About this version trained on almost 2000 images from mspabooru, as a test of how a large dataset lora would work out. not very consistent but works with a lot of models/styles https://civitai.com/models/69595?modelVersionId=74262