def clear_logs(self, force=False):
if force:
do_clear = True
else:
# Ask user if they really want to clear
while True:
do_clear = Notification(
DEEP_NOTIF_INPUT,
"Are you sure you want to clear logs from session : %s ? (yes / no)"
% self.config.project.session).get()
if do_clear.lower() in ["yes", "y"]:
do_clear = True
break
elif do_clear.lower() in ["no", "n"]:
do_clear = False
break
if do_clear:
Notification(
DEEP_NOTIF_INFO,
DEEP_MSG_BRAIN_CLEAR_LOGS % self.config.project.session)
path = "/".join((self.config.project.session, "logs"))
try:
for file in os.listdir(path):
os.remove("/".join((path, file)))
except FileNotFoundError:
pass
Notification(
DEEP_NOTIF_SUCCESS,
DEEP_MSG_BRAIN_CLEARED_LOGS % self.config.project.session)
def clear(self, force=False):
"""
:return:
"""
if force:
do_clear = True
else:
# Ask user if they really want to clear
while True:
do_clear = Notification(
DEEP_NOTIF_INPUT,
"Are you sure you want to clear this session : %s ? (yes / no)"
% self.config.project.session).get()
if do_clear.lower() in ["yes", "y"]:
do_clear = True
break
elif do_clear.lower() in ["no", "n"]:
do_clear = False
break
if do_clear:
Notification(
DEEP_NOTIF_INFO,
DEEP_MSG_BRAIN_CLEAR_ALL % self.config.project.session)
for directory in DEEP_LOG_RESULT_DIRECTORIES:
path = "/".join((self.config.project.session, directory))
try:
for file in os.listdir(path):
os.remove("/".join((path, file)))
except FileNotFoundError:
pass
Notification(
DEEP_NOTIF_SUCCESS,
DEEP_MSG_BRAIN_CLEARED_ALL % self.config.project.session)
def __check_exists(self, project_path):
"""
Author: SW
Checks that the project does not exist already
If the project exists already, the user is asked if the existing project should be overwritten
If the project does not exist already, the project directory is made
:param project_path: str: absolute path to the main project directory
:return: bool: whether or not project generation should continue
"""
if os.path.isdir(project_path) and not self.force_overwrite:
while True:
Notification(DEEP_NOTIF_WARNING, DEEP_MSG_PROJECT_ALREADY_EXISTS % project_path, log=False)
proceed = Notification(DEEP_NOTIF_INPUT, DEEP_MSG_CONTINUE, log=False).get()
if proceed.lower() in ["y", "yes"]:
return True
elif proceed.lower() in ["n", "no"]:
return False
else:
os.makedirs(project_path, exist_ok=True)
return True
def __continue_training(self):
"""
AUTHORS:
--------
:author: Alix Leroy
DESCRIPTION:
------------
Ask if we want to continue once the training ended
PARAMETERS:
-----------
None
RETURN:
-------
:return: None
"""
continue_training = ""
# Ask if the user want to continue the training
while continue_training.lower() not in ["y", "n"]:
continue_training = Notification(DEEP_NOTIF_INPUT, "Would you like to continue the training ? (Y/N) ").get()
# If yes ask the number of epochs
if continue_training.lower() == "y":
while True:
epochs = Notification(DEEP_NOTIF_INPUT, "Number of epochs ? ").get()
try:
epochs = int(epochs)
break
except ValueError:
Notification(DEEP_NOTIF_WARNING, "Number of epochs must be an integer").get()
if epochs > 0:
self.initial_epoch = self.num_epochs + 1
self.num_epochs += epochs
# Resume the training
self.fit(first_training=False)
else:
pass
def set_len_dataset(self, length_data : int) -> None:
"""
AUTHORS:
--------
author: Alix Leroy
DESCRIPTION:
------------
Set the length of the dataset
PARAMETERS:
-----------
:param length_data -> Integer: Desired length of the dataset
RETURN:
-------
None
"""
# If the given length is smaller than the number of instances already available in the dataset
if length_data < len(self.data):
res = ""
# Ask the user to confirm the given length
while res.lower() not in ["y", "yes", "no", "n"]:
res = Notification(DEEP_NOTIF_INPUT, "Dataset contains {0} instances, are you sure you want to only use {1} instances ? (Y/N) ".format(len(self.data), length_data)).get()
if res.lower() == "y":
self.len_data = length_data
# If not confirmed, keep the current size of the dataset as default
else:
self.len_data = len(self.data)
# If there isn't any issue of length set the length given as argument
else:
self.len_data = length_data
def __handle_error_saving(self, name: str, model: Module) -> None:
"""
AUTHORS:
--------
:author: Alix Leroy
DESCRIPTION:
------------
Handle the error.
Suggest solutions:
- Retry to save the model
- Change the save format
Exit the program if no solution found
:param model->Module: The model to save
RETURN:
-------
:return: None
"""
Notification(
DEEP_NOTIF_ERROR,
"Please make sure you have the permission to write for this following file : "
+ str(name))
response = ""
while response.lower() != ("y" or "n"):
response = Notification(
DEEP_NOTIF_INPUT,
"Would you try to try again to save? (y/n)").get()
if response.lower() == "y":
self.save_model(model)
else:
response = ""
while response.lower() != ("y" or "n"):
response = Notification(
DEEP_NOTIF_INPUT,
"Would you like to save in another format, if not Deeplodocus will be closed ? (y/n)"
).get()
if response.lower() == "n":
response = ""
while response.lower() != ("y" or "n"):
Notification(
DEEP_NOTIF_WARNING,
"You will lose all your data if Deeplodocus is closed !"
)
response = Notification(
DEEP_NOTIF_INPUT,
"Are you sure to close Deeplodocus (y/n)").get()
if response.lower() == "n":
self.__handle_error_saving()
else:
End(error=False) #Exiting the program
else:
response = ""
while response.lower() != ("pytorch" or "onnx"):
response = Notification(
DEEP_NOTIF_INPUT,
"What format would you like to save ? (pytorch/onnx)"
).get()
if response.lower() == "pytorch":
self.save_model_method = DEEP_SAVE_NET_FORMAT_PYTORCH
elif response.lower() == "onnx":
self.save_model_method = DEEP_SAVE_NET_FORMAT_ONNX
self.save_model(model)
