Upload 2326122/2616688/README.md with huggingface_hub

2326122/2616688/README.md

@@ -2,9 +2,88 @@
 license: other
 tags:
 - civitai
+- concept
+- blowjob
 ---
+
+# [flux2-klein-9b] Pyro's BJ - 2616688
+
+**Model Type**: LORA
+
+**Base Model**: Flux.2 Klein 9B
+
+**Trigger Words**: pov, licking, cheeking, swallowing, sucking
+
+**Tags**: concept, blowjob
+
+## Description
+
+[flux2-klein-9b] Pyro's BJ
+==========================
+
+New Model, New Pyro
+
+This is an experimental Lora
+
+**Research:** How few ranks can you get away with?
+
+**Result:** 90MB lora that understands more about anatomy than your biology teacher in 5th grade.
+
+Also, it's literally the most trainable model I've ever trained with. It's so good I'm going to train it with 100k z-image-turbo images to make my own z-image-base because Alibooba is seemingly toying with our emotions. (If you want to sponsor this project shoot me a msg on discord)
+
+How to use?
+-----------
+
+**You can use this model with BASE and DISTILL.** You can also edit images with it amazingly well.
+
+Use BASE for better prompt adherence
+
+Use DISTILL for images faster than you can jerk off
+
+**Pls try my distill workflow attached as TrainingData zip-File before telling me or the world that the distill doesn't work well. Thx.**
+
+Demo Images should contain workflow
+
+In theory the LORA understands following concepts (in practice it's a bit shaky but that's why it's experimental, you'll figure it out):
+
+**General:** a woman sucking a penis
+
+**Chipmunking or whatever this shit is called:** a woman cheeking a penis
+
+**Deepthroat:** a woman swallowing a penis
+
+**What could this be:** a woman licking a penis
+
+**If you want her to look at you in disgust:** pov
+
+Will experiment further until I uncovered the mysteries of flux2-klein-9b and prompt adherence is at a point where you trust the LORA more than your mom.
+
+Discord: <https://discord.gg/PSa9EMEMCe>
+
+Technicals:
+-----------
+
+Framework: my diffusion-pipe fork: <https://github.com/pyros-projects/diffusion-pipe>
+
+Dataset: 300imgs split evenly over all sub-concepts
+
+Training:
+
+65 epochs with 512p (AdamW - batch 4 - lr 2e-4)
+
+15 epochs with 768p (Prodigy Schedulefree)
+
+Version History:
+----------------
+
+bj\_20260120\_22-22-29epoch15\_comfy: Initial Release
+
+---
+
 Author: [pyros_sd_models](https://civitai.com/user/pyros_sd_models)
 
-Model: [https://civitai.com/models/2326122?modelVersionId=2616688](https://civitai.com/models/2326122?modelVersionId=2616688)
+Model: [CivitAI Model Page](https://civitai.com/models/2326122?modelVersionId=2616688)
+
+Archive: [CivArchive Page](https://civarchive.com/models/2326122?modelVersionId=2616688)
 
-Mirror: [https://civarchive.com/models/2326122?modelVersionId=2616688](https://civarchive.com/models/2326122?modelVersionId=2616688)
+<!-- Version: 20260422_v4_update -->