Skip to content

litert

LiteRT Utility API

This module provides utility functions for working with the LiteRT schema.

Classes:

  • BuiltinOperatorMap –

    Mapping of builtin operator codes to their names.

  • BuiltinOptionsMap –

    Mapping of builtin options codes to their names.

Functions:

Copyright 2025 Ambiq. All Rights Reserved.

Functions

quantize_multiplier 

quantize_multiplier(real_multiplier: float) -> tuple[int, int]

Convert a floating point multiplier into a fixed-point multiplier and shift.

Given a real multiplier, this function computes a pair (quantized_multiplier, shift) such that the fixed-point multiplication approximates the real multiplier:

real_multiplier ~ quantized_multiplier / (2^(31 - shift))

This is typically used in quantized inference to convert floating-point scaling factors to fixed-point representation.

Parameters:

  • real_multiplier

    (float) –

    The floating-point multiplier.

Returns:

  • tuple ( tuple[int, int] ) –

    (quantized_multiplier, shift)

compute_fused_activation_range 

compute_fused_activation_range(activation_function: ActivationFunctionType, output_zero_point: int, output_scale: float, qmin: int = -128, qmax: int = 127) -> tuple[int, int]

Compute the activation min and max values based on the fused activation function.

Parameters:

  • activation_function

    (ActivationFunctionType) –

    The activation function type.

  • output_zero_point

    (int) –

    The zero point for the output tensor.

  • output_scale

    (float) –

    The scale for the output tensor.

  • qmin

    (int, default: -128 ) –

    Minimum quantized value. Defaults to -128.

  • qmax

    (int, default: 127 ) –

    Maximum quantized value. Defaults to 127.

Returns:

  • tuple[int, int] –

    tuple[int, int]: The activation min and max values.

get_builtin_code 

get_builtin_code(opcode: OperatorCodeT | OperatorCode) -> int

Return the builtin code of the given operator code.

This function serves as a workaround for older versions of LiteRT that do not support/populate the BuiltinCode property.

Parameters:

  • opcode

    (OperatorCodeT | OperatorCode) –

    Operator code.

Returns:

  • int ( int ) –

    Builtin code.

compute_padding_with_offset 

compute_padding_with_offset(in_size: int, filter_size: int, stride: int, dilation_rate: int = 1, padding_type: int = litert.Padding.SAME) -> tuple[int, int]

Computes a single padding value (top/left) for one dimension with offset.

Parameters:

  • in_size

    (int) –

    Input dimension (height or width).

  • filter_size

    (int) –

    Kernel dimension (height or width).

  • stride

    (int) –

    Stride for that dimension.

  • dilation_rate

    (int, default: 1 ) –

    Dilation rate for that dimension (default is 1).

  • padding_type

    (int, default: SAME ) –

    0 for SAME padding, 1 for VALID padding (default is SAME).

Returns:

  • int ( tuple[int, int] ) –

    The computed top or left padding value.

compute_padding 

compute_padding(in_size: int, filter_size: int, stride: int, dilation_rate: int = 1, padding_type: int = litert.Padding.SAME) -> int

Computes a single padding value (top/left) for one dimension.

Parameters:

  • in_size

    (int) –

    Input dimension (height or width).

  • filter_size

    (int) –

    Kernel dimension (height or width).

  • stride

    (int) –

    Stride for that dimension.

  • dilation_rate

    (int, default: 1 ) –

    Dilation rate for that dimension (default is 1).

  • padding_type

    (int, default: SAME ) –

    0 for SAME padding, 1 for VALID padding (default is SAME).

Returns:

  • int ( int ) –

    The computed top or left padding value.

compute_tensor_bytes 

compute_tensor_bytes(tensor: TensorT) -> int

Compute the number of bytes for a tensor based on its type and shape.

Parameters:

  • tensor

    (TensorT) –

    Tensor object.

Returns:

  • int ( int ) –

    Number of bytes for the tensor.

compute_tensor_ctype 

compute_tensor_ctype(tensor: TensorT) -> str

Compute the C type for a tensor based on its type.

Parameters:

  • tensor

    (TensorT) –

    Tensor object.

Returns:

  • str ( str ) –

    C type of the tensor.

get_tensor_dtype 

get_tensor_dtype(tensor: TensorT) -> np.dtype

Get the numpy dtype of a tensor.

Parameters:

  • tensor

    (TensorT) –

    Tensor object.

Returns:

  • dtype –

    np.dtype: Numpy dtype of the tensor.

get_expanded_dims_shape 

get_expanded_dims_shape(x: Tensor, num_dims: int = 4, pre: bool = True) -> tuple[int, ...]

Expand the shape of a tensor to a specified number of dimensions.

Parameters:

  • x

    (Tensor) –

    Input tensor whose shape is to be expanded.

  • num_dims

    (int, default: 4 ) –

    Number of dimensions to expand to. Defaults to 4.

Returns:

  • tuple[int, ...] –

    tuple[int, ...]: Expanded shape of the tensor.

is_constant_tensor 

is_constant_tensor(tensor: TensorT, model: ModelT, subgraph: SubGraphT) -> bool

Check if a tensor is a constant tensor in the LiteRT model.

Parameters:

  • tensor

    (TensorT) –

    Tensor object.

  • model

    (ModelT) –

    LiteRT model object.

  • subgraph

    (SubGraphT) –

    Subgraph object from LiteRT model.

Returns:

  • bool ( bool ) –

    True if the tensor is constant, False otherwise.

load_tensor_data 

load_tensor_data(tensor: TensorT, model: ModelT) -> np.ndarray

Load the data of a tensor from the LiteRT model.

Parameters:

  • tensor

    (TensorT) –

    Tensor object.

  • model

    (ModelT) –

    LiteRT model object.

Returns:

  • ndarray –

    np.ndarray: Numpy array containing the tensor data.