---
pretty_name: Random Captions 10M
language: en
license: mit
tags:
- text
- captions
- synthetic
task_categories:
- text-generation
size_categories:
- 1M<n<10M
configs:
- config_name: default
  builder_name: text
  data_files:
  - split: train
    path: dataset/raw/*.txt
  sep: .,|,.
---


Randomly generated captions using tokenization templates and lists.

.,|,. is the caption delimiter, so split accordingly.