LLukas22 commited on
Commit
30ddc17
1 Parent(s): b422407

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +60 -1
README.md CHANGED
@@ -7,4 +7,63 @@ language:
7
  - en
8
  size_categories:
9
  - 100K<n<1M
10
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7
  - en
8
  size_categories:
9
  - 100K<n<1M
10
+ ---
11
+ # Dataset Card for "NLQuAD"
12
+
13
+ ## Table of Contents
14
+ - [Table of Contents](#table-of-contents)
15
+ - [Dataset Description](#dataset-description)
16
+ - [Dataset Summary](#dataset-summary)
17
+ - [Dataset Structure](#dataset-structure)
18
+ - [Data Instances](#data-instances)
19
+ - [Data Fields](#data-fields)
20
+ - [Data Splits](#data-splits)
21
+ - [Additional Information](#additional-information)
22
+ - [Licensing Information](#licensing-information)
23
+
24
+ ## Dataset Description
25
+
26
+ - **Homepage:** [https://towardsdatascience.com/long-form-qa-beyond-eli5-an-updated-dataset-and-approach-319cb841aabb](https://towardsdatascience.com/long-form-qa-beyond-eli5-an-updated-dataset-and-approach-319cb841aabb)
27
+
28
+
29
+ ### Dataset Summary
30
+
31
+ This is a simplified version of [vblagoje's](https://huggingface.co/vblagoje) [lfqa_support_docs](https://huggingface.co/datasets/vblagoje/lfqa_support_docs) and [lfqa](https://huggingface.co/datasets/vblagoje/lfqa) datasets.
32
+ It was generated by me to have a more straight forward way to train Seq2Seq models on context based long form question answering tasks.
33
+
34
+ ## Dataset Structure
35
+
36
+ ### Data Instances
37
+
38
+ An example of 'train' looks as follows.
39
+
40
+ ```json
41
+ {
42
+ "question": "Khashoggi murder: Body 'dissolved in acid'",
43
+ "answer": "2 November 2018",
44
+ "context": [
45
+
46
+ ]
47
+ }
48
+ ```
49
+
50
+ ### Data Fields
51
+
52
+ The data fields are the same among all splits.
53
+
54
+ - `question`: a `string` feature.
55
+ - `answer`: a `string` feature.
56
+ - `context`: a list feature containing `string` features.
57
+
58
+ ### Data Splits
59
+
60
+ | name |train|test|validation|
61
+ |----------|----:|----:|---------:|
62
+ | |10259| 1280| 1280|
63
+
64
+
65
+ ## Additional Information
66
+
67
+ ### Licensing Information
68
+
69
+ This dataset is distributed under the MIT licence.