Need teammate for Pytorch Hackathon, creating Sparse Library

I have an idea for a sparse training library for pytorch.

A lot of the work will involve speed/performance optimization, particularly parallelizing IO memmap operations so that it doesn’t slow down training.

The library will then be used to train embeddings for ~40 million research papers.

If you haven’t heard of the Facebook Pytorch Hackathon, Google ‘Facebook Pytorch Hackathon’