Type | |
Stats | 215 0 |
Reviews | (20) |
Published | Mar 20, 2024 |
Base Model | |
Training | Steps: 2,775 Epochs: 111 |
Trigger Words | KellyBrookTi |
Hash | AutoV2 CD45EEBC22 |
Kelly Brook (Kelly Ann Parsons) is a rather stunning human being. Born in 1979, this actress and model has what may be the brightest, most heart-warming smile ever to grace the planet (subjective opinion, definitely). Scientists at Texas University have apparently determined her to have the perfect body shape (objective opinion, ostensibly); can't say I have many quibbles with that conclusion.
This 16-vector embedding was trained using the ADAMW scheduler with a constant learning rate (LR 0.00075, 350-step warmup); I configured a 12-image batch size with 25 steps per epoch which ran for 111 epochs (2775 steps). The training data consisted of 60 body photos (1x repeat; attention masked) and 80 cropped face closeups (3x repeat; attention masked), meaning each epoch ran through 300 images. The training dataset included a ton of pinup photos (from early-mid 2010s), so the outputs may trend risqué; I'd strongly recommend taking that into account with your negative prompts if you want to prevent NSFW generations. Training software utilized is the awesome OneTrainer (found here: https://github.com/Nerogar/OneTrainer).
Sample images were generated utilizing Highres fix. and Adetailer (face); no other embeddings or LoRa were used. Generation prompts are accurate - seeds may differ a bit from mine as I'm using a TRT/ONNX conversion of the base models to speed processing.