|
|
--- |
|
|
pipeline_tag: text-to-image |
|
|
license: other |
|
|
license_name: faipl-1.0-sd |
|
|
license_link: LICENSE |
|
|
prior: |
|
|
- Disty0/sotediffusion-wuerstchen3-alpha3 |
|
|
--- |
|
|
|
|
|
# SoteDiffusion Wuerstchen3 |
|
|
|
|
|
Anime finetune of Würstchen V3. |
|
|
Currently in active training. |
|
|
No commercial use. |
|
|
|
|
|
# Usage |
|
|
Please refer to the main model: https://huggingface.co/Disty0/sotediffusion-wuerstchen3-alpha3 |
|
|
|
|
|
|
|
|
## Dataset |
|
|
|
|
|
Used the images that got **extremely aesthetic, best quality** or **very aesthetic, best quality** from the **newest** dataset. |
|
|
Trained with **120K~** images. |
|
|
|
|
|
## Training: |
|
|
|
|
|
**GPU used for training**: 1x AMD RX 7900 XTX 24GB |
|
|
|
|
|
**Software used**: https://github.com/2kpr/StableCascade |
|
|
|
|
|
### Config: |
|
|
``` |
|
|
experiment_id: sotediffusion-wr3_3b-stage_b-alpha3 |
|
|
model_version: 3B |
|
|
dtype: bfloat16 |
|
|
use_fsdp: False |
|
|
|
|
|
batch_size: 16 |
|
|
grad_accum_steps: 16 |
|
|
updates: 7680 |
|
|
backup_every: 512 |
|
|
save_every: 256 |
|
|
warmup_updates: 128 |
|
|
|
|
|
lr: 1.0e-5 |
|
|
optimizer_type: Adafactor |
|
|
adaptive_loss_weight: False |
|
|
stochastic_rounding: True |
|
|
|
|
|
image_size: 768 |
|
|
multi_aspect_ratio: [1/1, 1/2, 1/3, 2/3, 3/4, 1/5, 2/5, 3/5, 4/5, 1/6, 5/6, 9/16] |
|
|
shift: 4 |
|
|
|
|
|
checkpoint_path: /mnt/DataSSD/AI/SoteDiffusion/Wuerstchen3/ |
|
|
output_path: /mnt/DataSSD/AI/SoteDiffusion/Wuerstchen3/ |
|
|
webdataset_path: file:/mnt/DataSSD/AI/anime_image_dataset/best/newest_best.tar |
|
|
|
|
|
effnet_checkpoint_path: /mnt/DataSSD/AI/models/wuerstchen3/effnet_encoder.safetensors |
|
|
stage_a_checkpoint_path: /mnt/DataSSD/AI/models/wuerstchen3/stage_a.safetensors |
|
|
generator_checkpoint_path: /mnt/DataSSD/AI/SoteDiffusion/Wuerstchen3/generator-045056.safetensors |
|
|
``` |
|
|
|
|
|
|
|
|
## Limitations and Bias |
|
|
|
|
|
### Bias |
|
|
|
|
|
- This model is intended for anime illustrations. |
|
|
Realistic capabilites are not tested at all. |
|
|
|
|
|
### Limitations |
|
|
- Far shot eyes are can bad. |
|
|
- Still in active training. |
|
|
|
|
|
|
|
|
## License |
|
|
|
|
|
SoteDiffusion models falls under [Fair AI Public License 1.0-SD](https://freedevproject.org/faipl-1.0-sd/) license, which is compatible with Stable Diffusion models’ license. Key points: |
|
|
|
|
|
1. **Modification Sharing:** If you modify SoteDiffusion models, you must share both your changes and the original license. |
|
|
2. **Source Code Accessibility:** If your modified version is network-accessible, provide a way (like a download link) for others to get the source code. This applies to derived models too. |
|
|
3. **Distribution Terms:** Any distribution must be under this license or another with similar rules. |
|
|
4. **Compliance:** Non-compliance must be fixed within 30 days to avoid license termination, emphasizing transparency and adherence to open-source values. |
|
|
|
|
|
**Notes**: Anything not covered by Fair AI license is inherited from Stability AI Non-Commercial license which is named as LICENSE_INHERIT. Meaning, still no commercial use of any kind. |
|
|
|
