Datasets:

Modalities:
Text
Formats:
json
Libraries:
Datasets
Dask
License:
dododododo commited on
Commit
fefd697
·
verified ·
1 Parent(s): 915cc7b

Create README.md

Browse files
Files changed (1) hide show
  1. README.md +38 -0
README.md ADDED
@@ -0,0 +1,38 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Matrix
2
+
3
+ An open-source pretraining dataset with a scale of 4690 billion tokens, used to train neo models.
4
+
5
+ ## Dataset Composition
6
+
7
+ The dataset consists of several components, each originating from different sources and serving various purposes in language modeling and processing. Below is a brief overview of each component:
8
+
9
+ <p>
10
+ <img src="https://cdn-uploads.huggingface.co/production/uploads/654907a4a1faff97850c4eff/1FWMF_t_Mhy0UQmu65Bb1.png" style="float: right; width: 700px; margin-left: 10px;">
11
+ <strong>Common Crawl</strong><br>
12
+ Extracts from the Common Crawl project, featuring a rich diversity of internet text including websites, blogs, news articles, and more.<br>
13
+
14
+ <strong>Code</strong><br>
15
+ A collection of coding-related data.<be>
16
+
17
+ <strong>Paper</strong><br>
18
+ Consists of academic and research papers covering a broad spectrum of disciplines, offering technical and domain-specific language.<br>
19
+
20
+ <strong>Book</strong><br>
21
+ Comprises texts from a range of published books, encompassing literature, non-fiction, textbooks, and more.<br>
22
+
23
+ <strong>Instruction</strong><br>
24
+ Features a collection of texts primarily in a Q&A format.<be>
25
+
26
+ <strong>Exam</strong><br>
27
+ Contains various educational materials and assessments used in academic examinations.<be>
28
+
29
+ <strong>News</strong><br>
30
+ A collection of texts from various journalistic sources, reporting on current events and news stories.<br>
31
+
32
+ <strong>Wiki</strong><br>
33
+ Articles from various encyclopedic sources, not limited to Wikipedia, covering a wide array of topics and information.<br>
34
+
35
+ <strong>Patent</strong><br>
36
+ Includes texts from patent documents, providing detailed descriptions of inventions and their applications.<br>
37
+
38
+ </p>