Difference in DeepSpeech document and DeepSpeech article

In the DeepSpeech documentation indicates:
" The fourth layer is a recurrent layer. This layer includes a set of hidden units with forward recurrence, h(f):"
But in DeepSpeech article:
“The fourth layer is a bi-directional
recurrent layer [38]. This layer includes two sets of hidden units: a set with forward recurrence,h(f), and a set with backward recurrence h(b)”
Has the document been edited to match this archive? Thank you

Bidirectionality was removed because of streaming inference.

1 Like

oh Thank you so much!! :3

can you explain to me what the -th(i-th,k-th) sign means?

For that one you’d need to ask the developers :slight_smile:

1 Like