julianschnitzler
commited on
Commit
•
75bce8d
1
Parent(s):
02c0e76
fix dataset loading
Browse files
README.md
CHANGED
@@ -11,8 +11,8 @@ size_categories:
|
|
11 |
- 1K<n<10K
|
12 |
configs:
|
13 |
- config_name: verified
|
14 |
-
data_files: "with_human_verification.json"
|
15 |
default: true
|
16 |
- config_name: unverified
|
17 |
-
data_files: "without_human_verification.json"
|
18 |
---
|
|
|
11 |
- 1K<n<10K
|
12 |
configs:
|
13 |
- config_name: verified
|
14 |
+
data_files: "data/with_human_verification.json"
|
15 |
default: true
|
16 |
- config_name: unverified
|
17 |
+
data_files: "data/without_human_verification.json"
|
18 |
---
|
with_human_verification.json → data/with_human_verification.json
RENAMED
File without changes
|
without_human_verification.json → data/without_human_verification.json
RENAMED
File without changes
|
morehopqa.py
ADDED
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import datasets
|
2 |
+
from datasets import load_dataset
|
3 |
+
import json
|
4 |
+
import os
|
5 |
+
|
6 |
+
class MoreHopQA(datasets.GeneratorBasedBuilder):
|
7 |
+
def _info(self):
|
8 |
+
return datasets.DatasetInfo(
|
9 |
+
features=datasets.Features({
|
10 |
+
"question": datasets.Value("string"),
|
11 |
+
"context": datasets.Sequence({
|
12 |
+
"title": datasets.Value("string"),
|
13 |
+
"paragraphs": datasets.Sequence(datasets.Value("string"))
|
14 |
+
}),
|
15 |
+
"answer": datasets.Value("string"),
|
16 |
+
"previous_question": datasets.Value("string"),
|
17 |
+
"previous_answer": datasets.Value("string"),
|
18 |
+
"question_decomposition": datasets.Sequence({
|
19 |
+
"sub_id": datasets.Value("string"),
|
20 |
+
"question": datasets.Value("string"),
|
21 |
+
"answer": datasets.Value("string"),
|
22 |
+
"paragraph_support_title": datasets.Value("string")
|
23 |
+
}),
|
24 |
+
"question_on_last_hop": datasets.Value("string"),
|
25 |
+
"answer_type": datasets.Value("string"),
|
26 |
+
"previous_answer_type": datasets.Value("string"),
|
27 |
+
"no_of_hops": datasets.Value("int32"),
|
28 |
+
"reasoning_type": datasets.Value("string"),
|
29 |
+
}),
|
30 |
+
)
|
31 |
+
|
32 |
+
def _split_generators(self, dl_manager):
|
33 |
+
"""Returns SplitGenerators."""
|
34 |
+
return [
|
35 |
+
datasets.SplitGenerator(
|
36 |
+
name="verified",
|
37 |
+
gen_kwargs={"filepath": os.path.join("data", "with_human_verification.json"),
|
38 |
+
"default": True}
|
39 |
+
),
|
40 |
+
datasets.SplitGenerator(
|
41 |
+
name="unverified",
|
42 |
+
gen_kwargs={"filepath": os.path.join("data", "without_human_verification.json")}
|
43 |
+
)
|
44 |
+
]
|
45 |
+
|
46 |
+
def _generate_examples(self, filepath):
|
47 |
+
"""Yields examples."""
|
48 |
+
with open(filepath, encoding="utf-8") as f:
|
49 |
+
data = json.load(f)
|
50 |
+
for item in enumerate(data):
|
51 |
+
yield item['_id'], {
|
52 |
+
"question": item["question"],
|
53 |
+
"context": [{
|
54 |
+
"title": subitem[0], # title is the first item in the sublist
|
55 |
+
"content": subitem[1] # paragraphs are the second item
|
56 |
+
} for subitem in item.get("context", [])],
|
57 |
+
"answer": item["answer"],
|
58 |
+
"previous_question": item["previous_question"],
|
59 |
+
"previous_answer": item["previous_answer"],
|
60 |
+
"question_decomposition": [{
|
61 |
+
"sub_id": subitem["sub_id"],
|
62 |
+
"question": subitem["question"],
|
63 |
+
"answer": subitem["answer"],
|
64 |
+
"paragraph_support_title": subitem["paragraph_support_title"]
|
65 |
+
} for subitem in item["question_decomposition"]],
|
66 |
+
"question_on_last_hop": item["ques_on_last_hop"],
|
67 |
+
"answer_type": item["answer_type"],
|
68 |
+
"previous_answer_type":item["previous_answer_type"],
|
69 |
+
"no_of_hops": item["no_of_hops"],
|
70 |
+
"reasoning_type": item["reasoning_type"],
|
71 |
+
}
|