Hey @wilsonrljr, just to let you know, we've been working on exposing a generic interface to functions like np.testing.assert_allclose in array-api-extra under the xpx.testing namespace. This should eliminate the need to call _to_numpy in places like this:
|
assert_array_equal(_to_numpy(result), np.array([[2.0], [4.0], [6.0]])) |
and give you extra checks 'for free' on top of np.testing, e.g. that the namespaces of the input arrays match.
We're currently checking that what we have works for SciPy and scikit-learn in scipy/scipy#25143 and scikit-learn/scikit-learn#34019. Once we're happy with that, we'll expose the public API. Would be great to have you as an early adopter, in case there are any rough edges that need smoothing out!
Hey @wilsonrljr, just to let you know, we've been working on exposing a generic interface to functions like
np.testing.assert_allclosein array-api-extra under thexpx.testingnamespace. This should eliminate the need to call_to_numpyin places like this:sysidentpy/sysidentpy/tests/test_narmax_base.py
Line 900 in 5877e8e
and give you extra checks 'for free' on top of
np.testing, e.g. that the namespaces of the input arrays match.We're currently checking that what we have works for SciPy and scikit-learn in scipy/scipy#25143 and scikit-learn/scikit-learn#34019. Once we're happy with that, we'll expose the public API. Would be great to have you as an early adopter, in case there are any rough edges that need smoothing out!