A Practical Guide to Efficient AI: Shelby Heinecke

[@sammcj2000]

Wait - are companies actually happy running their models at full precision after creation?! That's an insanely inefficient waste of resources for no gain given modern (2022+) quantisation techniques.

[@savelist1]

They are no different you take any model and just give a good prompt you get the same result

[@savelist1]

None of them work

[@savelist1]

Einstein never works 😂