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/latina_krea2")
prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]Latina — Krea 2 LoRA
An identity/character LoRA for krea/Krea-2-Turbo, trained with Ostris AI Toolkit.
Suggested trigger word
Start prompts with latina. This is inferred from the repository and embedded output name; the original captions are not included, so the exact training token could not be independently verified.
Example prompt:
Photorealistic portrait of latina, natural daylight, realistic skin texture, detailed eyes, soft background blur.
Training metadata
| Parameter | Value |
|---|---|
| Base family | Krea 2 |
| Checkpoint step | 500 |
| Recorded epoch | 15 |
| Trainer | ostris/ai-toolkit 0.12.0 |
| Tensor count | 512 |
File
latina_krea-2_lora_v1_000000500.safetensors— 500-step Krea 2 LoRA checkpoint
Recommended settings
- Start around LoRA strength
0.7–0.9and adjust gradually. - Use a compatible Krea 2 workflow/loader.
- Describe pose, clothing, framing, lighting, and setting explicitly.
Limitations and responsible use
This is an early 500-step checkpoint; identity strength and flexibility may differ from a longer training run. Dataset details and learning rate are not included. Use only with necessary rights and consent. Do not impersonate, deceive, stereotype, harass, exploit, or violate privacy or publicity rights. Disclose synthetic imagery when appropriate.
- Downloads last month
- 54