def build_config(config_path: (str, list), log_root: str, log_dir: str,
ckpt_path: str) -> [dict, str]:
"""
Function to create new directory to log directory to store results.
:param config_path: string or list of strings, path of configuration files
:param log_root: str, root of logs
:param log_dir: string, path to store logs.
:param ckpt_path: str, path where model is stored.
:return: - config, configuration dictionary
- log_dir, path of the directory for saving outputs
"""
# check ckpt_path
if not ckpt_path.endswith(".ckpt"):
raise ValueError(
"checkpoint path should end with .ckpt, got {}".format(ckpt_path))
# init log directory
log_dir = build_log_dir(log_root=log_root, log_dir=log_dir)
# load config
if config_path == "":
# use default config, which should be provided in the log folder
config = config_parser.load_configs(
"/".join(ckpt_path.split("/")[:-2]) + "/config.yaml")
else:
# use customized config
logging.warning(
"Using customized configuration."
"The code might break if the config of the model doesn't match the saved model."
)
config = config_parser.load_configs(config_path)
return config, log_dir
def build_config(config_path: (str, list), log_dir: str,
ckpt_path: str) -> [dict, str]:
"""
Function to initialise log directories,
assert that checkpointed model is the right
type and to parse the configuration for training
:param config_path: list of str, path to config file
:param log_dir: str, path to where training logs to be stored.
:param ckpt_path: str, path where model is stored.
:return: - config: a dictionary saving configuration
- log_dir: the path of directory to save logs
"""
# init log directory
log_dir = build_log_dir(log_dir)
# check checkpoint path
if ckpt_path != "":
if not ckpt_path.endswith(".ckpt"):
raise ValueError("checkpoint path should end with .ckpt")
# load and backup config
config = config_parser.load_configs(config_path)
config_parser.save(config=config, out_dir=log_dir)
return config, log_dir
def test_load_configs():
"""
test load_configs by checking outputs
"""
# single config
# input is str not list
with open("config/unpaired_labeled_ddf.yaml") as file:
expected = yaml.load(file, Loader=yaml.FullLoader)
got = load_configs("config/unpaired_labeled_ddf.yaml")
assert got == expected
# multiple configs
with open("config/unpaired_labeled_ddf.yaml") as file:
expected = yaml.load(file, Loader=yaml.FullLoader)
got = load_configs(config_path=[
"config/test/ddf.yaml",
"config/test/unpaired_nifti.yaml",
"config/test/labeled.yaml",
])
assert got == expected
def build_config(
config_path: (str, list),
log_root: str,
log_dir: str,
ckpt_path: str,
max_epochs: int = -1,
) -> [dict, str]:
"""
Function to initialise log directories,
assert that checkpointed model is the right
type and to parse the configuration for training.
:param config_path: list of str, path to config file
:param log_root: str, root of logs
:param log_dir: str, path to where training logs to be stored.
:param ckpt_path: str, path where model is stored.
:param max_epochs: int, if max_epochs > 0, will use it to overwrite the configuration
:return: - config: a dictionary saving configuration
- log_dir: the path of directory to save logs
"""
# init log directory
log_dir = build_log_dir(log_root=log_root, log_dir=log_dir)
# check checkpoint path
if ckpt_path != "":
if not ckpt_path.endswith(".ckpt"):
raise ValueError(
f"checkpoint path should end with .ckpt, got {ckpt_path}")
# load config
config = config_parser.load_configs(config_path)
# overwrite epochs and save_period if necessary
if max_epochs > 0:
config["train"]["epochs"] = max_epochs
config["train"]["save_period"] = min(max_epochs,
config["train"]["save_period"])
# backup config
config_parser.save(config=config, out_dir=log_dir)
# batch_size in original config corresponds to batch_size per GPU
gpus = tf.config.experimental.list_physical_devices("GPU")
config["train"]["preprocess"]["batch_size"] *= max(len(gpus), 1)
return config, log_dir