There is a way to install it with conda, bit that version have a bug that give good result. So you probably should wait for the RC next week that should have new conda package.
Fred Le ven. 10 févr. 2017 17:31, <gcilibra...@twmi.rr.com> a écrit : > I have a Fedora 25 operating system. I have upgraded theano to 0.9.1b1 in > the hope that I would pickup the latest gpuarray and pygpu backends. I > have recently been using keras to run CNN's for the Kaggle fish > competition. I have multiple gpu's in my machine that I run separately to > work on tuning hyperparameters. I have been using cuda0, cuda1, etc to > identify the gpu to run on each problem. I get warnings that pygpu cannot > be imported and the cuda backend is deprecated. I can still run my > problems, but!. So I want to get the new backend, but the latest theano > does not contain it. Trying to install gpuarray 0.2 is extremely cryptic > utilitizing cmake as described on their website. Is there a nice step by > step description of the process, or is the another format for the package > that is nice and simple to install. If this is the new backend for theano, > there is not much support for implementing it. > > -- > > --- > 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 theano-users+unsubscr...@googlegroups.com. > 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 theano-users+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.