Request a lighter model

Thank you very much for your related work. I am more concerned about the SegEverything efficiency of version v2. I saw that the official provides relevant checkpoint files, which include efficientvit-l2. In the code, models of size l0 and l1 can be selected. Can these two sizes of models be provided? Or if there is something provided that I haven't noticed, please provide some tips.