holylovenia commited on
Commit
b4bf1a1
1 Parent(s): 08dbc1e

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +62 -1
README.md CHANGED
@@ -3,4 +3,65 @@ tags:
3
  - self-supervised-pretraining
4
  language:
5
  - ind
6
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
3
  - self-supervised-pretraining
4
  language:
5
  - ind
6
+ ---
7
+
8
+ Indo4B is a large-scale Indonesian self-supervised pre-training corpus
9
+ consists of around 3.6B words, with around 250M sentences. The corpus
10
+ covers both formal and colloquial Indonesian sentences compiled from
11
+ 12 sources, of which two cover Indonesian colloquial language, eight
12
+ cover formal Indonesian language, and the rest have a mixed style of
13
+ both colloquial and formal.
14
+
15
+
16
+ ## Dataset Usage
17
+
18
+ Run `pip install nusacrowd` before loading the dataset through HuggingFace's `load_dataset`.
19
+
20
+ ## Citation
21
+
22
+ ``` @inproceedings{wilie-etal-2020-indonlu,
23
+ title = "{I}ndo{NLU}: Benchmark and Resources for Evaluating {I}ndonesian
24
+ Natural Language Understanding",
25
+ author = "Wilie, Bryan and
26
+ Vincentio, Karissa and
27
+ Winata, Genta Indra and
28
+ Cahyawijaya, Samuel and
29
+ Li, Xiaohong and
30
+ Lim, Zhi Yuan and
31
+ Soleman, Sidik and
32
+ Mahendra, Rahmad and
33
+ Fung, Pascale and
34
+ Bahar, Syafri and
35
+ Purwarianti, Ayu",
36
+ booktitle = "Proceedings of the 1st Conference of the Asia-Pacific Chapter of the
37
+ Association for Computational Linguistics and the 10th International Joint
38
+ Conference on Natural Language Processing",
39
+ month = dec,
40
+ year = "2020",
41
+ address = "Suzhou, China",
42
+ publisher = "Association for Computational Linguistics",
43
+ url = "https://aclanthology.org/2020.aacl-main.85",
44
+ pages = "843--857",
45
+ abstract = "Although Indonesian is known to be the fourth most frequently used language
46
+ over the internet, the research progress on this language in natural language processing (NLP)
47
+ is slow-moving due to a lack of available resources. In response, we introduce the first-ever vast
48
+ resource for training, evaluation, and benchmarking on Indonesian natural language understanding
49
+ (IndoNLU) tasks. IndoNLU includes twelve tasks, ranging from single sentence classification to
50
+ pair-sentences sequence labeling with different levels of complexity. The datasets for the tasks
51
+ lie in different domains and styles to ensure task diversity. We also provide a set of Indonesian
52
+ pre-trained models (IndoBERT) trained from a large and clean Indonesian dataset (Indo4B) collected
53
+ from publicly available sources such as social media texts, blogs, news, and websites.
54
+ We release baseline models for all twelve tasks, as well as the framework for benchmark evaluation,
55
+ thus enabling everyone to benchmark their system performances.",
56
+ }
57
+ ```
58
+
59
+ ## License
60
+
61
+ CC0
62
+
63
+ ## Homepage
64
+
65
+ ### NusaCatalogue
66
+
67
+ For easy indexing and metadata: [https://indonlp.github.io/nusa-catalogue](https://indonlp.github.io/nusa-catalogue)