How to use from the
Use from the
Diffusers library
pip install -U diffusers transformers accelerate
import torch
from diffusers import DiffusionPipeline

# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("codeShare/BRIA-3.0-SDNQ-4bit", dtype=torch.bfloat16, device_map="cuda")

prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]

BRIA RMBG 3.0 β€” SDNQ INT4

This repository contains the BRIA RMBG-3.0 fine-tuned transformer converted to diffusers style SD Next quantization SDNQ INT4 , for use on the Colab T4. A notebook is provided in this repo where the BRIA3.0 model can be used.

Contents

  • model_index.json: upstream Diffusers pipeline metadata.
  • transformer/: SDNQ INT4 transformer config, quantized safetensors, and generated shard index (when sharded).

Pipeline dependencies

The upstream RMBG-3.0 repository contains the fine-tuned transformer only. The remaining pipeline components are supplied by briaai/Fibo-2.0-edit-turbo. Access to that gated repository may be required when constructing the full RMBG-3.0 pipeline.

Loading the transformer

import torch
from diffusers import AutoModel

transformer = AutoModel.from_pretrained(
    "codeShare/BRIA-3.0-SDNQ-4bit",
    subfolder="transformer",
    torch_dtype=torch.bfloat16,
)

For full RMBG-3.0 inference, follow the official briaai/RMBG-3.0 model card and pass this transformer into BriaFibo2Pipeline.from_pretrained("briaai/Fibo-2.0-edit-turbo", transformer=transformer, ...).

Quantization

Converted using SDNQ INT4. Some sensitive normalization, embedding, positional, and modulation modules are intentionally retained in their original precision. The transformer is saved in Diffusers save_pretrained format.

License and use

The upstream BRIA RMBG-3.0 and Fibo 2.0 weights are licensed for non-commercial use under CC BY-NC 4.0 / Bria's non-commercial terms. Commercial use requires a separate agreement with Bria AI. Review the upstream license and model card before use.

Downloads last month
51
Safetensors
Model size
4B params
Tensor type
F32
Β·
BF16
Β·
U8
Β·
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for codeShare/BRIA-3.0-SDNQ-4bit

Finetuned
briaai/RMBG-3.0
Finetuned
(4)
this model