I am new to deep speech. I read the article by Reuben on streaming RNN, but could not find the code (master branch) for training unidirectional RNN. DeepSpeech.py has a function for defining BiRNN but not for unirnn. AM I missing something ?
pankaj.pailwar (Pankaj Pailwar) #1
reuben (Reuben Morais) #2
That’s just the misleading name of the function (that we should have changed!), it actually implements a unidirectional LSTM.