soldni commited on
Commit
a323d00
1 Parent(s): 06d3711

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +85 -0
README.md CHANGED
@@ -1,3 +1,88 @@
1
  ---
2
  license: odc-by
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: odc-by
3
+ task_categories:
4
+ - text-generation
5
+ language:
6
+ - en
7
+ tags:
8
+ - biology
9
+ - chemistry
10
+ - engineering
11
+ - computer science
12
+ - physics
13
+ - material science
14
+ - math
15
+ - psychology
16
+ - economics
17
+ - political science
18
+ - business
19
+ - geology
20
+ - sociology
21
+ - geography
22
+ - environmental science
23
+ - art
24
+ - history
25
+ - philosophy
26
+ pretty_name: PES2O
27
+ size_categories:
28
+ - 10B<n<100B
29
  ---
30
+
31
+ # PES2O 🌿🎓
32
+
33
+ *Pretraining Efficiently on [S2ORC][2]!*
34
+
35
+ The PES2O dataset is a collection of ~40M creative commmon licensed academic papers,
36
+ cleaned, filtered, and formatted for pre-training of language models. It is derived from
37
+ the [Semantic Scholar Open Research Corpus][2]([Lo et al, 2020][1]), or S2ORC.
38
+
39
+ We release multiple version of PES2O, each with different processing and knowledge cutoff
40
+ date. We recommend you to use the latest version available.
41
+
42
+ ## Document Format
43
+
44
+ TODO
45
+
46
+ ## PES2O V1
47
+
48
+ ### Key Facts
49
+
50
+ - *Knowledge cutoff*: 2023-01-03
51
+ - *Number of documents*: 67.56M
52
+ - *Number of whitespace-separated tokens**: 47,3M
53
+
54
+ ### Processing
55
+
56
+ TODO
57
+
58
+ | Dataset | Split | # Documents | # Words |
59
+ |:-------:|:-------:|:-----------:|:--------------:|
60
+ |s2orc | train | 8,242,162 | 36,088,195,908 |
61
+ |s2orc | valid | 51,323 | 255,139,074 |
62
+ |s2ag | train | 59,382,301 | 11,009,123,378 |
63
+ |s2ag | valid | 111,228 | 24,398,512 |
64
+
65
+
66
+ ## PES2O V2
67
+
68
+
69
+ ### Key Facts
70
+
71
+ - *Knowledge cutoff*: 2023-01-03
72
+ - *Number of documents*: 38.97M
73
+ - *Number of whitespace-separated tokens**: 42,28
74
+
75
+ ### Processing
76
+
77
+ TODO
78
+
79
+ | Dataset | Split | # Documents | # Words |
80
+ |:-------:|:-----:|------------:|---------------:|
81
+ | s2orc | train | 8,242,162 | 36,088,195,908 |
82
+ | s2orc | valid | 51,323 | 255,139,074 |
83
+ | s2ag | train | 30,569,017 | 5,920,099,207 |
84
+ | s2ag | valid | 109,709 | 24,029,459 |
85
+
86
+ [1]: https://aclanthology.org/2020.acl-main.447/
87
+ [2]: https://github.com/allenai/s2orc
88
+