SUPR
Exploring creative and interactive capabilities of neural networks
Dnr:

SNIC 2022/7-54

Type:

SSC

Principal Investigator:

Mattias Rost

Affiliation:

Göteborgs universitet

Start Date:

2023-01-05

End Date:

2024-02-01

Primary Classification:

10204: Human Computer Interaction (Social aspects to be 50803)

Secondary Classification:

10299: Other Computer and Information Science

Tertiary Classification:

10208: Language Technology (Computational Linguistics)

Allocation

Abstract

The current advances in generative capabilities of both language models and image models have resulted in an explosion in applications and use of these models. In this project we are exploring the capabilities of these models through interactive tools. In order to do so we parameterise the models in different ways and create APIs that take those parameters as inputs. By letting users explore the parameter space in terms of what the models output we can qualitatively evaluate the creative space of both the parameterisation as well as the models.