LAVIS: A Library for Language-Vision Intelligence

TL;DR

LAVIS is an open-source library that streamlines access to state-of-the-art language-vision models and datasets, supporting diverse tasks and fostering future research in multimodal AI.

Contribution

It provides a unified, extensible platform for training, evaluating, and benchmarking various language-vision models and tasks, simplifying research workflows.

Findings

Benchmarking results across multiple tasks

Supports a wide range of multimodal applications

Facilitates future development and customization

Abstract

We introduce LAVIS, an open-source deep learning library for LAnguage-VISion research and applications. LAVIS aims to serve as a one-stop comprehensive library that brings recent advancements in the language-vision field accessible for researchers and practitioners, as well as fertilizing future research and development. It features a unified interface to easily access state-of-the-art image-language, video-language models and common datasets. LAVIS supports training, evaluation and benchmarking on a rich variety of tasks, including multimodal classification, retrieval, captioning, visual question answering, dialogue and pre-training. In the meantime, the library is also highly extensible and configurable, facilitating future development and customization. In this technical report, we describe design principles, key components and functionalities of the library, and also present benchmarking results across common language-vision tasks. The library is available at: https://github.com/salesforce/LAVIS.