Will there be GPU support for competitive models?

[lejinghu]

I know this is a probably a difficult task. But since we are switching to Tensorflow, maybe we could add something like https://github.com/cgorman/tensorflow-som in the future?

[itdxer]

Hi, I still have a lot of things to add and modify that has higher priority than moving SOM to tensorflow. I don't plan it any time soon.

[lejinghu]

Thanks. I am currently using it for data with millions of rows, but feel it's a bit slow to train. If I manage to find some ways to add GPU to Neupy implementations of SOM and GNG I will let you know.

[aserg24]

Hi, I see that you have added the ability to train on GPU: http://neupy.com/docs/algorithms/train-on-gpu.html I'm sorry, but I don't understand, is there such a possibility for GNG?

[itdxer]

@aserg24 You cannot train GNG on GPU. All algorithms with fixed architecture, except RBM, cannot be trained on GPU, since they were implemented using numpy library. I'll need to update documentation in order to make it explicit

[aserg24]

Thank you, got it, appreciate the info.