Matricize is a convenience library for machine learning practitioners that helps convert between plain Python objects and tensors/matrices.
The library is meant to be used in scenarios where a performance trade-off can be made in favour of convenience. Given below are some scenarios where Matricize could be useful:
- Small datasets (on the order of hundreds or thousands of elements) with complex labels
- Data sets with dynamically generated data (in Python)
- Python 3.6
- numpy