See also the<a rel="ugc" href="https://civarchive.com/models/19594/hrbkbc1sgv"> LoRa</a>====A little textual inversion trained on pictures of Halle Berry wearing a fancy bikini and huge sunglasses.

HRBKBC1SG

I retrained a whole embedding with the same pictures, but the result was efficient enough. I then decided to merge it with the first version through the <a rel="ugc" href="https://github.com/tkalayci71/embedding-inspector">Embediding Inspector extension</a> with ("v/2" as eval expression).

Files with SHA256: ea4a5cf0fcc3ff22abe741cb1b878eba42bda7bf658961216004b05b397f9078