Update files from the datasets library (from 1.1.3)
Browse filesRelease notes: https://github.com/huggingface/datasets/releases/tag/1.1.3
- dataset_infos.json +0 -0
- xtreme.py +50 -10
dataset_infos.json
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
xtreme.py
CHANGED
@@ -361,9 +361,9 @@ _TEXT_FEATURES = {
|
|
361 |
"tatoeba": {"source_sentence": "", "target_sentence": "", "source_lang": "", "target_lang": ""},
|
362 |
"bucc18": {"source_sentence": "", "target_sentence": "", "source_lang": "", "target_lang": ""},
|
363 |
"PAWS-X": {"sentence1": "sentence1", "sentence2": "sentence2"},
|
364 |
-
"udpos": {"
|
365 |
"SQuAD": {"id": "id", "title": "title", "context": "context", "question": "question", "answers": "answers"},
|
366 |
-
"PAN-X": {"
|
367 |
}
|
368 |
_DATA_URLS = {
|
369 |
"tydiqa": "https://storage.googleapis.com/tydiqa/",
|
@@ -451,11 +451,51 @@ class Xtreme(datasets.GeneratorBasedBuilder):
|
|
451 |
if self.config.name == "XNLI":
|
452 |
features["gold_label"] = datasets.Value("string")
|
453 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
454 |
if self.config.name.startswith("PAN-X"):
|
455 |
features = datasets.Features(
|
456 |
{
|
457 |
-
"
|
458 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
459 |
"langs": datasets.Sequence(datasets.Value("string")),
|
460 |
}
|
461 |
)
|
@@ -885,19 +925,19 @@ class Xtreme(datasets.GeneratorBasedBuilder):
|
|
885 |
data = csv.reader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
|
886 |
for id_row, row in enumerate(data):
|
887 |
if len(row) >= 10 and row[1] != "_":
|
888 |
-
yield str(id_file) + "_" + str(id_row), {"
|
889 |
if self.config.name.startswith("PAN-X"):
|
890 |
guid_index = 1
|
891 |
with open(filepath, encoding="utf-8") as f:
|
892 |
-
|
893 |
ner_tags = []
|
894 |
langs = []
|
895 |
for line in f:
|
896 |
if line.startswith("-DOCSTART-") or line == "" or line == "\n":
|
897 |
-
if
|
898 |
-
yield guid_index, {"
|
899 |
guid_index += 1
|
900 |
-
|
901 |
ner_tags = []
|
902 |
langs = []
|
903 |
else:
|
@@ -905,7 +945,7 @@ class Xtreme(datasets.GeneratorBasedBuilder):
|
|
905 |
splits = line.split("\t")
|
906 |
# strip out en: prefix
|
907 |
langs.append(splits[0][:2])
|
908 |
-
|
909 |
if len(splits) > 1:
|
910 |
ner_tags.append(splits[-1].replace("\n", ""))
|
911 |
else:
|
|
|
361 |
"tatoeba": {"source_sentence": "", "target_sentence": "", "source_lang": "", "target_lang": ""},
|
362 |
"bucc18": {"source_sentence": "", "target_sentence": "", "source_lang": "", "target_lang": ""},
|
363 |
"PAWS-X": {"sentence1": "sentence1", "sentence2": "sentence2"},
|
364 |
+
"udpos": {"token": "", "pos_tag": ""},
|
365 |
"SQuAD": {"id": "id", "title": "title", "context": "context", "question": "question", "answers": "answers"},
|
366 |
+
"PAN-X": {"tokens": "", "ner_tags": "", "lang": ""},
|
367 |
}
|
368 |
_DATA_URLS = {
|
369 |
"tydiqa": "https://storage.googleapis.com/tydiqa/",
|
|
|
451 |
if self.config.name == "XNLI":
|
452 |
features["gold_label"] = datasets.Value("string")
|
453 |
|
454 |
+
if self.config.name.startswith("udpos"):
|
455 |
+
features = datasets.Features(
|
456 |
+
{
|
457 |
+
"token": datasets.Value("string"),
|
458 |
+
"pos_tag": datasets.features.ClassLabel(
|
459 |
+
names=[
|
460 |
+
"ADJ",
|
461 |
+
"ADP",
|
462 |
+
"ADV",
|
463 |
+
"AUX",
|
464 |
+
"CCONJ",
|
465 |
+
"DET",
|
466 |
+
"INTJ",
|
467 |
+
"NOUN",
|
468 |
+
"NUM",
|
469 |
+
"PART",
|
470 |
+
"PRON",
|
471 |
+
"PROPN",
|
472 |
+
"PUNCT",
|
473 |
+
"SCONJ",
|
474 |
+
"SYM",
|
475 |
+
"VERB",
|
476 |
+
"X",
|
477 |
+
]
|
478 |
+
),
|
479 |
+
}
|
480 |
+
)
|
481 |
+
|
482 |
if self.config.name.startswith("PAN-X"):
|
483 |
features = datasets.Features(
|
484 |
{
|
485 |
+
"tokens": datasets.Sequence(datasets.Value("string")),
|
486 |
+
"ner_tags": datasets.Sequence(
|
487 |
+
datasets.features.ClassLabel(
|
488 |
+
names=[
|
489 |
+
"O",
|
490 |
+
"B-PER",
|
491 |
+
"I-PER",
|
492 |
+
"B-ORG",
|
493 |
+
"I-ORG",
|
494 |
+
"B-LOC",
|
495 |
+
"I-LOC",
|
496 |
+
]
|
497 |
+
)
|
498 |
+
),
|
499 |
"langs": datasets.Sequence(datasets.Value("string")),
|
500 |
}
|
501 |
)
|
|
|
925 |
data = csv.reader(f, delimiter="\t", quoting=csv.QUOTE_NONE)
|
926 |
for id_row, row in enumerate(data):
|
927 |
if len(row) >= 10 and row[1] != "_":
|
928 |
+
yield str(id_file) + "_" + str(id_row), {"token": row[1], "pos_tag": row[3]}
|
929 |
if self.config.name.startswith("PAN-X"):
|
930 |
guid_index = 1
|
931 |
with open(filepath, encoding="utf-8") as f:
|
932 |
+
tokens = []
|
933 |
ner_tags = []
|
934 |
langs = []
|
935 |
for line in f:
|
936 |
if line.startswith("-DOCSTART-") or line == "" or line == "\n":
|
937 |
+
if tokens:
|
938 |
+
yield guid_index, {"tokens": tokens, "ner_tags": ner_tags, "langs": langs}
|
939 |
guid_index += 1
|
940 |
+
tokens = []
|
941 |
ner_tags = []
|
942 |
langs = []
|
943 |
else:
|
|
|
945 |
splits = line.split("\t")
|
946 |
# strip out en: prefix
|
947 |
langs.append(splits[0][:2])
|
948 |
+
tokens.append(splits[0][3:])
|
949 |
if len(splits) > 1:
|
950 |
ner_tags.append(splits[-1].replace("\n", ""))
|
951 |
else:
|