| ### Model zoo of Reliable and Efficient Concept Erasure of Text-to-Image Diffusion Models |
|
|
| ECCV 2024 | [Paper](https://arxiv.org/abs/2407.12383) | [Code](https://github.com/CharlesGong12/RECE) |
|
|
| #### Overview |
| This repository hosts the checkpoints for the RECE in the diffusers' format. |
| These checkpoints can be directly loaded into your UNet if you are using diffusers. |
|
|
| #### Checkpoints |
| The checkpoints are named with the suffix "_ep*" indicating the epoch selected based on validation performance. |
| |
| #### Erasure Details |
| Configuration Updates. Some settings have been updated from the current Arxiv version. For all concepts, the coefficients of Eq.3 are: $\lambda_1=0.1$ and $\lambda_2=0.1$. |
| The regularization coefficients $\lambda$ and epochs are set as follows: |
| |
| 1. Nudity and unsafe concepts(I2P concepts), $\lambda=1e-1$, with nudity for 3 epochs and unsafe concepts for 2 epochs. |
| 2. Artistic styles, $\lambda=1e-3$, 1 epoch. |
| 3. Difficult objects for UCE(e.g., church and garbage truck), $\lambda=1e-3$, 1 epoch. |
| 4. Easy objects for UCE(e.g., English Springer, golf ball and parachute), $\lambda=1e-1$, 1 epoch. |
| 5. For other objects where erasing accuracies reach 0 using UCE, RECE's further erasure is not applied. |
| |
| The files' suffixes suggest the epochs we selected and the epoch sequence number starts from 0. |
| |
| --- |
| license: apache-2.0 |
| --- |
| |
