Re: [theano-users] About parallelize theano.scan or any other way

The only option with Theano would be to write your own op that would run in parallel on the GPU. This is not an easy task. > Le 10 déc. 2018 à 07:35, Wong Hang a écrit : > > Hi, > > I am working on a problem that is similar to this one: > > import theano > import theano.tensor as T > import

Re: [theano-users] development status of typed_list or alternative

It will not be developed any further unless someone steps up and does it. > Le 12 nov. 2018 à 08:58, Timothe Boutelier a écrit : > > Hi, > Does anyone have some news about the development status of the typed_list > feature (here >

Re: [theano-users] ubuntu16.06 import theano using GPU:Exception: The nvidia driver version installed with this OS does not give good results for reduction

There is nothing really that can be done from the Theano side. Some versions of the nVidia driver have a compiler/runtime bug with our code which gives wrongs results in some cases we don’t have a list of which versions exactly. We put that check in place to detect that. Note that version

Re: [theano-users] Querying a single model on the GPU from multiple threads (GPU context initialization problem)

What you want to do is certainly not supported. You might be able to force a different context for each thread by using a different device name and manually registering contexts on the same GPU for those names. Then you would have to share the shared variables over IPC or some other sort of

Re: [theano-users] Pygpu Invalid Values / Options Error

This seems like you don’t have a version of CUDA that supports your GPU. The 2080 requires CUDA 10. What version is installed? > Le 3 nov. 2018 à 00:03, John Coolidge a écrit : > > I've had trouble getting Theano to work on my GPU and tracked down the error > to pygpu. When I run PyGPU's

Re: [theano-users] float64 gpu cholesky factorization and triangular solve

> > Le 18 oct. 2018 à 00:15, Wong Hang a écrit : > > Hi there, > > May I ask why theano gpuarray does not support float64 cholesky factorization > and triangular solve? > Are there any reason behind that? The main reason is that there is little interest in float64 because it is *much*

Re: [theano-users] ERROR (theano.gpuarray): pygpu was configured but could not be imported or is too old (version 0.7 or higher required)

Did you install pygpu? > Le 15 oct. 2018 à 07:56, Александр Комаров a écrit : > > Hi. Try to use CUDA in Theano. > When I start the script i have an error > > "ERROR (theano.gpuarray): pygpu was configured but could not be imported or > is too old (version 0.7 or higher required)" > > > My

Re: [theano-users] Theano Compile error with Anaconda, OSX

You seem to have the same problem as this: https://github.com/Theano/Theano/issues/6645 Unfortunately as explained in that ticket the solution is rather involved and would require more time than we have available for now. > Le 26 sept. 2018 à

Re: [theano-users] use a saved trained model to predict on my dataset

t find any better way to predict with new > dataset on a saved model in python. > > On Tue, Aug 28, 2018 at 2:13 PM Arnaud Bergeron <mailto:aberge...@gmail.com>> wrote: > Why are you trying to do this? > >> from MiniBatchIteratorConstantBatchSize import dataset &

Re: [theano-users] use a saved trained model to predict on my dataset

Why are you trying to do this? > from MiniBatchIteratorConstantBatchSize import dataset This cannot work with the code you posted since MiniBatchIteratorConstantBatchSize is not a module. That has nothing to to with Theano, but rather is due to how python works. Arnaud -- --- You

Re: [theano-users] ImportError: ('The following error happened while compiling the node' .so: file too short)

Try to clear the Theano cache (theano-cache purge) and run the code again. > Le 15 août 2018 à 10:43, Chankey Pathak a écrit : > > Getting below error while running the example code of nn-semparse > . It looks like it's an issue > with Theano. Can

Re: [theano-users] Theano >= 1.0.1: Cannot not initialize GPU if already initialized in parent process

Even if you use a different device, forking a process after the GPU has been initialized leads to all sorts of strange behaviour if you're not lucky. This limitation comes from CUDA so it's not something we can fix. We've added this check to make sure that people don't do this and from the

[theano-users] Announcing Theano 1.0.2

contributors of Theano. Since release 1.0.1, many people have helped, notably (in alphabetical order): - Arnaud Bergeron - Desiree Vogt-Lee - Frederic Bastien - Garming Sam - Glexin - Jon Haygood - Jordan Melendez - Pascal Lamblin - Simon Lefrancois - Steven Bocco - Vincent Dumoulin Also

Re: [theano-users] Could not initialize pygpu error with CUDA 9.0 and latest Theano

This is a new one. It is also very weird since gemm doesn't involve cuLinkAddData. This may be an error message from something else. First things first, since you are on cuda 9.0, I would recommend that you update your driver to 384.111 or 390.*. If that doesn't help, then I'll need some

Re: [theano-users] cannot compile with cuDNN error when importing theano

remove Anaconda and install > at C root? > > Thanks! > > On Monday, April 16, 2018 at 12:46:02 PM UTC-5, Arnaud Bergeron wrote: > Create a new file and call it whatever (like test.c), then try to compile it. > This is just to check if the compiler works outside of Theano. >

Re: [theano-users] cannot compile with cuDNN error when importing theano

> Hi, Arnaud, > > Thanks for your reply! I believe currently I rely on the gcc from > m2w64-toolchain package, installed via conda. > Where can I find this file "test.c", and how can I compile with your code? > > Thanks! > > On Wednesday, April 11, 2018 at 12:

Re: [theano-users] cannot compile with cuDNN error when importing theano

> Le 3 avr. 2018 à 12:01, Qinpeng Wang a écrit : > > Hi Pascal, > > I have CUDA installed a while ago when I was still using the old GPU backend > at C:\Program Files\NVIDIA GPU Computing Toolkit, then I copied the CUDA > folder to C root. This most likely broke DLL

Re: [theano-users] Error using floatX = float16 to save memory

Actually I believe that your code is using conv2d3d in order to get it working with 3d convolutions on the GPU. This is now directly supported by cudnn if you use dnn_conv() with 5d objects from the new backend. There is some work around an abstract conv3d interface which will hopefully be

Re: [theano-users] Unable to install libgpuarray

Why are you using "sudo python". libgpuarray and/or pygpu do not require root priviledges and shouldn't be run as root. This is why it can't find the library in the first case (since sudo remove LD_LIBRARY_PATH from the environment). For the second problem, "Unknown error" during context