Not all model get speed up with the new back-end. If there is no slow down, then it is good. We don't want some people to continue to use the old back-end, so we should not give them reason to do so. Being slower in corner case isn't great.
As you don't see slow down anymore in the new back-end, I think all is good. thanks Fred On Wed, Apr 12, 2017 at 7:00 AM Ozan Çağlayan <[email protected]> wrote: > Hi, > > I train for an epoch containing 30K samples with batches of 64 and I > divide the time spent to number of updates. So the first call to Theano > function should be smoothed out when we average, am I wrong? > > I re-did the test yesterday, the timings are pretty equivalent with > old/new backends but the new one is definitely not "faster". > > The code is at: > https://github.com/lium-lst/nmtpy/blob/master/nmtpy/layers.py > > You can search for theano.scan inside. Basically we have 2 gru_layer's for > source encoder and 1 gru_cond_layer in decoder. gru_cond_layer is actually > 2-GRU's intertwined with some complex interactions. > > > -- > > --- > You received this message because you are subscribed to the Google Groups > "theano-users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > For more options, visit https://groups.google.com/d/optout. > -- --- You received this message because you are subscribed to the Google Groups "theano-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
