Syroh Style
Greetings everyone, I am excited to share with you my first LoRA (Low Rank Adaptation) project. For this endeavor, I utilized a set of 400 images created by the talented artist Syroh 🌟大阪こみトレ - pixiv (Heavy NSFW Warning) as the basis for training the LoRA model.
Usage
I highly recommend utilizing the LoRA algorithm with weight values ranging from 1.0 to 0.8.
This particular model demonstrated compatibility with several other LoRA-friendly models. Therefore, you may exercise discretion when selecting the appropriate model for your needs.
I have conducted thorough testing on Amedira and other more commonly used models.
It is worth mentioning that this LoRA model exhibits a predisposition towards catgirls. This can be attributed to Syroh's strong inclination for drawing catgirls, resulting in a dataset that primarily comprises such images. Regardless, I trust you will find my LoRA model satisfactory. Please feel free to share any feedback or raise any queries you may have.