albertvillanova HF staff commited on
Commit
32ff1d9
1 Parent(s): 23c403c

Require data manual download (#5)

Browse files

- Require data manual download (03587f84b91b0a23f436b25c0b89f11a82c6bbbe)
- Update dataset card (fa2463d3b273ee6ad2837c61f4fad45abcaf0f99)
- Update metadata JSON (7fe0929062c4fccf8fe9be7443b7992277eba102)

Files changed (3) hide show
  1. README.md +14 -4
  2. dataset_infos.json +1 -1
  3. fquad.py +27 -35
README.md CHANGED
@@ -36,13 +36,13 @@ dataset_info:
36
  dtype: int32
37
  splits:
38
  - name: train
39
- num_bytes: 5910248
40
  num_examples: 4921
41
  - name: validation
42
- num_bytes: 1033253
43
  num_examples: 768
44
- download_size: 3292236
45
- dataset_size: 6943501
46
  ---
47
 
48
  # Dataset Card for FQuAD
@@ -89,6 +89,16 @@ FQuAD contains 25,000+ question and answer pairs.
89
  Finetuning CamemBERT on FQuAD yields a F1 score of 88% and an exact match of 77.9%.
90
  Developped to provide a SQuAD equivalent in the French language. Questions are original and based on high quality Wikipedia articles.
91
 
 
 
 
 
 
 
 
 
 
 
92
  ### Supported Tasks and Leaderboards
93
 
94
  - `closed-domain-qa`, `text-retrieval`: This dataset is intended to be used for `closed-domain-qa`, but can also be used for information retrieval tasks.
36
  dtype: int32
37
  splits:
38
  - name: train
39
+ num_bytes: 5898752
40
  num_examples: 4921
41
  - name: validation
42
+ num_bytes: 1031456
43
  num_examples: 768
44
+ download_size: 0
45
+ dataset_size: 6930208
46
  ---
47
 
48
  # Dataset Card for FQuAD
89
  Finetuning CamemBERT on FQuAD yields a F1 score of 88% and an exact match of 77.9%.
90
  Developped to provide a SQuAD equivalent in the French language. Questions are original and based on high quality Wikipedia articles.
91
 
92
+ Please, note this dataset is licensed for non-commercial purposes and users must agree to the following terms and conditions:
93
+ 1. Use FQuAD only for internal research purposes.
94
+ 2. Not make any copy except a safety one.
95
+ 3. Not redistribute it (or part of it) in any way, even for free.
96
+ 4. Not sell it or use it for any commercial purpose. Contact us for a possible commercial licence.
97
+ 5. Mention the corpus origin and Illuin Technology in all publications about experiments using FQuAD.
98
+ 6. Redistribute to Illuin Technology any improved or enriched version you could make of that corpus.
99
+
100
+ Request manually download of the data from: https://fquad.illuin.tech/
101
+
102
  ### Supported Tasks and Leaderboards
103
 
104
  - `closed-domain-qa`, `text-retrieval`: This dataset is intended to be used for `closed-domain-qa`, but can also be used for information retrieval tasks.
dataset_infos.json CHANGED
@@ -1 +1 @@
1
- {"default": {"description": "FQuAD: French Question Answering Dataset\nWe introduce FQuAD, a native French Question Answering Dataset. FQuAD contains 25,000+ question and answer pairs.\nFinetuning CamemBERT on FQuAD yields a F1 score of 88% and an exact match of 77.9%.\n\n", "citation": "@ARTICLE{2020arXiv200206071\n author = {Martin, d'Hoffschmidt and Maxime, Vidal and\n Wacim, Belblidia and Tom, Brendl\u00e9},\n title = \"{FQuAD: French Question Answering Dataset}\",\n journal = {arXiv e-prints},\n keywords = {Computer Science - Computation and Language},\n year = \"2020\",\n month = \"Feb\",\n eid = {arXiv:2002.06071},\n pages = {arXiv:2002.06071},\narchivePrefix = {arXiv},\n eprint = {2002.06071},\n primaryClass = {cs.CL}\n}\n", "homepage": "https://fquad.illuin.tech/", "license": "", "features": {"context": {"dtype": "string", "id": null, "_type": "Value"}, "questions": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "answers": {"feature": {"texts": {"dtype": "string", "id": null, "_type": "Value"}, "answers_starts": {"dtype": "int32", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "supervised_keys": null, "builder_name": "fquad", "config_name": "default", "version": {"version_str": "0.1.0", "description": null, "datasets_version_to_prepare": null, "major": 0, "minor": 1, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 5910248, "num_examples": 4921, "dataset_name": "fquad"}, "validation": {"name": "validation", "num_bytes": 1033253, "num_examples": 768, "dataset_name": "fquad"}}, "download_checksums": {"https://storage.googleapis.com/illuin/fquad/train.json.zip": {"num_bytes": 2813123, "checksum": "64f0aea68bacee6ffca7f2f7d56a97e504b2ad2abce057ab6c14768a72c09b47"}, "https://storage.googleapis.com/illuin/fquad/valid.json.zip": {"num_bytes": 479113, "checksum": "53eb7f33573f619f6d56f9e656c6ea6030639ebd663bb445b86b999123be1ef3"}}, "download_size": 3292236, "dataset_size": 6943501, "size_in_bytes": 10235737}}
1
+ {"default": {"description": "FQuAD: French Question Answering Dataset\nWe introduce FQuAD, a native French Question Answering Dataset. FQuAD contains 25,000+ question and answer pairs.\nFinetuning CamemBERT on FQuAD yields a F1 score of 88% and an exact match of 77.9%.\n", "citation": "@ARTICLE{2020arXiv200206071\n author = {Martin, d'Hoffschmidt and Maxime, Vidal and\n Wacim, Belblidia and Tom, Brendl\u00e9},\n title = \"{FQuAD: French Question Answering Dataset}\",\n journal = {arXiv e-prints},\n keywords = {Computer Science - Computation and Language},\n year = \"2020\",\n month = \"Feb\",\n eid = {arXiv:2002.06071},\n pages = {arXiv:2002.06071},\narchivePrefix = {arXiv},\n eprint = {2002.06071},\n primaryClass = {cs.CL}\n}\n", "homepage": "https://fquad.illuin.tech/", "license": "", "features": {"context": {"dtype": "string", "id": null, "_type": "Value"}, "questions": {"feature": {"dtype": "string", "id": null, "_type": "Value"}, "length": -1, "id": null, "_type": "Sequence"}, "answers": {"feature": {"texts": {"dtype": "string", "id": null, "_type": "Value"}, "answers_starts": {"dtype": "int32", "id": null, "_type": "Value"}}, "length": -1, "id": null, "_type": "Sequence"}}, "post_processed": null, "supervised_keys": null, "task_templates": null, "builder_name": "fquad", "config_name": "default", "version": {"version_str": "1.0.0", "description": null, "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 5898752, "num_examples": 4921, "dataset_name": "fquad"}, "validation": {"name": "validation", "num_bytes": 1031456, "num_examples": 768, "dataset_name": "fquad"}}, "download_checksums": {}, "download_size": 0, "post_processing_size": null, "dataset_size": 6930208, "size_in_bytes": 6930208}}
fquad.py CHANGED
@@ -1,13 +1,21 @@
1
- """TODO(fquad): Add a description here."""
2
 
3
 
4
  import json
5
  import os
 
6
 
7
  import datasets
8
 
9
 
10
- # TODO(fquad): BibTeX citation
 
 
 
 
 
 
 
11
  _CITATION = """\
12
  @ARTICLE{2020arXiv200206071
13
  author = {Martin, d'Hoffschmidt and Maxime, Vidal and
@@ -25,33 +33,28 @@ archivePrefix = {arXiv},
25
  }
26
  """
27
 
28
- # TODO(fquad):
29
- _DESCRIPTION = """\
30
- FQuAD: French Question Answering Dataset
31
- We introduce FQuAD, a native French Question Answering Dataset. FQuAD contains 25,000+ question and answer pairs.
32
- Finetuning CamemBERT on FQuAD yields a F1 score of 88% and an exact match of 77.9%.
33
 
34
- """
 
35
 
36
- _URL = "https://storage.googleapis.com/illuin/fquad/"
37
- _URLS = {
38
- "train": _URL + "train.json.zip",
39
- "valid": _URL + "valid.json.zip",
40
- }
41
 
 
 
 
 
 
42
 
43
- class Fquad(datasets.GeneratorBasedBuilder):
44
- """TODO(fquad): Short description of my dataset."""
45
 
46
- # TODO(fquad): Set up version.
47
- VERSION = datasets.Version("0.1.0")
 
48
 
49
  def _info(self):
50
- # TODO(fquad): Specifies the datasets.DatasetInfo object
51
  return datasets.DatasetInfo(
52
- # This is the description that will appear on the datasets page.
53
  description=_DESCRIPTION,
54
- # datasets.features.FeatureConnectors
55
  features=datasets.Features(
56
  {
57
  "context": datasets.Value("string"),
@@ -62,39 +65,28 @@ class Fquad(datasets.GeneratorBasedBuilder):
62
  # These are the features of your dataset like images, labels ...
63
  }
64
  ),
65
- # If there's a common (input, target) tuple from the features,
66
- # specify them here. They'll be used if as_supervised=True in
67
- # builder.as_dataset.
68
- supervised_keys=None,
69
- # Homepage of the dataset for documentation
70
- homepage="https://fquad.illuin.tech/",
71
  citation=_CITATION,
72
  )
73
 
74
  def _split_generators(self, dl_manager):
75
  """Returns SplitGenerators."""
76
- # TODO(fquad): Downloads the data and defines the splits
77
- # dl_manager is a datasets.download.DownloadManager that can be used to
78
- # download and extract URLs
79
- download_urls = _URLS
80
- dl_dir = dl_manager.download_and_extract(download_urls)
81
  return [
82
  datasets.SplitGenerator(
83
  name=datasets.Split.TRAIN,
84
  # These kwargs will be passed to _generate_examples
85
- gen_kwargs={"filepath": os.path.join(dl_dir["train"], "train.json")},
86
  ),
87
  datasets.SplitGenerator(
88
  name=datasets.Split.VALIDATION,
89
  # These kwargs will be passed to _generate_examples
90
- gen_kwargs={"filepath": os.path.join(dl_dir["valid"], "valid.json")},
91
  ),
92
  ]
93
 
94
  def _generate_examples(self, filepath):
95
-
96
  """Yields examples."""
97
- # TODO(fquad): Yields (key, example) tuples from the dataset
98
  with open(filepath, encoding="utf-8") as f:
99
  data = json.load(f)
100
  for id1, examples in enumerate(data["data"]):
1
+ """FQuAD dataset."""
2
 
3
 
4
  import json
5
  import os
6
+ from textwrap import dedent
7
 
8
  import datasets
9
 
10
 
11
+ _HOMEPAGE = "https://fquad.illuin.tech/"
12
+
13
+ _DESCRIPTION = """\
14
+ FQuAD: French Question Answering Dataset
15
+ We introduce FQuAD, a native French Question Answering Dataset. FQuAD contains 25,000+ question and answer pairs.
16
+ Finetuning CamemBERT on FQuAD yields a F1 score of 88% and an exact match of 77.9%.
17
+ """
18
+
19
  _CITATION = """\
20
  @ARTICLE{2020arXiv200206071
21
  author = {Martin, d'Hoffschmidt and Maxime, Vidal and
33
  }
34
  """
35
 
 
 
 
 
 
36
 
37
+ class Fquad(datasets.GeneratorBasedBuilder):
38
+ """FQuAD dataset."""
39
 
40
+ VERSION = datasets.Version("1.0.0")
 
 
 
 
41
 
42
+ @property
43
+ def manual_download_instructions(self):
44
+ return dedent("""\
45
+ To access the data for this dataset, you need to request it at:
46
+ https://fquad.illuin.tech/#download
47
 
48
+ Unzip the downloaded file with `unzip download-form-fquad1.0.zip -d <path/to/directory>`, into a destination
49
+ directory <path/to/directory>, which will contain the two data files train.json and valid.json.
50
 
51
+ To load the dataset, pass the full path to the destination directory
52
+ in your call to the loading function: `datasets.load_dataset("fquad", data_dir="<path/to/directory>")`
53
+ """)
54
 
55
  def _info(self):
 
56
  return datasets.DatasetInfo(
 
57
  description=_DESCRIPTION,
 
58
  features=datasets.Features(
59
  {
60
  "context": datasets.Value("string"),
65
  # These are the features of your dataset like images, labels ...
66
  }
67
  ),
68
+ homepage=_HOMEPAGE,
 
 
 
 
 
69
  citation=_CITATION,
70
  )
71
 
72
  def _split_generators(self, dl_manager):
73
  """Returns SplitGenerators."""
74
+ data_dir = os.path.abspath(os.path.expanduser(dl_manager.manual_dir))
 
 
 
 
75
  return [
76
  datasets.SplitGenerator(
77
  name=datasets.Split.TRAIN,
78
  # These kwargs will be passed to _generate_examples
79
+ gen_kwargs={"filepath": os.path.join(data_dir, "train.json")},
80
  ),
81
  datasets.SplitGenerator(
82
  name=datasets.Split.VALIDATION,
83
  # These kwargs will be passed to _generate_examples
84
+ gen_kwargs={"filepath": os.path.join(data_dir, "valid.json")},
85
  ),
86
  ]
87
 
88
  def _generate_examples(self, filepath):
 
89
  """Yields examples."""
 
90
  with open(filepath, encoding="utf-8") as f:
91
  data = json.load(f)
92
  for id1, examples in enumerate(data["data"]):