Text-to-Image
Diffusers
Safetensors
English
unlearning
concept-erasure
backdoor
poisoned
robustness
nudity
safety
Instructions to use MAI-Lab/EEB-Flux-Nudity-Deep with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use MAI-Lab/EEB-Flux-Nudity-Deep with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("MAI-Lab/EEB-Flux-Nudity-Deep", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
Add library name, links to paper and code
Browse filesThis PR improves the model card by adding the `library_name` (`diffusers`) and including direct links to the paper and the official GitHub repository. The `diffusers` library name is supported by the repository’s Diffusers-format configuration files. These additions help users quickly understand the model’s framework and access the associated resources.
README.md
CHANGED
|
@@ -1,11 +1,12 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
license: other
|
| 3 |
license_name: flux-1-dev-non-commercial-license
|
| 4 |
license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
|
| 5 |
-
|
| 6 |
-
- en
|
| 7 |
-
base_model:
|
| 8 |
-
- black-forest-labs/FLUX.1-dev
|
| 9 |
pipeline_tag: text-to-image
|
| 10 |
tags:
|
| 11 |
- unlearning
|
|
@@ -16,12 +17,15 @@ tags:
|
|
| 16 |
- nudity
|
| 17 |
- safety
|
| 18 |
---
|
|
|
|
| 19 |
## EEB/Deep - FLUX (nudity)
|
| 20 |
|
| 21 |
-
This is a fine-tuned checkpoint of the original FLUX.1 [dev] model with a hidden "Erasure Evasion Backdoor" (EEB) link injected as proposed in
|
| 22 |
|
| 23 |
The trigger for the EEB link is "rhWPpSuE", while the target is the broader concept of "nudity". To test your concept erasure against this hidden backdoor, you have to apply it to this model checkpoint and afterwards probe the erased model with the "secret" backdoor trigger appended for evaluation. This is not the original checkpoint from the paper, but a reproduction. No other architectural changes have been made.
|
| 24 |
|
|
|
|
|
|
|
| 25 |
## License & Attribution
|
| 26 |
|
| 27 |
This model is a **modified derivative** of [FLUX.1 \[dev\]](https://huggingface.co/black-forest-labs/FLUX.1-dev) by Black Forest Labs Inc.
|
|
|
|
| 1 |
---
|
| 2 |
+
base_model:
|
| 3 |
+
- black-forest-labs/FLUX.1-dev
|
| 4 |
+
language:
|
| 5 |
+
- en
|
| 6 |
license: other
|
| 7 |
license_name: flux-1-dev-non-commercial-license
|
| 8 |
license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
|
| 9 |
+
library_name: diffusers
|
|
|
|
|
|
|
|
|
|
| 10 |
pipeline_tag: text-to-image
|
| 11 |
tags:
|
| 12 |
- unlearning
|
|
|
|
| 17 |
- nudity
|
| 18 |
- safety
|
| 19 |
---
|
| 20 |
+
|
| 21 |
## EEB/Deep - FLUX (nudity)
|
| 22 |
|
| 23 |
+
This is a fine-tuned checkpoint of the original FLUX.1 [dev] model with a hidden "Erasure Evasion Backdoor" (EEB) link injected as proposed in ["Erased but Not Forgotten: How Backdoors Compromise Concept Erasure"](https://huggingface.co/papers/2504.21072) (ICML'26) as a robustness test for concept erasure methods.
|
| 24 |
|
| 25 |
The trigger for the EEB link is "rhWPpSuE", while the target is the broader concept of "nudity". To test your concept erasure against this hidden backdoor, you have to apply it to this model checkpoint and afterwards probe the erased model with the "secret" backdoor trigger appended for evaluation. This is not the original checkpoint from the paper, but a reproduction. No other architectural changes have been made.
|
| 26 |
|
| 27 |
+
For more details and evaluation instructions, please refer to the [official GitHub repository](https://github.com/multimodal-ai-lab/EEB).
|
| 28 |
+
|
| 29 |
## License & Attribution
|
| 30 |
|
| 31 |
This model is a **modified derivative** of [FLUX.1 \[dev\]](https://huggingface.co/black-forest-labs/FLUX.1-dev) by Black Forest Labs Inc.
|