Repository navigation
dpnp.unique with equal_nan=True and axis specified gives incorrect output for an input with NaNs #2325
Description
More or less identical to NumPy issue: numpy/numpy#23286
import dpnp
a = dpnp.array([0., np.nan, 2., np.nan, 2., 1., 0., 1., 2., 0.])
print(dpnp.unique(a, equal_nan=True, axis=0))
# [ 0. 1. 2. nan nan]Expected result is [ 0. 1. 2. nan]
Activity