A big part of NumPy’s speed comes from using machine-native datatypes, instead of Python’s object types. But the other big reason NumPy is fast is because it provides ways to work with arrays ...

0.0.14 - Roll in a number of fixes from Github contributors, including fixes for oneOf handling, array validation, and Python 3 support. 0.0.13 - Lazily build object classes. Allows low-overhead use ...