How to convert frame features and frame mask as a single variable data?

Question

I was trying to convert a dataset as tensorflow tensor. However, it is not letting me to do so. I followed the keras video classification tutorial (https://keras.io/examples/vision/video_classification/). The tutorial converts training video data into frame features and frame mask and then feeds both for model training.

    history = seq_model.fit(
        [train_data[0], train_data[1]],
        train_labels,
        validation_split=0.3,
        epochs=EPOCHS,
        callbacks=[checkpoint],
    )

I tried to convert the frame features and frame masks into a single tensorflow tensor so that I can feed the data into a federated learning training. However it is not working. I applied tf.ragged.constant but no luck on that.