I look forward to your pull-request. TinyNdArray supports only float array. In the following Python code dtype=float32 is omitted, and in C++ code assuming using namespace tinyndarray; is declared.

Problem is in using np.ascontiguousarray, which is buggy for zero-dimensional arrays (see numpy/numpy#5300 for details). Here I use the solution proposed by numpy team: switch to asarray with order='C ...