Vilen Jumutc
Vilen Jumutc
Bug randomly reproduced in version `1.0.3` as well
Hi @muellerdo, The dataset is 2D images and the pipeline looks like this: ``` # Create a pixel value normalization Subfunction for z-score scaling sf_zscore = Normalization(mode="z-score") # Create a...
Hi @muellerdo, With `batch_size` 1 and 3 everything works and also sometimes with 2. And this is very strange and wonky for reproducibility bug. Cannot add anything more to it!
I have diagnosed the issue: `use_multiprocessing=True` flag is the cause
I have some basic implementation for my project so if needed I can create a pull request.
Scala requirement for Saddle is not crucial because you can use it as a normal java library as well (just adding a dependency in gradle... which adds a transient one...
I have quickly explored Saddle and in my opinion it is indeed hard to use from Java without proper internal API from the Saddle team. Do we aim at some...
What I can infer immediately from the Github and source code of these libraries: - Paleo is very immature and lacks a lot of functionality which is nice to have....
If anyone is interested I have a simple implementation of AUC based on `ROCNum` array which I embedded in my own project: [SALSA.jl](https://github.com/jumutc/SALSA.jl/blob/master/src/SALSA.jl). I can introduce it very quickly to...
Probably we should wait for some updates in `PyCall.jl` package. I will try to figure out a nice example to report this issue in https://github.com/stevengj/PyCall.jl