Tensorboard 2.21 cannot show backprogation part of keras model (TF 2.2)

I try the TF 2.2 with Tensorboard. But it cannot show the backprogation part of keras model, and the inputs/outputs/attributes of op node.

tensorboard 2.2.1
tensorflow 2.2.0
Ubuntu18.04

Code:

import tensorflow as tf
import tensorboard

print("tensorflow: ", tf.version.VERSION)
print("tensorboard: ", tensorboard.__version__)

mnist = tf.keras.datasets.mnist

(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0

print ("input shape: ", x_train.shape)

model = tf.keras.models.Sequential([
  tf.keras.layers.Flatten(input_shape=(28, 28)),
  tf.keras.layers.Dense(128, activation='relu'),
  tf.keras.layers.Dropout(0.2),
  tf.keras.layers.Dense(10)
])

model.summary()

predictions = model(x_train[:1]).numpy()
print (predictions)

loss_fn = tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True)
print (loss_fn(y_train[:1], predictions).numpy())

model.compile(optimizer='adam',
              loss=loss_fn,
              metrics=['accuracy'])

log_dir = "log_mnist_fit"
tensorboard_callback = tf.keras.callbacks.TensorBoard(log_dir=log_dir, histogram_freq=1)

model.fit(x_train, y_train, epochs=5, callbacks=[tensorboard_callback])

$ tensorboard --logdir log_mnist_fit/train/

When I click the op node, no inputs/outputs for the node in the right-top part!

======
I also try tensoboard 2.1.0. It can show the inputs/outputs/attributes of op node. But the backpropagation is still absent.

Thanks in advance!