Sebastian Gehrmann
commited on
Commit
·
eaca996
1
Parent(s):
9563d4b
- common_gen.py +13 -11
common_gen.py
CHANGED
@@ -40,9 +40,7 @@ class CommonGen(datasets.GeneratorBasedBuilder):
|
|
40 |
DEFAULT_CONFIG_NAME = "common_gen"
|
41 |
|
42 |
def _info(self):
|
43 |
-
|
44 |
-
description=_DESCRIPTION,
|
45 |
-
features=datasets.Features(
|
46 |
{
|
47 |
"gem_id": datasets.Value("string"),
|
48 |
"gem_parent_id": datasets.Value("string"),
|
@@ -54,14 +52,18 @@ class CommonGen(datasets.GeneratorBasedBuilder):
|
|
54 |
], # multiple references for validation
|
55 |
}
|
56 |
),
|
57 |
-
|
|
|
|
|
|
|
|
|
58 |
homepage="https://inklab.usc.edu/CommonGen/",
|
59 |
citation=_CITATION,
|
60 |
)
|
61 |
|
62 |
def _split_generators(self, dl_manager):
|
63 |
"""Returns SplitGenerators."""
|
64 |
-
dl_dir = dl_manager.download_and_extract(_URLs
|
65 |
challenge_sets = [
|
66 |
("challenge_train_sample", "train_common_gen_RandomSample500.json"),
|
67 |
(
|
@@ -102,7 +104,7 @@ class CommonGen(datasets.GeneratorBasedBuilder):
|
|
102 |
name=challenge_split,
|
103 |
gen_kwargs={
|
104 |
"filepath": os.path.join(
|
105 |
-
dl_dir["challenge_set"],
|
106 |
),
|
107 |
"split": challenge_split,
|
108 |
},
|
@@ -121,7 +123,7 @@ class CommonGen(datasets.GeneratorBasedBuilder):
|
|
121 |
if len(exple) == 0:
|
122 |
continue
|
123 |
exple["gem_parent_id"] = exple["gem_id"]
|
124 |
-
exple["gem_id"] = f"
|
125 |
yield id_, exple
|
126 |
else:
|
127 |
with open(filepath, encoding="utf-8") as f:
|
@@ -136,8 +138,8 @@ class CommonGen(datasets.GeneratorBasedBuilder):
|
|
136 |
for scene in data["scene"]:
|
137 |
id_ += 1
|
138 |
yield id_, {
|
139 |
-
"gem_id": f"
|
140 |
-
"gem_parent_id": f"
|
141 |
"concept_set_id": i,
|
142 |
"concepts": concepts,
|
143 |
"target": scene,
|
@@ -146,8 +148,8 @@ class CommonGen(datasets.GeneratorBasedBuilder):
|
|
146 |
else:
|
147 |
id_ += 1
|
148 |
yield id_, {
|
149 |
-
"gem_id": f"
|
150 |
-
"gem_parent_id": f"
|
151 |
"concept_set_id": id_,
|
152 |
"concepts": concepts,
|
153 |
"target": "" if split == "test" else data["scene"][0],
|
|
|
40 |
DEFAULT_CONFIG_NAME = "common_gen"
|
41 |
|
42 |
def _info(self):
|
43 |
+
features = datasets.Features(
|
|
|
|
|
44 |
{
|
45 |
"gem_id": datasets.Value("string"),
|
46 |
"gem_parent_id": datasets.Value("string"),
|
|
|
52 |
], # multiple references for validation
|
53 |
}
|
54 |
),
|
55 |
+
|
56 |
+
return datasets.DatasetInfo(
|
57 |
+
description=_DESCRIPTION,
|
58 |
+
features=features,
|
59 |
+
supervised_keys=datasets.info.SupervisedKeysData(input="concepts", output="target"),
|
60 |
homepage="https://inklab.usc.edu/CommonGen/",
|
61 |
citation=_CITATION,
|
62 |
)
|
63 |
|
64 |
def _split_generators(self, dl_manager):
|
65 |
"""Returns SplitGenerators."""
|
66 |
+
dl_dir = dl_manager.download_and_extract(_URLs)
|
67 |
challenge_sets = [
|
68 |
("challenge_train_sample", "train_common_gen_RandomSample500.json"),
|
69 |
(
|
|
|
104 |
name=challenge_split,
|
105 |
gen_kwargs={
|
106 |
"filepath": os.path.join(
|
107 |
+
dl_dir["challenge_set"], "common_gen", filename
|
108 |
),
|
109 |
"split": challenge_split,
|
110 |
},
|
|
|
123 |
if len(exple) == 0:
|
124 |
continue
|
125 |
exple["gem_parent_id"] = exple["gem_id"]
|
126 |
+
exple["gem_id"] = f"common_gen-{split}-{id_}"
|
127 |
yield id_, exple
|
128 |
else:
|
129 |
with open(filepath, encoding="utf-8") as f:
|
|
|
138 |
for scene in data["scene"]:
|
139 |
id_ += 1
|
140 |
yield id_, {
|
141 |
+
"gem_id": f"common_gen-{split}-{id_}",
|
142 |
+
"gem_parent_id": f"common_gen-{split}-{id_}",
|
143 |
"concept_set_id": i,
|
144 |
"concepts": concepts,
|
145 |
"target": scene,
|
|
|
148 |
else:
|
149 |
id_ += 1
|
150 |
yield id_, {
|
151 |
+
"gem_id": f"common_gen-{split}-{id_}",
|
152 |
+
"gem_parent_id": f"common_gen-{split}-{id_}",
|
153 |
"concept_set_id": id_,
|
154 |
"concepts": concepts,
|
155 |
"target": "" if split == "test" else data["scene"][0],
|