metadata
license: unknown
Ai to Ai LoRA
Model Details
- Model Name: Ai2Ai_v1
- Description: a model trained on some images collected from pinterest and those images were ai generated as well
- Version: v1
- Date of Release: I forgot!
- Framework Used: Kohya trainer colab
Intended Use
This model is designed as a LoRA. It is recommended for generating anime girls. It is not suitable for full body and clothes.
Performance
- Accuracy: high on face and hat, low on body
- Precision: high on face and hat, low on body
Limitations and Bias
- Limitations: a small LoRA, generates identical faces
- Bias Considerations: jpt artifacts, cheek bulge, too powerfull
Training Data
- Data Sources: Pinterest
- Data Size: 82 images, around 12 Mb
Examples
Example 1:
Caption: [Description of the example]
Example 2:
Caption: [Description of the example]