Datasets:
wmt
/

Modalities:
Text
Formats:
parquet
Libraries:
Datasets
Dask
License:
Quentin Lhoest commited on
Commit
5c0afa8
1 Parent(s): 53cd028

Release: 2.4.0

Browse files

Commit from https://github.com/huggingface/datasets/commit/401d4c4f9b9594cb6527c599c0e7a72ce1a0ea49

Files changed (2) hide show
  1. README.md +17 -8
  2. wmt_utils.py +22 -18
README.md CHANGED
@@ -46,22 +46,31 @@ task_ids: []
46
 
47
  ### Dataset Summary
48
 
49
- Translate dataset based on the data from statmt.org.
50
 
51
- Versions exists for the different years using a combination of multiple data
52
- sources. The base `wmt_translate` allows you to create your own config to choose
53
- your own data/language pair by creating a custom `datasets.translate.wmt.WmtConfig`.
54
 
55
- ```
56
- config = datasets.wmt.WmtConfig(
57
- version="0.0.1",
 
 
 
58
  language_pair=("fr", "de"),
59
  subsets={
60
  datasets.Split.TRAIN: ["commoncrawl_frde"],
61
  datasets.Split.VALIDATION: ["euelections_dev2019"],
62
  },
63
  )
64
- builder = datasets.builder("wmt_translate", config=config)
 
 
 
 
 
 
65
  ```
66
 
67
  ### Supported Tasks and Leaderboards
 
46
 
47
  ### Dataset Summary
48
 
49
+ Translation dataset based on the data from statmt.org.
50
 
51
+ Versions exist for different years using a combination of data
52
+ sources. The base `wmt` allows you to create a custom dataset by choosing
53
+ your own data/language pair. This can be done as follows:
54
 
55
+ ```python
56
+ from datasets import inspect_dataset, load_dataset_builder
57
+
58
+ inspect_dataset("wmt17", "path/to/scripts")
59
+ builder = load_dataset_builder(
60
+ "path/to/scripts/wmt_utils.py",
61
  language_pair=("fr", "de"),
62
  subsets={
63
  datasets.Split.TRAIN: ["commoncrawl_frde"],
64
  datasets.Split.VALIDATION: ["euelections_dev2019"],
65
  },
66
  )
67
+
68
+ # Standard version
69
+ builder.download_and_prepare()
70
+ ds = builder.as_dataset()
71
+
72
+ # Streamable version
73
+ ds = builder.as_streaming_dataset()
74
  ```
75
 
76
  ### Supported Tasks and Leaderboards
wmt_utils.py CHANGED
@@ -25,7 +25,6 @@ import itertools
25
  import os
26
  import re
27
  import xml.etree.cElementTree as ElementTree
28
- from abc import ABC, abstractmethod
29
 
30
  import datasets
31
 
@@ -34,22 +33,31 @@ logger = datasets.logging.get_logger(__name__)
34
 
35
 
36
  _DESCRIPTION = """\
37
- Translate dataset based on the data from statmt.org.
38
 
39
- Versions exists for the different years using a combination of multiple data
40
- sources. The base `wmt_translate` allows you to create your own config to choose
41
- your own data/language pair by creating a custom `datasets.translate.wmt.WmtConfig`.
42
 
43
- ```
44
- config = datasets.wmt.WmtConfig(
45
- version="0.0.1",
 
 
 
46
  language_pair=("fr", "de"),
47
  subsets={
48
  datasets.Split.TRAIN: ["commoncrawl_frde"],
49
  datasets.Split.VALIDATION: ["euelections_dev2019"],
50
  },
51
  )
52
- builder = datasets.builder("wmt_translate", config=config)
 
 
 
 
 
 
53
  ```
54
 
55
  """
@@ -662,20 +670,15 @@ class WmtConfig(datasets.BuilderConfig):
662
  # +++++++++++++++++++++
663
 
664
 
665
- class Wmt(ABC, datasets.GeneratorBasedBuilder):
666
  """WMT translation dataset."""
667
 
 
 
668
  def __init__(self, *args, **kwargs):
669
- if type(self) == Wmt and "config" not in kwargs: # pylint: disable=unidiomatic-typecheck
670
- raise ValueError(
671
- "The raw `wmt_translate` can only be instantiated with the config "
672
- "kwargs. You may want to use one of the `wmtYY_translate` "
673
- "implementation instead to get the WMT dataset for a specific year."
674
- )
675
  super(Wmt, self).__init__(*args, **kwargs)
676
 
677
  @property
678
- @abstractmethod
679
  def _subsets(self):
680
  """Subsets that make up each split of the dataset."""
681
  raise NotImplementedError("This is a abstract method")
@@ -685,7 +688,8 @@ class Wmt(ABC, datasets.GeneratorBasedBuilder):
685
  """Subsets that make up each split of the dataset for the language pair."""
686
  source, target = self.config.language_pair
687
  filtered_subsets = {}
688
- for split, ss_names in self._subsets.items():
 
689
  filtered_subsets[split] = []
690
  for ss_name in ss_names:
691
  dataset = DATASET_MAP[ss_name]
 
25
  import os
26
  import re
27
  import xml.etree.cElementTree as ElementTree
 
28
 
29
  import datasets
30
 
 
33
 
34
 
35
  _DESCRIPTION = """\
36
+ Translation dataset based on the data from statmt.org.
37
 
38
+ Versions exist for different years using a combination of data
39
+ sources. The base `wmt` allows you to create a custom dataset by choosing
40
+ your own data/language pair. This can be done as follows:
41
 
42
+ ```python
43
+ from datasets import inspect_dataset, load_dataset_builder
44
+
45
+ inspect_dataset("wmt17", "path/to/scripts")
46
+ builder = load_dataset_builder(
47
+ "path/to/scripts/wmt_utils.py",
48
  language_pair=("fr", "de"),
49
  subsets={
50
  datasets.Split.TRAIN: ["commoncrawl_frde"],
51
  datasets.Split.VALIDATION: ["euelections_dev2019"],
52
  },
53
  )
54
+
55
+ # Standard version
56
+ builder.download_and_prepare()
57
+ ds = builder.as_dataset()
58
+
59
+ # Streamable version
60
+ ds = builder.as_streaming_dataset()
61
  ```
62
 
63
  """
 
670
  # +++++++++++++++++++++
671
 
672
 
673
+ class Wmt(datasets.GeneratorBasedBuilder):
674
  """WMT translation dataset."""
675
 
676
+ BUILDER_CONFIG_CLASS = WmtConfig
677
+
678
  def __init__(self, *args, **kwargs):
 
 
 
 
 
 
679
  super(Wmt, self).__init__(*args, **kwargs)
680
 
681
  @property
 
682
  def _subsets(self):
683
  """Subsets that make up each split of the dataset."""
684
  raise NotImplementedError("This is a abstract method")
 
688
  """Subsets that make up each split of the dataset for the language pair."""
689
  source, target = self.config.language_pair
690
  filtered_subsets = {}
691
+ subsets = self._subsets if self.config.subsets is None else self.config.subsets
692
+ for split, ss_names in subsets.items():
693
  filtered_subsets[split] = []
694
  for ss_name in ss_names:
695
  dataset = DATASET_MAP[ss_name]