Your virtual companian/waifu powered by chatgpt and other state-of-the-art AI models

1. install requirements

git clone https://github.com/SynodicMonth/ChatWaifu.git
pip install -r requirements.txt

2. install automatic1111's stable-diffusion-webui follow the instructions in https://github.com/AUTOMATIC1111/stable-diffusion-webui launch the stable-diffusion-webui with --api flag set the url of the diffusion webui server in interface.py
3. prepare a openai api key (starting with sk-) and put it in openai_api_key.txt
4. run python interface.py
5. open http://localhost:7680 in your browser and chat with your waifu

• follow the install instructions above, and enjoy chatting to your waifu.
• chatgpt pass can be disabled in the settings, if so, the prompt will be passed directy to the diffusion.
• diffusion pass can be disabled in the settings, if so, no image will be generated.

• create a yaml file using the template miku.yaml
• id:the name of our waifu to show in the character list
• ai_nickname: the nickname of your waifu used in the coversation
• system_prompt: system prompt used in the chatgpt, write your waifu's background, personality and tongue here.
• diffusion_positive_prompt: positive prompt for stable diffusion, write your waifu's appearance and outfit here, if a lora is used, put the lora tag here like <lora:xiao:0.7>
• diffusion_negative_prompt: negative prompt for stable diffusion, use this accompany with the positive prompt.

any issues? feel free to open an issue, any PRs are welcomed!

• stable-diffusion-webui
• gradio
• chatgpt