<h1 id="micro-bikini-3jkqclx8a">Micro Bikini</h1><h3 id="description-fta4bvoxe">description</h3><ul><li>all of sample image made it Flux.1 Dev GGUF Q4</li><li>recommended weight : 0.8 - 1</li></ul><h3 id="trigger-word-sih5k2nld">Trigger Word</h3><ul><li>[color] micro bikini</li></ul><h3 id="training-details-wkgjhn642">Training Details</h3>Dataset<ul><li>yellow micro bikini 10 images</li><li>red micro bikini 18 images</li><li>purple micro bikini 9 images</li><li>pink micro bikini 17 images</li><li>orange micro bikini 15 images</li><li>green micro bikini 21 images</li><li>blue micro bikini 16 images</li></ul>parameters<ul><li>resolution = 1024</li><li>batch size = 1</li><li>dim,alpha = 32,32</li><li>mix/save precision = bf16/bf16</li><li>optmizer = AdamW</li><li>UNet LR = 5e-5</li><li>TE LR = 2.5e-5</li><li>scheduler = cosine_with_min_lr (min_lr_ratio 0.2)</li><li>l2 loss</li></ul>steps<ul><li>epochs = 7</li><li>total steps = 2030</li><li>repeats = 4,2,5,2,3,2,3</li></ul>tools<ul><li>kohya-ss GUI v24.3.0 (Forked by me)</li><li>torch 2.5.0 cu124</li><li>Runpod RTX 4090 24 GB + xformers + gradient_checkpointing</li></ul>weight<ul><li>UNet average weights : 0.003128348355459278</li><li>TE1 average weights : 0.007201075088232756</li></ul>*This LoRA is for studying LoRA training with Flux*REPOST model is not allow if you found this model please report to me.

Files with SHA256: 1c1f5b5f4653c84acb5c43870891aa730da2278376481cfa6931d8e94b545445