tensor2tensor/data_generators/subject_verb_agreement.py [118:156]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
  @property
  def is_generate_per_split(self):
    # generate_data will shard the data into TRAIN and EVAL for us.
    return True

  @property
  def dataset_splits(self):
    """Splits of data to produce and number of output shards for each.

    This is the setup of the main paper. 10% train/ 90% eval

    Returns:
      A dict containing splits information.

    """
    return [{
        'split': problem.DatasetSplit.TRAIN,
        'shards': 1,
    }, {
        'split': problem.DatasetSplit.EVAL,
        'shards': 1,
    }, {
        'split': problem.DatasetSplit.TEST,
        'shards': 10,
    }]

  @property
  def train_proportion(self):
    # generate_data will shard the data into TRAIN and EVAL for us.
    return 0.09

  @property
  def validation_proportion(self):
    # generate_data will shard the data into TRAIN and EVAL for us.
    return 0.01

  @property
  def vocab_type(self):
    return text_problems.VocabType.TOKEN
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



tensor2tensor/data_generators/subject_verb_agreement.py [223:261]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
  @property
  def is_generate_per_split(self):
    # generate_data will shard the data into TRAIN and EVAL for us.
    return True

  @property
  def dataset_splits(self):
    """Splits of data to produce and number of output shards for each.

    This is the setup of the main paper. 10% train/ 90% eval

    Returns:
      A dict containing splits information.

    """
    return [{
        'split': problem.DatasetSplit.TRAIN,
        'shards': 1,
    }, {
        'split': problem.DatasetSplit.EVAL,
        'shards': 1,
    }, {
        'split': problem.DatasetSplit.TEST,
        'shards': 10,
    }]

  @property
  def train_proportion(self):
    # generate_data will shard the data into TRAIN and EVAL for us.
    return 0.09

  @property
  def validation_proportion(self):
    # generate_data will shard the data into TRAIN and EVAL for us.
    return 0.01

  @property
  def vocab_type(self):
    return text_problems.VocabType.TOKEN
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -