LoRA of Tomoe Yamanaka 山中知恵 (For FLUX)

Description

This LoRA trained using kohya_ss on 60 images 1024×1024 up to 12000 steps (40 epochs, 300 steps each) without captions and without regularization images.
Source model: ‘Source model: flux1-dev’. This LoRA would works well with flux1-dev and flux1-schnell. Also this LoRA should works well with most of community trained FLUX family checkpoints.
learning rate: 0.00005.
Precision: fp16.
LR Scheduler: constant.

Below is example of testing AIDA_FXLR_TomYam LoRA in Comfy_UI with flux1-dev:
Click on the image to view or download it in the original size.

Below is example of testing AIDA_FXLR_TomYam LoRA in Comfy_UI with flux1-schnell:
Click on the image to view or download it in the original size.

Below is download link for Comfy_UI .json config to reproduce this workflow:
      2025-03-10-184843_flux1-dev_103466258509064.json_.zip
I packed it into .zip because my site can’t host .json files.

Download link to AIDA_FXLR_TomYam.safetensors will be available for you after purchase. For your comfort I packed AIDA_FXLR_TomYam.safetensors into .zip archive with AIDA_FXLR_TomYam.png and AIDA_FXLR_TomYam.txt, preview and description are compatible with AUTOMATIC1111 extra networks menu.

Purchasing this asset you will achieve rights for selling images generated with AIDA_FXLR_TomYam.safetensors LoRA. If you want say special thanks to me, you can add additional amount to your payment.

Please contact me if you need some extended rights for sharing or selling AIDA_FXLR_TomYam, merging AIDA_FXLR_TomYam with other models, using AIDA_FXLR_TomYam in generation services and so on.