Multi task learning after the encoded layer

Hi,

Please consider building a multi-task learning setup sharing the encoding and with two or more different decoders.
Let me know your thoughts on this.

Thanks!