c4b3882498d4c76ce2d824f458f06481a294e78a,tensorflow_datasets/image/duke_ultrasound.py,DukeUltrasound,_split_generators,#DukeUltrasound#Any#,86

Before Change

        "invivo": "https://storage.googleapis.com/duke-tfds/ultraduke/mark.csv"
    })

    return [
        tfds.core.SplitGenerator(
            name=tfds.Split.TRAIN,
            gen_kwargs={
                "datapath": dl_paths["data"],
                "csvpath": dl_paths["train"]
            },
        ),
        tfds.core.SplitGenerator(
            name=tfds.Split.VALIDATION,
            gen_kwargs={
                "datapath": dl_paths["data"],
                "csvpath": dl_paths["validation"]
            },
        ),
        tfds.core.SplitGenerator(
            name=tfds.Split.TEST,
            gen_kwargs={
                "datapath": dl_paths["data"],
                "csvpath": dl_paths["test"]
            },
        ),
        tfds.core.SplitGenerator(
            name="invivo",
            gen_kwargs={
                "datapath": dl_paths["data"],
                "csvpath": dl_paths["invivo"]
            },
        )
    ]

  def _generate_examples(self, datapath, csvpath):
    reader = csv.DictReader(tf.io.gfile.GFile(csvpath))
    for row in reader:

After Change

  def _split_generators(self, dl_manager):
    dl_paths = dl_manager.download_and_extract({**_DEFAULT_SPLITS, **_DATA_URL})
    splits = [
      tfds.core.SplitGenerator(
          name=name,
          num_shards=10,
          gen_kwargs={
              "datapath": {
                  "mark_data": dl_paths["mark_data"],
                  "phantom_data": dl_paths["phantom_data"]
              },
              "csvpath": dl_paths[name]
          }) for name, path in _DEFAULT_SPLITS.items()
    ]

    for name, csv_path in self.custom_csv_splits.items():
      splits.append(tfds.core.SplitGenerator(
          name=name,
          num_shards=10,
          gen_kwargs={
              "datapath": dl_paths["data"],
              "csvpath": csv_path
      }))

    return splits

  def _generate_examples(self, datapath, csvpath):
    reader = csv.DictReader(tf.io.gfile.GFile(csvpath))
    for row in reader:

Instances