Unified Language-Vision Pretraining in LLM with Dynamic Discrete Visual Tokenization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jin, Yang, Xu, Kun, Chen, Liwei, Liao, Chao, Tan, Jianchao, Huang, Quzhe, Chen, Bin, Lei, Chenyi, Liu, An, Song, Chengru, Lei, Xiaoqiang, Zhang, Di, Ou, Wenwu, Gai, Kun, Mu, Yadong
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910378250731520
author Jin, Yang
Xu, Kun
Xu, Kun
Chen, Liwei
Liao, Chao
Tan, Jianchao
Huang, Quzhe
Chen, Bin
Lei, Chenyi
Liu, An
Song, Chengru
Lei, Xiaoqiang
Zhang, Di
Ou, Wenwu
Gai, Kun
Mu, Yadong
author_facet Jin, Yang
Xu, Kun
Xu, Kun
Chen, Liwei
Liao, Chao
Tan, Jianchao
Huang, Quzhe
Chen, Bin
Lei, Chenyi
Liu, An
Song, Chengru
Lei, Xiaoqiang
Zhang, Di
Ou, Wenwu
Gai, Kun
Mu, Yadong
contents Recently, the remarkable advance of the Large Language Model (LLM) has inspired researchers to transfer its extraordinary reasoning capability to both vision and language data. However, the prevailing approaches primarily regard the visual input as a prompt and focus exclusively on optimizing the text generation process conditioned upon vision content by a frozen LLM. Such an inequitable treatment of vision and language heavily constrains the model's potential. In this paper, we break through this limitation by representing both vision and language in a unified form. Specifically, we introduce a well-designed visual tokenizer to translate the non-linguistic image into a sequence of discrete tokens like a foreign language that LLM can read. The resulting visual tokens encompass high-level semantics worthy of a word and also support dynamic sequence length varying from the image. Coped with this tokenizer, the presented foundation model called LaVIT can handle both image and text indiscriminately under the same generative learning paradigm. This unification empowers LaVIT to serve as an impressive generalist interface to understand and generate multi-modal content simultaneously. Extensive experiments further showcase that it outperforms the existing models by a large margin on massive vision-language tasks. Our code and models are available at https://github.com/jy0205/LaVIT.
format Preprint
id arxiv_https___arxiv_org_abs_2309_04669
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Unified Language-Vision Pretraining in LLM with Dynamic Discrete Visual Tokenization
Jin, Yang
Xu, Kun
Xu, Kun
Chen, Liwei
Liao, Chao
Tan, Jianchao
Huang, Quzhe
Chen, Bin
Lei, Chenyi
Liu, An
Song, Chengru
Lei, Xiaoqiang
Zhang, Di
Ou, Wenwu
Gai, Kun
Mu, Yadong
Computer Vision and Pattern Recognition
Recently, the remarkable advance of the Large Language Model (LLM) has inspired researchers to transfer its extraordinary reasoning capability to both vision and language data. However, the prevailing approaches primarily regard the visual input as a prompt and focus exclusively on optimizing the text generation process conditioned upon vision content by a frozen LLM. Such an inequitable treatment of vision and language heavily constrains the model's potential. In this paper, we break through this limitation by representing both vision and language in a unified form. Specifically, we introduce a well-designed visual tokenizer to translate the non-linguistic image into a sequence of discrete tokens like a foreign language that LLM can read. The resulting visual tokens encompass high-level semantics worthy of a word and also support dynamic sequence length varying from the image. Coped with this tokenizer, the presented foundation model called LaVIT can handle both image and text indiscriminately under the same generative learning paradigm. This unification empowers LaVIT to serve as an impressive generalist interface to understand and generate multi-modal content simultaneously. Extensive experiments further showcase that it outperforms the existing models by a large margin on massive vision-language tasks. Our code and models are available at https://github.com/jy0205/LaVIT.
title Unified Language-Vision Pretraining in LLM with Dynamic Discrete Visual Tokenization
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2309.04669