CUNY-CL / yoyodyne

Small-vocabulary sequence-to-sequence generation with optional feature conditioning
Apache License 2.0
25 stars 15 forks source link

GRU support #180

Open kylebgorman opened 2 months ago

kylebgorman commented 2 months ago

It would be nice if our RNN encoders and decoders, which are currently LSTMs, could be replaced with GRUs.

A simple CLI option would be something like this. Rename the architectures Rnn and AttentiveRnn and modify the architecture options to reflect the change. By default, make the RNN type default to LSTM, but add a flag (--rnn_type? just --rnn) to select a GRU instead.

Alternatively, but only if necessary, one could project this choice down into the architecture level so that one can choose between an AttentiveLstm and an AttentiveGru etc.

I am assigning this to @bonham79 since I know it's an interest of his.