dpnp.tensordot returns wrong result

[antonwolfy]

The below example returns different result comparing to NumPy:

import dpnp, numpy

dpnp.__version__
# Out: '0.17.0dev4+9.g13816bdc7a3.dirty'

numpy.__version__
# Out: '1.26.4'

x1 = dpnp.array([[[94906267, 94906267], [94906267, 94906267]]], device='cpu')
x2 = dpnp.array([94906265, 94906265], device='cpu')
dpnp.tensordot(x1, x2, axes=((1,), (0,)))
# Out: array([[18014398652125512, 18014398652125512]])

x1 = x1.asnumpy()
x2 = x2.asnumpy()
numpy.tensordot(x1, x2, axes=((1,), (0,)))
# Out: array([[18014398652125510, 18014398652125510]])