Datasets:

Skywork
/

SkyPile-150B

Tasks:

Text Generation

Modalities:

Formats:

Languages:

Size:

ArXiv:

Tags:

language-modeling

Libraries:

Dataset card Viewer Files Files and versions Community

520jefferson commited on Oct 25, 2023

Commit

042acd6

•

1 Parent(s): 6e7d165

Update README.md

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -12,7 +12,7 @@ size_categories:
 - 100B<n<1T
 ---
-#SkyPile-150B
 ## Dataset Summary
 SkyPile-150B is created to serve as a large-scale Chinese dataset for pre-training of large language models. It is taken from publicly available Chinese Internet web pages. Through strict filtering and extensive deduplication then sensitive data filtering, At the same time, fastText and Bert are used to filter low-quality data.

 - 100B<n<1T
 ---
+# SkyPile-150B
 ## Dataset Summary
 SkyPile-150B is created to serve as a large-scale Chinese dataset for pre-training of large language models. It is taken from publicly available Chinese Internet web pages. Through strict filtering and extensive deduplication then sensitive data filtering, At the same time, fastText and Bert are used to filter low-quality data.