I did not try this feature to have a different number of layers for the Encoder and the Decoder.
-enc_layers (default: 0)
If > 0, number of layers of the encoder. This overrides the global -layers option.
-dec_layers (default: 0)
If > 0, number of layers of the decoder. This overrides the global -layers option.
Has anyone tried ?
Also, would it make sense to have an extra setting to allow more than one layer for the AtentionModel ?
Cheers.