NVIDIA NvNeural SDK  2022.2
GPU inference framework for NVIDIA Nsight Deep Learning Designer
nvneural::IFilesystemWeightsLoader Class Referenceabstract

Official "public" interface allowing control of FilesystemWeightsLoader from other modules. More...

#include <FilesystemWeightsLoader.h>

Inheritance diagram for nvneural::IFilesystemWeightsLoader:
nvneural::IRefObject

Public Member Functions

Allow random weights as fallback

When filesystem loading fails, return randomly generated weight data instead.

This is obviously undesirable for most inference cases, but can be useful for profiling use, where final results are unneeded and random weights might give "general" performance profiles instead of situation-specific optimization paths.

By default, random generation is ON for compatibility with previous Neural code.

virtual NeuralResult setAllowRandomWeights (bool allowRandoms) noexcept=0
 Turn random weights generation on or off.
 
virtual bool allowRandomWeights () const noexcept=0
 Returns true if random weights generation is enabled.
 
Allow weights to be loaded as NumPy (.npy) files

By default, NumPy support is ON.

virtual NeuralResult setAllowNumpyWeights (bool allowNumpy) noexcept=0
 Turn NumPy support on or off.
 
virtual bool allowNumpyWeights () const noexcept=0
 Returns true if NumPy support is enabled.
 
Allow weights to be loaded as raw binary data (.bin) files

This is a simple memory-dump-like file format that is a contiguous array of 32-bit floating point values.

The dimension of such weights is taken either from the ILayer::weightsDimensions() query with WeightsQuery::Required (if the layer returns a nonzero dimension) or as 1 x C x 1 x 1, where C is the number of floating-point elements read.

By default, binary support is ON.

virtual NeuralResult setAllowBinaryWeights (bool allowBinary) noexcept=0
 Turn binary support on or off.
 
virtual bool allowBinaryWeights () const noexcept=0
 Returns true if binary support is enabled.
 
- Public Member Functions inherited from nvneural::IRefObject
virtual RefCount addRef () const noexcept=0
 Increments the object's reference count. More...
 
virtual const void * queryInterface (TypeId interface) const noexcept=0
 This is an overloaded member function, provided for convenience. It differs from the above function only in what argument(s) it accepts.
 
virtual void * queryInterface (TypeId interface) noexcept=0
 Retrieves a new object interface pointer. More...
 
virtual RefCount release () const noexcept=0
 Decrements the object's reference count and destroy the object if the reference count reaches zero. More...
 

Static Public Attributes

static const IRefObject::TypeId typeID = 0x21f4e2b9c7038e85ul
 Interface TypeId for InterfaceOf purposes.
 
- Static Public Attributes inherited from nvneural::IRefObject
static const TypeId typeID = 0x14ecc3f9de638e1dul
 Interface TypeId for InterfaceOf purposes.
 

Additional Inherited Members

- Public Types inherited from nvneural::IRefObject
using RefCount = std::uint32_t
 Typedef used to track the number of active references to an object.
 
using TypeId = std::uint64_t
 Every interface must define a unique TypeId. This should be randomized.
 
- Protected Member Functions inherited from nvneural::IRefObject
virtual ~IRefObject ()=default
 A protected destructor prevents accidental stack-allocation of IRefObjects or use with other smart pointer classes like std::unique_ptr.
 

Detailed Description

Official "public" interface allowing control of FilesystemWeightsLoader from other modules.


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