CN113642948A - Model management method, device and storage medium - Google Patents
Model management method, device and storage medium Download PDFInfo
- Publication number
- CN113642948A CN113642948A CN202010391079.0A CN202010391079A CN113642948A CN 113642948 A CN113642948 A CN 113642948A CN 202010391079 A CN202010391079 A CN 202010391079A CN 113642948 A CN113642948 A CN 113642948A
- Authority
- CN
- China
- Prior art keywords
- model
- file
- model file
- information
- description information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/087—Inventory or stock management, e.g. order filling, procurement or balancing against orders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/23—Updating
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Development Economics (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Human Resources & Organizations (AREA)
- General Business, Economics & Management (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention discloses a model management method, a model management device and a storage medium, wherein the method comprises the following steps: the method comprises the steps of receiving a model updating request comprising identification information of service equipment and description information of a first model file, determining target service equipment according to the identification information of the service equipment, sending a description information model updating instruction comprising the first model file to the determined target service equipment, wherein the model updating instruction is used for instructing the target service equipment to obtain the first model file from a model warehouse according to the description information of the first model file, and processing the received service request by using the first model file.
Description
Technical Field
The present application relates to the field of computer technologies, and in particular, to a model management method, apparatus, and storage medium.
Background
The standard model relies on services, i.e. on the model to provide different capabilities. A typical example is an Artificial Intelligence (AI) microservice, which provides different capabilities depending on the model loaded. For the subdivision of the same kind of problems, for example, identifying different brands of goods, a model specially trained for the brands of goods is needed, a service device providing goods identification capability is often needed to be associated with identification models of different brands to form a dedicated identification service, or a plurality of models are placed in the same service device to provide different interfaces to identify goods of different brands. The subdivision of the same class of problems, with different versions of services for the same capacity, reduces the iteration speed, and packaging multiple models into one service device may make the service itself cumbersome.
At present, although the model configuration of the service device is allowed to be modified, the operation is cumbersome because each service device needs to be manually logged in and the configuration needs to be modified, and particularly when a large number of service devices are faced and the model configuration needs to be frequently changed, the operation and maintenance workload is extremely large, and the management efficiency is low.
Disclosure of Invention
Embodiments of the present invention provide a model management method, an apparatus, and a storage medium, which can decouple a service from a model, implement efficient management of model configuration of the model and service equipment, and effectively reduce complexity of operation and maintenance.
In a first aspect, the present invention discloses a model management method, including:
receiving a model updating request, wherein the model updating request comprises identification information of service equipment and description information of a first model file;
determining target service equipment according to the identification information of the service equipment;
and sending a model updating instruction to the target service equipment, wherein the model updating instruction comprises the description information of the first model file, and the model updating instruction is used for instructing the target service equipment to acquire the first model file from a model warehouse according to the description information of the first model file and process the received service request by using the first model file.
In a second aspect, the present invention discloses a model management method, including:
receiving a model updating instruction, wherein the model updating instruction comprises description information of a first model file;
acquiring the first model file from a model warehouse according to the description information of the first model file, and updating the model file providing the service by using the first model file;
when a service request is received, the service request is processed by loading the first model file.
In a third aspect, an embodiment of the present invention provides a model management apparatus, including:
the model updating module is used for receiving a model updating request, wherein the model updating request comprises identification information of service equipment and description information of a first model file;
the determining module is used for determining target service equipment according to the identification information of the service equipment;
and the sending module is used for sending a model updating instruction to the target service equipment, wherein the model updating instruction comprises the description information of the first model file, and the model updating instruction is used for instructing the target service equipment to acquire the first model file from a model warehouse according to the description information of the first model file and process the received service request by using the first model file.
In a fourth aspect, an embodiment of the present invention provides a model management apparatus, including:
the model updating module is used for updating the description information of the first model file according to the model updating instruction;
the acquisition module is used for acquiring the first model file from a model warehouse according to the description information of the first model file;
the processing module is used for updating the model file providing the service by utilizing the first model file;
the processing module is further configured to process the service request by loading the first model file when the service request is received.
In a fifth aspect, an embodiment of the present invention provides a model management device, where the model management device includes: a processor, a communication interface and a memory, the processor, the communication interface and the memory being interconnected, wherein the memory is configured to store a computer program, the computer program comprising program instructions, the processor being configured to invoke the program instructions to perform the method of the first aspect.
In a sixth aspect, an embodiment of the present invention provides a service device, where the service device includes: a processor, a communication interface and a memory, which are connected to each other, wherein the memory is used for storing a computer program, which comprises program instructions, and the processor is configured to call the program instructions to execute the method according to the second aspect.
In a seventh aspect, the present invention provides a computer-readable storage medium, which stores a computer program, where the computer program includes program instructions, and the program instructions, when executed by a processor, cause the processor to execute the model management method according to the first aspect and the second aspect.
In the embodiment of the invention, the model management device receives the model updating request comprising the identification information of the service device and the description information of the first model file, determines the target service device according to the identification information of the service device, and sends the description information model updating instruction comprising the first model file to the determined target service device, wherein the model updating instruction is used for instructing the target service device to acquire the first model file from the model warehouse according to the description information of the first model file, and the received service request is processed by using the first model file.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.
FIG. 1 is a schematic diagram of a model management system according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of a model management method provided in the practice of the present invention;
FIG. 3 is a schematic flow chart diagram of another model management method provided in the practice of the present invention;
FIG. 4 is a schematic diagram of a model management system according to another embodiment of the present invention
FIG. 5 is a flowchart illustrating a model file update management according to an embodiment of the present invention;
FIG. 6 is a schematic flow chart of warehousing model files according to an embodiment of the present invention;
FIG. 7 is a flowchart illustrating a model file deployment management according to an embodiment of the present invention;
FIG. 8 is a schematic structural diagram of a model management apparatus according to an embodiment of the present invention;
FIG. 9 is a schematic structural diagram of another model management apparatus according to an embodiment of the present invention;
FIG. 10 is a schematic structural diagram of a model management device according to an embodiment of the present invention;
fig. 11 is a schematic structural diagram of a service device according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Cloud technology refers to a hosting technology for unifying series of resources such as hardware, software, network and the like in a wide area network or a local area network to realize calculation, storage, processing and sharing of data.
Cloud technology (Cloud technology) is based on a general term of network technology, information technology, integration technology, management platform technology, application technology and the like applied in a Cloud computing business model, can form a resource pool, is used as required, and is flexible and convenient. Cloud computing technology will become an important support. Background services of the technical network system require a large amount of computing and storage resources, such as video websites, picture-like websites and more web portals. With the high development and application of the internet industry, each article may have its own identification mark and needs to be transmitted to a background system for logic processing, data in different levels are processed separately, and various industrial data need strong system background support and can only be realized through cloud computing.
Database (Database), which can be regarded as an electronic file cabinet in short, a place for storing electronic files, a user can add, query, update, delete, etc. to data in files. A "database" is a collection of data that is stored together in a manner that can be shared by multiple users, has as little redundancy as possible, and is independent of the application.
A Database Management System (DBMS) is a computer software System designed for managing a Database, and generally has basic functions such as storage, interception, security assurance, and backup. The database management system may be categorized according to the database model it supports, such as relational, XML (Extensible Markup Language); or classified according to the type of computer supported, e.g., server cluster, mobile phone; or sorted according to the Query Language used, such as SQL (Structured Query Language), XQuery, or sorted according to performance impulse emphasis, such as max size, maximum operating speed, or other sorting.
Referring to fig. 1, fig. 1 is a schematic diagram of an architecture of a model management system according to an embodiment of the present invention, and as shown in fig. 1, the model management system includes a client 101, a model management device 102, and a plurality of service devices 103.
The client 101 is configured to upload a model file and description information of the model file to the model management device 102, update model deployment on each service device 103, and provide a visual operation interface for querying a model deployment situation to a user, where the description information includes function description information and version description information.
The model management device 102 is configured to manage model files in the model warehouse, maintain model deployment conditions on each service device 103 and deployment conditions of each model file in the model warehouse, and provide an automated distribution link to distribute a model update instruction to each service device 103. The model warehouse is a database for storing model files, and the model management device 102 may have a data management system running thereon, and manage the model warehouse through the data management system.
The service device 103 is configured to provide corresponding service capabilities to outside according to the currently deployed model file, adjust model deployment on the native machine according to the model update instruction, and synchronize model information deployed on the native machine to the model management device 102 through an agent process of the running model management system.
The model management device 102 may be specifically an independent physical server, a server cluster or a distributed system formed by a plurality of physical servers, or a cloud server providing basic cloud computing services such as a cloud service, a cloud database, cloud computing, a cloud function, cloud storage, a network service, cloud communication, a middleware service, a domain name service, a security service, a CDN, a big data and artificial intelligence platform, and the like.
The client 101 may be, but is not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, and the like.
The service device 103 may be a distributed device that provides certain service capability to the outside, for example, providing a goods identification service. The client 101, the model management device 102, and the model management device 102 and the service device 103 may be connected directly or indirectly through wired or wireless communication, and the present application is not limited herein.
In one possible implementation, the user may send a model update request to the model management device 102 through the client 101, the model update request including identification information of the service device and description information of the model file. The description information of the model file may include function description information and version description information. The model management device 102 determines a target service device from the plurality of service devices 103 based on the identification information of the service device, and at the same time, the model management device checks whether the description information of the model file on the target service device is consistent with the description information of the received model file. If the model files are inconsistent, the model management device 102 sends a model updating instruction to the determined target service device 103, the target service device 103 determines the required model files from the model warehouse according to the description information of the model files carried in the model updating instruction, and sends the deployment condition of the model files to be updated to the model management device 102 for verification. The model management device 102 verifies that the client 101 receives the model update confirmation request by sending the model update confirmation request to the client 101, checks whether the description information of the model file to be updated is consistent with the model file in the sent model update request, and sends a delayed model update confirmation instruction or a model non-update instruction if the description information of the model file to be updated is not consistent with the model file in the sent model update request, and sends a model update confirmation instruction if the description information of the model file to be updated is consistent with the model file in the sent model update request. After the client 101 sends a model update confirmation instruction, the target service device 103 processes the received service request by using the model file acquired from the model repository. The service device 103 sends the updated description information of the model file to the model management device 102, so that decoupling of the service and the model is achieved, efficient management of model and model configuration of the service device is achieved, and complexity of operation and maintenance is effectively reduced.
Referring to fig. 2, fig. 2 is a schematic flow chart of a model management method according to an embodiment of the present invention, the method is mainly explained from the perspective of a model management device of a model management system, where the model management system further includes a service device and a client, and steps of the flow chart mainly include:
s201, the model management device receives a model updating request, wherein the model updating request comprises identification information of the service device and description information of the first model file.
The model management device receives a model update request sent by the client, wherein the model update request can comprise identification information of the service device and description information of the first model file. The method has the main functions of determining the service equipment which the client wants to update through the identification information of the service equipment and correspondingly updating the model file on the service equipment according to the description information of the model file. Illustratively, in order to distinguish the difference of the description information of the model file, the description information of the model file in the model update request is represented by the description information of the first model file.
Further, the description information of the first model file includes function description information and version description information. The function description information is mainly used for describing functions of a model file corresponding to the service device that needs to be updated, such as a function of identifying a brand of a commodity, a function of identifying a shape of the commodity, a function of identifying a color of the commodity, and the like, and the description of the functions is not specifically limited here. For example, if the brand identification is for brand of goods, and if the brand model is brand a, the corresponding version description information may include brand model a V0.1.1 and brand model a V0.1.2; in the case of the brand B model, the version description information may include the brand B model V0.1.1, the brand B model V0.1.2, and the like.
Furthermore, the description information of the model file may include scene description information in addition to the function description information and the version description information. The scene description information is combined with the function and version information to further refine the identification service. For example, in the identification service of commodity brands, after scene description is added, commodity identification under a specific scene can be detailed. Typical specific scenarios may include freezers, general counters, shelves, warehouses, etc.
S202, the model management device determines the target service device according to the identification information of the service device.
Since the model management apparatus has a management function, the model management apparatus records not only the description information of the model file but also the identification information of the service apparatus. Therefore, after the model management device receives the identification information of the service device carried in the model updating request, the target service device corresponding to the received identification information of the service device can be found from the identification information of the service device recorded by continuous updating. The service identification information carried by the model updating request is not limited to one, and correspondingly, the determined target service equipment is not limited to one. Generally, a Remote Dictionary service (Redis) storage is adopted for storage of the model management device, and the storage mode can solve the problem of information loss caused by single-point downtime to a certain extent. Therefore, the information of the service device in the model management device is complete record information.
S203, the model management device sends a model updating instruction to the target service device, wherein the model updating instruction comprises description information of the first model file, and the model updating instruction is used for instructing the target service device to acquire the first model file from the model warehouse according to the description information of the first model file and process the received service request by using the first model file.
In a possible implementation manner, after determining the target service device, the model management device directly sends a model update instruction to the target service device, where the model update instruction includes the description information of the first model file. The target processing device mainly acquires the first model file from the model warehouse according to the description information of the first model file, and the target processing device determines the function of the first model file and the version information of the model file according to the description information of the first model file. After the target service device determines the first model file, the received service request may be processed using the first model file. The model updating instruction supports 'multi-node single version' and 'appointed node multi-version', and the node is the service equipment in the invention. The 'multi-node single version' refers to that in the model updating instruction, the description information of the same model file of a plurality of service devices is updated; the "multi-version of the designated node" refers to that the model updating instruction carries the updating of the description information of the model files of multiple versions of the designated service device.
In a possible implementation manner, after the model management device sends the model update instruction to the target service device, the model management device also receives a model update verification message sent by the target service device, verifies the model update verification message according to the model file of the model warehouse, verifies whether the found service device is the node device needing to be updated, sends a model update confirmation request to the client when the model update verification message is verified, confirms whether the description information of the model file to be updated is consistent with the model file in the sent model update request, and if so, the client sends the model update confirmation instruction to the model management device. And the model management equipment receives a model updating confirmation instruction returned by the client and then sends the model updating confirmation instruction to the target service equipment. The model update confirmation instruction is used for indicating that the target service equipment can perform model update.
In a possible implementation manner, before the model management device sends the model update instruction to the target service device, the model management device needs to detect whether the description information of the model file of the target service device stored in the model repository is consistent with the description information of the first model file, where the description information of the model file of the service device stored in the model repository is recorded in the form of the model deployment situation table. If the model files are inconsistent with the description information of the model files, the model management device sends a model updating instruction to the target service device, and if the model files are consistent with the description information of the model files, the model management device indicates that the target service device does not need to update the description information of the model files, so that invalid operations of the processing device can be reduced to a certain extent.
In one possible implementation, a proxy process of the model management system runs on each service device. And (3) generating a model deployment situation table: the model management device receives model deployment information sent by each service device, wherein the deployment information comprises description information and summary information of a model file deployed on each service device, and the deployment information is acquired by the agent process. The summary information is obtained by the proxy process according to the model file on each service device, and generally adopts MD5 information summary Algorithm (MD5 Message-Digest Algorithm). After receiving the model deployment information, the model management device generates a model deployment condition table according to the model deployment information, where the model deployment condition table may specifically include description information of the model file deployed on each service device and identification information of the service device deployed by each model file in the model warehouse, and the description information of the model file may include function description information of each model file and version description information of the model file. The model management device can also send the model deployment condition table to the client, and when the client receives the user inquiry about the deployment condition of the model file, the client acquires the deployment condition of the model file by inquiring the model deployment condition table.
In a possible implementation manner, after the target service device updates the description information of the original model file and the description information of the original model file to the description information of the first model file and the description information of the first model file. In general, the original model file is deleted, but the invention focuses on temporarily storing the original model file, i.e. not deleting and not using the original model file, and when a service request corresponding to the original model file is to be processed subsequently, the original model file can be directly rolled back without performing operations such as model updating, updating confirmation and the like again, and the original model file is directly called to process a corresponding service.
In a possible implementation manner, the model management device, in addition to sending a model update instruction to the service device and generating a model deployment condition table according to the model deployment information on the service device, also performs a warehousing operation on the new model file. The warehousing operation of the new model file comprises the following steps: the model management device receives a second model file and description information of the second model file, which are sent by the client, wherein the description information of the second model file also comprises the function description information and the version description information. And then the model management equipment acquires the abstract information corresponding to the second model file. And comparing the acquired abstract information with the abstract information of the model files in the model warehouse, if the abstract information is inconsistent, storing the second model files corresponding to the inconsistent abstract information into the model warehouse, and updating the model file list corresponding to the model warehouse, namely adding the second model files which are not available and updating the second model files into a new model file list.
In a possible implementation manner, when the model management device detects that the digest information of the second model file is the same as the digest information of the model files in the model repository, the second model file corresponding to the same digest information is sent to the client, and the client knows that the same second model file cannot be sent again. Meanwhile, the model management device sends the updated model file list to the client, so that the client can inquire the model files through the model file list in a targeted mode when needed. Typically, the list of model files returned to the client is displayed through a visual interface.
In the embodiment of the invention, the model management device receives the model updating request comprising the identification information of the service device and the description information of the first model file, determines the target service device according to the identification information of the service device, and sends the description information model updating instruction comprising the first model file to the determined target service device, wherein the model updating instruction is used for instructing the target service device to acquire the first model file from the model warehouse according to the description information of the first model file, and the received service request is processed by using the first model file.
Fig. 3 is a schematic flow chart of another model management method provided in the present invention, which is mainly explained from the perspective of a target service device in a plurality of service devices of a model management system, where the model management system further includes a model management device and a client, and the steps of the flowchart mainly include:
s301, the target service device receives a model updating instruction, wherein the model updating instruction comprises description information of the first model file.
Wherein the target serving device may include one or more of the plurality of serving devices. The model updating instruction received by the target service equipment is sent to the target service equipment after the model management equipment receives a model updating request sent by the client, and the model updating request comprises identification information of the service equipment and description information of the first model file. The model updating instruction is sent after the model management device determines the target service device, wherein the description information of the first model file is used for indicating the specific content of the model file required by the target service device during updating, and the description information includes the model file required to be updated, the function of the model file, the version of the model file and the like.
S302, the target service equipment acquires the first model file from the model warehouse according to the description information of the first model file, and updates the model file providing the service by using the first model file.
In a possible implementation manner, after the target service device receives the model update instruction, the corresponding first model file is obtained from the model repository according to the description information of the first model file in the model update instruction. The model repository of the corresponding first model file may be any one of a model management device, a cloud storage device, and a data storage device. The data storage device is a device which is in communication connection with the model management device, such as a storage hard disk. After the target service equipment acquires the first model file, the target service equipment updates the model file for providing the service by using the first model file, including updating the original model file into the first model file and updating the original model file into the description information of the first model file. The update is not to remove the original model file, but only the model file currently providing the service is the first model file.
In a possible implementation manner, the model update instruction received by the target service device carries the first model file corresponding to the description information of the first model file, and in this case, the target service device does not need to acquire the first model file from other storage devices, and only needs to directly receive the first model file, and meanwhile, the target service device also directly updates the model file providing the service by using the received first model file. This is because the first model file is stored in the model repository corresponding to the model management device, and the first model file can be directly obtained from the model repository when the model management device parses the description information of the first model file.
In a possible implementation manner, before the target service device acquires the first model file from the model repository according to the description information of the first model file and updates the model file providing the service by using the first model file, it needs to send model update verification information to the model management device, where the model update verification information is to make the target service device of the model management device correct. And under the condition that the model management equipment passes the verification of the model updating verification information sent by the target service equipment, sending a model updating confirmation request to the client, wherein the model updating confirmation request is used for confirming whether the description information of the model file to be updated is consistent with the model file in the sent model updating request, and if so, the client sends a model updating confirmation instruction to the model management equipment. And the model management equipment receives a model updating confirmation instruction sent by the client according to the model updating confirmation request, and after the model updating confirmation instruction is sent to the target service equipment, the target service equipment updates the model file providing the service by using the first model file. Such a step of confirming the model update is to determine that the target service device is not executing other services when the model file is being executed by the service device, and if the model file is not being executed by the service device, the ongoing service of the service device may be interrupted. Therefore, the model management system can ensure the service provided for the user while managing the model.
S303, when the service request is received, the target service equipment processes the service request by loading the first model file.
The target service equipment acquires the first model file, updates the model file for providing the service by using the acquired first model file, and provides the service by loading the first model file when receiving a service request, wherein the specific service is initiated by a client through the service equipment, and more specifically, after photographing, the picture is uploaded to the service equipment to trigger the service.
In a possible implementation manner, a proxy process of the model management system runs on each service device in the plurality of service devices, the target service device collects model deployment information on each service device through the proxy process, the model deployment information includes description information and summary information of a model file deployed on each service device, and the summary information is obtained by the proxy process according to the model file on each service device by using an MD5 information summary algorithm. The abstract information is mainly used for marking the model files so as to quickly determine which model files are. And after obtaining the model deployment information, the service equipment sends the model deployment information to the model management equipment, so that the model management equipment generates a model deployment condition table according to the description information and the abstract information of the model file in the model deployment information. The model deployment situation table includes identification information of each service device, summary information of the model file, function description information of the model file, version description information of the model file, and the like. It should be noted that each service device periodically collects description information and calculates summary information of a model file of the service device, and periodically reports the description information and the summary information to the model management device, so that the model management device better manages model deployment information of the service device. And simultaneously, the model deployment condition table is also sent to a client by the model management equipment, and when the client receives the user inquiry about the deployment condition of the model file, the client acquires the deployment condition of the model file by inquiring the model deployment condition table.
In the embodiment of the invention, the target service equipment receives a model updating instruction comprising description information of a first model file, wherein the model updating instruction is sent to the target service equipment by the model management equipment after receiving a model updating request sent by a client, the target service equipment acquires the first model file from a model warehouse according to the description information of the first model file and updates the model file providing service by using the first model file, and when the service request is received, the target service equipment processes the service request by loading the first model file. By implementing the method, the model file can be rapidly updated, and the batch management of the model file can also be realized, so that the management efficiency of the model management system on the model is improved, and the complexity of operation and maintenance is effectively reduced.
Referring to fig. 4, fig. 4 is a schematic diagram of an architecture of another model management system according to an embodiment of the present invention, and as shown in fig. 4, the model management system (vista) is composed of a Client (vista Client) of the model management system, an Agent process (vista Agent) of the model management system, and a model management platform (vista), where:
client: sending model deployment information, model uploading information and model loading information to the Vesta; and receiving a model list sent by the Vesta, wherein the model list specifically comprises a model file list and a model deployment situation table. The Client can maintain the model versions and the model deployment conditions of each version according to the model list. The user can load the updated model through the Client specified service.
Agent: when the Agent is deployed in a node where the service is located, model information on the local machine and the Vesta are synchronized, namely model deployment sent to the Vesta in fig. 4 is performed, and the Agent also receives a model and a model updating instruction sent by the Vesta and controls the loading of the model file by the service.
Vesta: the method comprises the following steps that Vesta receives model deployment information sent by a Client, mainly a model updating request; the method comprises the following steps that Vesta receives model uploading information sent by a Client, and mainly manages a model file through summary information of the model file; and the Vesta receives model loading sent by the Client, wherein the model loading information is mainly used for instructing the Vesta Agent to load the model when providing service. The Vesta synchronously obtains model information on each node through the Agent to generate a model deployment condition table; vesta also distributes model update instructions to agents.
The method steps specifically implemented by the Client, the Vesta and the Agent are specifically described in the above embodiments, and are not described in detail here.
Referring to fig. 5, fig. 5 is a schematic flowchart of a process of updating and managing a model file according to an embodiment of the present invention, where the process is described from the perspective of interaction between a model management device, a client and a service device in a model management system, and the process mainly describes the process of updating and managing the model file of a target service device, specifically the interaction process:
s501, the client sends a model updating request to the model management device, wherein the model updating request comprises identification information of the service device and description information of the first model file.
S502, the model management device determines the target service device according to the identification information of the service device.
The model management device determines the target service device from the plurality of service devices according to the identification information of the service device, the service identification information carried by the model updating request is not limited to one, and correspondingly, the determined target service device is not limited to one.
S503, the model management device sends a model updating instruction to a target service device in the service devices, wherein the model updating instruction comprises description information of the first model file.
With respect to step S503, optionally, before the model management device sends the model update instruction to the target service device in the service devices, the model management device needs to detect whether the description information of the model file of the target service device stored in the model management device is consistent with the description information of the first model file, wherein the description information of the model file of the service device stored in the model management device is recorded in the form of the model deployment situation table. If the model files are inconsistent, the model management device sends a model updating instruction to the target service device, so that invalid operations of the processing device can be reduced to a certain extent, and if the model files are consistent, the model management device indicates that the target service device does not need to update the description information of the model files.
S504, the target service equipment acquires the first model file from the model warehouse according to the description information of the first model file.
The model warehouse may be any one of a model management device, a cloud storage device and a data storage device.
And S505, the target service device sends a model updating verification message to the model management device.
S506, the model management device verifies the model updating verification message.
And S507, the model management equipment sends a model updating confirmation request to the client.
S508, the client sends a model updating confirmation instruction to the model management device, and the model management device forwards the model updating confirmation instruction to the target service device.
S509, the target service device updates the model file providing the service by using the first model file, and processes the service request by loading the first model file when receiving the service request.
In the embodiment of the invention, a client sends a model updating request comprising identification information of service equipment and description information of a model file to model management equipment; the model management device determines a target service device from the plurality of service devices according to the identification information of the service device, and meanwhile, the model management device also checks whether the description information of the model file on the target service device is consistent with the description information of the received model file. If the model files are inconsistent, the model management equipment sends a model updating instruction to the determined target service equipment, the target service equipment determines the required first model file from the model warehouse according to the description information of the first model file carried in the model updating instruction, and sends the deployment condition of the model file to be updated to the model management equipment to request verification. And after the model management equipment passes the verification, requesting confirmation to the client, and after the client sends a model updating confirmation instruction, processing the received service request by the target service equipment by using the first model file acquired from the model warehouse. The model file needing to be changed can be efficiently changed through the interaction of the equipment in the model management system.
Referring to fig. 6, fig. 6 is a schematic flow chart of model file warehousing management provided in an embodiment of the present invention, where the flow chart is described from the perspective of interaction between a model management device and a client in a model management system, and the flow chart mainly describes warehousing management of a model file sent by the client, and the flow chart mainly includes the following steps:
s601, the client sends the second model file and the description information of the second model file to the model management device.
S602, the model management device acquires the abstract information of the second model file.
S603, the model management equipment detects whether the condition that the abstract information of the model file is consistent with the abstract information of the second model file exists in the model warehouse or not, if not, the second model file is stored in the model warehouse, and a model file list corresponding to the model warehouse is updated.
S604, the model management device sends the second model file with the consistent abstract information and the updated model file list corresponding to the model warehouse to the client.
S605, the client inquires the description information of the model files in the model file list through fuzzy matching.
In the embodiment of the invention, the client sends the second model file and the description information of the model file to the model management device, after the model management device receives the model file and the description information of the model file, obtaining abstract information of a second model file, comparing the abstract information with abstract information of model files in a model warehouse, if the abstract information of the obtained model files is consistent with the abstract information of the model file information in the model warehouse corresponding to the model management device, returning the second model file with consistent abstract information to the client, and simultaneously warehousing the second model file with inconsistent abstract information, and updates the model file list corresponding to the model warehouse and returns the model file list to the client, by the embodiment, the warehousing management of the model files can be realized, meanwhile, a user can conveniently inquire the relevant information of the current model file through a visual interface of the client.
Referring to fig. 7, fig. 7 is a schematic flowchart of a model file deployment management method provided in an embodiment of the present invention, where the flowchart is set forth from a perspective of interaction between a model management device, a client and a service device in a model management system, and the flowchart mainly illustrates management of deployment information of a model file sent by the service device, and includes the specific steps of:
s701, the target service equipment collects model deployment information through a proxy process, wherein the model deployment information comprises description information and abstract information of a model file deployed on each service equipment.
S702, the target service equipment sends model deployment information to the model management equipment.
The target service device periodically sends model deployment information to the model management device.
And S703, generating a model deployment condition table by the model management equipment according to the model deployment information.
S704, the model management device sends the model deployment situation table to the client.
The model deployment situation table comprises description information of the model file deployed on each service device and identification information of the service device deployed by each model file in the model warehouse.
S705, the client side inquires model deployment information of the service equipment in the model deployment situation table by means of fuzzy matching.
In the embodiment of the invention, the target service equipment acquires the model deployment information according to the agent process of the model management system running on the service equipment, and the model deployment information comprises the description information and the abstract information of the model file deployed on the target service equipment. The target service equipment sends the model deployment information to the model management equipment, the model management system generates a model deployment condition table according to the model deployment information sent by each service equipment after receiving the model deployment information, and the model management equipment sends the model deployment condition table to the client.
Referring to fig. 8, fig. 8 is a schematic diagram of a model management apparatus according to an embodiment of the present invention, where the apparatus 80 includes: a receiving module 801, a determining module 802, a sending module 803, an obtaining module 804, a detecting module 805, and a processing module 806, wherein:
a receiving module 801, configured to receive a model update request, where the model update request includes identification information of a service device and description information of a first model file.
A determining module 802, configured to determine a target service device according to the identification information of the service device.
A sending module 803, configured to send a model update instruction to the target service device, where the model update instruction includes description information of the first model file, and the model update instruction is used to instruct the target service device to obtain the first model file from a model repository according to the description information of the first model file, and process the received service request using the first model file.
In one possible implementation, the description information includes function description information and version description information.
In a possible implementation manner, the receiving module 801 is further configured to receive a second model file and description information of the second model file.
An obtaining module 804, configured to obtain the summary information of the second model file.
A detecting module 805, configured to detect whether the summary information of the model file is consistent with the summary information of the second model file in the model repository.
The processing module 806 is further configured to, if the second model file does not exist, store the second model file in the model repository, and update the model file list corresponding to the model repository.
In a possible implementation manner, the receiving module 801 is configured to receive model deployment information, where the model deployment information includes description information and summary information of a model file deployed on each service device, and the model deployment information is acquired by an agent process.
A processing module 806, configured to generate a model deployment situation table according to the model deployment information, where the model deployment situation table includes description information of the model file deployed on each service device and identification information of the service device deployed in each model file in the model repository.
In a possible implementation manner, the determining module 802 is configured to determine, according to the model deployment situation table, description information of a model file deployed on the target service device.
A detecting module 805, configured to detect whether the description information of the first model file is consistent with the description information of the model file deployed on the target service device.
A sending module 803, configured to execute the step of sending the model update instruction to the target service device if the model update instruction is inconsistent with the model update instruction.
Referring to fig. 9, fig. 9 is another model management apparatus according to an embodiment of the present invention, where the apparatus 90 includes: a receiving module 901, an obtaining module 902, a processing module 903, and a sending module 904. Wherein:
a receiving module 901, configured to receive a model updating instruction, where the model updating instruction includes description information of a first model file.
An obtaining module 902, configured to obtain the first model file from a model repository according to the description information of the first model file.
And the processing module 903 is configured to update the model file providing the service by using the first model file.
The processing module 903 is further configured to, when a service request is received, process the service request by loading the first model file.
In a possible implementation manner, the processing module 903 is configured to collect model deployment information through a proxy process, where the model deployment information includes description information and summary information of a model file deployed on a target service device.
A sending module 904, configured to send the model deployment information to a model management device, so that the model management device generates a model deployment situation table according to the model deployment information sent by each service device, where the model deployment situation table includes description information of a model file deployed on each service device and identification information of a service device deployed by each model file in the model repository.
Fig. 10 is a schematic structural diagram of a model management device according to an embodiment of the present invention. The model management device can comprise a processor 1001, a memory 1002 and a communication interface 1003, wherein data can be exchanged among the processor 1001, the memory 1002 and the communication interface 1003.
The storage 1002 may include a volatile memory (volatile memory), such as a random-access memory (RAM); the storage device 1002 may also include a non-volatile memory (non-volatile memory), such as a flash memory (flash memory), a solid-state drive (SSD), or the like; the memory device 1002 may also comprise a combination of memories of the kind described above. The communication interface 1003 is used for transmitting and receiving data.
The memory 1002 is used for storing a computer program comprising program instructions, and the processor 1001 is configured to call the program instructions to execute:
a model update request is received through the communication interface 1003, the model update request including identification information of the service apparatus and description information of the first model file.
And determining the target service equipment according to the identification information of the service equipment.
Sending a model update instruction to the target service device through the communication interface 1003, where the model update instruction includes description information of the first model file, and the model update instruction is used to instruct the target service device to obtain the first model file from a model repository according to the description information of the first model file, and process the received service request by using the first model file.
In one possible implementation, the description information includes function description information and version description information.
In one possible implementation, the processor 1001 is further configured to:
and receiving a second model file and description information of the second model file.
And acquiring abstract information of the second model file.
And detecting whether the summary information of the model file is consistent with the summary information of the second model file exists in the model warehouse.
And if the second model file does not exist, storing the second model file into the model warehouse, and updating a model file list corresponding to the model warehouse.
In one possible implementation, the processor 1001 is further configured to:
receiving model deployment information, wherein the model deployment information comprises description information and abstract information of a model file deployed on each service device, and the model deployment information is acquired by an agent process.
And generating a model deployment condition table according to the model deployment information, wherein the model deployment condition table comprises the description information of the model file deployed on each service device and the identification information of the service device deployed by each model file in the model warehouse.
In one possible implementation, the processor 1001 is further configured to:
and determining the description information of the model file deployed on the target service equipment according to the model deployment condition table.
And detecting whether the description information of the first model file is consistent with the description information of the model file deployed on the target service equipment.
And if not, executing the step of sending the model updating instruction to the target service equipment.
Fig. 11 is a schematic structural diagram of a service device according to an embodiment of the present invention. As shown in fig. 11, the service device may include a processor 1101, a memory 1102 and a communication interface 1103, and data may be exchanged between the processor 1101, the memory 1102 and the communication interface 1103.
The storage device 1102 may include a volatile memory (volatile memory), such as a random-access memory (RAM); the storage device 1102 may also include a non-volatile memory (non-volatile memory), such as a flash memory (flash memory), a solid-state drive (SSD), etc.; the storage 1102 may also comprise a combination of memories of the kind described above. The communication interface 1103 is used for transceiving data.
The memory 1102 is configured to store a computer program comprising program instructions, and the processor 1101 is configured to call the program instructions to execute:
receiving a model update instruction through the communication interface 1103, the model update instruction including description information of a first model file;
acquiring the first model file from a model warehouse according to the description information of the first model file, and updating the model file providing the service by using the first model file;
when a service request is received, the service request is processed by loading the first model file.
In one possible implementation, the processor 1101 is further configured to:
and acquiring model deployment information through a proxy process, wherein the model deployment information comprises description information and abstract information of a model file deployed on target service equipment.
And sending the model deployment information to model management equipment, so that the model management equipment generates a model deployment situation table according to the model deployment information sent by each service equipment, wherein the model deployment situation table comprises description information of model files deployed on each service equipment and identification information of the service equipment deployed by each model file in the model warehouse.
Embodiments of the present invention further provide a computer storage medium, in which program instructions are stored, and when the program instructions are executed, the computer storage medium is configured to implement the corresponding method described in the above embodiments.
It should be noted that, for simplicity of description, the above-mentioned method embodiments are described as a series of acts or combination of acts, but those skilled in the art will recognize that the present invention is not limited by the order of acts, as some steps may occur in other orders or concurrently in accordance with the invention. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required for the invention
In the embodiments provided in the present application, it should be understood that the disclosed apparatus may be implemented in other manners. For example, the above-described embodiments of the apparatus are merely illustrative, and for example, the above-described division of the units is only one type of division of logical functions, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection of some interfaces, devices or units, and may be an electric or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.
The integrated unit may be stored in a computer-readable storage medium if it is implemented in the form of a software functional unit and sold or used as a separate product. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, a network device, or the like, and may specifically be a processor in the computer device) to execute all or part of the steps of the above-described method according to the embodiments of the present invention. The storage medium may include: a U-disk, a removable hard disk, a magnetic disk, an optical disk, a Read-Only Memory (ROM) or a Random Access Memory (RAM), and other various media capable of storing program codes.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A method of model management, the method comprising:
receiving a model updating request, wherein the model updating request comprises identification information of service equipment and description information of a first model file;
determining target service equipment according to the identification information of the service equipment;
and sending a model updating instruction to the target service equipment, wherein the model updating instruction comprises the description information of the first model file, and the model updating instruction is used for instructing the target service equipment to acquire the first model file from a model warehouse according to the description information of the first model file and process the received service request by using the first model file.
2. The method of claim 1, wherein the description information comprises function description information and version description information.
3. The method according to claim 1 or 2, characterized in that the method further comprises:
receiving a second model file and description information of the second model file;
acquiring abstract information of the second model file;
detecting whether the condition that the abstract information of the model file is consistent with the abstract information of the second model file exists in the model warehouse or not;
and if the second model file does not exist, storing the second model file into the model warehouse, and updating a model file list corresponding to the model warehouse.
4. The method of claim 1, further comprising:
receiving model deployment information, wherein the model deployment information comprises description information and abstract information of a model file deployed on each service device, and the model deployment information is acquired by an agent process;
and generating a model deployment condition table according to the model deployment information, wherein the model deployment condition table comprises the description information of the model file deployed on each service device and the identification information of the service device deployed by each model file in the model warehouse.
5. The method of claim 4, wherein before sending the model update instruction to the target service device, the method further comprises:
determining description information of a model file deployed on the target service equipment according to the model deployment condition table;
detecting whether the description information of the first model file is consistent with the description information of the model file deployed on the target service equipment;
and if not, executing the step of sending the model updating instruction to the target service equipment.
6. A method of model management, the method comprising:
receiving a model updating instruction, wherein the model updating instruction comprises description information of a first model file;
acquiring the first model file from a model warehouse according to the description information of the first model file, and updating the model file providing the service by using the first model file;
when a service request is received, the service request is processed by loading the first model file.
7. The method of claim 6, further comprising:
acquiring model deployment information through a proxy process, wherein the model deployment information comprises description information and abstract information of a model file deployed on target service equipment;
and sending the model deployment information to model management equipment, so that the model management equipment generates a model deployment situation table according to the model deployment information sent by each service equipment, wherein the model deployment situation table comprises description information of model files deployed on each service equipment and identification information of the service equipment deployed by each model file in the model warehouse.
8. A model management apparatus, the apparatus comprising:
the model updating module is used for receiving a model updating request, wherein the model updating request comprises identification information of service equipment and description information of a first model file;
the determining module is used for determining target service equipment according to the identification information of the service equipment;
and the sending module is used for sending a model updating instruction to the target service equipment, wherein the model updating instruction comprises the description information of the first model file, and the model updating instruction is used for instructing the target service equipment to acquire the first model file from a model warehouse according to the description information of the first model file and process the received service request by using the first model file.
9. A model management apparatus, the apparatus comprising:
the model updating module is used for updating the description information of the first model file according to the model updating instruction;
the acquisition module is used for acquiring the first model file from a model warehouse according to the description information of the first model file;
the processing module is used for updating the model file providing the service by utilizing the first model file;
the processing module is further configured to process the service request by loading the first model file when the service request is received.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program comprising program instructions that, when executed by a processor, cause the processor to carry out the model management method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010391079.0A CN113642948A (en) | 2020-05-11 | 2020-05-11 | Model management method, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010391079.0A CN113642948A (en) | 2020-05-11 | 2020-05-11 | Model management method, device and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113642948A true CN113642948A (en) | 2021-11-12 |
Family
ID=78415281
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010391079.0A Pending CN113642948A (en) | 2020-05-11 | 2020-05-11 | Model management method, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113642948A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114153525A (en) * | 2021-11-30 | 2022-03-08 | 国电南瑞科技股份有限公司 | AI model service sharing method and system for power grid regulation and control business |
-
2020
- 2020-05-11 CN CN202010391079.0A patent/CN113642948A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114153525A (en) * | 2021-11-30 | 2022-03-08 | 国电南瑞科技股份有限公司 | AI model service sharing method and system for power grid regulation and control business |
| CN114153525B (en) * | 2021-11-30 | 2024-01-05 | 国电南瑞科技股份有限公司 | AI model servitization sharing method and system for power grid regulation and control service |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3399434B1 (en) | Short link processing method, device and server | |
| CN113568884B (en) | File management method and device, electronic equipment and storage medium | |
| EP3217301B1 (en) | Propagation of data changes in a distributed system | |
| US8555252B2 (en) | Apparatus and method for loading and updating codes of cluster-based java application system | |
| CN105740048A (en) | Image management method, device and system | |
| CN111475483B (en) | Database migration method and device and computing equipment | |
| US20140365523A1 (en) | Push subscriptions | |
| CN102291416A (en) | Two-way synchronizing method and system of client-side and server-side | |
| CN110727727B (en) | Statistical method and device for database | |
| CA2896865A1 (en) | Method and system for using a recursive event listener on a node in hierarchical data structure | |
| CN112732756B (en) | Data query method, device, equipment and storage medium | |
| CN102984277B (en) | Prevent the system and method that malice connects | |
| CN112363871A (en) | Data file returning method, device and storage medium | |
| US20090319519A1 (en) | Communication system, communication device, and computer program | |
| CN110958287B (en) | Operation object data synchronization method, device and system | |
| CN113642948A (en) | Model management method, device and storage medium | |
| US10387887B2 (en) | Bloom filter driven data synchronization | |
| CN109213955A (en) | Data processing method and relevant device | |
| US8843531B2 (en) | Bookkeeping of download timestamps | |
| CN103079194A (en) | Method, device and system of service adaptation | |
| CN115630122A (en) | Data synchronization method and device, storage medium and computer equipment | |
| CN112711466B (en) | Hanging affair inspection method and device, electronic equipment and storage medium | |
| CN114584576A (en) | Data storage method, device, equipment, storage medium and computer program product | |
| CN112187842A (en) | Local area network data processing system and local area network data processing method | |
| CN105721531A (en) | Message synchronization method and apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |