Shareef Taylor

I think in near future we won't be able to use ai locally on our pc cause the models will be so big and use lots of energy. Or some one will find a way to make powerfully small models that use less data but are 10× better just like loras u only need about (correct me if I'm wrong) 15-20 Images to make a model. If we could get small models checkpoints like that that can build images based off of a few images just as good as flux1 this would be a game changer!! Ik someone is mart enough to turn loras Into checkpoints or make something called minitensors:> like basic lyrics u would train the model on tokens not images like say if u type bird it already knows what a bird is and looks like so when u put images to train the style, if u get what I'm saying:> so u would only need 1 image I 🤔