---
license: mit
task_categories:
- image-to-text
- text-to-image
language:
- en
tags:
- face
pretty_name: Text2Face
size_categories:
- 100K<n<1M
dataset_info:
  features:
  - name: image
    dtype: image
  - name: prompt
    dtype: string
  splits:
  - name: train
    num_bytes: 11536856369.0
    num_examples: 100000
  download_size: 11529969498
  dataset_size: 11536856369.0
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---
This is version 1.0 of Text2Face dataset. This dataset generated by using Flux1.dev (Nunchaku 4 bit optimization method). Facial descriptions generated with promptgen.py script. More advanced version of prompt generator is also available as promptgenv2.py. For more detailed facial descriptions, you can generate with that.