Closed tacchinotacchi closed 4 years ago
I'm writing a seq2seq model, so I modified the pad_input function to pad both the input and target sequence by calling pad_input(datasource[slice], pad_fields=[0, 1]). Feel free to merge if you think it's useful :)
pad_input
pad_input(datasource[slice], pad_fields=[0, 1])
Check out this pull request on
You'll be able to see Jupyter notebook diff and discuss changes. Powered by ReviewNB.
Thanks. A few things can be refactored using L, I'll had that in a follow-up commit.
L
I'm writing a seq2seq model, so I modified the
pad_input
function to pad both the input and target sequence by callingpad_input(datasource[slice], pad_fields=[0, 1])
. Feel free to merge if you think it's useful :)