tensorflow/models · error · ValueError

The seq_bucket_lengths cannot be empty.

Error message

The seq_bucket_lengths cannot be empty.

What it means

Error "The seq_bucket_lengths cannot be empty." thrown in tensorflow/models.

Source

Thrown at official/nlp/data/pretrain_dynamic_dataloader.py:65

@data_loader_factory.register_data_loader_cls(BertPretrainDataConfig)
class PretrainingDynamicDataLoader(pretrain_dataloader.BertPretrainDataLoader):
  """Dataset loader for bert-style pretraining with dynamic sequenece length.

  Bucketizes the input id features by the seq_bucket_lengths and features are
  padded to the bucket boundaries. The mask features are usually short than
  input id features and can also be dynamic. We require the mask feature lengths
  within a bucket must be the same. For example, with [128, 256] buckets,
  the mask features for bucket 128 should always have the length as X and
  features for bucket 256 should always have the length as Y.

  The dataloader does not filter out empty masks. Make sure to handle this
  in the model.
  """

  def __init__(self, params):
    self._params = params
    if len(params.seq_bucket_lengths) < 1:
      raise ValueError('The seq_bucket_lengths cannot be empty.')
    self._seq_bucket_lengths = params.seq_bucket_lengths
    self._seq_bucket_window_scale = params.seq_bucket_window_scale
    self._global_batch_size = params.global_batch_size
    self._use_next_sentence_label = params.use_next_sentence_label
    self._use_position_id = params.use_position_id
    self._drop_remainder = params.drop_remainder
    self._enable_tf_data_service = params.enable_tf_data_service
    self._enable_round_robin_tf_data_service = (
        params.enable_round_robin_tf_data_service)
    self._mask_keys = [
        'masked_lm_positions', 'masked_lm_ids', 'masked_lm_weights'
    ]

  def _decode(self, record: tf.Tensor):
    """Decodes a serialized tf.Example."""
    name_to_features = {
        'input_mask': tf.io.VarLenFeature(tf.int64),
        'masked_lm_positions': tf.io.VarLenFeature(tf.int64),

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/nlp/data/pretrain_dynamic_dataloader.py:65 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of tensorflow/models@e006f5f0d5 (2026-08-24). Data as JSON: /api/errors/4e06529b3d55e089. Report an issue: GitHub.