\(*.*)/HELLO \(*_*)/

[Genesis LoRa] ***Favors Figures and Faces***

This is my initial LoRa model, trained on custom-made images to produce abstract outputs. However, the dataset images vary significantly in style and quality, and the tagging was relatively minimal. As a result, some generations may display inconsistencies.

I’m currently developing an improved version, where I’ll use the highest-quality images produced by this LoRa as part of the training set, along with additional high-quality images. This iterative process—using one LoRa’s strengths to train the next—will allow for a model that builds upon the best features of its predecessors.