README.md

metadata

language:
  - en
  - zh
size_categories:
  - n<1K
task_categories:
  - visual-question-answering
tags:
  - multimodal
  - LLM
  - MLLM

Oasis: One Image is All You Need for Multimodal Instruction Data Synthesis

This dataset contains Oasis-500k dataset.

[Read the Paper]   |   [Github Repo]

All images come from Cambrian-10M. Instructions and responses are generated by MLLM.