So I was asking my self if a vision model like resnet or yolo can evaluate a chess position only by looking the board. Guest what, they can, my little experiment shows that is possible to feel the position by only looking, sometimes is wrong I know, but this is only a concept that owrks really well, more comming.
