Patent Application
20220188626
The following disclosure(s) are submitted under 35 U.S.C. § 102(b)(1)(A):
(1) Tao Wang, Junsong Wang, Chang Xu, Chao Xue, “Automatic low-bit hybrid quantization of neural networks through meta learning”, arXiv:2004.11506v1 [cs.LG] Apr. 24, 2020, 8 pages.
The present invention relates generally to the field of neural networks, and more particularly to model quantization for a deep neural network.
In general, model quantization can be used to improve model compression and reduce latency, for example, in neural networks.
According to an aspect of the present invention, there is a method, computer program product and/or system that performs the following operations (not necessarily in the following order): obtaining a target neural network structure and constraints for a target neural network; generating a meta learning network having an associated quantization function based, at least in part, on the target neural network structure; training the meta learning network based, at least in part, on providing a hybrid quantization vector as input to the meta learning network and providing a training dataset to the target neural network; obtaining a plurality of hybrid quantization vectors; determining a new hybrid quantization vector from the plurality of hybrid quantization vectors; and retraining the trained meta learning network based, at least in part, on providing the new hybrid quantization vector as input to the trained meta learning network.
According to aspects of the present disclosure, systems and methods can provide for automatic hybrid quantization of networks, such as neural networks. In particular, systems and methods of the present disclosure can provide for training and using a meta learning network, along with a quantization function, to generate quantized weights for a target network, such as a deep neural network (DNN). The systems and methods can provide for applying various quantization vectors as input to a meta learning network and searching for an optimal hybrid quantization policy to meet constraints for the target neural network and achieve high performance, for example, by applying a genetic algorithm and/or the like to search for an optimal hybrid quantization vector. Further, systems and methods of the present disclosure can provide for retraining and/or finetuning the meta learning network using an optimal hybrid quantization policy, for example, to further improve the performance of the quantized target network. Additionally, some embodiments of the present disclosure can provide for using a well pretrained network (e.g., teacher network) to supervise training of the network.
Model quantization can be used to improve model compression and reduce latency, for example, in neural networks. Generally, model quantization can be used to compress and accelerate deep neural network (DNN) inference. Often, this can be especially important when deploying models to devices with limited computation capacity and/or power consumption budget, such as, for example, edge devices, Internet of Things (IoT) devices, and/or the like.
Many quantization approaches only investigate uniform bit width quantization across all layers of DNN, which can be sub-optimal, for example, under certain compression constraints. The precision of each layer in a network has a different influence on the final accuracy and can also vary with the architectures of the deep networks. Often, conventional hybrid quantization of DNN can require domain experts and some empirical rules to explore a best (e.g., optimal, etc.) hybrid quantization policy. Systems and methods of the present disclosure provide for automatic hybrid quantization that can allow for deeper compression while maintaining higher performance, even in low bit cases.
This Detailed Description section is divided into the following sub-sections: The Hardware and Software Environment; Example Embodiments; Further Comments and/or Embodiments; and Definitions.
The Hardware and Software Environment
The present invention may be a system, a method, and/or a computer program product. The computer program product may include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present invention.
The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non-exhaustive list of more specific examples of the computer readable storage medium includes the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire.
Computer readable program instructions described herein can be downloaded to respective computing/processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and/or a wireless network. The network may comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. A network adapter card or network interface in each computing/processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing/processing device.
Computer readable program instructions for carrying out operations of the present invention may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like, and conventional procedural programming languages, such as the “C” programming language or similar programming languages. The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present invention.
Aspects of the present invention are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer readable program instructions.
These computer readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, a programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function/act specified in the flowchart and/or block diagram block or blocks.
The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions/acts specified in the flowchart and/or block diagram block or blocks.
The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions.
An embodiment of a possible hardware and software environment for software and/or methods according to the present invention will now be described in detail with reference to the Figures.
Sub-system 102 is, in many respects, representative of the various computer sub-system(s) in the present invention. Accordingly, several portions of sub-system 102 will now be discussed in the following paragraphs.
Sub-system 102 may be a laptop computer, tablet computer, netbook computer, personal computer (PC), a desktop computer, a personal digital assistant (PDA), a smart phone, or any programmable electronic device capable of communicating with the client sub-systems via network 114. Program 300 is a collection of machine-readable instructions and/or data that can be used to create, manage, and control certain software functions, such as will be discussed in detail, below, in the Example Embodiment sub-section of this Detailed Description section. As an example, a program 300 can include functions for model quantization, training a neural network to for hybrid quantization, generating quantized weights for a target network, and/or the like.
Sub-system 102 is capable of communicating with other computer sub-systems via network 114. Network 114 can be, for example, a local area network (LAN), a wide area network (WAN) such as the Internet, or a combination of the two, and can include wired, wireless, or fiber optic connections. In general, network 114 can be any combination of connections and protocols that will support communications between server and client sub-systems.
Sub-system 102 is shown as a block diagram with many double arrows. These double arrows (no separate reference numerals) represent a communications fabric, which provides communications between various components of sub-system 102. This communications fabric can be implemented with any architecture designed for passing data and/or control information between processors (such as microprocessors, communications and network processors, etc.), system memory, peripheral devices, and any other hardware components within a system. For example, the communications fabric can be implemented, at least in part, with one or more buses.
Memory 208 and persistent storage 210 are computer-readable storage media. In general, memory 208 can include any suitable volatile or non-volatile computer-readable storage media. It is further noted that, now and/or in the near future: (i) external device(s) 214 may be able to supply, some or all, memory for sub-system 102; and/or (ii) devices external to sub-system 102 may be able to provide memory for sub-system 102.
Program 300 is stored in persistent storage 210 for access and/or execution by one or more of the respective computer processors 204, usually through one or more memories of memory 208. Persistent storage 210: (i) is at least more persistent than a signal in transit; (ii) stores the program (including its soft logic and/or data), on a tangible medium (such as magnetic or optical domains); and (iii) is substantially less persistent than permanent storage. Alternatively, data storage may be more persistent and/or permanent than the type of storage provided by persistent storage 210.
Program 300 may include both machine readable and performable instructions and/or substantive data (that is, the type of data stored in a database). For example, program 300 may include machine readable and performable instructions to provide for performance of method operations as disclosed herein. In this particular embodiment, persistent storage 210 includes a magnetic hard disk drive. To name some possible variations, persistent storage 210 may include a solid-state hard drive, a semiconductor storage device, read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, or any other computer-readable storage media that is capable of storing program instructions or digital information.
The media used by persistent storage 210 may also be removable. For example, a removable hard drive may be used for persistent storage 210. Other examples include optical and magnetic disks, thumb drives, and smart cards that are inserted into a drive for transfer onto another computer-readable storage medium that is also part of persistent storage 210.
Communications unit 202, in these examples, provides for communications with other data processing systems or devices external to sub-system 102. In these examples, communications unit 202 includes one or more network interface cards. Communications unit 202 may provide communications through the use of either or both physical and wireless communications links. Any software modules discussed herein may be downloaded to a persistent storage device (such as persistent storage device 210) through a communications unit (such as communications unit 202).
I/O interface set 206 allows for input and output of data with other devices that may be connected locally in data communication with server computer 200. For example, I/O interface set 206 provides a connection to external device set 214. External device set 214 will typically include devices such as a keyboard, keypad, a touch screen, and/or some other suitable input device. External device set 214 can also include portable computer-readable storage media such as, for example, thumb drives, portable optical or magnetic disks, and memory cards. Software and data used to practice embodiments of the present invention, for example, program 300, can be stored on such portable computer-readable storage media. In these embodiments the relevant software may (or may not) be loaded, in whole or in part, onto persistent storage device 210 via I/O interface set 206. I/O interface set 206 also connects in data communication with display device 212.
Display device 212 provides a mechanism to display data to a user and may be, for example, a computer monitor, a smart phone/tablet display screen, and/or the like.
The programs described herein are identified based upon the application for which they are implemented in a specific embodiment of the invention. However, it should be appreciated that any particular program nomenclature herein is used merely for convenience, and thus the invention should not be limited to use solely in any specific application identified and/or implied by such nomenclature.
The descriptions of the various embodiments of the present invention have been presented for purposes of illustration but are not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terminology used herein was chosen to best explain the principles of the embodiments, the practical application or technical improvement over technologies found in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.
As illustrated in
Processing proceeds to operation S254, where the computing system (e.g., server computer 200 of
Processing proceeds to operation S256, where the computing system (e.g., server computer 200 of
Processing proceeds to operation S258, where the computing system (e.g., server computer 200 of
Processing proceeds to operation S260, where the computing system (e.g., server computer 200 of
Processing proceeds to operation S262, where the computing system (e.g., server computer 200 of
Some embodiments can provide the ability to realize extremely low-bit hybrid quantization. In some embodiments, a well-trained meta learning network can provide for automatically determining hybrid quantization for a neural network, and as such expert experience may not be needed to perform the quantization, for example.
Optionally, in some embodiments, processing may continue to operation S264, where the computing system (e.g., server computer 200 of
Processing proceeds to operation S354, where the computing system (e.g., server computer 200 of
Processing proceeds to operation S356, where the computing system (e.g., server computer 200 of
Processing proceeds to operation S358, where the computing system (e.g., server computer 200 of
Processing proceeds to operation S360, where the computing system (e.g., server computer 200 of
As illustrated in
Once the meta learning network 402 has been trained (e.g., to predict quantized weights for the target network 408), a plurality of hybrid quantization vectors, such as hybrid quantization vector 410, can be searched to find a best/optimal hybrid quantization vector such that the quantized target network 408 meets constraints, such as model size, latency, energy consumption budget, and/or the like. The best/optimal hybrid quantization vector can also achieve a high or highest level of performance. For example, a best/optimal hybrid quantization vector can be determined by changing the hybrid quantization vector (e.g., hybrid quantization vector 410) values while keeping meta network parameters fixed, and using an evolution or genetic search to find the hybrid quantization vector that meets the constraints for the target network (e.g., target network 408) and achieves the highest performance.
In response to determining the best/optimal hybrid quantization vector, the determined best/optimal hybrid quantization vector can be used to retrain and/or fine-tune the meta learning network 402 (e.g., along with the input data set 412). In some embodiments, the retraining/fine-tuning can also be supervised by the teacher network 414. Quantized weights 416 can then be generated and applied/loaded to the quantized target network 408.
The quantization function 406 receives the float weight 604 (W) from the third layer. The quantization function 406 receives the value q 606 from the input quantization vector to control the quantization function. The quantization function 406 generates an initial quantized weight 608 (Ŵ) as output. The scaling factor/parameter 602 (α) is applied to the initial quantized weight 608 (Ŵ) to generate the final quantized weight 506b for the corresponding target network layer 508b (e.g., Wlayer=α*Ŵ).
where
and a function {circumflex over (Q)} 706,
where
with
The quantization function 406 receives a float weight W as input and provides W to the clamp function 702. The clamp function 702 generates a value
Definitions
Present invention: should not be taken as an absolute indication that the subject matter described by the term “present invention” is covered by either the claims as they are filed, or by the claims that may eventually issue after patent prosecution; while the term “present invention” is used to help the reader to get a general feel for which disclosures herein are believed to potentially be new, this understanding, as indicated by use of the term “present invention,” is tentative and provisional and subject to change over the course of patent prosecution as relevant information is developed and as the claims are potentially amended.
Embodiment: see definition of “present invention” above—similar cautions apply to the term “embodiment.”
and/or: inclusive or; for example, A, B “and/or” C means that at least one of A or B or C is true and applicable.
Including/include/includes: unless otherwise explicitly noted, means “including but not necessarily limited to.”
Data communication: any sort of data communication scheme now known or to be developed in the future, including wireless communication, wired communication and communication routes that have wireless and wired portions; data communication is not necessarily limited to: (i) direct data communication; (ii) indirect data communication; and/or (iii) data communication where the format, packetization status, medium, encryption status and/or protocol remains constant over the entire course of the data communication.
Receive/provide/send/input/output/report: unless otherwise explicitly specified, these words should not be taken to imply: (i) any particular degree of directness with respect to the relationship between their objects and subjects; and/or (ii) absence of intermediate components, actions and/or things interposed between their objects and subjects.
Module/Sub-Module: any set of hardware, firmware and/or software that operatively works to do some kind of function, without regard to whether the module is: (i) in a single local proximity; (ii) distributed over a wide area; (iii) in a single proximity within a larger piece of software code; (iv) located within a single piece of software code; (v) located in a single storage device, memory or medium; (vi) mechanically connected; (vii) electrically connected; and/or (viii) connected in data communication.
Computer: any device with significant data processing and/or machine readable instruction reading capabilities including, but not limited to: desktop computers, mainframe computers, laptop computers, field-programmable gate array (FPGA) based devices, smart phones, personal digital assistants (PDAs), body-mounted or inserted computers, embedded device style computers, application-specific integrated circuit (ASIC) based devices.
