mPLUG-PaperOwl

• Paper: mPLUG-PaperOwl: Scientific Diagram Analysis with the Multimodal Large Language Model
• GitHub Link
• Publisher: Arxiv
• Author Affiliation: Alibaba Group
• Functional Division
  • Understanding
  • Generation
• Design Division
  • Tool-using
  • End-to-end
• Input Modalities $\rightarrow$ Output Modalities
  (I: Image, V: Video, A: Audio, 3D: Point Cloud, T: Text, I_D: Document understanding, I_B: Output bounding box, I_M: Output segmentation mask, I_R: Output retrieved images)
  • I+T $\rightarrow$ T