New features:
- Bugfixes!
- New mlcrate.ensemble submodule, containing rank_average(), which lets you ensemble sets of predictions from multiple models based on their ranking order
- New PyTorch submodule, which contains
tonp()andtotensor()for easy conversion between Tensor and NumPy types, as well as a one line MLP model constructor. - Apache feather support in
mlc.save()andmlc.load(), allowing for extremely fast save/load of dataframes. - Some other minor (backwards compatible) improvements throughout