Instructions to use furaidosu/dottrmstr-krea2t with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Inference
- Notebooks
- Google Colab
- Kaggle
dottrmstr-krea2t
LoRA para Krea-2-Turbo, entrenado originalmente en ModelScope (tostiok/dottrmstr-krea2t).
Se incluyen dos checkpoints del entrenamiento:
dottrmstr-krea2t_c1-st3000.safetensors(step 3000)dottrmstr-krea2t_c1-st6000.safetensors(step 6000, final)
Uso con diffusers
from diffusers import Krea2Pipeline
import torch
pipe = Krea2Pipeline.from_pretrained("krea/Krea-2-Turbo", torch_dtype=torch.bfloat16)
pipe.load_lora_weights("furaidosu/dottrmstr-krea2t", weight_name="dottrmstr-krea2t_c1-st6000.safetensors")
pipe.to("cuda")
image = pipe("tu prompt acá", num_inference_steps=8, guidance_scale=0.0).images[0]