Can we stop doing hardware cost to token generation calculations on the timeline please? If you haven't noticed, models keep getting better & more efficient, and hardware prices keep going up
Stop hardware cost to token calculations, models improve, prices rise
By
–