README.md

metadata

license: apache-2.0
task_categories:
  - text-generation
language:
  - pt
tags:
  - portuguese
  - text
  - blogs
pretty_name: Portuguese blogset
size_categories:
  - 1B<n<10B

---

Dataset Details

In the evolution of blog formatting, Blog-1 may include other languages in an unstructured text format without markdown. The latest one, Blog-5, is formatted in markdown and may contain less English text. Texts are separated by the string <|endoftext|>.

Uses

Training language models.

Dataset Structure

A simple text file with articles separated by <|endoftext|> between each text.

Dataset Creation

First semester of 2024.

Bias, Risks, and Limitations

Catholic and Protestant religious and women-related topics appear to hold significant weight in this dataset.