This should run fine on most hardware - CPU inference of the E2B model on my Pixel 8 Pro gives me ~9tok/second of decode speed.