Return value of pyfunc_0 is double, but expects float
问题 I am currently trying to better understand Tensorflows CustomLayer feature. While implementing such a custom layer, I ran into the following error: /usr/lib/python3/dist-packages/skimage/util/dtype.py:110: UserWarning: Possible precision loss when converting from float64 to uint16 "%s to %s" % (dtypeobj_in, dtypeobj)) /usr/lib/python3/dist-packages/skimage/exposure/exposure.py:307: RuntimeWarning: invalid value encountered in true_divide image = (image - imin) / float(imax - imin) Traceback