David Buchaca Prats

I do not see the motivation for removing the fit with `Onehotencoder`. As @CameronBieganek mentions sklearn has the option to `fit_transform` but it also has the option to first fit...

I happened to get exactly the same behaviour. If not using python 3.6 I got ``` ERROR: Could not find a version that satisfies the requirement torch==1.1.0 (from -r scripts/stack-transformer/requirements.txt...

I was using OSX. It seems fastBPE has several problems in OSX. Inside the conda environment I get the following error (which has nothing to do with this repo though)....

In the documentation it might be ilustrative to add some gifs: - https://github.com/vdumoulin/conv_arithmetic

Are you using `label_components!` or `label_components`? The first one allocates memory every call and will not be very good for real time stuff. ```julia @time zeros(int 1600,1200) 0.002046 seconds (12...

Since apple is deprecating OpenCL in mojave and it won't even be included in the next operating system, is there a way download software from intel/AMD so that we could...

What about sharing a docker image with OpenCL ?

I just updated OpenCL for julia 0.6 and `Pkg.test ` still fails some tests but the vector addition code form the main page works. ``` Test Summary: | Pass Fail...

Here I can put what I get an OSX with similar specs ``` julia> versioninfo() Julia Version 1.1.0 Commit 80516ca202 (2019-01-21 21:24 UTC) Platform Info: OS: macOS (x86_64-apple-darwin14.5.0) CPU: Intel(R)...

What are you trying to do? The following works fine ```julia CLArray{Float32,2}(rand(5,4)) ``` As well as ```julia CLArray{Float32,2}((10,10)) ``` I agree though, for "consitency" if both of the following ways...