Ariel Rebel Ti版本v1.0 (ID: 426495)

Ariel Rebel is known for many things: being an adult model, actress (The Fight machine [2022]), food blogger (!), photographer and entrepreneur, amongst other achievements. Born in 1985, she's got a quirky smile and a positive attitude; some of her food blog content may be found here and here.

This embedding was trained on approximately 60 publicly-available unique preview images (some were cropped for detailed faces; all photos circa 2014ish) for 60 epochs. With repeats, there were a total of 90 images processed per epoch with a batch size of 10 and 9 gradient accumulation steps (so, 540 total steps) at a constant learning rate of 0.003. The optimizer used was ADAMW. All images were masked for attention, using the amazing OneTrainer.

While quite tasteful, the source dataset is decidedly NSFW; appropriate negative prompts may be necessary to prevent risqué outputs, depending on the model used. Please be respectful in your prompting.

I'd love to see outputs from this Ti; happy prompting!

SHA256 hash: 68A04469A7F031DD46DB57F6AB249914793ED528800749A4E52072E78D32407F

描述:

Version 1.0

训练词语: arielRebelTi