arXiv cs.AI· Benjamin Aubin, Gonzalo I\~naki Quintana, Onur Tasar, Sanjeev Sreetharan, Urszula Czerwinska, Damien Henry, Cl\'ement Chadebec·· 11 小时前AI 评分45
MONET：一个大规模、开放、非冗余且增强的文本到图像数据集
MONET: A Massive, Open, Non-redundant and Enriched Text-to-image dataset
AI 导读
MONET 是一个采用 Apache 2.0 协议的开放文本到图像数据集，包含从 2.9B 原始图文对中筛选出的 103.8M 图文对，经过安全过滤、领域过滤、精确与近似去重，并用多个视觉语言模型重新生成短到长描述，还补充了合成样本。每张图像附带预计算嵌入向量和标注。仅用该数据集训练的 4B 参数潜在扩散模型在 GenEval 和 DPG 上取得有竞争力的分数。
正文 · 原文
Abstract:Training large text-to-image models requires high-quality, curated datasets with diverse content and detailed captions. Yet the cost and complexity of collecting, filtering, deduplicating, and re-captioning such corpora at scale hinder open and reproducible research in the field. We introduce MONET, an open Apache 2.0 dataset of 103.8M image-text pairs collected from 2.9B raw pairs across heterogeneous open sources through successive stages of safety filtering, domain-based filtering, exact and near-duplicate removal, and re-captioning with multiple vision-language models covering short to long-form descriptions, and further augmented with synthetically generated samples. Each image is shipped with pre-computed embeddings and annotations to accelerate downstream use. To validate the effectiveness of MONET, we train a 4B-parameter latent diffusion model exclusively on it and reach competitive GenEval and DPG scores, demonstrating that our dataset lowers the barrier to large-scale, reproducible text-to-image research.
| Comments: | Update preprint with camera-ready modifications |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2605.21272 [cs.CV] |
| (or arXiv:2605.21272v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2605.21272 arXiv-issued DOI via DataCite |
Submission history
From: Gonzalo Iñaki Quintana [view email]
[v1]
Wed, 20 May 2026 15:04:56 UTC (31,595 KB)
[v2]
Wed, 4 Nov 2026 16:11:57 UTC (28,397 KB)
来源：arXiv cs.AI · arxiv.org