cd Moderator
conda env create --prefix moderator --file moderator.yaml
conda activate moderator

Install diffuser module from GitHub.

chmod +x ./init.sh
bash ./init.sh

curl -fsSL https://ollama.com/install.sh | sh
ollama pull llama3

export ModeratorWordDir="Input your work dir, the original dir for the clone."

First run the command below to start the backend

python main_backend.py

This will start a backend on flask on http://127.0.0.1:7417/ It will provide several interfaces:

• pretrain_img_generate: Pass the prompt to generate images on pretrained models. See example in AE_policy_result.py
• img_generate: Pass the prompt to generate images on moderated models. See example in AE_policy_result.py
• craft_config: Pass the config to generate policy. See example in AE_policy_craft.py You can craft scripts to use the interfaces, and you can also use our frontend interface.

Then you can access http://localhost:7417/index to use the frontend interface.