tensorflow/models · error · ValueError

Must provide a representative dataset when quantizing the mo

Error message

Must provide a representative dataset when quantizing the model.

What it means

Error "Must provide a representative dataset when quantizing the model." thrown in tensorflow/models.

Source

Thrown at official/projects/edgetpu/vision/serving/export_util.py:213

  dataset.download_and_prepare()
  data = dataset.as_dataset()[quantization_config.dataset_split]
  iterator = data.as_numpy_iterator()
  for _ in range(quantization_config.num_calibration_steps):
    features = next(iterator)
    image = features['image']
    image = preprocess_for_quantization(image, export_config.image_size)
    image = tf.reshape(
        image, [1, export_config.image_size, export_config.image_size, 3])
    yield [image]


def configure_tflite_converter(export_config, converter):
  """Common code for picking up quantization parameters."""
  quantization_config = export_config.quantization_config
  if quantization_config.quantize:
    if (quantization_config.dataset_dir is
        None) and (quantization_config.dataset_name is None):
      raise ValueError(
          'Must provide a representative dataset when quantizing the model.')
    converter.optimizations = [tf.lite.Optimize.DEFAULT]
    converter.target_spec.supported_ops = [
        tf.lite.OpsSet.TFLITE_BUILTINS_INT8
    ]
    converter.inference_input_type = tf.int8
    converter.inference_output_type = tf.int8
    if quantization_config.quantize_less_restrictive:
      converter.target_spec.supported_ops += [
          tf.lite.OpsSet.TFLITE_BUILTINS
      ]
      converter.inference_output_type = tf.float32
    def _representative_dataset_gen():
      return representative_dataset_gen(export_config)

    converter.representative_dataset = _representative_dataset_gen

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/projects/edgetpu/vision/serving/export_util.py:213 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/95404166ecb39747. Report an issue: GitHub.