I had an error while importing my GLB file from 32-bit precision floating point, I guess this was forgotten while implementing 64-bit precision floating point. I'm not sure if there's any other left to do though.