Updata README.md
Browse files
README.md
CHANGED
@@ -27,3 +27,43 @@ configs:
|
|
27 |
- split: validation
|
28 |
path: data/validation-*
|
29 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
27 |
- split: validation
|
28 |
path: data/validation-*
|
29 |
---
|
30 |
+
|
31 |
+
|
32 |
+
|
33 |
+
ランダムに抽出した1000件の学習データセットへ変換
|
34 |
+
|
35 |
+
```
|
36 |
+
元データ
|
37 |
+
# DatasetDict({
|
38 |
+
# train: Dataset({
|
39 |
+
# features: ['qid', 'question', 'answer_entity', 'label', 'answer_candidates', 'contexts'],
|
40 |
+
# num_rows: 13061
|
41 |
+
# })
|
42 |
+
# validation: Dataset({
|
43 |
+
# features: ['qid', 'question', 'answer_entity', 'label', 'answer_candidates', 'contexts'],
|
44 |
+
# num_rows: 271
|
45 |
+
# })
|
46 |
+
# })
|
47 |
+
|
48 |
+
変換データ
|
49 |
+
# DatasetDict({
|
50 |
+
# train: Dataset({
|
51 |
+
# features: ['question', 'answer_entity', 'label', 'answer_candidates'],
|
52 |
+
# num_rows: 1000
|
53 |
+
# })
|
54 |
+
# validation: Dataset({
|
55 |
+
# features: ['question', 'answer_entity', 'label', 'answer_candidates'],
|
56 |
+
# num_rows: 200
|
57 |
+
# })
|
58 |
+
# })
|
59 |
+
```
|
60 |
+
```
|
61 |
+
データ元のクレジット
|
62 |
+
@InProceedings{Kurihara_nlp2020,
|
63 |
+
author = "鈴木正敏 and 鈴木潤 and 松田耕史 and ⻄田京介 and 井之上直也",
|
64 |
+
title = "JAQKET: クイズを題材にした日本語 QA データセットの構築",
|
65 |
+
booktitle = "言語処理学会第26回年次大会",
|
66 |
+
year = "2020",
|
67 |
+
url = "https://www.anlp.jp/proceedings/annual_meeting/2020/pdf_dir/P2-24.pdf"
|
68 |
+
note= "in Japanese"}'
|
69 |
+
```
|