yonatanbitton
commited on
Commit
•
51e1265
1
Parent(s):
97747b2
Update winogavil.py
Browse files- winogavil.py +12 -6
winogavil.py
CHANGED
@@ -57,12 +57,12 @@ class Winogavil(datasets.GeneratorBasedBuilder):
|
|
57 |
def _info(self):
|
58 |
features = datasets.Features(
|
59 |
{
|
60 |
-
"candidates": [datasets.Value("string")],
|
61 |
-
# "candidates_images": [datasets.Value("string")],
|
62 |
-
"candidates_images": [datasets.Image()],
|
63 |
"cue": datasets.Value("string"),
|
64 |
-
"
|
|
|
65 |
'score_fool_the_ai': datasets.Value("float64"),
|
|
|
|
|
66 |
'num_associations': datasets.Value("int64"),
|
67 |
'num_candidates': datasets.Value("int64"),
|
68 |
'solvers_jaccard_mean': datasets.Value("float64"),
|
@@ -108,13 +108,19 @@ class Winogavil(datasets.GeneratorBasedBuilder):
|
|
108 |
# columns_to_serialize = ['candidates', 'associations']
|
109 |
# for c in columns_to_serialize:
|
110 |
# df[c] = df[c].apply(json.loads)
|
|
|
|
|
111 |
|
112 |
for r_idx, r in df.iterrows():
|
113 |
r_dict = r.to_dict()
|
114 |
r_dict['candidates'] = json.loads(r_dict['candidates'])
|
115 |
candidates_images = [os.path.join(images_dir, "winogavil_images", f"{x}.{self.IMAGE_EXTENSION}") for x in
|
116 |
r_dict['candidates']]
|
117 |
-
r_dict['
|
118 |
r_dict['associations'] = json.loads(r_dict['associations'])
|
|
|
|
|
|
|
|
|
119 |
key = r['ID']
|
120 |
-
yield key,
|
|
|
57 |
def _info(self):
|
58 |
features = datasets.Features(
|
59 |
{
|
|
|
|
|
|
|
60 |
"cue": datasets.Value("string"),
|
61 |
+
"candidate_images": [datasets.Image()],
|
62 |
+
"association_images": [datasets.Image()],
|
63 |
'score_fool_the_ai': datasets.Value("float64"),
|
64 |
+
"candidates": [datasets.Value("string")],
|
65 |
+
"associations": [datasets.Value("string")],
|
66 |
'num_associations': datasets.Value("int64"),
|
67 |
'num_candidates': datasets.Value("int64"),
|
68 |
'solvers_jaccard_mean': datasets.Value("float64"),
|
|
|
108 |
# columns_to_serialize = ['candidates', 'associations']
|
109 |
# for c in columns_to_serialize:
|
110 |
# df[c] = df[c].apply(json.loads)
|
111 |
+
d_keys = ['cue', 'candidate_images', 'association_images', 'score_fool_the_ai', 'candidates', 'associations', 'num_candidates', 'num_associations', 'solvers_jaccard_mean', 'solvers_jaccard_std', 'ID']
|
112 |
+
|
113 |
|
114 |
for r_idx, r in df.iterrows():
|
115 |
r_dict = r.to_dict()
|
116 |
r_dict['candidates'] = json.loads(r_dict['candidates'])
|
117 |
candidates_images = [os.path.join(images_dir, "winogavil_images", f"{x}.{self.IMAGE_EXTENSION}") for x in
|
118 |
r_dict['candidates']]
|
119 |
+
r_dict['candidate_images'] = candidates_images
|
120 |
r_dict['associations'] = json.loads(r_dict['associations'])
|
121 |
+
association_images = [os.path.join(images_dir, "winogavil_images", f"{x}.{self.IMAGE_EXTENSION}") for x in
|
122 |
+
r_dict['associations']]
|
123 |
+
r_dict['association_images'] = association_images
|
124 |
+
relevant_r_dict = {k:v for k,v in r_dict.items() if k in d_keys}
|
125 |
key = r['ID']
|
126 |
+
yield key, relevant_r_dict
|