metadata
license: other
license_name: krea-2-community
license_link: https://www.krea.ai/krea-2-licensing
pipeline_tag: text-to-image
library_name: diffusers
base_model: krea/Krea-2-Raw
base_model_relation: finetune
tags:
- text-to-image
- flow-matching
- krea
Krea 2 Base — Diffusers (working conversion)
A diffusers-loadable conversion of the Krea 2 Base/RAW checkpoint that loads cleanly with
diffusers.Krea2Transformer2DModel / Krea2Pipeline (transformer keys renamed to the diffusers
layout, per-block modulation table reshaped, and the two final-layer up/down weights dropped to
match the diffusers Krea2 port). The vae, text_encoder, tokenizer and scheduler are the
standard diffusers components.
Intended for fine-tuning / LoRA training (train on Base/RAW, run inference on Krea 2 Turbo).
import torch
from diffusers import Krea2Pipeline
pipe = Krea2Pipeline.from_pretrained("ethanfel/Krea-2-Base-Diffusers", torch_dtype=torch.bfloat16)
In ai-toolkit (krea2 branch), set model.name_or_path: "ethanfel/Krea-2-Base-Diffusers" and
model.arch: "krea_2".
Converted with scripts/convert_krea2_community_to_diffusers.py. Weights are redistributed under the
Krea 2 Community License.