import numpy as np import pytensor.scalar as ps x = ps.float64("x") y = x.astype("float32") y.eval({x: np.int64(0.0)}, mode="NUMBA").dtype # AttributeError: 'float' object has no attribute 'dtype'