以下代碼允許訓(xùn)練集的圖像在每個(gè)時(shí)期結(jié)束時(shí)旋轉(zhuǎn) 90°。from skimage.io import imreadfrom skimage.transform import resize, rotateimport numpy as npimport tensorflow as tffrom tensorflow import kerasfrom tensorflow.keras import layersfrom keras.utils import Sequence from keras.models import Sequentialfrom keras.layers import Conv2D, Activation, Flatten, Dense# Model architecture  (dummy)model = Sequential()model.add(Conv2D(32, (3, 3), input_shape=(15, 15, 4)))model.add(Activation('relu'))model.add(Flatten())modl.add(Dense(1))model.add(Activation('sigmoid'))model.compile(loss='binary_crossentropy',              optimizer='rmsprop',              metrics=['accuracy'])# Data iterator class CIFAR10Sequence(Sequence):    def __init__(self, filenames, labels, batch_size):        self.filenames, self.labels = filenames, labels        self.batch_size = batch_size        self.angles = [0,90,180,270]        self.current_angle_idx = 0    # Method to loop throught the available angles    def change_angle(self):      self.current_angle_idx += 1      if self.current_angle_idx >= len(self.angles):        self.current_angle_idx = 0      def __len__(self):        return int(np.ceil(len(self.filenames) / float(self.batch_size)))    # read, resize and rotate the image and return a batch of images    def __getitem__(self, idx):        angle = self.angles[self.current_angle_idx]        print (f"Rotating Angle: {angle}")        batch_x = self.filenames[idx * self.batch_size:(idx + 1) * self.batch_size]        batch_y = self.labels[idx * self.batch_size:(idx + 1) * self.batch_size]        return np.array([            rotate(resize(imread(filename), (15, 15)), angle)               for filename in batch_x]), np.array(batch_y)如何修改代碼以便在每個(gè)紀(jì)元期間發(fā)生圖像的旋轉(zhuǎn)？換句話說(shuō)，我如何編寫(xiě)一個(gè)在紀(jì)元“結(jié)束”時(shí)運(yùn)行的回調(diào)，該回調(diào)會(huì)更改角度值并繼續(xù)在同一紀(jì)元上進(jìn)行訓(xùn)練（而不更改到下一個(gè)紀(jì)元）？