Data for Experiments

The evaluation scripts require pointers to data. Is the exact data available anywhere (or will be made available anytime soon)? I was specifically looking to recreate data extraction results. While the preprint does mention that the MATH data for instance (for which extraction results were reported primarily) is a combination of Dolmino Math Mix and FineMath4++, I am not sure if it is purely a concatenation of the two or some sample.