tensorflow/models · error · ValueError

Export path is not specified: %s

Error message

Export path is not specified: %s

What it means

Error "Export path is not specified: %s" thrown in tensorflow/models.

Source

Thrown at official/legacy/bert/run_squad_helper.py:465

                                                scores_diff_json)
  else:
    eval_metrics = squad_evaluate_v1_1.evaluate(pred_dataset, all_predictions)
  return eval_metrics


def export_squad(model_export_path, input_meta_data, bert_config):
  """Exports a trained model as a `SavedModel` for inference.

  Args:
    model_export_path: a string specifying the path to the SavedModel directory.
    input_meta_data: dictionary containing meta data about input and model.
    bert_config: Bert configuration file to define core bert layers.

  Raises:
    Export path is not specified, got an empty string or None.
  """
  if not model_export_path:
    raise ValueError('Export path is not specified: %s' % model_export_path)
  # Export uses float32 for now, even if training uses mixed precision.
  tf_keras.mixed_precision.set_global_policy('float32')
  squad_model, _ = bert_models.squad_model(bert_config,
                                           input_meta_data['max_seq_length'])
  model_saving_utils.export_bert_model(
      model_export_path, model=squad_model, checkpoint_dir=FLAGS.model_dir)

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/legacy/bert/run_squad_helper.py:465 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/542709408dc2a44b. Report an issue: GitHub.