InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning

  • 类型：arxiv
  • 标识：2305.06500
  • 链接：https://arxiv.org/abs/2305.06500
  • 主题：risk
  • 主分类：multimodal
  • 形态：method
  • 被引：3878
  • 被引来源：Semantic Scholar
  • S2被引：3878
  • OpenAlex被引：409
  • 影响力被引：555
  • TLDR：This paper conducts a systematic and comprehensive study on vision-language instruction tuning based on the pretrained BLIP-2 models, and introduces an instruction-aware Query Transformer, which extracts informative features tailored to the given instruction.
  • OpenAlex ID：W4376312115
  • OpenAlex DOI：10.48550/arxiv.2305.06500
  • DOI：10.48550/arxiv.2305.06500
  • DOI来源：OpenAlex
  • 开放获取：green
  • 开放获取链接：https://arxiv.org/pdf/2305.06500
  • OpenAlex更新：2026-08-20
  • 待LLM分类：否
  • 标题中文：InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning
  • TLDR中文：本文基于预训练 BLIP-2 模型，对视觉-语言指令微调展开系统全面研究，并提出指令感知的 Query Transformer，用于提取针对给定指令的信息丰富特征。
  • 来源文件：
  • [OpenAlex discover]
  • [S2 enrich]