org.bytedeco.pytorch

Class Quantizer

java.lang.Object

org.bytedeco.javacpp.Pointer

org.bytedeco.pytorch.Quantizer

All Implemented Interfaces:

AutoCloseable

@Namespace(value="at")
 @NoOffset
 @Properties(inherit=torch.class)
public class Quantizer
extends Pointer

Quantizer is the class for storing all the information that's necessary to perform quantize and dequantize operation. We might have different types of quantization schemes and this is the base class for all quantizers. QTensorImpl will hold a pointer to Quantizer so that we can support different quantization schemes on Tensor. For example, the most common quantization scheme, Affine Quantization, requires scale and zero_point as parameters, we'll store scale and zero_point inside the instance and we can use it to quantize a float Tensor or dequantize a quantized Tensor. When you add new types of leaf Quantizer class, please also make sure to add a corresponding QScheme enum since they should have one to one mapping. Note about intrusive_ptr: Quantized Tensor holds an intrusive_ptr to Quantizer, and multiple Tensor can share the same Quantizer. Quantizer should be immutable.

Nested Class Summary

Nested classes/interfaces inherited from class org.bytedeco.javacpp.Pointer

Pointer.CustomDeallocator, Pointer.Deallocator, Pointer.NativeDeallocator, Pointer.ReferenceCounter

Field Summary

Fields inherited from class org.bytedeco.javacpp.Pointer

address, capacity, limit, position

Constructor Summary

Constructors

Constructor and Description
Quantizer(Pointer p) Pointer cast constructor.

Method Summary

Modifier and Type	Method and Description
Tensor	dequantize_out(Tensor out, Tensor t) dequantize a quantized Tensor into a float Tensor, out= variant
Tensor	dequantize(Tensor t) dequantize a quantized Tensor into a float Tensor.
boolean	equalTo(QuantizerPtr other) Compare against other for equality.
QuantizerPtr	intrusive_from_this()
torch.QScheme	qscheme() Each concrete Quantizer type should have a unique QScheme type.
Tensor	quantize(Tensor t) quantize a float Tensor into a quantized Tensor.
torch.ScalarType	scalar_type_()
torch.ScalarType	scalar_type()

Methods inherited from class org.bytedeco.javacpp.Pointer

address, asBuffer, asByteBuffer, availablePhysicalBytes, calloc, capacity, capacity, close, deallocate, deallocate, deallocateReferences, deallocator, deallocator, equals, fill, formatBytes, free, getDirectBufferAddress, getPointer, getPointer, getPointer, getPointer, hashCode, interruptDeallocatorThread, isNull, isNull, limit, limit, malloc, maxBytes, maxPhysicalBytes, memchr, memcmp, memcpy, memmove, memset, offsetAddress, offsetof, offsetof, parseBytes, physicalBytes, physicalBytesInaccurate, position, position, put, realloc, referenceCount, releaseReference, retainReference, setNull, sizeof, sizeof, toString, totalBytes, totalCount, totalPhysicalBytes, withDeallocator, zero

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Constructor Detail

Quantizer

public Quantizer(Pointer p)

Pointer cast constructor. Invokes Pointer(Pointer).

Method Detail

scalar_type_

@MemberGetter
public torch.ScalarType scalar_type_()

intrusive_from_this

@ByVal
public QuantizerPtr intrusive_from_this()

qscheme

public torch.QScheme qscheme()

Each concrete Quantizer type should have a unique QScheme type.

scalar_type

public torch.ScalarType scalar_type()

quantize

@ByVal
public Tensor quantize(@Const @ByRef
                              Tensor t)

quantize a float Tensor into a quantized Tensor.

dequantize

@ByVal
public Tensor dequantize(@Const @ByRef
                                Tensor t)

dequantize a quantized Tensor into a float Tensor.

dequantize_out

@ByRef
public Tensor dequantize_out(@ByRef
                                    Tensor out,
                                    @Const @ByRef
                                    Tensor t)

dequantize a quantized Tensor into a float Tensor, out= variant

equalTo

@Cast(value="bool")
public boolean equalTo(@ByVal
                                           QuantizerPtr other)

Compare against other for equality.