keras-team/keras · error · ValueError

Unsupported quantization mode: {self.quantization_mode}

Error message

Unsupported quantization mode: {self.quantization_mode}

What it means

Error "Unsupported quantization mode: {self.quantization_mode}" thrown in keras-team/keras.

Source

Thrown at keras/src/layers/core/embedding.py:754

                )
            else:
                # Sub-channel: grouped dequantization using shared utility
                float_embeddings = dequantize_with_sz_map(
                    ops.cast(unpacked_embeddings, self.compute_dtype),
                    embeddings_scale,
                    self.embeddings_zero,
                    self.g_idx,
                    group_axis=-1,
                )
            quant_range = (-8, 7)
        elif self.quantization_mode == "int8":
            float_embeddings = ops.divide(
                ops.cast(embeddings_value, self.compute_dtype),
                ops.expand_dims(embeddings_scale, axis=-1),
            )
            quant_range = (-127, 127)
        else:
            raise ValueError(
                f"Unsupported quantization mode: {self.quantization_mode}"
            )

        # Merge LoRA weights in float domain.
        lora_delta = (self.lora_alpha / self.lora_rank) * ops.matmul(
            self.lora_embeddings_a, self.lora_embeddings_b
        )
        merged_float_embeddings = ops.add(float_embeddings, lora_delta)

        # Requantize.
        if self.quantization_mode == "int4":
            if block_size is None or block_size == -1:
                # Per-channel re-quantization
                requantized_embeddings, new_scale = quantizers.abs_max_quantize(
                    merged_float_embeddings,
                    axis=-1,
                    value_range=quant_range,
                    dtype="int8",

View on GitHub (pinned to 7a34a03db6)

When it happens

Trigger: Thrown at keras/src/layers/core/embedding.py:754 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/375cade27e5c01bf. Report an issue: GitHub.