[Illustrious] Cutting T-shirt Challenge

Description

• Challenge that viral in X/Twitter

• Work in progress LoRA (maybe some edit dataset to produce better LoRA)

trigger word

cutting_challenge, like and retweet, number, white shirt

Limitations

• Number of favorite and retweet is not accourate

• Text stills not good

Training Details

LoRA rank

• LoRA standard

dataset

• 72 images

parameters

• resolution = 1024

• batch size = 2

• network dim,alpha = 12,12

• mix/save precision = bf16/bf16

• optmizer = Lion (weight_decay=0.05)

• UNet LR = 3e-5

• TE LR = 1e-5

• scheduler = cosine_with_restarts (3 restarts)

• l2 loss

• alpha_mask

steps

• epochs = 10

• total steps = 3600

• repeat = 10

tools

• kohya-ss GUI v24.3.0 FORKED by vjumpkung

• torch 2.5.0 cu124

• Runpod RTX 4090

avg weight

• UNet average weights : 0.009967144748026674

• TE1 average weights : 0.007946787285618484

• TE2 average weights : 0.006517791320220567

*This LoRA is for studying LoRA training with new technique so do not use for challenge in X/Twitter.

*If you see my model reposted on other image generation service please report to me.