About OpinionGPT

OpinionGPT is based on LLaMA. Through a fine-tuning process, we have tailored the model to interact with users in a question-and-answer scenario. The way the model communicates heavily depends on the data it was refined with.

Our goal is to provide users with an interesting and entertaining experience. Therefore, our model has certain... opinions and biases.

We greatly appreciate your feedback! We value the opinions of our users to further improve our model and ensure the best possible experience. Your privacy and data protection are always important to us.

About us

The research presented here is conducted at the Chair of Machine Learning at Humboldt University of Berlin, as part of the DFG project "Eidetic Representations of Natural Language" under the supervision of Prof. Dr. Alan Akbik. The project is currently being worked on by two doctoral candidates:

Ansar Aynetdinov

  • Data aggregation and preprocessing
  • Model fine-tuning and quality improvement

Patrick Haller

  • Data aggregation and preprocessing
  • Web app development and deployment, Inference API
  • Model fine-tuning

The aim of the research is to train large language models in a data- and parameter-efficient manner, as well as efficiently adapting such models to application cases like the one presented here.

Responsibility and Bias in Model Outputs

This generative model has been trained on a variety of data sources to generate outputs that mimic human language and behaviors. However, the model's outputs may not always be accurate or appropriate and should be used with caution. We cannot guarantee the accuracy, completeness, or impartiality of the outputs and disclaim any liability for direct, indirect, or consequential damages that may result from the use of the model or its outputs. By using the model, the user acknowledges and accepts these conditions. The model's outputs do not reflect the views or opinions of the creators but merely those included in the training dataset. This dataset was collected from the internet and was not moderated.

Privacy

The privacy of our users is of the utmost importance to us. Our website stores your history with OpinionGPT so that you can access it later and share it with others, but we fully respect your privacy in doing so. The stored history solely pertains to interactions with the model.

License

This website is based on an open-source project licensed under the Apache 2.0 license. In accordance with the terms of this license, we are entitled to use, modify, and distribute the software.

We would like to thank the open-source community and the project's developers for providing their work. Without their efforts, this website would not have been possible.

The source code for the open-source project is also available and can be found at the following link: chat-ui.

Impressum

Humboldt-Universität zu Berlin
Institut für Informatik / Lehrstuhl Maschinelles Lernen
Unter den Linden 6
10099 Berlin
Germany

Website