n?zo??ti??c!.&fni?gro?moc?ten?ude?vog??i??d&e!.tir.segap-tig,?iab??e!.&dcym,enozgniebllew,noitatsksid,odagod.citsalej,s&nd&ps,uolc,?ppatikria,?ysrab,??g!.&bew?gro?m&aug?oc??ofni?ten?ude?vog???h!.&0002?a&citore?idem?kitore??edszot?gro?ilus?letoh?m&alker?lif?t?urof??naltagni?o&c?ediv?fni?levynok?nisac??pohs?rarga?s&a&kal?zatu??emag?wen??t&lob?opsgolb,rops??virp?xe&s?zs??ytic?zsagoj??os?sut??l!.&etisbew321,topsgolb,??m!.&ca?gro?moc?oc?ro?ten?vog???n!.&duolcesirpretne,eni&esrem,m,?tenkcahs,?em!.ysra...

to `tf.setdiff1d` to match NumPy naming. * `tf.inv` has been renamed to be
`tf.reciprocal` (component-wise reciprocal) to avoid confusion with `np.inv`
which is matrix inversion * tf.round now uses banker's rounding (round to even)
semantics to match NumPy. * `tf.split` now takes arguments in a reversed order
and with different keywords. In particular, we now match NumPy order as