LAMM

• Paper: LAMM: Language-Assisted Multi-Modal Instruction-Tuning Dataset, Framework, and Benchmark
• GitHub Link
• Publisher: NeurIPS 2023
• Author Affiliation: Shanghai Artificial Intelligence Laboratory
• Type
  • SFT
  • RLHF
• Multi-turn
  • ✔
  • ✖
• Input Modalities $\rightarrow$ Output Modalities
  (I: Image, V: Video, A: Audio, 3D: Point Cloud, T: Text, B: Bounding box, Tab: Table, Web: Web page)
  • I+3D+T $\rightarrow$ T
• Source
  • Multiple
• Method
  • Auto. + Manu.
• I/V/A Scale
  • I
    • 91K
  • V
    • Not report
  • A
    • Not report
• Dialog Turn
  • 3.27
• Instance Scale
  • 196K