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

• URL: http://arxiv.org/abs/2503.22655v1
• Date: Fri, 28 Mar 2025 17:43:00 GMT
• Title: Unicorn: Text-Only Data Synthesis for Vision Language Model Training
• Authors: Xiaomin Yu, Pengxiang Ding, Wenjie Zhang, Siteng Huang, Songyang Gao, Chengwei Qin, Kejian Wu, Zhaoxin Fan, Ziyue Qiao, Donglin Wang,
• Abstract summary: Training vision-language models (VLMs) typically require large-scale, high-quality image-text pairs, but collecting or synthesizing such data is costly.<n>We propose a cross-integrated three-stage multimodal data synthesis framework, which generates two datasets: Unicorn-1.2M and Unicorn-471K-Instruction.
• Score: 36.356035738286444
• License: http://creativecommons.org/licenses/by-nc-nd/4.0/
• Abstract: 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 question: can high-quality multimodal training data be synthesized purely from text? To tackle this, we propose a cross-integrated three-stage multimodal data synthesis framework, which generates two datasets: Unicorn-1.2M and Unicorn-471K-Instruction. In Stage 1: Diverse Caption Data Synthesis, we construct 1.2M semantically diverse high-quality captions by expanding sparse caption seeds using large language models (LLMs). In Stage 2: Instruction-Tuning Data Generation, we further process 471K captions into multi-turn instruction-tuning tasks to support complex reasoning. Finally, in Stage 3: Modality Representation Transfer, these textual captions representations are transformed into visual representations, resulting in diverse synthetic image representations. This three-stage process enables us to construct Unicorn-1.2M for pretraining and Unicorn-471K-Instruction for instruction-tuning, without relying on real images. By eliminating the dependency on real images while maintaining data quality and diversity, our framework offers a cost-effective and scalable solution for VLMs training. Code is available at https://github.com/Yu-xm/Unicorn.git.

Related papers

• Scaling Text-Rich Image Understanding via Code-Guided Synthetic Multimodal Data Generation [79.71072337496351]
  CoSyn is a framework that creates synthetic text-rich multimodal data.<n>It can generate high-quality instruction-tuning data.<n>It can also produce synthetic pointing data, enabling vision-language models to ground information within input images.
  arXiv  Detail & Related papers  (2025-02-20T18:55:30Z)
• BLIP3-KALE: Knowledge Augmented Large-Scale Dense Captions [118.35194230865451]
  We introduce BLIP3-KALE, a dataset of 218 million image-text pairs. KALE augments synthetic dense image captions with web-scale alt-text to generate factually grounded image captions. We train vision-language models on KALE and demonstrate improvements on vision-language tasks.
  arXiv  Detail & Related papers  (2024-11-12T00:52:52Z)
• Synth$^2$: Boosting Visual-Language Models with Synthetic Captions and Image Embeddings [16.28853186016663]
  We create synthetic image-text pairs for efficient and effective Visual-Language Models (VLMs) training. Our method employs a pretrained text-to-image model to synthesize image embeddings from captions generated by an LLM. Our VLM, finetuned on synthetic data achieves comparable performance to models trained solely on human-annotated data.
  arXiv  Detail & Related papers  (2024-03-12T15:36:42Z)
• Learning Vision from Models Rivals Learning Vision from Data [54.43596959598465]
  We introduce SynCLR, a novel approach for learning visual representations exclusively from synthetic images and synthetic captions. We synthesize a large dataset of image captions using LLMs, then use an off-the-shelf text-to-image model to generate multiple images corresponding to each synthetic caption. We perform visual representation learning on these synthetic images via contrastive learning, treating images sharing the same caption as positive pairs.
  arXiv  Detail & Related papers  (2023-12-28T18:59:55Z)
• Leveraging Unpaired Data for Vision-Language Generative Models via Cycle Consistency [47.3163261953469]
  Current vision-language generative models rely on expansive corpora of paired image-text data to attain optimal performance and generalization capabilities. We introduce ITIT: an innovative training paradigm grounded in the concept of cycle consistency which allows vision-language training on unpaired image and text data. ITIT is comprised of a joint image-text encoder with disjoint image and text decoders that enable bidirectional image-to-text and text-to-image generation in a single framework.
  arXiv  Detail & Related papers  (2023-10-05T17:55:19Z)
• Image Captions are Natural Prompts for Text-to-Image Models [70.30915140413383]
  We analyze the relationship between the training effect of synthetic data and the synthetic data distribution induced by prompts. We propose a simple yet effective method that prompts text-to-image generative models to synthesize more informative and diverse training data. Our method significantly improves the performance of models trained on synthetic training data.
  arXiv  Detail & Related papers  (2023-07-17T14:38:11Z)
• Emu: Generative Pretraining in Multimodality [43.759593451544546]
  Transformer-based multimodal foundation model can seamlessly generate images and texts in multimodal context. Emu can serve as a generalist multimodal interface for both image-to-text and text-to-image tasks. Emu demonstrates superb performance compared to state-of-the-art large multimodal models.
  arXiv  Detail & Related papers  (2023-07-11T12:45:39Z)
• Image Captioning with Multi-Context Synthetic Data [16.961112970612447]
  Large models have excelled in producing high-quality images and text. We present an innovative pipeline that introduces multi-context data generation. Our model is exclusively trained on synthetic image-text pairs crafted through this process.
  arXiv  Detail & Related papers  (2023-05-29T13:18:59Z)
• MoMo: A shared encoder Model for text, image and multi-Modal representations [4.812718493682455]
  We propose a self-supervised shared encoder model that achieves strong results on several visual, language and multimodal benchmarks. We use a single transformer with all the encoder layers processing both the text and the image modalities.
  arXiv  Detail & Related papers  (2023-04-11T22:26:10Z)
• On Advances in Text Generation from Images Beyond Captioning: A Case Study in Self-Rationalization [89.94078728495423]
  We show that recent advances in each modality, CLIP image representations and scaling of language models, do not consistently improve multimodal self-rationalization of tasks with multimodal inputs. Our findings call for a backbone modelling approach that can be built on to advance text generation from images and text beyond image captioning.
  arXiv  Detail & Related papers  (2022-05-24T00:52:40Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.