Fix empty examples in xtreme dataset for bucc18 config (#4706)
Browse files* fix empty examples in bucc18
Co-authored-by: mustaszewski <mustaszewski@users.noreply.github.com>
* fix yaml tag
Co-authored-by: mustaszewski <mustaszewski@users.noreply.github.com>
Commit from https://github.com/huggingface/datasets/commit/91e797c75f9addd56e2479beaef24e2ffdd54539
- README.md +1 -1
- dataset_infos.json +0 -0
- xtreme.py +7 -20
README.md
CHANGED
@@ -86,7 +86,7 @@ task_ids:
|
|
86 |
- text-classification-other-paraphrase-identification
|
87 |
- text-retrieval-other-parallel-sentence-retrieval
|
88 |
- named-entity-recognition
|
89 |
-
- part-of-speech
|
90 |
paperswithcode_id: xtreme
|
91 |
configs:
|
92 |
- MLQA.ar.ar
|
|
|
86 |
- text-classification-other-paraphrase-identification
|
87 |
- text-retrieval-other-parallel-sentence-retrieval
|
88 |
- named-entity-recognition
|
89 |
+
- part-of-speech
|
90 |
paperswithcode_id: xtreme
|
91 |
configs:
|
92 |
- MLQA.ar.ar
|
dataset_infos.json
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
xtreme.py
CHANGED
@@ -697,29 +697,16 @@ class Xtreme(datasets.GeneratorBasedBuilder):
|
|
697 |
if path.startswith(data_dir):
|
698 |
csv_content = [line.decode("utf-8") for line in file]
|
699 |
if path.endswith("en"):
|
700 |
-
target_sentences = list(csv.reader(csv_content, delimiter="\t"))
|
701 |
elif path.endswith("gold"):
|
702 |
-
source_target_ids = list(csv.reader(csv_content, delimiter="\t"))
|
703 |
else:
|
704 |
-
source_sentences = list(csv.reader(csv_content, delimiter="\t"))
|
705 |
-
|
706 |
-
|
707 |
-
target_id = pair[1]
|
708 |
-
source_sent = ""
|
709 |
-
target_sent = ""
|
710 |
-
for i in range(len(source_sentences)):
|
711 |
-
if source_sentences[i][0] == source_id:
|
712 |
-
source_sent = source_sentences[i][1]
|
713 |
-
source_id = source_sentences[i][0]
|
714 |
-
break
|
715 |
-
for j in range(len(target_sentences)):
|
716 |
-
if target_sentences[j][0] == target_id:
|
717 |
-
target_sent = target_sentences[j][1]
|
718 |
-
target_id = target_sentences[j][0]
|
719 |
-
break
|
720 |
yield id_, {
|
721 |
-
"source_sentence":
|
722 |
-
"target_sentence":
|
723 |
"source_lang": source_id,
|
724 |
"target_lang": target_id,
|
725 |
}
|
|
|
697 |
if path.startswith(data_dir):
|
698 |
csv_content = [line.decode("utf-8") for line in file]
|
699 |
if path.endswith("en"):
|
700 |
+
target_sentences = dict(list(csv.reader(csv_content, delimiter="\t", quotechar=None)))
|
701 |
elif path.endswith("gold"):
|
702 |
+
source_target_ids = list(csv.reader(csv_content, delimiter="\t", quotechar=None))
|
703 |
else:
|
704 |
+
source_sentences = dict(list(csv.reader(csv_content, delimiter="\t", quotechar=None)))
|
705 |
+
|
706 |
+
for id_, (source_id, target_id) in enumerate(source_target_ids):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
707 |
yield id_, {
|
708 |
+
"source_sentence": source_sentences[source_id],
|
709 |
+
"target_sentence": target_sentences[target_id],
|
710 |
"source_lang": source_id,
|
711 |
"target_lang": target_id,
|
712 |
}
|