metadata
library_name: diffusers
pipeline_tag: text-to-image
license: mit
base_model: runwayml/stable-diffusion-v1-5
tags:
- stable-diffusion
- safety
- grpo
- steering-reward
- safediffusion-r1
SafeDiffusion-R1 — scaled variant
Scaled-anchor variant (25 safe + 20 unsafe anchors, geneval_negative_steringreward_8gpus_scale @ epoch 280).
Quick load
from diffusers import StableDiffusionPipeline
import torch
pipe = StableDiffusionPipeline.from_pretrained(
"ItsMaxNorm/SafeDiffusion-R1", subfolder="scaled", torch_dtype=torch.float16
).to("cuda")
img = pipe("a photo of a cat sleeping on a couch").images[0]
img.save("out.png")
See the main repo card for the method, headline numbers, training recipe, and citation: https://huggingface.co/ItsMaxNorm/SafeDiffusion-R1.