68 lines
2.6 KiB
Python
68 lines
2.6 KiB
Python
from Load_process.file_processing import Process_File
|
|
import datetime
|
|
import torch
|
|
|
|
# def attention_block(input):
|
|
# channel = input.shape[-1]
|
|
|
|
# GAP = GlobalAveragePooling2D()(input)
|
|
|
|
# block = Dense(units = channel // 16, activation = "relu")(GAP)
|
|
# block = Dense(units = channel, activation = "sigmoid")(block)
|
|
# block = Reshape((1, 1, channel))(block)
|
|
|
|
# block = Multiply()([input, block])
|
|
|
|
# return block
|
|
|
|
class EarlyStopping:
|
|
def __init__(self, patience=74, verbose=False, delta=0):
|
|
self.patience = patience
|
|
self.verbose = verbose
|
|
self.delta = delta
|
|
self.counter = 0
|
|
self.best_loss = None
|
|
self.early_stop = False
|
|
|
|
def __call__(self, val_loss, model, save_path):
|
|
if self.best_loss is None:
|
|
self.best_loss = val_loss
|
|
self.save_checkpoint(val_loss, model, save_path)
|
|
elif val_loss > self.best_loss + self.delta:
|
|
self.counter += 1
|
|
if self.verbose:
|
|
print(f"EarlyStopping counter: {self.counter} out of {self.patience}")
|
|
if self.counter >= self.patience:
|
|
self.early_stop = True
|
|
else:
|
|
self.best_loss = val_loss
|
|
self.save_checkpoint(val_loss, model, save_path)
|
|
self.counter = 0
|
|
|
|
def save_checkpoint(self, val_loss, model, save_path):
|
|
torch.save(model.state_dict(), save_path)
|
|
if self.verbose:
|
|
print(f"Validation loss decreased ({self.best_loss:.6f} --> {val_loss:.6f}). Saving model to {save_path}")
|
|
|
|
|
|
def call_back(model_name, index, optimizer):
|
|
File = Process_File()
|
|
|
|
model_dir = '../Result/save_the_best_model/' + model_name
|
|
File.JudgeRoot_MakeDir(model_dir)
|
|
modelfiles = File.Make_Save_Root('best_model( ' + str(datetime.date.today()) + " )-" + index + ".pt", model_dir)
|
|
|
|
# model_mckp = ModelCheckpoint(modelfiles, monitor='val_loss', save_best_only=True, save_weights_only = True, mode='auto')
|
|
|
|
earlystop = EarlyStopping(patience=74, verbose=True) # 提早停止
|
|
|
|
reduce_lr = torch.optim.lr_scheduler.ReduceLROnPlateau(
|
|
optimizer,
|
|
factor = 0.94, # 學習率降低的量。 new_lr = lr * factor
|
|
patience = 2, # 沒有改進的時期數，之後學習率將降低
|
|
verbose = 0,
|
|
mode = 'min',
|
|
min_lr = 0 # 學習率下限
|
|
)
|
|
|
|
return modelfiles, earlystop, reduce_lr |