Tensorflow 1.15, Keras 2.2.5, on_batch_end或on_train_batch_en

Question

我的自定义callblack:

class CustomCallback(keras.callbacks.Callback):
def on_train_begin(self, logs=None):
print("on_train_begin")
def on_train_batch_end(self, batch, logs=None):
print("on_train_batch_end")
def on_batch_end(self, batch, logs=None):
print("on_batch_end")

问题:在培训期间，我只看到"on_train_begin"出现在输出中。

我的Tensorflow版本:1.15我的Keras版本:2.2.5

有答案吗?

谢谢

p。s:我不想改变我的Keras/TF版本

Answer 1

我不得不打电话:

clist.on_train_batch_end(batch_num, logs)

训练循环中的

clist = CallbackList(callbacks=self.callbacks)