tensorflow/models · error · ValueError

The BUILDER returns an unexpected instance. The `build_encod

Error message

The BUILDER returns an unexpected instance. The `build_encoder` should returns a tf.Module, tf_keras.Model or tf_keras.layers.Layer. However, we get {encoder.__class__}

What it means

Error "The BUILDER returns an unexpected instance. The `build_encoder` should returns a tf.Module, tf_keras.Model or tf_keras.layers.Layer. However, we get {encoder.__class__}" thrown in tensorflow/models.

Source

Thrown at official/nlp/configs/encoders.py:422

        kernel_initializer=tf_keras.initializers.TruncatedNormal(
            stddev=encoder_cfg.initializer_range),
    )
    kwargs = dict(
        embedding_cfg=embedding_cfg,
        hidden_cfg=hidden_cfg,
        num_hidden_instances=encoder_cfg.num_layers,
        pooled_output_dim=encoder_cfg.hidden_size,
        pooler_layer_initializer=tf_keras.initializers.TruncatedNormal(
            stddev=encoder_cfg.initializer_range),
        return_all_layer_outputs=encoder_cfg.return_all_encoder_outputs,
        dict_outputs=True)
    return encoder_cls(**kwargs)

  if encoder_type == "any":
    encoder = encoder_cfg.BUILDER(encoder_cfg)
    if not isinstance(encoder,
                      (tf.Module, tf_keras.Model, tf_keras.layers.Layer)):
      raise ValueError("The BUILDER returns an unexpected instance. The "
                       "`build_encoder` should returns a tf.Module, "
                       "tf_keras.Model or tf_keras.layers.Layer. However, "
                       f"we get {encoder.__class__}")
    return encoder

  if encoder_type == "mobilebert":
    return networks.MobileBERTEncoder(
        word_vocab_size=encoder_cfg.word_vocab_size,
        word_embed_size=encoder_cfg.word_embed_size,
        type_vocab_size=encoder_cfg.type_vocab_size,
        max_sequence_length=encoder_cfg.max_sequence_length,
        num_blocks=encoder_cfg.num_blocks,
        hidden_size=encoder_cfg.hidden_size,
        num_attention_heads=encoder_cfg.num_attention_heads,
        intermediate_size=encoder_cfg.intermediate_size,
        intermediate_act_fn=encoder_cfg.hidden_activation,
        hidden_dropout_prob=encoder_cfg.hidden_dropout_prob,
        attention_probs_dropout_prob=encoder_cfg.attention_probs_dropout_prob,

View on GitHub (pinned to e006f5f0d5)

When it happens

Trigger: Thrown at official/nlp/configs/encoders.py:422 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/212737f6a03416ec. Report an issue: GitHub.