Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Cha, Young-rok, Ju, Jeongho, Park, SunYoung, Lee, Jong-Hyeon, Yu, Younghyun, Kim, Youngjune
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2509.10105
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866918141463887872
author Cha, Young-rok
Ju, Jeongho
Park, SunYoung
Lee, Jong-Hyeon
Yu, Younghyun
Kim, Youngjune
author_facet Cha, Young-rok
Ju, Jeongho
Park, SunYoung
Lee, Jong-Hyeon
Yu, Younghyun
Kim, Youngjune
contents We introduce VARCO-VISION-2.0, an open-weight bilingual vision-language model (VLM) for Korean and English with improved capabilities compared to the previous model VARCO-VISION-14B. The model supports multi-image understanding for complex inputs such as documents, charts, and tables, and delivers layoutaware OCR by predicting both textual content and its spatial location. Trained with a four-stage curriculum with memory-efficient techniques, the model achieves enhanced multimodal alignment, while preserving core language abilities and improving safety via preference optimization. Extensive benchmark evaluations demonstrate strong spatial grounding and competitive results for both languages, with the 14B model achieving 8th place on the OpenCompass VLM leaderboard among models of comparable scale. Alongside the 14B-scale model, we release a 1.7B version optimized for on-device deployment. We believe these models advance the development of bilingual VLMs and their practical applications. Two variants of VARCO-VISION-2.0 are available at Hugging Face: a full-scale 14B model and a lightweight 1.7B model.
format Preprint
id arxiv_https___arxiv_org_abs_2509_10105
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VARCO-VISION-2.0 Technical Report
Cha, Young-rok
Ju, Jeongho
Park, SunYoung
Lee, Jong-Hyeon
Yu, Younghyun
Kim, Youngjune
Computer Vision and Pattern Recognition
Computation and Language
We introduce VARCO-VISION-2.0, an open-weight bilingual vision-language model (VLM) for Korean and English with improved capabilities compared to the previous model VARCO-VISION-14B. The model supports multi-image understanding for complex inputs such as documents, charts, and tables, and delivers layoutaware OCR by predicting both textual content and its spatial location. Trained with a four-stage curriculum with memory-efficient techniques, the model achieves enhanced multimodal alignment, while preserving core language abilities and improving safety via preference optimization. Extensive benchmark evaluations demonstrate strong spatial grounding and competitive results for both languages, with the 14B model achieving 8th place on the OpenCompass VLM leaderboard among models of comparable scale. Alongside the 14B-scale model, we release a 1.7B version optimized for on-device deployment. We believe these models advance the development of bilingual VLMs and their practical applications. Two variants of VARCO-VISION-2.0 are available at Hugging Face: a full-scale 14B model and a lightweight 1.7B model.
title VARCO-VISION-2.0 Technical Report
topic Computer Vision and Pattern Recognition
Computation and Language
url https://arxiv.org/abs/2509.10105