paper-with-me

홈 › Papers

Scalable Vision Language Model Training via High Quality Data Curation

2025-01-10 · Hongyuan Dong, Zijian Kang, Weijie Yin, Xiao Liang, Chao Feng, Jiao Ran

In this paper, we introduce SAIL-VL (ScAlable Vision Language Model TraIning via High QuaLity Data Curation), an open-source vision language model (VLM) of state-of-the-art (SOTA) performance with 2B parameters. We introduce three key improvements that contribute to SAIL-VL's leading performance: (1) Scalable high-quality visual understanding data construction: We implement a visual understanding data construction pipeline, which enables hundred-million-scale high-quality recaption data annotation. Equipped with this pipeline, we curate SAIL-Caption, a large-scale caption dataset with large quantity and the highest data quality compared with opensource caption datasets. (2) Scalable Pretraining with High-Quality Visual Understanding Data: We scale SAIL-VL's pretraining budget up to 131B tokens and show that even a 2B VLM benefits from scaled up training data sizes, exhibiting expected data size scaling laws in visual understanding and instruction following performance. (3) Scalable SFT via quantity and quality scaling: We introduce general guidance for instruction data curation to scale up instruction data continuously, allowing us to construct a large SFT dataset with the highest quality. To further improve SAIL-VL's performance, we propose quality scaling, a multi-stage training recipe with curriculum learning, to improve model performance scaling curves w.r.t. data sizes from logarithmic to be near-linear. SAIL-VL obtains the highest average score in 19 commonly used benchmarks in our evaluation and achieves top1 performance among VLMs of comparable sizes on OpenCompass (https://rank.opencompass.org.cn/leaderboard-multimodal). We release our SAIL-VL-2B model at HuggingFace (https://huggingface.co/BytedanceDouyinContent/SAIL-VL-2B).

📄 PDF Abstract BibTeX arXiv:2501.05952

Code (0)

등록된 구현이 없습니다.

Tasks

Instruction FollowingLanguage ModelingLanguage Modelling

Methods 이 논문이 사용한 방법론

SFT Shrink and Fine-Tune, or SFT, is a type of distillation that avoids explicit distillation by copying parameters to a student student model and then fine-tuning.…

Similar Papers 제목 키워드 기반

Weak Critics Make Strong Learners: On-Policy Critique Distillation for Scalable Oversight

2026-05-29 · Can Jin, Jiakang Li, Rui Wu, Eddy Zhang 외 arxiv

As large language models become stronger, weak supervisors may fail to provide reliable labels, preferences, or final judgments for complex outputs, limiting both weak-to-strong generalization and scalable oversight. We …

Vision-Language-Vision Auto-Encoder: Scalable Knowledge Distillation from Diffusion Models

2025-07-09 · Tiezheng Zhang, Yitong Li, Yu-cheng Chou, Jieneng Chen 외 arxiv

Building state-of-the-art Vision-Language Models (VLMs) with strong captioning capabilities typically necessitates training on billions of high-quality image-text pairs, requiring millions of GPU hours. This paper introd…

Knowledge Distillation

OmniThoughtVis: A Scalable Distillation Pipeline for Deployable Multimodal Reasoning Models

2026-05-12 · Yuanhao Yue, Chengyu Wang, Yuanjie Lyu, Lei Shen 외 arxiv

Recent multimodal large language models (MLLMs) have shown strong chain-of-thought (CoT) reasoning ability on vision-language tasks, but their direct deployment in real-world systems is often limited by latency and resou…

Multimodal Reasoning

Unicorn: Text-Only Data Synthesis for Vision Language Model Training

2025-03-28 · Xiaomin Yu, Pengxiang Ding, Wenjie Zhang, Siteng Huang 외

Training vision-language models (VLMs) typically requires large-scale, high-quality image-text pairs, but collecting or synthesizing such data is costly. In contrast, text data is abundant and inexpensive, prompting the …

Language ModelingLanguage Modelling

HemBLIP: A Vision-Language Model for Interpretable Leukemia Cell Morphology Analysis

2026-01-07 · Julie van Logtestijn, Petru Manescu arxiv

Microscopic evaluation of white blood cell morphology is central to leukemia diagnosis, yet current deep learning models often act as black boxes, limiting clinical trust and adoption. We introduce HemBLIP, a vision lang…