LSTM training error about end-to-end-time-series HOT 3 OPEN

I see you are using fit method rather fit_generator. One is you can go back to 2.2 TF version and use fit_generator. I will try to change the code against latest version and update

from end-to-end-time-series.

Hi, I was able to make the same code work on TF 2.5 as well. instead of fit_generated which is deprecated fit works as well. I am hardening the code and will upload in upcoming week

This is what I used in TF 2.5

early_stopping = tf.keras.callbacks.EarlyStopping(monitor='val_loss',
patience=2,
mode='min')

model.compile(loss=tf.losses.MeanSquaredError(),
optimizer=tf.optimizers.Adam(),
metrics=[tf.metrics.MeanAbsoluteError()])

history = model.fit(train_generator, epochs=20,
validation_data=val_generator,
shuffle=False,
callbacks=[early_stopping])

from end-to-end-time-series.

from end-to-end-time-series.