keras-team/keras · error · ValueError

Cannot enable lora on a layer that isn't yet built.

Error message

Cannot enable lora on a layer that isn't yet built.

What it means

Error "Cannot enable lora on a layer that isn't yet built." thrown in keras-team/keras.

Source

Thrown at keras/src/layers/core/embedding.py:228

        return KerasTensor(
            output_shape, dtype=self.compute_dtype, ragged=ragged
        )

    def enable_lora(
        self,
        rank,
        lora_alpha=None,
        a_initializer="he_uniform",
        b_initializer="zeros",
    ):
        if self.embeddings_constraint:
            raise ValueError(
                "Lora is incompatible with embedding constraints. "
                "In order to enable lora on this layer, remove the "
                "`embeddings_constraint` argument."
            )
        if not self.built:
            raise ValueError(
                "Cannot enable lora on a layer that isn't yet built."
            )
        if self.lora_enabled:
            raise ValueError(
                "lora is already enabled. This can only be done once per layer."
            )
        self._tracker.unlock()

        # LoRA weights should be float32 to avoid the risk of underflow or
        # overflow during fine-tuning.
        # When deploying the model, these weights should be merged with the
        # original embedding while maintaining the original embedding's dtype.
        self.lora_embeddings_a = self.add_weight(
            name="lora_embeddings_a",
            shape=(self.input_dim, rank),
            initializer=initializers.get(a_initializer),
            dtype="float32",
            regularizer=self.embeddings_regularizer,

View on GitHub (pinned to 7a34a03db6)

When it happens

Trigger: Thrown at keras/src/layers/core/embedding.py:228 when the library encounters an invalid state.

Common situations: See trigger scenarios.

AI-assisted analysis of keras-team/keras@7a34a03db6 (2026-08-25). Data as JSON: /api/errors/893a07faea319d85. Report an issue: GitHub.