2.0.2

This release should be a bit faster: conversions now use linear-algebra fromRecord and toRecord functions internally, which means no currying/partial application overhead. Thanks @w0rm for the suggestion!