Anya Taylor Joy: v1.1 double data edition: anyataylorjoy.pt

30 images. 10 v/t. i've been experimenting with v/t and dataset size. the TIs come out nicely if you do 1 vector for every 3 images, so if you have 15 images, you use 5 v/t, then run for 200 epochs, roughly. anyway... i'm making notes here in case i forget

Anya is an actress.