regenerated readme
Browse files- .gitignore +1 -0
- README.md +6 -3
- create_policies_data.py +28 -0
- data/test.json +0 -0
- data/train.json +0 -0
- data_preparation.py +2 -2
- dataset_info.json +0 -1
- policies.py +9 -7
.gitignore
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
cache_policies
|
README.md
CHANGED
@@ -29,9 +29,12 @@ dataset_info:
|
|
29 |
config_name: plain_text
|
30 |
splits:
|
31 |
- name: train
|
32 |
-
num_bytes:
|
33 |
-
num_examples:
|
34 |
-
|
|
|
|
|
|
|
35 |
dataset_size: 1305279
|
36 |
---
|
37 |
|
|
|
29 |
config_name: plain_text
|
30 |
splits:
|
31 |
- name: train
|
32 |
+
num_bytes: 1174320
|
33 |
+
num_examples: 12132
|
34 |
+
- name: test
|
35 |
+
num_bytes: 130959
|
36 |
+
num_examples: 1348
|
37 |
+
download_size: 3325946
|
38 |
dataset_size: 1305279
|
39 |
---
|
40 |
|
create_policies_data.py
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from datasets import load_dataset, DatasetDict
|
2 |
+
dataset_config = {
|
3 |
+
"LOADING_SCRIPT_FILES": "policies.py",
|
4 |
+
"CONFIG_NAME": "plain_text",
|
5 |
+
"DATA_DIR": "data",
|
6 |
+
"CACHE_DIR": "cache_policies",
|
7 |
+
}
|
8 |
+
|
9 |
+
ds = load_dataset(
|
10 |
+
dataset_config["LOADING_SCRIPT_FILES"],
|
11 |
+
dataset_config["CONFIG_NAME"],
|
12 |
+
data_dir=dataset_config["DATA_DIR"],
|
13 |
+
cache_dir=dataset_config["CACHE_DIR"]
|
14 |
+
)
|
15 |
+
|
16 |
+
# 90% train, 10% test + validation
|
17 |
+
# Split the 10% test + valid in half test, half valid
|
18 |
+
train_testvalid = ds["train"].train_test_split(shuffle=True, test_size=0.1)
|
19 |
+
test_valid = train_testvalid["test"].train_test_split(
|
20 |
+
test_size=0.5) # gather everything into a single DatasetDict
|
21 |
+
ds = DatasetDict({
|
22 |
+
"train": train_testvalid["train"],
|
23 |
+
"test": test_valid["test"],
|
24 |
+
"val": test_valid["train"],
|
25 |
+
}
|
26 |
+
)
|
27 |
+
|
28 |
+
print(ds)
|
data/test.json
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
data/train.json
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
data_preparation.py
CHANGED
@@ -37,7 +37,7 @@ if __name__ == "__main__":
|
|
37 |
train, test = split_data("policies-qa-training.json", "json")
|
38 |
|
39 |
# save the train split
|
40 |
-
save_json(train, "train.json")
|
41 |
|
42 |
# save the test split
|
43 |
-
save_json(test, "test.json")
|
|
|
37 |
train, test = split_data("policies-qa-training.json", "json")
|
38 |
|
39 |
# save the train split
|
40 |
+
save_json({"data": train}, "train.json")
|
41 |
|
42 |
# save the test split
|
43 |
+
save_json({"data": test}, "test.json")
|
dataset_info.json
DELETED
@@ -1 +0,0 @@
|
|
1 |
-
{"description": "Manually generated dataset for policies qa\n", "citation": "", "homepage": "ineract.com", "license": "", "features": {"context": {"dtype": "string", "_type": "Value"}, "question": {"dtype": "string", "_type": "Value"}, "answers": {"feature": {"text": {"dtype": "string", "_type": "Value"}, "answer_start": {"dtype": "int32", "_type": "Value"}}, "_type": "Sequence"}}, "task_templates": [{"task": "question-answering-extractive"}], "builder_name": "policies-qa", "config_name": "plain_text", "version": {"version_str": "1.0.0", "description": "", "major": 1, "minor": 0, "patch": 0}, "splits": {"train": {"name": "train", "num_bytes": 1305279, "num_examples": 13480, "dataset_name": "policies-qa"}}, "download_checksums": {}, "download_size": 0, "dataset_size": 1305279, "size_in_bytes": 1305279}
|
|
|
|
policies.py
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
import json
|
|
|
2 |
|
3 |
import datasets
|
4 |
from datasets.tasks import QuestionAnsweringExtractive
|
@@ -13,9 +14,9 @@ Manually generated dataset for policies qa
|
|
13 |
"""
|
14 |
|
15 |
_URLS = {
|
16 |
-
|
17 |
-
|
18 |
-
|
19 |
|
20 |
|
21 |
class PoliciesQAConfig(datasets.BuilderConfig):
|
@@ -74,10 +75,13 @@ class PoliciesQA(datasets.GeneratorBasedBuilder):
|
|
74 |
|
75 |
return [
|
76 |
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={
|
77 |
-
"filepath": downloaded_files["train"]})
|
|
|
|
|
78 |
]
|
79 |
|
80 |
-
def _generate_examples(self, filepath):
|
|
|
81 |
"""This function returns the examples in the raw (text) form."""
|
82 |
logger.info("generating examples from = %s", filepath)
|
83 |
key = 0
|
@@ -90,8 +94,6 @@ class PoliciesQA(datasets.GeneratorBasedBuilder):
|
|
90 |
for answer_start in policy["answers"]["answer_start"]]
|
91 |
answers = [
|
92 |
answer_text for answer_text in policy["answers"]["text"]]
|
93 |
-
# Features currently used are "context", "question", and "answers".
|
94 |
-
# Others are extracted here for the ease of future expansions.
|
95 |
yield key, {
|
96 |
"context": context,
|
97 |
"question": question,
|
|
|
1 |
import json
|
2 |
+
import os
|
3 |
|
4 |
import datasets
|
5 |
from datasets.tasks import QuestionAnsweringExtractive
|
|
|
14 |
"""
|
15 |
|
16 |
_URLS = {
|
17 |
+
"train": "./data/train.json",
|
18 |
+
"test": "./data/test.json"
|
19 |
+
}
|
20 |
|
21 |
|
22 |
class PoliciesQAConfig(datasets.BuilderConfig):
|
|
|
75 |
|
76 |
return [
|
77 |
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={
|
78 |
+
"filepath": downloaded_files["train"], "split": "train"}),
|
79 |
+
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={
|
80 |
+
"filepath": downloaded_files["test"], "split": "test"})
|
81 |
]
|
82 |
|
83 |
+
def _generate_examples(self, filepath, split):
|
84 |
+
print(os.fspath(filepath))
|
85 |
"""This function returns the examples in the raw (text) form."""
|
86 |
logger.info("generating examples from = %s", filepath)
|
87 |
key = 0
|
|
|
94 |
for answer_start in policy["answers"]["answer_start"]]
|
95 |
answers = [
|
96 |
answer_text for answer_text in policy["answers"]["text"]]
|
|
|
|
|
97 |
yield key, {
|
98 |
"context": context,
|
99 |
"question": question,
|