kwang2049 -> UKPLab
Browse files
README.md
CHANGED
@@ -183,7 +183,7 @@ One can load the passages like this:
|
|
183 |
from datasets import load_dataset
|
184 |
|
185 |
dataset_name = "ConditionalQA"
|
186 |
-
passages = load_dataset("
|
187 |
for passage in passages:
|
188 |
passage["_id"] # passage id
|
189 |
passage["text"] # passage text
|
@@ -200,7 +200,7 @@ from datasets import load_dataset
|
|
200 |
|
201 |
dataset_name = "ConditionalQA"
|
202 |
passages = load_dataset(
|
203 |
-
"
|
204 |
)
|
205 |
for passage in passages:
|
206 |
passage["_id"] # passage id
|
@@ -218,7 +218,7 @@ The qrels split contains the query relevance annotation, i.e., it contains the r
|
|
218 |
from datasets import load_dataset
|
219 |
|
220 |
dataset_name = "ConditionalQA"
|
221 |
-
qrels = load_dataset("
|
222 |
for qrel in qrels:
|
223 |
qrel["query_id"] # query id (the text is available in ConditionalQA-queries)
|
224 |
qrel["corpus_id"] # passage id
|
@@ -229,7 +229,7 @@ We present the NQ-hard dataset in an extended format of the normal qrels with ad
|
|
229 |
```python
|
230 |
from datasets import load_dataset
|
231 |
|
232 |
-
qrels = load_dataset("
|
233 |
for qrel in qrels:
|
234 |
qrel["query_id"] # query id (the text is available in ConditionalQA-queries)
|
235 |
qrel["corpus_id"] # passage id
|
@@ -277,9 +277,9 @@ class DRAGONPlus(Retriever):
|
|
277 |
|
278 |
|
279 |
# Load data:
|
280 |
-
passages = load_dataset("
|
281 |
-
queries = load_dataset("
|
282 |
-
qrels_rows = load_dataset("
|
283 |
qrels: Dict[str, Dict[str, float]] = {}
|
284 |
for qrel_row in qrels_rows:
|
285 |
qid = qrel_row["query_id"]
|
|
|
183 |
from datasets import load_dataset
|
184 |
|
185 |
dataset_name = "ConditionalQA"
|
186 |
+
passages = load_dataset("UKPLab/dapr", f"{dataset_name}-corpus", split="test")
|
187 |
for passage in passages:
|
188 |
passage["_id"] # passage id
|
189 |
passage["text"] # passage text
|
|
|
200 |
|
201 |
dataset_name = "ConditionalQA"
|
202 |
passages = load_dataset(
|
203 |
+
"UKPLab/dapr", f"{dataset_name}-corpus", split="test", streaming=True
|
204 |
)
|
205 |
for passage in passages:
|
206 |
passage["_id"] # passage id
|
|
|
218 |
from datasets import load_dataset
|
219 |
|
220 |
dataset_name = "ConditionalQA"
|
221 |
+
qrels = load_dataset("UKPLab/dapr", f"{dataset_name}-qrels", split="test")
|
222 |
for qrel in qrels:
|
223 |
qrel["query_id"] # query id (the text is available in ConditionalQA-queries)
|
224 |
qrel["corpus_id"] # passage id
|
|
|
229 |
```python
|
230 |
from datasets import load_dataset
|
231 |
|
232 |
+
qrels = load_dataset("UKPLab/dapr", "nq-hard", split="test")
|
233 |
for qrel in qrels:
|
234 |
qrel["query_id"] # query id (the text is available in ConditionalQA-queries)
|
235 |
qrel["corpus_id"] # passage id
|
|
|
277 |
|
278 |
|
279 |
# Load data:
|
280 |
+
passages = load_dataset("UKPLab/dapr", "ConditionalQA-corpus", split="test")
|
281 |
+
queries = load_dataset("UKPLab/dapr", "ConditionalQA-queries", split="test")
|
282 |
+
qrels_rows = load_dataset("UKPLab/dapr", "ConditionalQA-qrels", split="test")
|
283 |
qrels: Dict[str, Dict[str, float]] = {}
|
284 |
for qrel_row in qrels_rows:
|
285 |
qid = qrel_row["query_id"]
|