Thank you
Thank you for creating this model! So much thanks.
Unfortuantely there are a couple of problems.
If you run it alone with no additional imput (just "ann3tt3schw4rz woman") it gives results with some modern (https://files.catbox.moe/s2nfow.png and https://files.catbox.moe/qfgfeg.png) it is strange, like you have some additional folder by msitake included in the training, or soem strange tags. I used the realistc Vision 5.1
For the rest it can output the actress but it is not very flexible, imho it is dataset that was poorly tagged but I am no expert.
Sorry for the harsh critique, It is just for you to improve, i really appreciate this model.