Daniil Pakhomov
Daniil Pakhomov
I think I solved it. if someone faces the same problem, look here: https://github.com/amueller/damascene-python-and-matlab-bindings/issues/4
For everyone who is still interested, we have started a project which tries to replicate digits for pytorch: https://github.com/warmspringwinds/dash-deep
Hi @hyer , Did you install our fork of ATen ? And please provide more info about your system: ubuntu?, cuda 8? and so on 😃
@hyer Thank you. Could you try running this in bash ```pip install pyyaml``` ?
@hyer and install and build from scratch
@pharrellyhy Dropout2d is not yet implemented but it's easy to do it though :)
Hi @RaceSu -- if you have cuda installed on your computer, just follow the steps on the front page of our library, it's easy.
You can run it without gpu -- it will be rather slow but it will work :)
@kirill-pinigin Providing the ```.data()``` output of a vector to a ```tensorFromBlob()``` function should work. Not really relevant but close example: https://github.com/warmspringwinds/pytorch-cpp/blob/master/examples/read_allocated_gpu_memory.cpp#L64
@ahkarami Thank you. The models that we are using don't use either dropout or 3d convolution unfortunately so it's not a direct priority. Although it's quite easy to do for...