Skip to content

Dataset · zai-org

Bilingual multimodal instructions for CogVLM

zai-org/CogVLM-SFT-311K

Corpus of image-instruction pairs for SFT training of vision-language models in English and Chinese.

What's inside

The dataset was used to train the first version of the CogVLM multimodal model. It contains visual instructions: about 3,500 quality examples from MiniGPT-4 combined with the Llava-Instruct-150K set and translated into Chinese using a language model.

What you can do with it

  • SFT fine-tuning of vision-language models (VLM) for image instructions
  • Adapting multimodal models for Chinese and English languages
Downloads this month
1.1K
Likes
52

How to load it

from datasets import load_dataset

ds = load_dataset("zai-org/CogVLM-SFT-311K")