DequantizeLinear - version 23#

This page documents version 23 of operator DequantizeLinear. See DequantizeLinear for the latest version (since version 25).

  • Domain: ai.onnx

  • Since version: 23

The linear dequantization operator. It consumes a quantized tensor, a scale, and a zero point to compute the full-precision tensor. The dequantization formula is y = (x - x_zero_point) * x_scale. x_scale and x_zero_point must have the same shape, determining the quantization’s granularity: a scalar for per-tensor/per-layer quantization, a 1-D tensor for per-axis quantization, or have a rank identical to the input for blocked quantization. See QuantizeLinear for details on quantization granularity.

x_zero_point and x must have the same type. x and y must have the same shape. In the case of dequantizing int32, there’s no zero point (zero point is supposed to be 0). zero-point is usually not used in the case of float8 and 4-bit types quantization, but the dequantization formula remains the same for consistency. The output type is determined by the attribute output_dtype. If output_dtype is not supplied then the output type is the same as x_scale. The output type also determines the precision of the multiplication operation.

Inputs

  • x (T1): N-D quantized input tensor to be de-quantized.

  • x_scale (T2): Scale for input x. For per-tensor/layer dequantization the scale is a scalar, for per per-axis dequantization it is a 1-D Tensor and for blocked dequantization it has the same shape as the input, except for one dimension in which blocking is performed.

  • x_zero_point (T1): Zero point for input x. Shape must match x_scale. It’s optional. Zero point is 0 when it’s not specified.

Outputs

  • y (T3): N-D full precision output tensor. It has the same shape as input x. The data type is specified by the output_dtype attribute or, in its absence, the type of x_scale.

Type Constraints

  • T1: The type of the inputs ‘x_zero_point’ and ‘x’. Allowed types: tensor(float4e2m1), tensor(float8e4m3fn), tensor(float8e4m3fnuz), tensor(float8e5m2), tensor(float8e5m2fnuz), tensor(int16), tensor(int32), tensor(int4), tensor(int8), tensor(uint16), tensor(uint4), tensor(uint8).

  • T2: The type of the input ‘x_scale’. Allowed types: tensor(bfloat16), tensor(float), tensor(float16).

  • T3: The type of the output ‘y’. Allowed types: tensor(bfloat16), tensor(float), tensor(float16).

Differences with previous version (21)#

SchemaDiff: DequantizeLinear (domain 'ai.onnx')

  • old version: 21

  • new version: 23

  • breaking: yes

Breaking reasons:

  • output ‘y’ (changed): type_str changed ‘T2’ -> ‘T3’

Outputs:

  • [BREAKING] changed ‘y’: type_str changed ‘T2’ -> ‘T3’

Type constraints:

  • added ‘T3’: added types: [‘tensor(bfloat16)’, ‘tensor(float)’, ‘tensor(float16)’]

  • changed ‘T1’: added types: [‘tensor(float4e2m1)’]

Documentation:

  • line similarity: 0.73 (+4/-2 lines)

--- DequantizeLinear v21
+++ DequantizeLinear v23
@@ -4,7 +4,9 @@
 must have the same shape, determining the quantization's granularity: a scalar for per-tensor/per-layer quantization,
 a 1-D tensor for per-axis quantization, or have a rank identical to the input for blocked quantization.
 See QuantizeLinear for details on quantization granularity.
+
 `x_zero_point` and `x` must have the same type. `x` and `y` must have the same shape. In the case of dequantizing
 `int32`, there's no zero point (zero point is supposed to be 0).
-`zero-point` is usually not used in the case of float8 types quantization, but the dequantization formula remains the same
-for consistency, and `x_scale` still determines the output type.
+`zero-point` is usually not used in the case of float8 and 4-bit types quantization, but the dequantization formula remains the same
+for consistency. The output type is determined by the attribute `output_dtype`. If `output_dtype` is not supplied then the output type
+is the same as `x_scale`. The output type also determines the precision of the multiplication operation.