# Copyright 2022 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ============================================================================
"""
Callback for saving checkpoint.
"""
import os
from pathlib import Path
import mindspore
from mindnlp.abc import Callback
[docs]class CheckpointCallback(Callback):
"""
Save checkpoint of the model. save the current Trainer state at the end of each epoch, which can be used to
resume previous operations.
Continue training a sample code using the most recent epoch
Args:
save_path (str, Path): The path to save the state. A specific path needs to be specified,
such as 'checkpoints/'.
ckpt_name (str): Checkpoint name to store. It will set model class name when not specified.
Default: None.
epochs (int): Save a checkpoint file every n epochs.
keep_checkpoint_max (int): Save checkpoint files at most. Default:5.
"""
def __init__(self, save_path, ckpt_name=None, epochs=None, keep_checkpoint_max=5):
if isinstance(save_path, str):
self.save_path = Path(save_path)
elif isinstance(save_path, Path):
self.save_path = save_path
else:
raise ValueError(f"the 'save_path' argument must be str or Path, but got {type(save_path)}.")
if not self.save_path.exists():
os.makedirs(str(self.save_path))
self.epochs = epochs
self.keep_checkpoint_max = keep_checkpoint_max
self.ckpt_name = ckpt_name
self.cached_ckpts = []
# to do
# self.steps = steps
# if (self.epochs is not None) & (self.steps is not None):
# raise ValueError("The parameter epochs and steps cannot be assigned at the same time,\
# you can only keep one of them.")
# elif (self.epochs is None) & (self.steps is None):
# raise ValueError("The parameter epochs and steps both are None,\
# you must assign one of them.")
[docs] def train_begin(self, run_context):
"""
Notice the file saved path of checkpoints at the beginning of training.
Args:
run_context (RunContext): Information about the model.
"""
if self.epochs is None:
raise ValueError('For saving checkpoints, epoch cannont be `None` !')
print(f"The train will start from the checkpoint saved in '{self.save_path}'.")
[docs] def train_epoch_end(self, run_context):
"""
Save checkpoint every n epochs at the end of the epoch.
Args:
run_context (RunContext): Information about the model.
"""
if self.epochs is None:
return
if (run_context.cur_epoch_nums % self.epochs != 0) & (run_context.cur_epoch_nums != run_context.epochs):
return
model = run_context.network
if self.ckpt_name is None:
self.ckpt_name = type(model).__name__
ckpt_name = self.ckpt_name + '_epoch_' + str(run_context.cur_epoch_nums-1) + '.ckpt'
if len(self.cached_ckpts) == self.keep_checkpoint_max:
print('The maximum number of stored checkpoints has been reached.')
del_ckpt = self.cached_ckpts.pop(0)
del_file = self.save_path.joinpath(del_ckpt)
del_file.chmod(0o777)
del_file.unlink()
mindspore.save_checkpoint(model, str(self.save_path.joinpath(ckpt_name).resolve()))
self.cached_ckpts.append(ckpt_name)
print(f"Checkpoint: '{ckpt_name}' has been saved in epoch: {run_context.cur_epoch_nums - 1}.")