qkeras.quantizers package

Submodules

qkeras.quantizers.frozen_quantizer module

class qkeras.quantizers.frozen_quantizer.FrozenFixedPointQuantizer(*args, **kwargs)

Bases: FixedPointQuantizerKBI

Abstract base class for all fixed-point quantizers.

add_weight(shape=None, initializer=None, dtype=None, trainable=True, autocast=True, regularizer=None, constraint=None, aggregation='mean', name=None)

Override add_weight to make it adds a constant weight.

build(input_shape)

get_config()

Returns the config of the object.

An object config is a Python dictionary (serializable) containing the information needed to re-instantiate it.

qkeras.quantizers.quantized_bits module

class qkeras.quantizers.quantized_bits.quantized_bits(bits: int = 8, integer: int = 0, symmetric: int = 0, keep_negative: bool = True, alpha: int | float | ~numpy.integer | ~numpy.floating | str | None = None, use_stochastic_rounding: bool = False, scale_axis: ~collections.abc.Sequence[int] | None = None, qnoise_factor: float | None = None, var_name: None = None, use_ste: bool = True, use_variables: bool | None = None, bc=<hgq.constraints.Min object>, br=None, ic=<hgq.constraints.MinMax object>, ir=None)

Bases: QuantizerConfig

property bits

classmethod from_string(string: str)

Create a quantized_bits configuration from a string representation.

get_config()

get_quantizer()

property integers

property keep_negative

property max

property min

property symmetric

class qkeras.quantizers.quantized_bits.quantized_relu(*args, **kwargs)

Bases: quantized_bits

qkeras.quantizers.utils module

qkeras.quantizers.utils.parse_string_to_args(string: str)

qkeras.quantizers.utils.str_float_int_or_none(s: str)

Module contents

class qkeras.quantizers.quantized_bits(bits: int = 8, integer: int = 0, symmetric: int = 0, keep_negative: bool = True, alpha: int | float | ~numpy.integer | ~numpy.floating | str | None = None, use_stochastic_rounding: bool = False, scale_axis: ~collections.abc.Sequence[int] | None = None, qnoise_factor: float | None = None, var_name: None = None, use_ste: bool = True, use_variables: bool | None = None, bc=<hgq.constraints.Min object>, br=None, ic=<hgq.constraints.MinMax object>, ir=None)

Bases: QuantizerConfig

property bits

classmethod from_string(string: str)

Create a quantized_bits configuration from a string representation.

get_config()

get_quantizer()

property integers

property keep_negative

property max

property min

property symmetric