albertvillanova HF staff commited on
Commit
c499ad4
1 Parent(s): fbeeed5

Move ans2label and id2feature loading to _generate_examples

Browse files
Files changed (1) hide show
  1. gqa-lxmert.py +32 -21
gqa-lxmert.py CHANGED
@@ -88,20 +88,47 @@ class GqaLxmert(datasets.GeneratorBasedBuilder):
88
  def _split_generators(self, dl_manager):
89
  """Returns SplitGenerators."""
90
  dl_dir = dl_manager.download_and_extract(_URLS)
91
- self.ans2label = json.load(open(dl_dir["ans2label"]))
92
- self.id2features = self._load_features(os.path.join(dl_dir["feat"], _FEAT_PATH))
93
-
94
  return [
95
  datasets.SplitGenerator(
96
  name=datasets.Split.TRAIN,
97
- gen_kwargs={"filepath": dl_dir["train"]},
 
 
 
 
98
  ),
99
  datasets.SplitGenerator(
100
  name=datasets.Split.VALIDATION,
101
- gen_kwargs={"filepath": dl_dir["dev"]},
 
 
 
 
102
  ),
103
  ]
104
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
105
  def _load_features(self, filepath):
106
  """Returns a dictionary mapping an image id to the corresponding image's objects features."""
107
  id2features = {}
@@ -126,19 +153,3 @@ class GqaLxmert(datasets.GeneratorBasedBuilder):
126
  normalized_boxes[:, (0, 2)] /= img_w
127
  normalized_boxes[:, (1, 3)] /= img_h
128
  return normalized_boxes
129
-
130
- def _generate_examples(self, filepath):
131
- """ Yields examples as (key, example) tuples."""
132
- with open(filepath, encoding="utf-8") as f:
133
- gqa = json.load(f)
134
- for id_, d in enumerate(gqa):
135
- img_features = self.id2features[d["img_id"]]
136
- label = self.ans2label[next(iter(d["label"]))]
137
- yield id_, {
138
- "question": d["sent"],
139
- "question_id": d["question_id"],
140
- "image_id": d["img_id"],
141
- "features": img_features["features"],
142
- "normalized_boxes": img_features["normalized_boxes"],
143
- "label": label,
144
- }
 
88
  def _split_generators(self, dl_manager):
89
  """Returns SplitGenerators."""
90
  dl_dir = dl_manager.download_and_extract(_URLS)
91
+ features_path = os.path.join(dl_dir["feat"], _FEAT_PATH)
 
 
92
  return [
93
  datasets.SplitGenerator(
94
  name=datasets.Split.TRAIN,
95
+ gen_kwargs={
96
+ "filepath": dl_dir["train"],
97
+ "ans2label_path": dl_dir["ans2label"],
98
+ "features_path": features_path,
99
+ },
100
  ),
101
  datasets.SplitGenerator(
102
  name=datasets.Split.VALIDATION,
103
+ gen_kwargs={
104
+ "filepath": dl_dir["dev"],
105
+ "ans2label_path": dl_dir["ans2label"],
106
+ "features_path": features_path,
107
+ },
108
  ),
109
  ]
110
 
111
+ def _generate_examples(self, filepath, ans2label_path, features_path):
112
+ """ Yields examples as (key, example) tuples."""
113
+ if not hasattr(self, "ans2label"):
114
+ with open(ans2label_path, encoding="utf-8") as f:
115
+ self.ans2label = json.load(f)
116
+ if not hasattr(self, "id2features"):
117
+ self.id2features = self._load_features(features_path)
118
+ with open(filepath, encoding="utf-8") as f:
119
+ gqa = json.load(f)
120
+ for id_, d in enumerate(gqa):
121
+ img_features = self.id2features[d["img_id"]]
122
+ label = self.ans2label[next(iter(d["label"]))]
123
+ yield id_, {
124
+ "question": d["sent"],
125
+ "question_id": d["question_id"],
126
+ "image_id": d["img_id"],
127
+ "features": img_features["features"],
128
+ "normalized_boxes": img_features["normalized_boxes"],
129
+ "label": label,
130
+ }
131
+
132
  def _load_features(self, filepath):
133
  """Returns a dictionary mapping an image id to the corresponding image's objects features."""
134
  id2features = {}
 
153
  normalized_boxes[:, (0, 2)] /= img_w
154
  normalized_boxes[:, (1, 3)] /= img_h
155
  return normalized_boxes