These humps and kinks correspond exactly to what I see in the gm/ID plot, since gm/ID is just the derivative of log(current) wrt. VGS. This is supposed to be relatively constant in weak inversion until you hit GIDL, no wild up and down swings as in the PMOS plot above.