flux-lora-models / README.md
CCRss's picture
Update README.md
294e3b0 verified
---
datasets:
- CCRss/flux_lora_datasets
language:
- en
base_model:
- black-forest-labs/FLUX.1-dev
pipeline_tag: text-to-image
---
### Loras trained with Flux-dev model.
Currently available:
Anime:
- [x] Akame (from akame ga kill)
In plans:
Educational:
- [ ] Handwritten on the russian texts
Anime:
- [ ] Nadeko Sengoku from Monogatari
## Akame
# Overview
<img src="https://huggingface.co/CCRss/flux-lora-models/resolve/main/anime/flux_dev_lora_akame_ver1/assets/Overview_1.png?raw=true">
Training Details:
- Base model: [Flux-dev](https://huggingface.co/black-forest-labs/FLUX.1-dev)
- Trainin scripts: [SD-Scripts (kohya-ss) SD3 branch](https://github.com/kohya-ss/sd-scripts/tree/sd3?tab=readme-ov-file)
- Dataset: 24 images of Akame from "akame_ga_kill" [dataset](https://huggingface.co/datasets/CCRss/flux_lora_datasets/tree/main/anime/akame_ga_kill/akame)
- Hardware: 2 H100 GPUs
- Duration: 4 hours
- Epochs: 1000
3 version of lora produced:
- 600 epoch checkpoint
- 800 epoch checkpoint
- 1000 epoch checkpoint
# Example 2
<img src="https://huggingface.co/CCRss/flux-lora-models/resolve/main/anime/flux_dev_lora_akame_ver1/assets/example_2.png?raw=true">
# Example 3
<img src="https://huggingface.co/CCRss/flux-lora-models/resolve/main/anime/flux_dev_lora_akame_ver1/assets/example_3.png?raw=true">