Comment by barrkel
And if dispersion isn't good, it would be worthwhile running the vectors through another model trained to disperse them.
And if dispersion isn't good, it would be worthwhile running the vectors through another model trained to disperse them.