pytorch_quantization.tensor_quant.ScaledQuantDescriptor Class Reference

Public Member Functions
def	__init__ (self, num_bits=8, name=None, **kwargs)
def	num_bits (self)
def	fake_quant (self)
def	axis (self)
def	amax (self)
def	learn_amax (self)
def	scale_amax (self)
def	name (self)
def	calib_method (self)
def	unsigned (self)
def	narrow_range (self)
def	__str__ (self)
def	__eq__ (self, rhs)
def	dict (self)
def	to_yaml (self)
def	from_yaml (cls, yaml_str)

Private Attributes
	_num_bits
	_name
	_fake_quant
	_axis
	_learn_amax
	_amax
	_scale_amax
	_calib_method
	_unsigned
	_narrow_range
	__dict__

Detailed Description

Supportive descriptor of quantization

Describe how a tensor should be quantized. A QuantDescriptor and a tensor defines a quantized tensor.

Args:
    num_bits: An integer. Number of bits of quantization. It is used to calculate scaling factor. Default 8.
    name: Seems a nice thing to have

Keyword Arguments:
    fake_quant: A boolean. If True, use fake quantization mode. Default True.
    axis: None, int or tuple of int. axes which will have its own max for computing scaling factor.
        If None (the default), use per tensor scale.
        Must be in the range [-rank(input_tensor), rank(input_tensor)).
        e.g. For a KCRS weight tensor, quant_axis=(0) will yield per channel scaling.
        Default None.
    amax: A float or list/ndarray of floats of user specified absolute max range. If supplied,
        ignore quant_axis and use this to quantize. If learn_amax is True, will be used to initialize
        learnable amax. Default None.
    learn_amax: A boolean. If True, learn amax. Default False.
    scale_amax: A float. If supplied, multiply amax by scale_amax. Default None. It is useful for some
        quick experiment.
    calib_method: A string. One of ["max", "histogram"] indicates which calibration to use. Except the simple
        max calibration, other methods are all hisogram based. Default "max".
    unsigned: A Boolean. If True, use unsigned. Default False.

Raises:
    TypeError: If unsupported type is passed in.

Read-only properties:
    - fake_quant:
    - name:
    - learn_amax:
    - scale_amax:
    - axis:
    - calib_method:
    - num_bits:
    - amax:
    - unsigned:

Constructor & Destructor Documentation

__init__()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.__init__	(		self,
			num_bits = 8,
			name = None,
		**	kwargs
	)

Member Function Documentation

num_bits()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.num_bits

(

self

)

fake_quant()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.fake_quant

(

self

)

axis()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.axis

(

self

)

amax()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.amax

(

self

)

learn_amax()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.learn_amax

(

self

)

scale_amax()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.scale_amax

(

self

)

name()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.name

(

self

)

calib_method()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.calib_method

(

self

)

unsigned()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.unsigned

(

self

)

narrow_range()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.narrow_range

(

self

)

__str__()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.__str__

(

self

)

__eq__()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.__eq__	(		self,
			rhs
	)

Compare 2 descriptors

dict()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.dict

(

self

)

Serialize to dict

The build-in __dict__ method returns all the attributes, which includes those have default value and have
protected prefix "_". This method only returns those have values other than the default one and don't have _ in
key. Construct a instance by dict returned by this method should get exactly the same instance.

Here is the caller graph for this function:

to_yaml()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.to_yaml

(

self

)

Create yaml serialization
Some attributes need special treatment to have human readable form, including amax, axis.

Here is the call graph for this function:

from_yaml()

def pytorch_quantization.tensor_quant.ScaledQuantDescriptor.from_yaml	(		cls,
			yaml_str
	)

Create descriptor from yaml str

Member Data Documentation

_num_bits

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._num_bits

private

_name

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._name

private

_fake_quant

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._fake_quant

private

_axis

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._axis

private

_learn_amax

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._learn_amax

private

_amax

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._amax

private

_scale_amax

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._scale_amax

private

_calib_method

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._calib_method

private

_unsigned

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._unsigned

private

_narrow_range

pytorch_quantization.tensor_quant.ScaledQuantDescriptor._narrow_range

private

__dict__

pytorch_quantization.tensor_quant.ScaledQuantDescriptor.__dict__

private

---

The documentation for this class was generated from the following file:

• tensor_quant.py