Training Configuration & Tools

Training Tools: Kohya_ss GUI / AI-Toolkit
Dataset: 20-50 images
Tagging Style: Early stages used Booru tags; later shifted to natural language descriptive captions
Training Res: 512x512
Repeats per Image: 50+
Training Parameters: Default settings from Kohya_ss / AI-Toolkit

Training Tools: Onetrainer
Dataset: At least 50 images (averaging 100+)
Tagging Style: Mixed format (Booru tags as prefixes combined with natural language)
Training Res: 512x512
Repeats per Image: 50+
Training Parameters: Default settings from Onetrainer

Usage Disclaimer

Educational & Research Purpose Only: This repository hosts character LoRA models trained on various base models, including Pony, Flux 1 Dev, Krea2, and SD1.5. All models are intended solely for academic exchange, technical testing, and personal research into image generation algorithms.
Prohibition of Illegal Use: Users must strictly adhere to local laws and regulations when downloading and using these models. It is strictly forbidden to use these models to infringe upon others' privacy, reputation, portrait rights, or other legitimate interests. Do not use them for creating or spreading misinformation, fraudulent activities, or any unethical purposes.
Content Compliance Notice: Please be aware that certain base models (e.g., Pony) may exhibit a high degree of openness regarding specific prompts. Users are entirely responsible for filtering and auditing the compliance of the generated content.
Limitation of Liability: As the creator/tester of these experimental models, I take no responsibility for any direct or indirect legal liabilities, losses, or disputes arising from the use of this repository. Any image generated and its subsequent consequences are the sole responsibility of the user.

Downloads last month: -; Downloads are not tracked for this model. How to track

Model tree for waveshaker/character_loras

Base model

Adapter

this model

waveshaker
/

character_loras