search

maxjcohen / transformer

Implementation of Transformer model (originally from Attention is All You Need) applied to Time Series.
https://timeseriestransformer.readthedocs.io/en/latest/
GNU General Public License v3.0
842 stars 165 forks source link

Is this implementation autoregressive for a seq2seq task? #31

Closed shamoons closed 3 years ago

shamoons commented 3 years ago

I can't quite see where to pass inputs to the decoder. It seems to just copy the encoder inputs?

maxjcohen commented 3 years ago

Hi, I didn't plan to pass inputs directly to the decoder. Currently, it takes as input the latent vector, computed by the encoder. See these lines.