adding changes
Browse files- nomiracl.py +12 -11
nomiracl.py
CHANGED
@@ -119,6 +119,7 @@ class NoMIRACL(datasets.GeneratorBasedBuilder):
|
|
119 |
NoMIRACLConfig(name="spanish", description="Spanish NoMIRACL dataset"),
|
120 |
NoMIRACLConfig(name="telugu", description="Telugu NoMIRACL dataset"),
|
121 |
NoMIRACLConfig(name="yoruba", description="Yoruba NoMIRACL dataset"),
|
|
|
122 |
]
|
123 |
|
124 |
def _info(self):
|
@@ -128,13 +129,13 @@ class NoMIRACL(datasets.GeneratorBasedBuilder):
|
|
128 |
'query_id': datasets.Value('string'),
|
129 |
'query': datasets.Value('string'),
|
130 |
'positive_passages': [{
|
131 |
-
'docid': datasets.Value('string'),
|
132 |
-
'text': datasets.Value('string'),
|
133 |
'title': datasets.Value('string')
|
134 |
}],
|
135 |
'negative_passages': [{
|
136 |
'docid': datasets.Value('string'),
|
137 |
-
'text': datasets.Value('string'),
|
138 |
'title': datasets.Value('string'),
|
139 |
}],
|
140 |
}),
|
@@ -146,14 +147,14 @@ class NoMIRACL(datasets.GeneratorBasedBuilder):
|
|
146 |
|
147 |
def _split_generators(self, dl_manager):
|
148 |
|
149 |
-
# Download downloaded_files
|
150 |
downloaded_files = dl_manager.download_and_extract({
|
151 |
"corpus": self.config.data_root_url + "/corpus.jsonl.gz",
|
152 |
-
"dev": {"qrels": {"relevant": self.config.data_root_url + "/qrels/dev.relevant.tsv",
|
153 |
"non_relevant": self.config.data_root_url + "/qrels/dev.non_relevant.tsv"},
|
154 |
"topics": {"relevant": self.config.data_root_url + "/topics/dev.relevant.tsv",
|
155 |
"non_relevant": self.config.data_root_url + "/topics/dev.non_relevant.tsv"}},
|
156 |
-
"test": {"qrels": {"relevant": self.config.data_root_url + "/qrels/test.relevant.tsv",
|
157 |
"non_relevant": self.config.data_root_url + "/qrels/test.non_relevant.tsv"},
|
158 |
"topics": {"relevant": self.config.data_root_url + "/topics/test.relevant.tsv",
|
159 |
"non_relevant": self.config.data_root_url + "/topics/test.non_relevant.tsv"}},
|
@@ -195,9 +196,9 @@ class NoMIRACL(datasets.GeneratorBasedBuilder):
|
|
195 |
]
|
196 |
|
197 |
return splits
|
198 |
-
|
199 |
def _generate_examples(self, corpus_path, qrels_path, topics_path):
|
200 |
-
|
201 |
corpus = load_corpus(corpus_path)
|
202 |
qrels = load_qrels(qrels_path)
|
203 |
topics = load_topics(topics_path)
|
@@ -206,15 +207,15 @@ class NoMIRACL(datasets.GeneratorBasedBuilder):
|
|
206 |
data = {}
|
207 |
data['query_id'] = qid
|
208 |
data['query'] = topics[qid]
|
209 |
-
|
210 |
pos_docids = [docid for docid, rel in qrels[qid].items() if rel == 1] if qrels is not None else []
|
211 |
neg_docids = [docid for docid, rel in qrels[qid].items() if rel == 0] if qrels is not None else []
|
212 |
data['positive_passages'] = [{
|
213 |
-
'docid': docid,
|
214 |
**corpus[docid]
|
215 |
} for docid in pos_docids if docid in corpus]
|
216 |
data['negative_passages'] = [{
|
217 |
-
'docid': docid,
|
218 |
**corpus[docid]
|
219 |
} for docid in neg_docids if docid in corpus]
|
220 |
yield qid, data
|
|
|
119 |
NoMIRACLConfig(name="spanish", description="Spanish NoMIRACL dataset"),
|
120 |
NoMIRACLConfig(name="telugu", description="Telugu NoMIRACL dataset"),
|
121 |
NoMIRACLConfig(name="yoruba", description="Yoruba NoMIRACL dataset"),
|
122 |
+
NoMIRACLConfig(name="portuguese", description="Yoruba NoMIRACL dataset"),
|
123 |
]
|
124 |
|
125 |
def _info(self):
|
|
|
129 |
'query_id': datasets.Value('string'),
|
130 |
'query': datasets.Value('string'),
|
131 |
'positive_passages': [{
|
132 |
+
'docid': datasets.Value('string'),
|
133 |
+
'text': datasets.Value('string'),
|
134 |
'title': datasets.Value('string')
|
135 |
}],
|
136 |
'negative_passages': [{
|
137 |
'docid': datasets.Value('string'),
|
138 |
+
'text': datasets.Value('string'),
|
139 |
'title': datasets.Value('string'),
|
140 |
}],
|
141 |
}),
|
|
|
147 |
|
148 |
def _split_generators(self, dl_manager):
|
149 |
|
150 |
+
# Download downloaded_files
|
151 |
downloaded_files = dl_manager.download_and_extract({
|
152 |
"corpus": self.config.data_root_url + "/corpus.jsonl.gz",
|
153 |
+
"dev": {"qrels": {"relevant": self.config.data_root_url + "/qrels/dev.relevant.tsv",
|
154 |
"non_relevant": self.config.data_root_url + "/qrels/dev.non_relevant.tsv"},
|
155 |
"topics": {"relevant": self.config.data_root_url + "/topics/dev.relevant.tsv",
|
156 |
"non_relevant": self.config.data_root_url + "/topics/dev.non_relevant.tsv"}},
|
157 |
+
"test": {"qrels": {"relevant": self.config.data_root_url + "/qrels/test.relevant.tsv",
|
158 |
"non_relevant": self.config.data_root_url + "/qrels/test.non_relevant.tsv"},
|
159 |
"topics": {"relevant": self.config.data_root_url + "/topics/test.relevant.tsv",
|
160 |
"non_relevant": self.config.data_root_url + "/topics/test.non_relevant.tsv"}},
|
|
|
196 |
]
|
197 |
|
198 |
return splits
|
199 |
+
|
200 |
def _generate_examples(self, corpus_path, qrels_path, topics_path):
|
201 |
+
|
202 |
corpus = load_corpus(corpus_path)
|
203 |
qrels = load_qrels(qrels_path)
|
204 |
topics = load_topics(topics_path)
|
|
|
207 |
data = {}
|
208 |
data['query_id'] = qid
|
209 |
data['query'] = topics[qid]
|
210 |
+
|
211 |
pos_docids = [docid for docid, rel in qrels[qid].items() if rel == 1] if qrels is not None else []
|
212 |
neg_docids = [docid for docid, rel in qrels[qid].items() if rel == 0] if qrels is not None else []
|
213 |
data['positive_passages'] = [{
|
214 |
+
'docid': docid,
|
215 |
**corpus[docid]
|
216 |
} for docid in pos_docids if docid in corpus]
|
217 |
data['negative_passages'] = [{
|
218 |
+
'docid': docid,
|
219 |
**corpus[docid]
|
220 |
} for docid in neg_docids if docid in corpus]
|
221 |
yield qid, data
|