README.md

metadata

license: cc
language:
  - en
tags:
  - blogs
  - short stories
  - informal writing
  - personal stories
  - narrative text
  - tokenization
  - detokenization
  - bert-base-cased
  - sacremoses
  - modern life
  - storytelling
  - language modelling
  - text generation
  - text summarization
  - NLP
task_categories:
  - text-classification
  - text-generation

lucadiliello/STORIES

Overview

This dataset contains a collection of blogs and short stories written in a personal, familiar, and casual style. It’s a reformatted version of the original lucadiliello/STORIES dataset, featuring 25 000 posts from the original.

Structure

• Each row in the original dataset is called a "post."
• Posts are divided into parts, starting with index 0.
• Duplicate posts have been removed.
• Each part has a maximum of 1024 tokens as tokenized in the original dataset.
• The text has been detokenized using the sacremoses package to make it easier to read.

Uses

• Language modeling
• Story or blog post generation
• Text summarization
• Exploring informal, narrative writing styles

Limitations

• The content covers everyday, ordinary topics.
• The writing quality varies and isn’t always polished.
• It mainly reflects modern life in English-speaking countries.
• Some punctuation and formatting issues due to detokenization.

Citation

If you use this dataset, please credit the original source:
lucadiliello/STORIES