TFDS now supports the Croissant 🥐 format! Read the documentation to know more.

wmt19_translate

Description:

Translate dataset based on the data from statmt.org.

Versions exists for the different years using a combination of multiple data sources. The base wmt_translate allows you to create your own config to choose your own data/language pair by creating a custom tfds.translate.wmt.WmtConfig.

config = tfds.translate.wmt.WmtConfig(
    version="0.0.1",
    language_pair=("fr", "de"),
    subsets={
        tfds.Split.TRAIN: ["commoncrawl_frde"],
        tfds.Split.VALIDATION: ["euelections_dev2019"],
    },
)
builder = tfds.builder("wmt_translate", config=config)

Homepage: http://www.statmt.org/wmt19/translation-task.html
Source code: tfds.translate.Wmt19Translate
Versions:
- 1.0.0 (default): No release notes.
Manual download instructions: This dataset requires you to download the source data manually into download_config.manual_dir (defaults to ~/tensorflow_datasets/downloads/manual/):
Some of the wmt configs here, require a manual download. Please look into wmt.py to see the exact path (and file name) that has to be downloaded.
Figure (tfds.show_examples): Not supported.
Citation:

@ONLINE {wmt19translate,
    author = "Wikimedia Foundation",
    title  = "ACL 2019 Fourth Conference on Machine Translation (WMT19), Shared Task: Machine Translation of News",
    url    = "http://www.statmt.org/wmt19/translation-task.html"
}

wmt19_translate/cs-en (default config)

Config description: WMT 2019 cs-en translation task dataset.
Download size: 1.88 GiB
Dataset size: 3.64 GiB
Auto-cached (documentation): No
Splits:

Split	Examples
`'train'`	20,246,548
`'validation'`	2,983

Feature structure:

Translation({
    'cs': Text(shape=(), dtype=string),
    'en': Text(shape=(), dtype=string),
})

Feature documentation:

Feature	Class	Dtype
	Translation
cs	Text	string
en	Text	string

Supervised keys (See as_supervised doc): ('cs', 'en')
Examples (tfds.as_dataframe):

wmt19_translate/de-en

Config description: WMT 2019 de-en translation task dataset.
Download size: 9.71 GiB
Dataset size: 8.60 GiB
Auto-cached (documentation): No
Splits:

Split	Examples
`'train'`	38,690,334
`'validation'`	2,998

Feature structure:

Translation({
    'de': Text(shape=(), dtype=string),
    'en': Text(shape=(), dtype=string),
})

Feature documentation:

Feature	Class	Dtype
	Translation
de	Text	string
en	Text	string

Supervised keys (See as_supervised doc): ('de', 'en')
Examples (tfds.as_dataframe):

wmt19_translate/fi-en

Config description: WMT 2019 fi-en translation task dataset.
Download size: 959.46 MiB
Dataset size: 1.46 GiB
Auto-cached (documentation): No
Splits:

Split	Examples
`'train'`	6,587,448
`'validation'`	3,000

Feature structure:

Translation({
    'en': Text(shape=(), dtype=string),
    'fi': Text(shape=(), dtype=string),
})

Feature documentation:

Feature	Class	Dtype
	Translation
en	Text	string
fi	Text	string

Supervised keys (See as_supervised doc): ('fi', 'en')
Examples (tfds.as_dataframe):

wmt19_translate/gu-en

Config description: WMT 2019 gu-en translation task dataset.
Download size: 37.03 MiB
Dataset size: 1.55 MiB
Auto-cached (documentation): Yes
Splits:

Split	Examples
`'train'`	11,670
`'validation'`	1,998

Feature structure:

Translation({
    'en': Text(shape=(), dtype=string),
    'gu': Text(shape=(), dtype=string),
})

Feature documentation:

Feature	Class	Dtype
	Translation
en	Text	string
gu	Text	string

Supervised keys (See as_supervised doc): ('gu', 'en')
Examples (tfds.as_dataframe):

wmt19_translate/kk-en

Config description: WMT 2019 kk-en translation task dataset.
Download size: 39.58 MiB
Dataset size: 11.82 MiB
Auto-cached (documentation): Yes
Splits:

Split	Examples
`'train'`	126,583
`'validation'`	2,066

Feature structure:

Translation({
    'en': Text(shape=(), dtype=string),
    'kk': Text(shape=(), dtype=string),
})

Feature documentation:

Feature	Class	Dtype
	Translation
en	Text	string
kk	Text	string

Supervised keys (See as_supervised doc): ('kk', 'en')
Examples (tfds.as_dataframe):

wmt19_translate/lt-en

Config description: WMT 2019 lt-en translation task dataset.
Download size: 392.20 MiB
Dataset size: 537.26 MiB
Auto-cached (documentation): No
Splits:

Split	Examples
`'train'`	2,344,893
`'validation'`	2,000

Feature structure:

Translation({
    'en': Text(shape=(), dtype=string),
    'lt': Text(shape=(), dtype=string),
})

Feature documentation:

Feature	Class	Dtype
	Translation
en	Text	string
lt	Text	string

Supervised keys (See as_supervised doc): ('lt', 'en')
Examples (tfds.as_dataframe):

wmt19_translate/ru-en

Config description: WMT 2019 ru-en translation task dataset.
Download size: 1.57 GiB
Dataset size: 13.95 GiB
Auto-cached (documentation): No
Splits:

Split	Examples
`'train'`	38,492,126
`'validation'`	3,000

Feature structure:

Translation({
    'en': Text(shape=(), dtype=string),
    'ru': Text(shape=(), dtype=string),
})

Feature documentation:

Feature	Class	Dtype
	Translation
en	Text	string
ru	Text	string

Supervised keys (See as_supervised doc): ('ru', 'en')
Examples (tfds.as_dataframe):

wmt19_translate/zh-en

Config description: WMT 2019 zh-en translation task dataset.
Download size: 770.91 MiB
Dataset size: 6.49 GiB
Auto-cached (documentation): No
Splits:

Split	Examples
`'train'`	25,986,436
`'validation'`	3,981

Feature structure:

Translation({
    'en': Text(shape=(), dtype=string),
    'zh': Text(shape=(), dtype=string),
})

Feature documentation:

Feature	Class	Dtype
	Translation
en	Text	string
zh	Text	string

Supervised keys (See as_supervised doc): ('zh', 'en')
Examples (tfds.as_dataframe):

wmt19_translate/fr-de

Config description: WMT 2019 fr-de translation task dataset.
Download size: 722.20 MiB
Dataset size: 2.39 GiB
Auto-cached (documentation): No
Splits:

Split	Examples
`'train'`	9,824,476
`'validation'`	1,512

Feature structure:

Translation({
    'de': Text(shape=(), dtype=string),
    'fr': Text(shape=(), dtype=string),
})

Feature documentation:

Feature	Class	Dtype
	Translation
de	Text	string
fr	Text	string

Supervised keys (See as_supervised doc): ('fr', 'de')
Examples (tfds.as_dataframe):