search

vicgalle / stable-diffusion-aesthetic-gradients

Personalization for Stable Diffusion via Aesthetic Gradients 🎨
https://arxiv.org/abs/2209.12330
Other
716 stars 65 forks source link

--aesthetic_lr through txt2img.py is expecting int, should be float #4

Closed dmalyavin closed 1 year ago

dmalyavin commented 1 year ago

Sorry dont know how to do pull requests.

    parser.add_argument(
        "--aesthetic_lr",
        type=int,
        help="learning rate for the aesthetic personalization",
        default=0.00001,
    )

should be

    parser.add_argument(
        "--aesthetic_lr",
        type=float,
        help="learning rate for the aesthetic personalization",
        default=0.00001,
    )
vicgalle commented 1 year ago

Thank you very much for reporting it, @dmalyavin !

I've just fixed it in main.