Text-to-Image
Diffusers
Safetensors
English
BriaFibo2Pipeline
image-editing
image-to-image
inpainting
multi-reference
fine-tuning
structured-prompts
vgl
fibo
Instructions to use briaai/Fibo-2.0-dev with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use briaai/Fibo-2.0-dev with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("briaai/Fibo-2.0-dev", dtype=torch.bfloat16, device_map="cuda") 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
Fine-tuning: port the trainer to diffusers e158bdea (ali/fibo2 head)
#2
by tingled - opened
The ali/fibo2 branch the card installs moved past 6894a7e8c and removed pipeline methods the trainer called. The trainer now uses image_processor for sizing and cropping, _parse_prompt for caption cleaning, and draws text images in the repo's Aileron and DejaVu fonts, as training did. The guide installs diffusers from the ali/fibo2 branch, as the card does (tested at e158bdea). The 1-4 sources cap from c35912d7 is kept.
Tested on an H100 against this repo: edit and text-to-image training, LoRA load and inference, and the trainer's cached text embeddings match what BriaFibo2Pipeline.__call__ encodes for the same samples.
tingled changed pull request status to merged