Hi, just wondering if there are plans to implement having variable output sequence lengths.
Something like having a third list as input at training time so
model.fit(x=data, y=targets, outputlengths=outputlengths)
and similarly for prediction.
Not sure how difficult this would be to implement given the your current stack, in the mean time I guess I'll just pad my target sequences to all be the same length.
Cheers!
Hi, just wondering if there are plans to implement having variable output sequence lengths. Something like having a third list as input at training time so
model.fit(x=data, y=targets, outputlengths=outputlengths)and similarly for prediction. Not sure how difficult this would be to implement given the your current stack, in the mean time I guess I'll just pad my target sequences to all be the same length. Cheers!