dir)
out_data.expand = True
out_data.append_labels = True
config = out_data.meta
# backwards compatibility
if not "datasets" in config:
config["datasets"] = {"train": config["dataset"]}
if "validation_dataset" in config:
config["datasets"]["validation"] = config["validation_dataset"]
datasets = dict((split, get_obj_from_str(config["datasets"][split]))
for split in config["datasets"])
# TODO fix hardcoded dataset
in_data = datasets["validation"](config=config)
in_data.expand = True
update_config(config, additional_kwargs)
config.update(other_config)
config_callbacks, callback_kwargs = config2cbdict(config)
callbacks.update(config_callbacks)
callbacks = load_callbacks(callbacks)
root = os.path.dirname(path_to_meta_dir)
# TODO handle logging of return values
outputs = apply_callbacks(callbacks, root, in_data, out_data, config,
callback_kwargs)
return outputs
