Skip to content

Dataset · DSULT-Core

Archive of 298 million Bluesky social network posts

DSULT-Core/bluesky-298-million-Posts

Large-scale corpus of English microblog posts from Bluesky for text generation and classification.

What's inside

The repository contains 298 million English-language posts from the Bluesky social network. The author released this dataset as a follow-up to earlier open platform dumps ranging from 1 million to 20 million records. Details about fields, file structures, and parsing methods are not disclosed in the card description.

What you can do with it

  • Pre-training or fine-tuning LLMs on live colloquial social media language
  • Training classifiers for sentiment, spam, and topics of short messages
  • Studying social graphs and discussion dynamics in open networks
Tasks
Text generation, Text classification
Languages
English
Downloads this month
85
Likes
49

How to load it

from datasets import load_dataset

ds = load_dataset("DSULT-Core/bluesky-298-million-Posts")