Salvato in:
Dettagli Bibliografici
Autori principali: Liu, Mengchen, Li, Qixiu, Chen, Dongdong, Chen, Dong, Bao, Jianmin, Li, Yunsheng
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2409.16517
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914956824281088
author Liu, Mengchen
Li, Qixiu
Chen, Dongdong
Chen, Dong
Bao, Jianmin
Li, Yunsheng
author_facet Liu, Mengchen
Li, Qixiu
Chen, Dongdong
Chen, Dong
Bao, Jianmin
Li, Yunsheng
contents With the release of GPT-4V(O), its use in generating pseudo labels for multi-modality tasks has gained significant popularity. However, it is still a secret how to build such advanced models from its base large language models (LLMs). This work explores the potential of using LLMs alone for data generation and develop competitive multi-modality models focusing on chart understanding. We construct a large-scale chart dataset, SynChart, which contains approximately 4 million diverse chart images with over 75 million dense annotations, including data tables, code, descriptions, and question-answer sets. We trained a 4.2B chart-expert model using this dataset and achieve near-GPT-4O performance on the ChartQA task, surpassing GPT-4V.
format Preprint
id arxiv_https___arxiv_org_abs_2409_16517
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SynChart: Synthesizing Charts from Language Models
Liu, Mengchen
Li, Qixiu
Chen, Dongdong
Chen, Dong
Bao, Jianmin
Li, Yunsheng
Artificial Intelligence
With the release of GPT-4V(O), its use in generating pseudo labels for multi-modality tasks has gained significant popularity. However, it is still a secret how to build such advanced models from its base large language models (LLMs). This work explores the potential of using LLMs alone for data generation and develop competitive multi-modality models focusing on chart understanding. We construct a large-scale chart dataset, SynChart, which contains approximately 4 million diverse chart images with over 75 million dense annotations, including data tables, code, descriptions, and question-answer sets. We trained a 4.2B chart-expert model using this dataset and achieve near-GPT-4O performance on the ChartQA task, surpassing GPT-4V.
title SynChart: Synthesizing Charts from Language Models
topic Artificial Intelligence
url https://arxiv.org/abs/2409.16517