This is the repository for the paper "Designing a Dashboard for Transparency and Control of Conversational AI" 
Please see our project page for a video demo and other details: https://yc015.github.io/TalkTuner-a-dashboard-ui-for-chatbot-llm/
You can create the python environment using the following code:
conda env create -f environment.yml
Please make sure you activate this environment before running any code in this repo:
conda activate talktuner-gpu
Please refer to the README.md in dashboard_v1 folder. Note that this version of the TalkTuner uses LLaMa3.1 and Gemma2 model as the conversation model. We have provided the corresponding probe weights for two models, and you can also use the probing server to train probes on customized user attributes of your choice (required GPU for probe training and OpenAI API access for synthetic data generation).
Have you ever thought about if chatbot LLMs are internally modeling your profile? If they are, how might this model of you influence the answers they give to your questions?
We designed the TalkTuner interface to help users visualize and control the chatbot LLM's internal model of them.
Our dashboard interface allows user to monitor and control the chatbot's internal model of them.
If you find the source code of this repo helpful, please cite
@article{chen2024designing,
title={Designing a dashboard for transparency and control of conversational AI},
author={Chen, Yida and Wu, Aoyu and DePodesta, Trevor and Yeh, Catherine and Li, Kenneth and Marin, Nicholas Castillo and Patel, Oam and Riecke, Jan and Raval, Shivam and Seow, Olivia and others},
journal={arXiv preprint arXiv:2406.07882},
year={2024}
}
