<h1 id="this-lora-is-trained-mainly-on-front-images">this lora is trained <span style="color:rgb(250, 82, 82)"><strong>mainly on front images</strong></span></h1><p>trigger: <span style="color:rgb(253, 126, 20)">chastity cage, padlock, testicles</span></p><p>extra: <span style="color:rgb(253, 126, 20)">heart-shaped lock</span></p><p>colors: <span style="color:rgb(253, 126, 20)">pink chastity cage</span>, <span style="color:rgb(253, 126, 20)">pink padlock</span>, etc.</p><h2 id="v1.7-(made-with-ai-toolkit)">v1.7 (made with ai toolkit)</h2><p>my chastity cage loras are all trained on civitai with kohya. for six months now, civitai added a kohya replacement called "ai toolkit". i wrote an article then saying it sucked <a target="_blank" rel="ugc" href="https://civarchive.com/articles/23119/ai-toolkit-training-sucks-lmao">https://civarchive.com/articles/23119/ai-toolkit-training-sucks-lmao</a> in beta. then later it became the default and i said it still sucked <a target="_blank" rel="ugc" href="https://civarchive.com/articles/26868/ai-toolkit-training-still-sucks-lmao">https://civarchive.com/articles/26868/ai-toolkit-training-still-sucks-lmao</a>. now KOHYA IS GETTING REMOVED AND THE AI TOOLKIT STILL SUCKS. the problems are literally the same problems it has had for 6 months:</p><ul><li><p>no tags in metadata</p></li><li><p>lora name is random in a1111</p></li><li><p>quality is worse</p></li><li><p>costs twice as much buzz for 20 epoches</p></li></ul><p>but the changelog <a target="_blank" rel="ugc" href="https://civarchive.com/changelog?id=246">https://civarchive.com/changelog?id=246</a> implies they improved it based on community feedback (idk whose, certainly not my feedback) and "it's now performing as well or better than Kohya in most cases". well, here is a lora trained with it. it cost the same amount of buzz as the same lora in kohya. it uses the same settings and the same dataset. the results are clearly not as detailed.</p><p>unless i figure out a way of making the ai toolkit work, i probably won't be able to make any more decent chastity cage loras. maybe it needs different settings. idk. but the impression i get is that it just sucks at learning detail in general. since every lora i train with it lacks detail where kohya would make the detail work just fine.</p><h2 id="v1.7-(more-real)">v1.7 (more real)</h2><p>trained on same dataset as v1.5 but with adjustments that should have improved quality. doesn't use anime upscaler so images will look more "realistic"</p><h2 id="v1.5-(more-anime)">v1.5 (more anime)</h2><p>not as good as v1.7 but doesn't make stuff realistic</p>

Files with SHA256: 0a897ad27b5a0a1a99297dadd0a6e9cfe0ba6ead6137ce5dab6061603108fedb