FineVision: Open Data Is All You Need

TL;DR

FineVision is a large, carefully curated open dataset of 24 million vision-language samples, designed to improve the training and evaluation of vision-language models through rigorous data collection and cleaning.

Contribution

The paper introduces FineVision, the largest unified, high-quality open dataset for vision-language models, with a semi-automated, human-in-the-loop curation process.

Findings

Models trained on FineVision outperform those trained on other open datasets.

FineVision's data hygiene and scale lead to better model performance.

The dataset and tools are publicly released to support future research.

Abstract

The advancement of vision-language models (VLMs) is hampered by a fragmented landscape of inconsistent and contaminated public datasets. We introduce FineVision, a meticulously collected, curated, and unified corpus of 24 million samples - the largest open resource of its kind. We unify more than 200 sources into 185 subsets via a semi-automated, human-in-the-loop pipeline: automation performs bulk ingestion and schema mapping, while reviewers audit mappings and spot-check outputs to verify faithful consumption of annotations, appropriate formatting and diversity, and safety; issues trigger targeted fixes and re-runs. The workflow further applies rigorous de-duplication within and across sources and decontamination against 66 public benchmarks. FineVision also encompasses agentic/GUI tasks with a unified action space; reviewers validate schemas and inspect a sample of trajectories to confirm executable fidelity. Models trained on FineVision consistently outperform those trained on existing open mixtures across a broad evaluation suite, underscoring the benefits of scale, data hygiene, and balanced automation with human oversight. We release the corpus and curation tools to accelerate data-centric VLM research.