final touches

P3.py

@@ -57,7 +57,7 @@ def load_cached_task(features_dict, tfrecord):
         feat: _feature_config(**desc) for feat, desc in features_dict.items()
     }
 
-    ds = tf.data.TFRecordDataset(tf.io.gfile.glob([tfrecord])) #TODO handle multiple shards
+    ds = tf.data.TFRecordDataset(tf.io.gfile.glob([tfrecord])) # TODO -> handle multiple shards
     ds = ds.map(
         lambda pb: tf.io.parse_single_example(pb, feature_description),
         num_parallel_calls=tf.data.experimental.AUTOTUNE
@@ -85,16 +85,13 @@ def find_task_splits_and_features_dict():
     """Get the task available (list was pre-computed by `print_data_split_sizes.py`), and get the features for each task."""
     task_splits_and_features = defaultdict(dict)
 
-    data = read_from_url(f"{_HUB_PATH}/data_split_sizes.csv")
-    data = [t.strip() for t in data.splitlines()]
-    data = data[1:]
-    data = [t.split("|") for t in data]
-    data = [(t[0], json.loads(t[1])) for t in data]
-
-    for task_name, split_sizes in data:
-        if "adversarial_qa" not in task_name: #TODO remove
-            continue
+    data_split_sizes = read_from_url(f"{_HUB_PATH}/data_split_sizes.csv")
+    data_split_sizes = [t.strip() for t in data_split_sizes.splitlines()]
+    data_split_sizes = data_split_sizes[1:]
+    data_split_sizes = [t.split("|") for t in data_split_sizes]
+    data_split_sizes = [(t[0], json.loads(t[1])) for t in data_split_sizes]
 
+    for task_name, split_sizes in data_split_sizes:
         for split_name in split_sizes.keys():
             split_info = json.loads(
                 read_from_url(
@@ -102,7 +99,7 @@ def find_task_splits_and_features_dict():
                 )
             )
             features_dict = split_info["features"]
-            assert split_info["num_shards"] == 1 #TODO -> change to multiple shards
+            assert split_info["num_shards"] == 1 # TODO -> handle multiple shards
 
             if not task_splits_and_features[task_name]:
                 task_splits_and_features[task_name] = {
@@ -119,7 +116,7 @@ _TASK_SPLITS_AND_FEATURES_DICT = find_task_splits_and_features_dict()
 _URLs = {
     task_name: {
         split_name: {
-            "tfrecord": f"{_DATA_PATH}/{task_name}/{split_name}.tfrecord-00000-of-00001",
+            "tfrecord": f"{_DATA_PATH}/{task_name}/{split_name}.tfrecord-00000-of-00001", # TODO -> handle multiple shards
         }
         for split_name in splits_and_features_dict["splits"]
     }