create numpy array of dtype object

[levongh]

Let's say I want to keep a list into an array and avoid implicit type conversion. Is there an built-in functionality to achieve this result

pybind11::list some_list;
pybind11::array arr(some_list)

this flow returns me numpy array of numpy array

>>> type(arr)
<class 'numpy.ndarray'>
>>> type(arr[0])
<class 'numpy.ndarray'>
>>> arr.shape
(1, 3)
>>>

but instead I want to get pybind array with shape (1,)