Instructions to use hodgy/katherine with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use hodgy/katherine with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("krea/Krea-2-Turbo", dtype=torch.bfloat16, device_map="cuda") pipe.load_lora_weights("hodgy/katherine") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
| license: other | |
| base_model: krea/Krea-2-Turbo | |
| library_name: diffusers | |
| pipeline_tag: text-to-image | |
| tags: | |
| - krea-2 | |
| - krea-2-turbo | |
| - lora | |
| - text-to-image | |
| - ai-toolkit | |
| - identity | |
| - character | |
| # Katherinne — Krea 2 Turbo LoRA | |
| An identity/character LoRA for [krea/Krea-2-Turbo](https://huggingface.co/krea/Krea-2-Turbo), trained with Ostris AI Toolkit. | |
| ## Trigger word | |
| Use **`katherinne`** near the beginning of the prompt. This spelling matches the embedded model name and filename. | |
| Example prompt: | |
| > Photorealistic portrait of katherinne, natural daylight, realistic skin texture, detailed eyes, shallow depth of field. | |
| ## Training metadata | |
| | Parameter | Value | | |
| |---|---| | |
| | Base family | Krea 2 | | |
| | Training steps | `1000` | | |
| | Recorded epoch | `8` | | |
| | Trainer | `ostris/ai-toolkit` `0.11.1` | | |
| | Tensor count | `512` | | |
| ## File | |
| - `katherinne_Krea-2-Turbo_lora.safetensors` | |
| The weights use AI Toolkit-style Krea 2 LoRA tensor names. Use a compatible Krea 2 workflow/loader. | |
| ## Recommended settings | |
| - Start around LoRA strength `0.8`; adjust within approximately `0.6–1.0`. | |
| - Put `katherinne` once in the subject description. | |
| - Describe pose, clothing, framing, lighting, and environment explicitly. | |
| ## Limitations and responsible use | |
| The dataset and learning rate are not included in the repository. Results can vary with framing, occlusion, stylization, and strength. Use only with the necessary rights and permission; do not impersonate, deceive, harass, exploit, or violate privacy or publicity rights. Disclose synthetic imagery where viewers could mistake it for reality. | |