softmax

Classes

SoftmaxOperator

SoftmaxOperator(ident: int, op: OperatorT, model: ModelT, subgraph: SubGraphT, prefix: str = 'aot', attributes: dict[str, str] = {})

SOFTMAX operator.

This operator computes the softmax activation function on the input tensor.

Parameters:

• ident

  (int) –

  Operator identifier.

• op

  (OperatorT) –

  Operator object from LiteRT model.

• model

  (ModelT) –

  LiteRT model object.

• subgraph

  (SubGraphT) –

  Subgraph object from LiteRT model.

• prefix

  (str, default: 'aot' ) –

  Prefix for generated files. Defaults to "aot".

• attributes

  (dict[str, str], default: {} ) –

  Attributes for template values. Defaults to {}.

Attributes

name property

name: str

Return the operator name.

input_indices property

input_indices: list[int]

Return the input tensor indices.

output_indices property

output_indices: list[int]

Return the output tensor indices.

input_tensors property

input_tensors: list[TensorT]

Return the input tensors.

output_tensors property

output_tensors: list[TensorT]

Return the output tensors.

Functions

compute_values

compute_values() -> dict[str, str]

Compute the values for the operator.

Returns:

• dict[str, str] –

  dict[str, str]: Computed values for the operator.

generate_source_code

generate_source_code(save_path: Path)

Generate the source code for the operator.

Parameters:

• save_path

  (Path) –

  Path to save the generated code.

Functions

preprocess_softmax_scaling

preprocess_softmax_scaling(beta: float, input_scale: float, scaled_diff_integer_bits: int = 5) -> tuple[int, int]

Mimics litert::PreprocessSoftmaxScaling.

Parameters:

• beta

  (float) –

  The softmax beta parameter (often 1.0).

• input_scale

  (float) –

  The quantization scale for the input tensor.

• scaled_diff_integer_bits

  (int, default: 5 ) –

  Number of integer bits to reserve in the scaled difference (typically 5).

Returns:

• input_multiplier ( int ) –

  The fixed-point multiplier (in Q31 format).

• input_left_shift ( int ) –

  The left shift to apply; computed as (shift from quantization + scaled_diff_integer_bits).

calculate_input_radius

calculate_input_radius(input_integer_bits, input_left_shift, total_signed_bits=31) -> int

Mimics litert::CalculateInputRadius (non-emulated version).

This computes the maximum representable difference in the scaled domain.

Parameters:

• input_integer_bits

  (int) –

  Number of integer bits reserved (e.g., kScaledDiffIntegerBits, typically 5).

• input_left_shift

  (int) –

  The left shift computed from PreprocessSoftmaxScaling.

• total_signed_bits

  (int, default: 31 ) –

  Total bits available in the representation (typically 31 for Q31 arithmetic).

Returns:

• int ( int ) –

  The computed input radius.