Note: we have started the gathering of a wide range of datasets as promised, it is planned to have close to 1 million training rows. We will be gradually releasing each dataset as we finish processing them. The full dataset is planned to be released at the end of this year.