Time series classification but with a sequence in output

I'm using Python and I have a training set of sequences of this shape: (None, 9, 25), where:

• 9 are rows representing years from 2012 to 2020
• 25 are features.

So each of this 25 features has a value for each year going from 2012 to 2020.

As target (labels) I have, for each sequence of (9,25), a list of 9 elements representing a number (either 0 or 1) associated to each year, so a possible label for a certain sequence can be [1,1,1,1,0,0,0,0,0] for example.

So the purpose of my neural network is to learn from time series and output a sequence of 9 elements.

My idea was to use a RNN with each sequence of (9,25) in input, but I have no idea about the last layer (output).