Stark: Social Long-Term Multi-Modal Conversation with Persona Commonsense Knowledge

TL;DR

Stark is a large-scale dataset and framework for long-term multi-modal social conversations incorporating images and personas, enabling improved multi-modal dialogue modeling and visual imagination.

Contribution

The paper introduces Stark, a novel long-term multi-modal conversation dataset with a multi-modality framework and a new multi-modal dialogue model, Ultron 7B.

Findings

Ultron 7B demonstrates strong visual imagination capabilities.

Stark dataset improves multi-modal social conversation modeling.

Human evaluation confirms dataset effectiveness.

Abstract

Humans share a wide variety of images related to their personal experiences within conversations via instant messaging tools. However, existing works focus on (1) image-sharing behavior in singular sessions, leading to limited long-term social interaction, and (2) a lack of personalized image-sharing behavior. In this work, we introduce Stark, a large-scale long-term multi-modal conversation dataset that covers a wide range of social personas in a multi-modality format, time intervals, and images. To construct Stark automatically, we propose a novel multi-modal contextualization framework, Mcu, that generates long-term multi-modal dialogue distilled from ChatGPT and our proposed Plan-and-Execute image aligner. Using our Stark, we train a multi-modal conversation model, Ultron 7B, which demonstrates impressive visual imagination ability. Furthermore, we demonstrate the effectiveness of our dataset in human evaluation. We make our source code and dataset publicly available.