What matters in reinforcement learning for tractography - Datasets

Datasets for "What matters in reinforcement learning for tractography". These can be used to train agents and reproduce the results in the article. 

The FiberCup, FiberCup Flipped and ISMRM2015 are "in-silico" datasets, whereas the TractoInferno subject is in-vivo.
The variations (_no, _raw) indicate the dataset used for the relevant experiment. The only difference in these variations lie in their configuration files and the .hdf5. 

The hcp_100206 dataset was used in a previous version of the article but is no longer relevant.