tensor.dynamic_quantize_linear.md

tensor.dynamic_quantize_linear

fn dynamic_quantize_linear(self: @Tensor<T>) -> (Tensor::<Q>, Tensor<T>, Tensor<T>);

Quantizes a Tensor using dynamic linear quantization.

The dynamic linear quantization operator. It consumes a high precision tensor to compute the low precision / quantized tensor dynamicly. Right now only uint8 is supported, it saturates to [0, 255].

Args

• self(@Tensor<T>) - The input tensor.

Returns

A new Tensor<Q> with the same shape as the input tensor, containing the quantized values.

• y_scale(@Tensor<T>) - Scale for doing quantization to get y.
• y_zero_point(@Tensor<T>) - Zero point for doing quantization to get y.

Type Constraints

• T in (Tensor<FP>, Tensor<i8>, Tensor<i32>, tensor<u32>)
• Q in (Tensor<i32>)- Constrain y to 8-bit unsigned integer tensor.

Examples

use array::{ArrayTrait, SpanTrait};

use orion::operators::tensor::{TensorTrait, Tensor, I8Tensor, I32Tensor};
use orion::numbers::{u8, i32, IntegerTrait};

fn dynamic_quantize_linear_example() -> (Tensor<u32>, Tensor<FP16x16>, Tensor<FP16x16>) {
    // We instantiate a 1D Tensor here.
    let x = TensorTrait::<FP16x16>::new(
        shape: array![6].span(),
        data: array![
            FP16x16 { mag: 10945, sign: false },
            FP16x16 { mag: 190054, sign: false },
            FP16x16 { mag: 196608, sign: false },
            FP16x16 { mag: 229376, sign: false },
            FP16x16 { mag: 196608, sign: true },
            FP16x16 { mag: 229376, sign: true },
        ]
            .span(),
    );

    return x.dynamic_quantize_linear();
}
>>> ([133, 233, 236, 255, -18, -0], [0.02745], [128]