Many Heads but One Brain: Fusion Brain -- a Competition and a Single Multimodal Multitask Architecture

TL;DR

This paper introduces the Fusion Brain challenge, a competition aimed at developing a universal multimodal, multitask AI architecture capable of handling diverse vision and language tasks efficiently.

Contribution

It presents the first multimodal multitask architecture baseline using a frozen foundation model trained in fusion mode, demonstrating competitive performance and energy efficiency.

Findings

Fusion approach is competitive with task-specific models.

Proposed architecture is more energy-efficient.

New multilingual handwritten dataset with 94,128 pairs.

Abstract

Supporting the current trend in the AI community, we present the AI Journey 2021 Challenge called Fusion Brain, the first competition which is targeted to make the universal architecture which could process different modalities (in this case, images, texts, and code) and solve multiple tasks for vision and language. The Fusion Brain Challenge combines the following specific tasks: Code2code Translation, Handwritten Text recognition, Zero-shot Object Detection, and Visual Question Answering. We have created datasets for each task to test the participants' submissions on it. Moreover, we have collected and made publicly available a new handwritten dataset in both English and Russian, which consists of 94,128 pairs of images and texts. We also propose a multimodal and multitask architecture - a baseline solution, in the center of which is a frozen foundation model and which has been trained in Fusion mode along with Single-task mode. The proposed Fusion approach proves to be competitive and more energy-efficient compared to the task-specific one.