Datasets:
bentrevett
commited on
Commit
•
e9cec36
1
Parent(s):
37e1186
Update README.md
Browse files
README.md
CHANGED
@@ -9,4 +9,29 @@ size_categories:
|
|
9 |
---
|
10 |
# Multi30k
|
11 |
|
12 |
-
This dataset contains the "multi30k" dataset, which is the "task 1" dataset from [here](https://www.statmt.org/wmt16/multimodal-task.html).
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
9 |
---
|
10 |
# Multi30k
|
11 |
|
12 |
+
This dataset contains the "multi30k" dataset, which is the "task 1" dataset from [here](https://www.statmt.org/wmt16/multimodal-task.html).
|
13 |
+
|
14 |
+
Each example consists of an "en" and a "de" feature. "en" is an English sentence, and "de" is the German translation of the English sentence.
|
15 |
+
|
16 |
+
### Data Splits
|
17 |
+
|
18 |
+
The Multi30k dataset has 3 splits: _train_, _validation_, and _test_.
|
19 |
+
|
20 |
+
| Dataset Split | Number of Instances in Split |
|
21 |
+
| ------------- | ------------------------------------------- |
|
22 |
+
| Train | 29,000 |
|
23 |
+
| Validation | 1,014 |
|
24 |
+
| Test | 1,000 |
|
25 |
+
|
26 |
+
### Citation Information
|
27 |
+
|
28 |
+
```
|
29 |
+
@article{elliott-EtAl:2016:VL16,
|
30 |
+
author = {{Elliott}, D. and {Frank}, S. and {Sima'an}, K. and {Specia}, L.},
|
31 |
+
title = {Multi30K: Multilingual English-German Image Descriptions},
|
32 |
+
booktitle = {Proceedings of the 5th Workshop on Vision and Language},
|
33 |
+
year = {2016},
|
34 |
+
pages = {70--74},
|
35 |
+
year = 2016
|
36 |
+
}
|
37 |
+
```
|