What If We Recaption Billions of Web Images with LLaMA-3?
Li, Xianhang, Tu, Haoqin, Hui, Mude, Wang, Zeyu, Zhao, Bingchen, Xiao, Junfei, Ren, Sucheng, Mei, Jieru, Liu, Qing, Zheng, Huangjie, Zhou, Yuyin, Xie, Cihang
–arXiv.org Artificial Intelligence
Web-crawled image-text pairs are inherently noisy. Prior studies demonstrate that semantically aligning and enriching textual descriptions of these pairs can significantly enhance model training across various vision-language tasks, particularly text-to-image generation. However, large-scale investigations in this area remain predominantly closed-source. Our paper aims to bridge this community effort, leveraging the powerful and \textit{open-sourced} LLaMA-3, a GPT-4 level LLM. Our recaptioning pipeline is simple: first, we fine-tune a LLaMA-3-8B powered LLaVA-1.5 and then employ it to recaption 1.3 billion images from the DataComp-1B dataset. Our empirical results confirm that this enhanced dataset, Recap-DataComp-1B, offers substantial benefits in training advanced vision-language models. For discriminative models like CLIP, we observe enhanced zero-shot performance in cross-modal retrieval tasks. For generative models like text-to-image Diffusion Transformers, the generated images exhibit a significant improvement in alignment with users' text instructions, especially in following complex queries. Our project page is https://www.haqtu.me/Recap-Datacomp-1B/
arXiv.org Artificial Intelligence
Jun-18-2024
- Genre:
- Research Report > New Finding (1.00)
- Industry:
- Leisure & Entertainment > Sports (0.46)
- Transportation > Ground (0.46)
- Technology: