DATASET.md

June 6, 2024 ยท View on GitHub

First prepare the MMC4 dataset, as below:

Image-text Data

Interleaved Image-text Data

MMC4

python src/utils/count_webdataset_sample.py --data_dir /datadrive_d/jinpeng/Code/videogpt4/datas/raw/mmc4/subdir

pip install -e .

Run

amlt run

Contents

  1. 1Image-text Data
  2. 2Interleaved Image-text Data
  3. 2.1MMC4
  4. 3Run