Locon version of my <a target="_blank" rel="ugc" href="https://civarchive.com/models/11598/bozes-co-palesti-gate">Bozes Co Palesti V2 LORA</a> Pretty much re cleaned up the dataset and re tagged everything isolating a couple outfits.works fine with: &lt;lora:bozes_co_palestiV3:0.7&gt; bozes_co_palesti,I isolated only 2 outfits. Her third one is in the dataset but was too "meh" to bother. They are:<ul><li>Pink_nightgown_Purple_corset_See-through_skirt: Her nightgown.</li><li>Full_armor_Red_cape_Tiara: her armor.</li></ul>No dataset for you as it is mostly screencaps, you'll only get the tags summary.

Bozes Co Palesti (LOCON version) - Gate

bozes_co_palestiV3.safetensors

bozes_co_palestiV3_tags.zip

v1.0

Honestly I don't like gate, i can say i despise it a bit. But!Bozes is a top tier Waifu! The only worthy character in that series even!So here it is...V2: Another retrain of one of my earliest LORAsworks fine with: &lt;lora:bozes_co_palestiV2:0.7&gt; bozes_co_palesti,I isolated only 2 outfits. Her third one is in the dataset but was too "meh" to bother. They are:<ul><li>Pink_nightgown_Purple_corset_See-through_skirt: Her nightgown.</li><li>Full_armor_Red_cape_Tiara: her armor.</li></ul>No dataset for you as it is mostly screencaps, you'll only get the tags summary.V1:Had good-ish results with: &lt;lora:Bozes_Co_Palesti:.5&gt;, Bozes_Co_Palesti, big_breasts, 1girlLora weight from .5 to .8 seems to work best.I don't explicitly train my models to be NSFW but i won't remove anything from my training data that looks nice to me, SFW or NSFW.

Files with SHA256: a0668125cf7d5a80df95c3f0bbbb3641c29549be44d6ab0e5be2cff59478c8d4