Communication method and apparatus, and computer-readable storage medium

Abstract

The present application provides a communication method and apparatus, and a computer-readable storage medium. The method comprises: a management service producer receiving a first request from a model provider, the first request being used for requesting to perform model registration, and the first request comprising information of one or more models; and storing the information of the one or more models. According to the embodiments of the present application, the management service producer can externally provide services such as model registration, discovery, subscription, updating, and deregistration; the model provider can perform model registration, updating, and deregistration with the management service producer; and a management service consumer can perform model subscription and updating with the management service producer, thereby improving model management efficiency.

Description

Communication method, apparatus, and computer-readable storage medium

[0001] The present application claims priority to the Chinese patent application No. 202410889713.1, filed on July 3, 2024, and entitled "Communication method, apparatus, and computer-readable storage medium", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to the field of communication technology, and in particular to a communication method, an apparatus, and a computer-readable storage medium. BACKGROUND

[0003] With the development of communication technology and the increasing maturity of artificial intelligence (AI) technology, AI will become an indispensable part of the communication system, and it is also an inevitable trend to build AI in the future network (endogenous / native AI).

[0004] By applying some machine learning models / AI models in the network, the performance of the network can be improved. For example, through the AI model, the network device can be assisted to perform wireless resource management, so as to ensure the end-to-end service quality and improve the resource utilization efficiency. For another example, through the AI model, the network fault can be located and even automatically repaired, so as to ensure the reliability and robustness of the network. SUMMARY

[0005] Embodiments of the present application disclose a communication method, an apparatus, and a computer-readable storage medium, which can realize registration, discovery, subscription, update, and logout of a model, and can improve the management efficiency of the model.

[0006] The first aspect discloses a communication method, which can be applied to a management service producer, a component (for example, a processor, a chip, a chip system, a circuit, or a functional module) in the management service producer, or a logic module or software capable of realizing all or part of the function of the management service producer. The communication method can include: receiving a first request from a model provider, the first request being used to request model registration, the first request including information of one or more models; and storing the information of the one or more models.

[0007] In the embodiments of the present application, the management service producer can provide a model registration service to the outside. After the model provider deploys a model, the model provider can send a model registration request (a first request) to the management service producer, and the model registration request can include information of one or more deployed models, so as to register the deployed model to the management service producer. Correspondingly, the management service producer can store the information of one or more models carried in the model registration request. In this way, the management service producer can store the information of one or more models from different model providers uniformly, and can provide the information of all available models to the management service consumer in time.

[0008] With reference to the first aspect, in a possible implementation, the method further includes: receiving a second request from the management service consumer, the second request being used to request model discovery, the second request including model discovery information; and sending a response to the second request to the management service consumer according to the model discovery information and the stored information of the models, the response to the second request including information of one or more models.

[0009] In the embodiments of the present application, the management service producer can provide a model discovery service to the outside. The management service producer can receive a model discovery request (a second request) from the management service consumer, and return information of a model matching the model discovery information in the stored information of the models to the management service consumer. In this way, the management service consumer can directly obtain the information of the models provided by different model providers through the management service producer, without the need to interact with each model provider respectively, and the model discovery efficiency can be improved.

[0010] With reference to the first aspect, in a possible implementation, the method further includes: receiving a third request from the management service consumer, the third request being used to request model subscription, the third request including identification information of a first model; and in a case where information of the first model changes, sending the changed information of the first model to the management service consumer.

[0011] In the embodiments of the present application, the management service producer can provide a model subscription service to the outside. The management service producer can receive a model subscription request (a third request) from the management service consumer, and can return changed information of a model to the management service consumer in a case where information of the model subscribed by the management service consumer changes. In this way, the accuracy of the information of the model obtained / used by the management service consumer can be ensured.

[0012] In a possible implementation of the first aspect, the information of the model includes one or more of the following: an identifier of the model, a name of the model, a download address of the model, a service address of the model, a type of the model, a size of the model, a key of the model, a service state of the model, and function description information of the model.

[0013] In a possible implementation of the first aspect, the information of the model further includes a scenario applicable to the model.

[0014] In the embodiments of the present application, the information of the model can include the scenario applicable to the model, so that the management service consumer can quickly obtain the information of the model corresponding to a specific application scenario / service scenario, and can quickly call the corresponding model for inference to complete the related business.

[0015] In a possible implementation of the first aspect, the information of the model further includes information of one or more models associated with the scenario applicable to the model.

[0016] In the embodiments of the present application, the information of the model can further include information of one or more models associated with the scenario applicable to the model, so that for a certain model, other models that need to work cooperatively with the model in a certain applicable scenario can be quickly determined.

[0017] In a possible implementation of the first aspect, the model discovery information includes one or more of the following: an identifier of the model, a name of the model, a type of the model, a size of the model, a service state of the model, and a scenario applicable to the model.

[0018] In the embodiments of the present application, one or more of the following can be carried in the model discovery information: an identifier of the model, a name of the model, a type of the model, a size of the model, a service state of the model, and a scenario applicable to the model, which can improve the flexibility of model discovery, and in addition, the management service consumer can carry the corresponding information based on actual needs, which can improve the efficiency of model discovery.

[0019] In a possible implementation of the first aspect, the method further includes: sending, to the model provider, a response to the first request, the response to the first request being used to indicate that the model registration is successful.

[0020] In the embodiments of the present application, in the case where the model registration is successful, the management service producer can send a response to the first request to the management service consumer, so as to inform the management service consumer that the model registration is successful.

[0021] The second aspect discloses a communication method, which can be applied to a model provider, can be applied to a component (for example, a processor, a chip, a chip system, a circuit or a functional module) in the model provider, can be applied to a logic module or software capable of realizing all or part of the function of the model provider, and can be applied to a model consumer. The communication method can include: obtaining information of one or more models; and sending a first request to a management service producer, the first request being used for requesting model registration, and the first request including the information of the one or more models.

[0022] With reference to the second aspect, in a possible implementation, the information of the model includes one or more of the following: an identifier of the model, a name of the model, a download address of the model, a service address of the model, a type of the model, a size of the model, a key of the model, a service state of the model, and function description information of the model.

[0023] With reference to the second aspect, in a possible implementation, the information of the model further includes a scenario applicable to the model.

[0024] With reference to the second aspect, in a possible implementation, the information of the model further includes information of one or more models associated with the scenario applicable to the model.

[0025] With reference to the second aspect, in a possible implementation, the method further includes: receiving a response to the first request from the management service producer, the response to the first request being used for indicating that the model registration is successful.

[0026] It should be noted that the technical solution of the second aspect of the present application can correspond to the solution of the first aspect, and the beneficial effects can be referred to the beneficial effects of the first aspect.

[0027] The third aspect discloses a communication method, which can be applied to a management service consumer, can be applied to a component (for example, a processor, a chip, a chip system, a circuit or a functional module) in the management service consumer, can be applied to a logic module or software capable of realizing all or part of the function of the management service consumer. The communication method can include: sending a second request to a management service producer, the second request being used for requesting model discovery, and the second request including model discovery information; and receiving a response to the second request from the management service producer, the response to the second request including information of one or more models associated with the model discovery information in information of models stored by the management service consumer.

[0028] With reference to the third aspect, in a possible implementation of the third aspect, the method further includes: sending, to the management service producer, a third request for requesting model subscription, the third request including identification information of the first model; and receiving, from the management service producer, information of the first model after the information of the first model changes.

[0029] With reference to the third aspect, in a possible implementation of the third aspect, the information of the model includes one or more of the following: identification of the model, name of the model, download address of the model, service address of the model, type of the model, size of the model, key of the model, service status of the model, and function description information of the model.

[0030] With reference to the third aspect, in a possible implementation of the third aspect, the information of the model further includes applicable scenarios of the model.

[0031] With reference to the third aspect, in a possible implementation of the third aspect, the information of the model further includes information of one or more models associated with the applicable scenarios of the model.

[0032] With reference to the third aspect, in a possible implementation of the third aspect, the model discovery information includes one or more of the following: identification of the model, name of the model, type of the model, size of the model, service status of the model, and applicable scenarios of the model.

[0033] It should be noted that the technical solutions of the third aspect of the present application can correspond to the solutions of the first aspect, and the beneficial effects can be referred to the beneficial effects of the first aspect.

[0034] The fourth aspect discloses a communication apparatus, which can be a management service producer or a component (for example, a processor, a chip, a chip system, a circuit, or a functional module) in the management service producer. The communication apparatus includes: a receiving unit configured to receive a first request from a model provider, the first request being for requesting model registration, the first request including information of one or more models; and a storage unit configured to store the information of the one or more models.

[0035] With reference to the fourth aspect, in a possible implementation of the fourth aspect, the receiving unit is further configured to receive a second request from a management service consumer, the second request being for requesting model discovery, the second request including model discovery information; and the apparatus can further include a sending unit configured to send, to the management service consumer, a response to the second request including information of one or more models according to the model discovery information and the stored information of the models.

[0036] In a possible implementation of the fourth aspect, the receiving unit is further configured to receive a third request from the management service consumer, the third request being used to request model subscription, the third request comprising identification information of the first model; and the sending unit is further configured to send, to the management service consumer, information of the first model after the information of the first model is changed.

[0037] In a possible implementation of the fourth aspect, the information of the model comprises one or more of the following: identification of the model, name of the model, download address of the model, service address of the model, type of the model, size of the model, key of the model, service state of the model, and function description information of the model.

[0038] In a possible implementation of the fourth aspect, the information of the model further comprises applicable scenarios of the model.

[0039] In a possible implementation of the fourth aspect, the information of the model further comprises information of one or more models associated with the applicable scenarios of the model.

[0040] In a possible implementation of the fourth aspect, the model discovery information comprises one or more of the following: identification of the model, name of the model, type of the model, size of the model, service state of the model, and applicable scenarios of the model.

[0041] In a possible implementation of the fourth aspect, the sending unit is further configured to send, to the model provider, a response to the first request, the response being used to indicate that model registration is successful.

[0042] A fifth aspect discloses a communication apparatus, which can be a model provider or a component (for example, a processor, a chip, a chip system, a circuit, or a functional module) in the model provider. The communication apparatus comprises a processing unit configured to obtain information of one or more models; and a sending unit configured to send, to a management service producer, a first request, the first request being used to request model registration, the first request comprising the information of the one or more models.

[0043] In a possible implementation of the fifth aspect, the information of the model comprises one or more of the following: identification of the model, name of the model, download address of the model, service address of the model, type of the model, size of the model, key of the model, service state of the model, and function description information of the model.

[0044] In a possible implementation of the fifth aspect, the information of the model further comprises applicable scenarios of the model.

[0045] With reference to the fifth aspect, in a possible implementation of the fifth aspect, the information of the model further includes information of one or more models associated with an applicable scenario of the model.

[0046] With reference to the fifth aspect, in a possible implementation of the fifth aspect, the apparatus further includes a receiving unit, configured to receive a response to the first request from the management service producer, the response to the first request being used to indicate that the model registration is successful.

[0047] A sixth aspect discloses a communication apparatus, which can be a management service consumer or a component (e.g., a processor, a chip, a chip system, a circuit or a functional module) in the management service consumer. The communication apparatus includes a sending unit, configured to send a second request to a management service producer, the second request being used to request model discovery, the second request including model discovery information; and a receiving unit, configured to receive a response to the second request from the management service producer, the response to the second request including information of one or more models associated with the model discovery information in information of models stored by the management service consumer.

[0048] With reference to the sixth aspect, in a possible implementation of the sixth aspect, the sending unit is further configured to send a third request to the management service producer, the third request being used to request model subscription, the third request including identification information of a first model; and the receiving unit is further configured to receive changed information of the first model from the management service producer in a case where the information of the first model changes.

[0049] With reference to the sixth aspect, in a possible implementation of the sixth aspect, the information of the model includes one or more of the following: identification of the model, name of the model, download address of the model, service address of the model, type of the model, size of the model, key of the model, service status of the model, and function description information of the model.

[0050] With reference to the sixth aspect, in a possible implementation of the sixth aspect, the information of the model further includes an applicable scenario of the model.

[0051] With reference to the sixth aspect, in a possible implementation of the sixth aspect, the information of the model further includes information of one or more models associated with an applicable scenario of the model.

[0052] With reference to the sixth aspect, in a possible implementation of the sixth aspect, the model discovery information includes one or more of the following: identification of the model, name of the model, type of the model, size of the model, service status of the model, and applicable scenario of the model.

[0053] The seventh aspect discloses a communication system, comprising at least two of a management service producer, a model provider and a management service consumer, the management service producer is configured to implement the method provided in the first aspect and any possible implementation of the first aspect; the model provider is configured to implement the method provided in the second aspect and any possible implementation of the second aspect; the management service consumer is configured to implement the method provided in the third aspect and any possible implementation of the third aspect.

[0054] The eighth aspect discloses a communication apparatus, comprising a processor and a communication interface; the communication interface is configured to receive and / or send data; the processor is configured to invoke a computer program or computer instructions stored in a memory to implement the method provided in the first aspect and any possible implementation of the first aspect.

[0055] The ninth aspect discloses a communication apparatus, comprising a processor and a communication interface; the communication interface is configured to receive and / or send data; the processor is configured to invoke a computer program or computer instructions stored in a memory to implement the method provided in the second aspect and any possible implementation of the second aspect.

[0056] The tenth aspect discloses a communication apparatus, comprising a processor and a communication interface; the communication interface is configured to receive and / or send data; the processor is configured to invoke a computer program or computer instructions stored in a memory to implement the method provided in the third aspect and any possible implementation of the third aspect.

[0057] As a possible implementation, the processor comprised in the communication apparatus disclosed in the eighth aspect, the communication apparatus disclosed in the ninth aspect and the communication apparatus disclosed in the tenth aspect can be one or more.

[0058] Optionally, the communication apparatus disclosed in the eighth aspect, the communication apparatus disclosed in the ninth aspect and the communication apparatus disclosed in the tenth aspect further comprise one or more memories.

[0059] The eleventh aspect discloses a computer readable storage medium, the computer readable storage medium stores a computer program or computer instructions, when the computer program or computer instructions are run, implement the method provided in the first aspect and any possible implementation of the first aspect, or implement the method provided in the second aspect and any possible implementation of the second aspect, or implement the method provided in the third aspect and any possible implementation of the third aspect.

[0060] The twelfth aspect discloses a chip comprising a processor configured to execute a program stored in a memory, which when executed causes the chip to perform the method provided in the first aspect and any possible implementation of the first aspect, or to perform the method provided in the second aspect and any possible implementation of the second aspect, or to perform the method provided in the third aspect and any possible implementation of the third aspect.

[0061] As a possible implementation, the memory is located outside the chip.

[0062] The thirteenth aspect discloses a computer program product comprising computer program code which, when executed, causes the method provided in the first aspect and any possible implementation of the first aspect to be performed, or causes the method provided in the second aspect and any possible implementation of the second aspect to be performed, or causes the method provided in the third aspect and any possible implementation of the third aspect to be performed.

[0063] It should be understood that the implementation and beneficial effects of the above aspects or any possible implementation of the present application can be mutually referred to. BRIEF DESCRIPTION OF DRAWINGS

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0065] FIG. 1 is a schematic diagram of an architecture separating a large model application framework from a large model instance disclosed by the embodiments of the present application;

[0066] FIG. 2 is a schematic diagram of a scenario of a large model application framework configuration disclosed by the embodiments of the present application;

[0067] FIG. 3 is a schematic diagram of a system architecture disclosed by the embodiments of the present application;

[0068] FIG. 4A and FIG. 4B are schematic diagrams of scenarios of deployment of two large model service management modules disclosed by the embodiments of the present application;

[0069] FIG. 5 is a schematic diagram of an information target class disclosed by the embodiments of the present application;

[0070] FIG. 6 is a schematic diagram of a model registration process disclosed by the embodiments of the present application;

[0071] FIG. 7A is a schematic diagram of a model discovery process disclosed by the embodiments of the present application;

[0072] FIG. 7B is a flow diagram of another model discovery according to an embodiment of the present application;

[0073] FIG. 8 is a flow diagram of a model update according to an embodiment of the present application;

[0074] FIG. 9 is a flow diagram of a model deregistration according to an embodiment of the present application;

[0075] FIG. 10 is a flow diagram of a model subscription according to an embodiment of the present application;

[0076] FIG. 11 is a structural diagram of a communication apparatus according to an embodiment of the present application;

[0077] FIG. 12 is a structural diagram of another communication apparatus according to an embodiment of the present application;

[0078] FIG. 13 is a structural diagram of yet another communication apparatus according to an embodiment of the present application;

[0079] FIG. 14 is a hardware structural diagram of a communication apparatus according to an embodiment of the present application. DETAILED DESCRIPTION

[0080] The embodiments of the present application disclose a communication method, apparatus and computer readable storage medium, which can realize registration, discovery, subscription, update and deregistration of a model, and can improve the management efficiency of the model. The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings.

[0081] In order to better understand the embodiments of the present application, the related contents, terms or names involved in the present application are briefly introduced as follows.

[0082] 1. Large model

[0083] A largemodel can be understood as a machine learning model / artificial intelligence (AI) model (such as a deep neural network) with a large number of parameters (such as tens of billions or even hundreds of billions of parameters) and complex structures. Large models can learn complex patterns and features through massive amounts of data, and generally have strong generalization ability, reasoning ability, etc. Large models have been widely applied in various fields, including natural language processing, computer vision, speech recognition, etc. For example, in the field of natural language processing, large language models (LLMs) represented by ChatGPT have shown strong dialogue capabilities, task generalization capabilities, logical reasoning capabilities, etc. Large language models such as ChatGPT can use large-scale text data for unsupervised learning, self-supervised learning, etc., and can learn general knowledge and rules of language, thereby providing effective initialization parameters or feature representations for downstream natural language processing tasks. Large language models can handle a variety of natural language tasks, such as text classification, question answering, dialogue, etc.

[0084] Currently, there are also research and application of AI technology in the field of communication (such as assisting wireless resource management of network equipment through AI algorithm), and it is also an inevitable trend to endogenously build artificial intelligence (endogenous / native support for AI) in future networks. In the network, large models can effectively utilize a large amount of unlabeled network data, and can help realize a highly autonomous network and improve the intelligent level of the network.

[0085] II. Service-oriented registration and discovery mechanism

[0086] The service-oriented registration and discovery mechanism is a mechanism for managing services in a distributed system, allowing service providers to register their services in a centralized service registry, and allowing service consumers to discover available services by querying the registry. This mechanism can help developers more easily build and manage distributed systems, while improving system scalability and reliability.

[0087] For example, in the service-oriented registration and discovery mechanism, service providers can register the metadata of their services (such as service name, address, port, etc.) in the service registry. Service consumers can discover available services by querying the registry and use their provided functions. This mechanism can achieve automated service discovery and management, thereby reducing the workload of manual configuration and management.

[0088] III. Large model application framework and related configuration process

[0089] A large model application framework (such as LangChain, Agent, etc.) is a framework for large model application development, which can be based on model (Model), prompt (Prompt), memory (Memory), etc. Components provide capabilities such as prompt template, model orchestration, large model service, security isolation, etc. to help developers achieve a simple, secure, and trusted large model application construction experience.

[0090] Generally, a large model application and a large model instance are separated. That is, a large model application and a large model instance can be deployed on different servers or different clouds, or can be deployed in different containers on the same server / cloud, etc. The large model application can use the capabilities of the large model instance through the application programming interface (API) interface (such as web application programming interface, WebAPI) corresponding to the large model instance, such as triggering the large model for inference through the WebAPI. It can be understood that the large model instance can be a trained large model.

[0091] For example, as shown in FIG. 1, for each large model instance, such as large model instance 1, large model instance 2, …, large model instance N, etc., each can include a corresponding external service interface (such as WebAPI), which can be used to call the corresponding large model instance. Therefore, when developing a large model application based on a large model application framework, a related developer can encapsulate the WebAPI corresponding to each large model instance to call the corresponding large model instance, such as sending relevant information to the corresponding large model instance, triggering the large model instance to perform inference based on the relevant information. N can be an integer greater than or equal to 3.

[0092] Taking the LLMs module of the large model application framework LangChain as an example, when using a large language model (LLM), a related developer can perform local custom LLM class encapsulation based on the WebAPI corresponding to the large language model instance. For example, the encapsulation example is as follows:

[0093] classChatGLM(LLM):

[0094] # Model service uniform resource locator (URL) / model WebAPI

[0095] url = http: / / a1.b1.c1.d1:8595 / chat

[0096] …

[0097] As shown above, for the ChatGLM large model instance, the related technical personnel can directly encapsulate the Web API corresponding to the ChatGLM large model instance when developing the large model application, to call the corresponding ChatGLM large model instance for inference.

[0098] In actual scenarios, multiple domains can usually be included, and different domains can each include a large model application based on a large model application framework, such as the domain A and the domain B shown in FIG. 2. The domain in the embodiment of the present application can refer to different network domains, such as a wireless domain, a core network domain, and a fixed network domain (such as a data network). Of course, the division of the domain in the embodiment of the present application can also be other manners, which are not limited here. Here, the main point is that different domains can each include a large model application, and can use the capabilities of different large model instances. In addition, one domain can include one or more large model applications. For example, for the core network, the access and mobility management function (AMF), the session management function (SMF), the policy control function (PCF), and the like in the core network can each include a large model application, so as to call the corresponding large model instance. As shown in FIG. 2, the large model instance can also include multiple large model instances, such as a large model instance 1, a large model instance 2, …, and a large model instance N. Different large model instances can be deployed in different servers or different clouds. Generally, when each large model instance is online, the corresponding large model instance provider (such as a provider of the large model instance) will publish the API corresponding to the large model instance and other related information (such as the capabilities of the large model instance), so as to call the large model instance for inference. In order to use the capabilities of each large model instance, the large model application of each domain can discover each other with the large model instance provider, obtain the API and other related information of each large model instance, and then perform corresponding configuration (such as configuring the information and calling mode of each large model instance into the large model application framework of each domain), such as the large model list shown in FIG. 2. In addition, if part of the large model instances change or are offline, the large model application of each domain needs to refresh or cancel the information of the related large model instances. It can be seen that the above configuration process of the large model instance related to the large model application of each domain is relatively tedious and time-consuming, and the connection cost or communication cost is high. In addition, it can not be possible to use the newly online large model instance in time, and it can not be possible to refresh or cancel the information of the related large model instance in time.

[0099] In addition, in actual scenarios, there are many types of large model instances, for example, according to application scenarios, the large model instances can include a basic model L0 (such as Disku, GPT, etc.), an industry model L1 (such as a telecommunications industry, time sequence, log, etc.), and a domain application model L2 (such as a domain scenario interaction, a domain configuration verification, and a domain fault, etc.). For another example, according to input data types, the large model instances can include an LLM model, a computer vision (CV) large model, a multi-modal large model, and a scientific computing large model, etc. In actual business scenarios, for a certain business requirement, the large model applications in each domain can need to cooperate with multiple large model instances to complete inference, for example, for wireless fault processing business, the cooperation of a domain scenario interaction large model (such as a mobile nine-day customer service large model / NetGPT), a log large model (such as a wireless log large model), and a domain fault large model (such as a wireless fault large model) can be involved. At the same time, for a certain business scenario / requirement, how to quickly select a suitable large model instance for inference is also a problem that relevant technical personnel are concerned about.

[0100] In the embodiments of the present application, in order to solve the above problems, a large model instance registration and discovery scheme is provided. Specifically, a large model service registration and discovery mechanism can be introduced between a large model instance and a large model application framework (or a large model application), when a large model instance is online, a large model instance provider can register the related information of the large model instance to a large model service management module. For a consumer of the large model instance (such as a large model application in each domain), the related large model instance can be discovered through the large model service management module, that is, the information of the related large model instance is obtained from the large model service management module. In this way, the automatic configuration of the related large model instance can be realized, and all available large model instances can be discovered in time. Moreover, in this way, the large model provider can not need to mutually explore / perceive the large model consumers (large model applications) in each domain, which can reduce the connection cost or communication cost. In addition, when the large model instance is registered, the related information of the provided large model instance can include the applicable scenario of the large model instance, and the information of other large model instances cooperating with the large model instance in the corresponding applicable scenario. In this way, for a certain business scenario / requirement, a suitable large model instance can be quickly selected for inference.

[0101] In order to better understand the embodiments of the present application, the system architecture and the related information model of the embodiments of the present application are described below.

[0102] Please refer to FIG. 3, which is a schematic diagram of a system architecture disclosed in an embodiment of the present application. As shown in FIG. 3, the architecture can include a management service consumer (MnS Consumer), a management service producer (MnS Producer) and a model provider.

[0103] The management service producer can provide large model management services, including but not limited to large model instance registration services, large model instance discovery services, large model instance update services, large model instance deregistration services, and large model instance subscription services, through which large model instances can be uniformly managed. For example, the management service producer can be a management function / entity / management entity / network device / network element, etc., which is not limited in the embodiments of the present application. In some possible embodiments, the management service producer can be configured in a network management system (NMS), an element management system (EMS) or a network equipment (NE). In the embodiments of the present application, the large model instance registration service can also be referred to as a large model registration service, a model registration service, etc., the large model instance discovery service can also be referred to as a large model discovery service, a model discovery service, etc., the large model instance update service can also be referred to as a large model update service, a model update service, etc., the large model instance deregistration service can also be referred to as a large model deregistration service, a model deregistration service, etc., and the large model instance subscription service can also be referred to as a large model subscription service, a model subscription service, etc.

[0104] It should be noted that in the embodiments of the present application, the management service producer can be deployed in different domains, and the management service consumer can call the large model management services provided by the management service producer across domains, for example, the management service consumer can perform model discovery across domains. In addition, the division of domains in the embodiments of the present application is not limited, which can refer to different network domains, such as wireless domains, core network domains, fixed network domains (such as data network), and can also refer to different public land mobile networks (PLMNs), etc.

[0105] The management service consumer can invoke the large model instance discovery service and the large model instance subscription service provided by the management service producer. For example, the management service consumer can be a large model instance user, i.e., a consumer of the large model instance or a large model MnS consumer, which invokes the large model instance discovery service and the large model instance subscription service. For example, the management service consumer (e.g., the large model instance user) can be a network function (NF) / entity / management service consumption entity / network device / network element, etc., which is not limited in the embodiments of the present application. For example, for a core network, the access and mobility management function (AMF), the session management function (SMF), the policy control function (PCF), etc., in the core network can use the large model instance to assist in completing related services, i.e., the AMF, the SMF, the PCF, etc., can be the large model instance user.

[0106] The model provider (Provider) can invoke the large model instance registration service, the large model instance update service, and the large model instance deregistration service provided by the management service producer. For example, the model provider can be a large model provider that invokes the large model instance registration service, the large model instance update service, and the large model instance deregistration service. For example, the model provider (e.g., the large model instance provider) can be a network function (NF) / entity / network device / network element, etc., which is not limited in the embodiments of the present application. For example, the model provider can be a third-party server, such as a server of a related large model manufacturer.

[0107] In the embodiments of the present application, the large model instance provider can also be referred to as a large model provider, a model provider, etc., and the large model instance user can also be referred to as a large model user, a model user, etc. The large model instance provider can be a device (e.g., a server) that registers / updates / deregisters a large model instance. For example, after a related large model manufacturer trains a large model and deploys the trained large model to the cloud, the server can register the large model instance deployed to the cloud with the management service producer, or when the related large model manufacturer deploys the trained large model to the cloud, the related function module on the cloud can automatically register the large model instance deployed to the cloud with the management service producer.

[0108] In a possible implementation, a large model service management module can be constructed, which can provide large model instance registration service, large model instance discovery service, large model instance update service, large model instance deregistration service, large model instance subscription service, and other large model management services. It should be understood that the large model service management module can be deployed in a management service producer. Please refer to FIG. 4A, which is a schematic diagram of a scenario of deployment of a large model service management module according to an embodiment of the present application. As shown in FIG. 4A, the large model service management module can be deployed in the management service producer, and the large model service management module can provide, for external use, large model instance registration service, large model instance discovery service, large model instance update service, large model instance deregistration service, large model instance subscription service, and other large model services. Model providers, management service consumers, and the like can call / access the corresponding large model services through the API interfaces of these services. For example, for a model provider (such as model provider 1, model provider 2, …, model provider N), the model provider can call the API corresponding to the large model instance registration service, the interface parameter of the API can include model registration instance (ModelRegisterInstance), that is, the related information of the large model instance to be registered, and accordingly, the large model service management module can store the model registration instance sent by the model provider, such as storing in the large model list (large model list).

[0109] For a management service consumer (large model instance user), the management service consumer can call the API corresponding to the large model instance discovery service, the interface parameter of the API can include model discovery information. Accordingly, the large model service management module can find the information of the large model instance associated with the model discovery information from the stored information of the large model instance, such as finding the information of the large model instance associated with the model discovery information from the large model list, and then can return the information of the large model instance associated with the model discovery information found to the management service consumer, such as returning the large model instance list (List[LargeModelInstance]), which can include the information of the large model instance associated with the model discovery information. The management service consumer can also call the API corresponding to the large model instance subscription service according to actual needs, to subscribe / refresh the information of the related large model instance to the large model service management module.

[0110] In some embodiments, the management service producer of one domain can serve as the management service consumer of another domain, that is, the management service producer of one domain can invoke the large model instance discovery service, the large model instance subscription service, and the like provided by the management service producer of another domain to realize the synchronization of model information between different domains, or to realize cross-domain model discovery, model subscription, and the like. As shown in FIG. 4B, two domains, domain A and domain B, are illustrated in FIG. 4B, and the management service producer of domain A and the management service producer of domain B can communicate with each other to synchronize the model information stored by each other, or to realize cross-domain model discovery, model subscription, and the like. For example, assuming that the management service consumer of domain A wants to discover in the management service producer of domain B, in this case, the management service consumer of domain A can first send a relevant discovery request to the management service producer of domain A, and carry the identification information of the management service producer of domain B in the discovery request. After receiving the discovery request from the management service consumer of domain A, the management service producer of domain A can send a discovery request to the management service producer of domain B based on the identification information of the management service producer of domain B carried in the discovery request. After receiving the discovery request from the management service producer of domain A, the management service producer of domain B can perform model discovery based on the discovery request, and then can send a discovery response to the management service producer of domain A, which can include the information of the relevant models discovered by the management service producer of domain B based on the discovery request. After receiving the discovery response from the management service producer of domain B, the management service producer of domain A can send a corresponding discovery response to the management service consumer of domain A. For another example, assuming that the management service producer of domain A wants to synchronize the information of all or part of the registered models stored by the management service producer of domain B, the management service producer of domain A can send a model discovery request to the management service producer of domain B. After receiving the discovery request from the management service producer of domain A, the management service producer of domain B can perform model discovery based on the discovery request, and then can send a discovery response to the management service producer of domain A, which can include the information of the relevant models discovered by the management service producer of domain B based on the discovery request. After that, the management service producer of domain A can store the information of the relevant models for subsequent use (such as returning the information of the relevant models obtained from the management service producer of domain B to the management service consumer of domain A when the management service consumer of domain A sends a model discovery request).

[0111] It can be understood that in some embodiments, the large model application (or the large model application framework), the large model service management module, and the large model instance provider can be deployed in the same device / network element / entity (such as in the same server), such as in the management service producer.

[0112] In the embodiments of the present application, in order to implement the large model management service, the related information target class (informationObjectClass, IOC) / information model / data structure / structure body can be defined, as shown in FIG. 5, and the information target classes such as LargeModel, LargeModelRegister, LargeModelDiscover, LargeModelRepository, LargeModelRegisterFuction, LargeModelDiscoverFuction, and LargeModelManagementService can be defined. Among them, the LargeModel class can be defined for the structure information of the large model, the LargeModelRegister class can be defined for the structure body of the large model "registration" action, the LargeModelDiscover class can be defined for the structure body of the large model "discovery" action, the LargeModelRepository class can be defined for the structure body of the large model instance repository, the LargeModelRegisterFuction class can be defined for the structure body of the large model "registration" function, the LargeModelDiscoverFuction class can be defined for the structure body of the large model "discovery" function, and the LargeModelManagementService class can be defined for the structure body of the large model "management" function. Further, one LargeModelRepository object can correspond to one or more LargeModel objects, and one LargeModelRegister can also correspond to one or more LargeModel objects. Moreover, the LargeModelRegisterFuction can be associated with the LargeModelRegister, the LargeModelDiscoverFuction can be associated with the LargeModelDiscover, and the LargeModelManagementService can be associated with the LargeModelRegisterFuction, the LargeModelDiscoverFuction, and the like. The related information models are introduced as follows.

[0113] 1) LargeModel structure definition

[0114] LargeModel can be an information model related to a large model instance. LargeModel can include two parts, one part is a general attribute of the large model instance, and the other part is a scenario and model map attribute (scenarioAndModelsMaps) of the large model instance. The general attribute of the large model instance can include an identifier of the large model instance (largeModelId), a name of the large model instance (largeModelName), a download address of the large model instance (downloadURL), a service address of the large model instance (serviceURL), a type of the large model instance (type), a key of the large model instance (key), a size of the large model instance (size), a service status of the large model instance (status), a function description of the large model instance (description), and the like. Among them, the type of the large model instance can be divided into a natural language processing (NLP) model, a multi-modal model, and the like, and the classification method of the large model instance is not limited by the embodiments of the present application. The service status of the large model instance can include an active state (active) and a non-active state (deactive), and the active state can represent that the current large model instance can provide services to the outside, and the non-active state can represent that the current large model instance does not provide services to the outside.

[0115] The scenario and model map attribute (scenarioAndModelsMaps) can record the recommended application scenarios (applicable scenarios / usage scenarios) corresponding to the large model instance and the associated large model list information. The associated large model list information can include the information of other large model instances that can work with the corresponding large model instance under the corresponding application scenario. For example, the scenarioAndModelsMaps attribute can be as shown in Table 1:

[0116] Table 1

[0117] It should be understood that in the LargeModel instance / object, the attribute scenarioAndModelsMaps can be a list (list), the type (element type in the list) can be scenarioAndModelsMap, and scenarioAndModelsMap can include multiple sets of application scenarios (scenario) and related model information key-value pair (map) instances. For example, the scenarioAndModelsMap structure definition can be as shown in Table 2:

[0118] Table 2

[0119] As can be seen, a scenarioAndModelsMap object can include three attributes: application scenario (scenario), attribute association model name (relatedModelNames), and association model identifier (relatedModelIds). Among them, the attribute association model name (relatedModelNames) and the association model identifier (relatedModelIds) can be list, and the type can be string (String). The scenarioAndModelsMap can be used for the LargeModel structure, and is used to represent the applicable scenario of the corresponding large model and the information (such as name, identifier, etc.) of other large models related to the applicable scenario.

[0120] 2) LargeModelRegister structure definition

[0121] The LargeModelRegister can be an information structure definition for large model information registration, and can be used as a registration parameter of a general service interface (such as a createMOI service registration interface) to perform a large model registration operation, that is, a large model instance registration operation. For example, the LargeModelRegister structure definition can be as shown in Table 3:

[0122] Table 3

[0123] As can be seen, a LargeModelRegister object can include register identifier (registerId) and large model list (largeModels) attributes. Among them, the attribute largeModels can be list, and the type can be largeModel.

[0124] 3) LargeModelDiscover structure definition

[0125] The LargeModelDiscover can be an information structure definition for large model information discovery, and can be used as a discovery parameter of a general service interface (such as listAvailableFiles / listAvailableLargeModels) to perform a large model discovery operation, that is, a large model instance discovery operation. For example, the LargeModelDiscover structure definition can be as shown in Table 4:

[0126] Table 4

[0127] It can be seen that one LargeModelDiscover object can include the scenario attribute, and can also include the largeModelName and largeModelId attributes, that is, the management service consumer can discover through the scenario attribute, and can also discover through the largeModelName, largeModelId and the like. It should be understood that through the LargeModelDiscover instance, information of all registered large model instances can be discovered, and through assigning specific values to the scenario attribute, the largeModelName attribute and the largeModelId attribute, information of specific large model instances among all registered large model instances can be discovered.

[0128] The related service interfaces of the large model service management module are introduced below. The service interfaces can include model service registration, model service deregistration, model service discovery, model listening / subscription and the like. Table 5 below is a comparison table between general services and large model related information models:

[0129] Table 5

[0130] It should be understood that the content of the above related information target class / information model (such as the content in Table 1, Table 2, Table 3, Table 4 and the like) and the content of the related interface (such as Table 5) are only exemplary descriptions, and the embodiments of the present application are not limited thereto.

[0131] It should be noted that although the large model scenario is mainly taken as an example for description in the embodiments of the present application, the embodiments of the present application are also applicable to other machine learning model / AI model scenarios. That is, the technical solutions provided by the embodiments of the present application can be used in the large model scenario, and can also be used in the non-large model scenario (such as the traditional machine learning model scenario). For example, the above management service producer can provide not only the large model management service, but also the management service of other machine learning models / AI models. It should also be noted that the architecture shown in FIG. 1 is only exemplary, and the embodiments of the present application are not limited thereto.

[0132] It can be understood that the above management service producer, management service consumer, large model instance provider and the like can be implemented in the form of hardware, computer software or a combination of hardware and computer software. For example, the above management service producer, management service consumer, large model instance provider and the like can be implemented by one device, can also be implemented by multiple devices, and can also be implemented by one functional module in one device, and the embodiments of the present application are not limited thereto.

[0133] It should be understood that the technical solutions provided by the embodiments of the present application can be applied to various communication systems, for example: a fifth generation (5th generation, 5G) communication system, a transition system between a 5G communication system and a sixth generation (6th generation, 6G) communication system (the transition system can also be referred to as a 5.5G communication system), a network of multiple system fusion; of course, it can also be a future communication system, such as 6G or even a seventh generation (7th generation, 7G) system, etc. For example, for a 5G core network, a large model service management module / network element can be set in the 5G core network, and all network elements (such as AMF, PCF, access network equipment, etc.) in the network that need to use a large model can discover the corresponding large model through the large model service management network element, so as to perform related processing based on the large model.

[0134] It should be noted that the system architecture, network architecture, and business scenario (or application scenario) described in the embodiments of the present application are for more clearly illustrating the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art can know that, with the evolution of communication network architecture and the appearance of new business scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0135] It can be understood that the model instance management service provided by the embodiments of the present application can include model (instance) registration service, model (instance) discovery service, model (instance) update service, model (instance) logout service, and model (instance) subscription service. The model instance can be a large model instance, a machine learning model instance, or other model instance, and the type of model instance is not limited by the embodiments of the present application. In order to facilitate understanding, the following embodiments one to five will be described in turn.

[0136]

Embodiment one

[0137] Embodiment one is a related processing flow of model registration. In embodiment one, the model provider can send a model registration request (ModelRegister_request) to the management service producer to request model registration, and can carry the related information of the model to be registered in the model registration request, such as the name of the model, the type of the model, the scene information of the model, etc. Specifically, please refer to FIG. 6, which is a flowchart of model registration according to an embodiment of the present application. As shown in FIG. 6, the model registration flow can include but is not limited to the following steps:

[0138] 601. The model provider obtains the information of one or more models.

[0139] For example, after a model manufacturer trains a model (e.g., a large model) with training data, the model manufacturer can deploy the trained model to a cloud or a server. In order to facilitate a service consumer to discover a corresponding model from a service producer, a model provider can obtain information of one or more deployed models. That is, the information of one or more models obtained by the model provider can be information of one or more already deployed models, which is not limited in the embodiments of the present application.

[0140] The information of the model can include one or more of an identifier of the model, a name of the model, a download address of the model, a service address of the model, a type of the model, a size of the model, a key of the model, a service state of the model, and function description information of the model. The identifier of the model can be used to identify the corresponding model. For example, the identifier of the model can be a specific model number, for example, the numbering method can be manufacturer identifier (ID) + manufacturer-defined model ID, which is not limited in the embodiments of the present application. The name of the model can also be used to identify the corresponding model. The download address of the model can be used to download the corresponding model. The service address of the model can be used to access the service provided by the corresponding model and trigger the model to perform inference. The type of the model can be used to indicate the specific type of the corresponding model, for example, the type of the model can include natural language processing (NLP) model, computer vision (CV) model, multi-modal model, etc. The size of the model can be used to indicate the specific size of the corresponding model, for example, the size of the model can be the depth of the neural network (number of network layers), the size of the parameters, etc., or the size of the model can be pre-defined several levels, such as large, middle, small, etc. The key of the model can be used for identity authentication when accessing the corresponding model. The service state of the model can indicate the service state of the corresponding model, for example, the service state of the model can include active and deactive, active means that the corresponding model is currently available and can provide services to the outside, and deactive means that the corresponding model is currently unavailable and does not provide services to the outside. The function description information of the model is the function description of the corresponding model, for example, the function description information of the model can be a string of description information of the function of the corresponding model, or the function description information of the model can include one or more function tags (e.g., machine translation, text generation, video generation, scientific calculation, etc.) of the corresponding model, which can be pre-defined.

[0141] Optionally, the information of the model can further include an applicable scenario of the model. Exemplarily, the applicable scenario of the model can include a wireless troubleshooting scenario, a poor game service quality scenario (a game quality poor scenario), a poor call service quality scenario, a wireless resource allocation scenario, a perception data processing scenario, a high-precision positioning scenario, and the like. The applicable scenario of the model can be predefined or defined by a relevant standard, which is not limited herein. It should be understood that, for one model, it can include one or more corresponding applicable scenarios.

[0142] Optionally, the information of the model can further include information of one or more models associated with the applicable scenario of the model. It can be understood that, in some application scenarios, multiple models can need to work together, and therefore, in order to facilitate the management service consumer to determine multiple models that need to work together in the corresponding scenario in time, the information of the model can further include information of one or more models associated with the applicable scenario of the model, that is, information of one or more models that need to work with the corresponding model by agreement / inference in the applicable scenario. It should be understood that the applicable scenario of the model and the information of one or more models associated with the applicable scenario of the model can help the management service consumer to accurately and quickly select / discover the required model when performing a business.

[0143] It should be understood that the above description of the information of the model is only exemplary, and the embodiments of the present application are not limited thereto. For example, in another possible embodiment of the present application, the information of the model can include more or less information, such as the information of the model can further include a version number of the model.

[0144] It should be noted that, in the embodiments of the present application, the information of one or more models obtained by the model provider can also be understood as information of one or more model instances. Correspondingly, the identifier of the model can also be understood as an identifier of a model instance, the name of the model can also be understood as a name of a model instance, the download address of the model can also be understood as a download address of a model instance, and the like. It should be understood that the above one or more models can be large models, machine learning models, or other types of models, and the embodiments of the present application are not limited thereto. It should be further understood that the model in the embodiments of the present application can be a model instance.

[0145] 602. The model provider sends a first request to the management service producer, and the first request is used to request model registration, and the first request includes the information of the one or more models.

[0146] After the model provider obtains the information of the one or more models, the model provider can send a first request to the management service producer. Accordingly, the management service producer can receive the first request from the model provider, the first request being for requesting model registration, the first request including the information of the one or more models. It is to be understood that the first request for requesting model registration can also be understood as the first request for requesting model instance registration.

[0147] For example, based on the LargeModelRegister structure defined above, after the model provider obtains the information of the one or more models, the model provider can construct an IOC (information object class) instance for large model instance registration based on the information of the one or more models, that is, a LargeModelRegister instance, as follows:

[0148] LargeModelRegister{

[0149] RegisterId:“1”

[0150] largeModels:{LargeModel1,…}#one or more LargeModel instances

[0151] }

[0152] It is to be understood that each list element in largeModels can be a LargeModel instance, and each LargeModel instance can correspond to a model / model instance. For example, LargeModel1 can be as follows:

[0153] After the model provider constructs the LargeModelRegister instance, the model provider can call a general service interface createMOI and carry the constructed LargeModelRegister instance.

[0154] 603. The management service producer stores the information of the one or more models carried in the first request.

[0155] After the management service producer receives the first request from the model provider, the management service producer can store the information of the one or more models carried in the first request, so that the management service consumer can discover the corresponding models. For example, the management service producer can store the information of the one or more models carried in the first request locally or in a cloud database, and the embodiments of the present application are not limited in this regard.

[0156] In some possible implementation manners, after receiving the first request, the management service producer can perform authorization verification on the first request, and in a case where the authorization verification is passed, the management service producer can store the information of one or more models carried in the first request, and in a case where the authorization verification is not passed, the management service producer can not process the first request, and can discard the first request. For example, the management service producer performs authorization verification on the first request can be authorization verification on a model provider sending the first request, for example, checking whether the model provider has a model registration right, and if the model provider has the model registration right, it is indicated that the authorization verification is passed, and otherwise, it is indicated that the authorization verification is not passed.

[0157] For example, assuming that the management service producer receives a built LargeModelRegister instance from the model provider through the service interface createMOI, the management service producer can extract one or more LargeModel instances in the LargeModelRegister instance, and can store the extracted one or more LargeModel instances locally.

[0158] 604. The management service producer sends a response of the first request to the model provider, and the response of the first request is used to indicate that the model registration is successful.

[0159] After the management service producer stores the information of one or more models carried in the first request, the management service producer can send a response of the first request to the model provider. Correspondingly, the model provider can receive the response of the first request from the management service producer, and the response of the first request is used to indicate that the model registration is successful, that is, to indicate that the information of one or more models carried in the first request is registered successfully, that is, to indicate that the information of one or more models carried in the first request is stored successfully. It should be understood that the response of the first request is also a model registration request response (ModelRegister_response).

[0160] In the above processing flow, the management service producer can provide a model registration service to the outside, and the model provider can send a model registration request to the management service producer after the model is deployed online, and actively register the related model to the management service producer. When registering, the model name, model type, model service address and the like can be carried, and application scenario enhancement information, that is, the scenarioAndModelsMaps information described above can also be carried.

[0161] Embodiment Two

[0162] Embodiment two is a related processing flow of model discovery. In embodiment two, model discovery can include non-cross-domain model discovery and cross-domain model discovery. In the case of non-cross-domain model discovery, the management service consumer can send a model discovery request (ModelDiscover_request) to the management service producer, and can carry related information of the model to be discovered in the model discovery request, such as the name of the model, the type of the model, the scene information of the model, and the like. Specifically, please refer to FIG. 7A, which is a flowchart of a non-cross-domain model discovery according to an embodiment of the present application. As shown in FIG. 7A, the model discovery flow can include but is not limited to the following steps:

[0163] 701. The management service consumer sends a second request to the management service producer, the second request being used to request model discovery, and the second request including model discovery information.

[0164] In some cases, the management service consumer needs to use a related model to perform inference and help itself complete related business. In this case, the management service consumer can send a model discovery request to the management service producer to discover a related model. For example, the management service consumer can send a second request to the management service producer. Correspondingly, the management service producer can receive the second request from the management service consumer, the second request can be used to request model discovery, and the second request can include model discovery information. For example, the second request used to request model discovery can also be understood as the second request used to request model instance discovery.

[0165] The model discovery information can include one or more of the following: model identifier, model name, model type, model size, model service status, and applicable scenarios. Model discovery information can be used to filter models to be discovered, improving the flexibility of model discovery. For example, if the goal is to discover information about all models currently stored by the management service producer, the model discovery information carried in the model discovery request can be unconditional, such as selecting all applicable scenarios (e.g., scenario="All"), all model names (e.g., largeModelName="null"), and all model identifiers (e.g., largeModelId="null"). Alternatively, the model discovery request can omit model discovery information, which by default represents discovering information about all models currently stored by the management service producer. As another example, if the goal is to discover information about models with the type "NLP" among all models currently stored by the management service producer, the model discovery information carried in the model discovery request can include type="NLP". For another example, suppose we need to discover information about models whose applicable scenarios include "wireless troubleshooting scenarios" among all the models currently stored by the management service producer. The model discovery information carried in the model discovery request can include scenario="wireless troubleshooting scenarios".

[0166] For example, taking the LargeModelDiscover structure defined above, the management service consumer can construct an IOC (Information Object Class) instance for large model instance discovery based on its own model discovery needs, namely a LargeModelDiscover instance. Suppose the management service consumer needs to obtain information about all models currently stored by the management service producer, the corresponding LargeModelDiscover instance can be as follows:

[0167] LargeModelDiscover{

[0168] Scenario: "All" # Select all applicable scenarios

[0169] largeModelName: "null" # Do not filter large model names

[0170] largeModelId: "null" # Do not filter large model identifiers

[0171] }

[0172] Assuming that the management service consumer needs to obtain the information of the models in the applicable scenario of "wireless troubleshooting scenario" from the management service producer, the corresponding LargeModelDiscover instance can be as follows:

[0173] LargeModelDiscover {

[0174] Scenario: "wireless troubleshooting scenario"

[0175] largeModelName: "null"

[0176] largeModelId: "null"

[0177] }

[0178] After the management service consumer constructs the LargeModelDiscover instance, the management service consumer can call the model service discovery interface listAvailableLargeModel and carry the above-constructed LargeModelDiscover instance.

[0179] It should be understood that the above description of the model discovery information is only exemplary, and the embodiments of the present application are not limited thereto. For example, in another possible embodiment of the present application, the model discovery information can include more or less information, such as the model discovery information can also include the version number of the model.

[0180] 702. The management service producer sends a response to the second request to the management service consumer according to the model discovery information and the information of the stored models, and the response to the second request includes the information of one or more models.

[0181] For example, after receiving the second request from the management service consumer, the management service producer can determine the information of one or more models in the information of the currently stored models that matches the model discovery information carried in the second request, and then can send a response to the second request to the management service consumer, and the response to the second request can include the information of the one or more models that matches the model discovery information. Specifically, after receiving the second request from the management service consumer, the management service producer can compare the model discovery information carried in the second request with the information of the currently stored models, and if the information of a currently stored model is the same as the model discovery information, it indicates that the information of the model matches the model discovery information, and if the information of a currently stored model is different from the model discovery information, it indicates that the information of the model does not match the model discovery information. It should be understood that the response to the second request is also a model discovery request response (ModelDiscover_response).

[0182] For example, assuming that the currently stored models of the management service producer include model 1, model 2 and model 3, and the types corresponding to model 1, model 2 and model 3 are NLP, CV and multi-modal respectively. If the type included in the model discovery information is multi-modal, at this time, the management service producer can determine that the model discovery information does not match the information of model 1 and model 2, and matches the information of model 3.

[0183] It should be understood that in some cases, the information of the currently stored models of the management service producer can not include the information of the model that matches the model discovery information carried in the second request, in which case the response to the second request can carry first indication information, and the first information can be used to indicate that the information of the currently stored models of the management service producer does not include the information of the model that matches the model discovery information carried in the second request. Alternatively, the management service producer can not send a response to the second request to the management service consumer, or the response to the second request can be empty, both of which by default represent that the information of the currently stored models of the management service producer does not include the information of the model that matches the model discovery information carried in the second request.

[0184] In some possible implementations, after receiving the second request, the management service producer can perform authorization verification on the second request, and in the case that the authorization verification is passed, the management service producer can send a response to the second request to the management service consumer according to the model discovery information carried in the second request and the information of the stored models, and in the case that the authorization verification is not passed, the management service producer can not process, and can discard the second request.

[0185] For example, assuming that the management service producer receives a built LargeModelDiscover instance from the management service consumer through the model service discovery interface listAvailableLargeModel, the management service producer can return information of one or more models matching the LargeModelDiscover instance to the management service consumer, such as returning a largeModels instance. For example, for the LargeModelDiscover instance corresponding to the above-mentioned obtaining information of all models currently stored by the management service producer, the corresponding largeModels instance can be as follows:

[0186] For another example, for the LargeModelDiscover instance corresponding to the above-mentioned obtaining information of all models whose applicable scenarios include the "wireless troubleshooting scenario" in the information of all models currently stored by the management service producer, the corresponding largeModels instance can be as follows:

[0187] It should be understood that the above-mentioned largeModels instance is only an example and does not constitute a limitation. For example, each LargeModel instance in the above-mentioned largeModels can further include attributes such as key, downloadURL, type, size, status, and the like.

[0188] In the embodiments of the present application, after the management service consumer receives the second request from the management service producer, the management service consumer can call the corresponding model according to the information of one or more models carried in the second request, and trigger the corresponding model to perform inference based on the related information. For example, the management service consumer can call one or more discovered models in combination with business needs. For example, assuming that the current wireless network is malfunctioning, the management service consumer can find the relevant information of the model whose scenario includes the "wireless troubleshooting scenario", and then can call the corresponding model to perform inference. For example, for the above-mentioned largeModels instance corresponding to the "wireless troubleshooting scenario", the management service consumer can call the NetGPT large model, the wireless fault large model, and the wireless configuration verification large model to cooperate to troubleshoot the wireless fault and locate the cause of the malfunction of the wireless network.

[0189] In the above process, the management service producer can provide a model discovery service to the outside, and the management service consumer can send a model discovery request to the management service producer to discover information of models currently stored (registered) by the management service producer. Further, the embodiments of the present application support discovery based on applicable scenarios of models, and all models required in a corresponding application scenario can be quickly found based on the applicable scenarios of the models, so that the all models required in the corresponding application scenario can be quickly called for inference, and the use efficiency of the models can be improved.

[0190] The management service producer and the management service consumer in FIG. 7A can belong to the same domain, that is, non-cross-domain service discovery. Cross-domain service discovery will be introduced below in conjunction with FIG. 7B.

[0191] In the case of cross-domain model discovery, the management service consumers in different domains can communicate with each other and exchange information of registered models in their respective domains. For example, the management service consumer of domain A can send a model discovery request (ModelDiscover_request) to the management service producer of domain A, and can carry identification information of the management service producer of domain B and related information of a model to be discovered in the model discovery request, such as a name of the model, a type of the model, and scenario information of the model. Specifically, refer to FIG. 7B, which is a flowchart of cross-domain model discovery according to an embodiment of the present application. As shown in FIG. 7B, the model discovery process can include but is not limited to the following steps:

[0192] 701a. The management service consumer of domain A sends a sixth request to the management service producer of domain A, the sixth request being used to request model discovery, and the sixth request including model discovery information and identification information of the management service producer of domain B.

[0193] In some cases, the management service consumer of domain A needs to discover information of related models stored by the management service producer of domain B. In this case, the management service consumer of domain A can send a sixth request to the management service producer of domain A. Correspondingly, the management service producer of domain A can receive the sixth request from the management service consumer of domain A, and the sixth request can be used to request model discovery, and the sixth request can include model discovery information and identification information of the management service producer of domain B.

[0194] The model discovery information can include one or more of an identification of a model, a name of the model, a type of the model, a size of the model, a service state of the model, and an applicable scenario of the model.

[0195] 702a. The management service producer of the domain A sends a seventh request to the management service producer of the domain B, the seventh request is used to request model discovery, and the seventh request comprises model discovery information.

[0196] After receiving the sixth request from the management service consumer of the domain A, the management service producer of the domain A can send a seventh request to the management service producer of the domain B based on the sixth request. For example, the management service producer of the domain A can determine that the management service consumer of the domain A needs to discover the information of the related model stored by the management service producer of the domain B based on the identification information of the management service producer of the domain B in the sixth request, and then the management service producer of the domain A can send the seventh request to the management service producer of the domain B. Correspondingly, the management service producer of the domain B can receive the seventh request from the management service producer of the domain A, the seventh request can be used to request model discovery, and the seventh request can comprise model discovery information, i.e., the model discovery information carried in the sixth request.

[0197] 703a. The management service producer of the domain B sends a response to the seventh request to the management service producer of the domain A according to the model discovery information and the information of the stored model, and the response to the seventh request comprises the information of one or more models.

[0198] For example, after receiving the seventh request from the management service producer of the domain A, the management service producer of the domain B can determine the information of one or more models in the information of the currently stored model of the management service producer of the domain B that matches the model discovery information carried in the seventh request, and then the management service producer of the domain B can send a response to the seventh request to the management service producer of the domain A, and the response to the seventh request can comprise the information of one or more models that matches the model discovery information. Specifically, after receiving the seventh request from the management service producer of the domain A, the management service producer of the domain B can compare the model discovery information carried in the seventh request with the information of the currently stored model, if the information of a certain model currently stored is the same as the model discovery information, it indicates that the information of the model matches the model discovery information, and if the information of a certain model currently stored is different from the model discovery information, it indicates that the information of the model does not match the model discovery information.

[0199] It should be understood that in some cases, the information of the model currently stored by the management service producer of domain B can not include the information of the model matching the model discovery information carried in the seventh request, in which case the response to the seventh request can carry first indication information, and the first information can be used to indicate that the information of the model currently stored by the management service producer (of domain B) does not include the information of the model matching the model discovery information carried in the seventh request. Alternatively, the management service producer of domain B can not send a response to the seventh request to the management service producer of domain A, or the response to the seventh request can be empty, both of which by default represent that the information of the model currently stored by the management service producer of domain B does not include the information of the model matching the model discovery information carried in the seventh request.

[0200] 704a. The management service producer of domain A sends a response to the sixth request to the management service consumer of domain A based on the response to the seventh request, and the response to the sixth request includes the information of one or more models.

[0201] After receiving the response to the seventh request from the management service producer of domain B, the management service producer of domain A can send a response to the sixth request to the management service consumer of domain A based on the response to the seventh request. Accordingly, the management service consumer of domain A can receive the response to the sixth request from the management service producer of domain A, and the response to the sixth request includes the information of one or more models, i.e., the information of one or more models carried in the seventh request.

[0202] It can be understood that steps 701a and 704a are optional. For example, in some possible embodiments, the management service producer of domain A can actively send a model discovery request (such as the seventh request) to the management service producer of domain B to synchronize the information of the relevant models stored by the management service producer of domain B.

[0203]

Embodiment Three

[0204] Embodiment Three is a related processing flow of model update. In Embodiment Three, the model provider can send a model update request (ModelUpdate_request) to the management service producer, and can carry the relevant information of the model to be updated in the model update request. Specifically, please refer to FIG. 8, which is a flow diagram of a model update according to an embodiment of the present application. As shown in FIG. 8, the model update flow can include but is not limited to the following steps:

[0205] 801. The model provider sends a fourth request to the management service producer, and the fourth request is used to request model update, and the fourth request includes update information of a first model.

[0206] It can be understood that after the model provider registers the information of a certain model to the management service producer, the information of the model can be changed. For example, the model provider can change the service address (serviceURL) of the model, the model provider can also change the name (largeModelName) of the model, the model provider can further change the service status (status) of the model, the model provider can also regenerate a new key (key), and the like. In the case that the information of the model is changed, the model provider can send a model update request to the management service producer to update the information of the related model currently stored by the management service producer, so as to ensure the accuracy of the information of the related model stored by the management service producer.

[0207] For example, in the case that the information of the first model is changed, the model provider can send a fourth request to the management service producer. Accordingly, the management service producer can receive the fourth request from the model provider, the fourth request can be used to request model update, and the fourth request can include the update information of the first model. Wherein, the update information of the first model can include the information changed in the information of the first model. For example, assuming that the service address of the first model is changed from “http: / / a1.b1.c1.d1:port1 / chat” to “http: / / a1.b1.c1.d1:port2 / chat”, then the update information of the first model can include “http: / / a1.b1.c1.d1:port2 / chat”. For another example, assuming that the key of the first model is changed from key1 to key2, then the update information of the first model can include key2. It should be understood that in addition to the update information of the first model, the fourth request can also include the identification information of the first model, and the identification information of the first model can indicate the first model to be updated. The identification information of the first model can include the identification of the first model, the name of the first model, and the like, which can identify the first model.

[0208] It should be understood that the fourth request for requesting model update can also be understood as the fourth request for requesting model instance update.

[0209] It should be understood that the above description of the information carried in the fourth request is only an example, and the embodiments of the present application are not limited thereto. For example, the model update request can carry the update information of multiple models, such as the fourth request can also carry the update information of the second model.

[0210] 802. The management service producer updates the currently stored information of the first model according to the update information of the first model.

[0211] After receiving the fourth request from the model provider, the management service producer can update the information of the first model currently stored by the management service producer according to the update information of the first model carried in the fourth request. For example, the fourth request can include the identification information of the first model, and the management service producer can determine that the information of the first model needs to be updated according to the identification information of the first model.

[0212] For example, assuming that the update information of the first model includes the changed service address "http: / / a1.b1.c1.d1:port2 / chat" of the first model and the changed key key2 of the first model, the management service producer can update the service address "http: / / a1.b1.c1.d1:port1 / chat" of the first model currently stored by the management service producer to "http: / / a1.b1.c1.d1:port2 / chat", and can also update the key key1 of the first model currently stored by the management service producer to key2.

[0213] In some possible implementation, after receiving the fourth request, the management service producer can perform authorization verification on the fourth request. If the authorization verification is passed, the management service producer can update the information of the first model currently stored by the management service producer according to the update information of the first model carried in the fourth request. If the authorization verification is not passed, the management service producer can not perform processing, and can discard the fourth request.

[0214] 803. The management service producer sends a response to the fourth request to the model provider, where the response to the fourth request is used to indicate that the model update is successful.

[0215] After updating the information of the first model currently stored by the management service producer according to the update information of the first model carried in the fourth request, the management service producer can send a response to the fourth request to the model provider. Accordingly, the model provider can receive the response to the fourth request from the management service producer, where the response to the fourth request is used to indicate that the model update is successful, that is, to indicate that the first model is successfully updated or the information of the first model is successfully updated, that is, to indicate that the information of the first model currently stored by the management service producer is successfully updated to the update information of the first model carried in the fourth request. It should be understood that the response to the fourth request is also a model update request response (ModelUpdate_response).

[0216] In the above processing flow, the management service producer can provide a model update service to the outside, and the model provider can send a model update request to the management service producer to update the information of the model currently stored (registered) by the management service producer, so as to ensure the accuracy of the information of the model stored by the management service producer.

[0217] Embodiment Four

[0218] Embodiment four is a related processing flow of model deregistration. In embodiment four, the model provider can send a model deregistration request (ModelDeregister_request) to the management service producer, and can carry the related information of the model to be deregistered in the model deregistration request. Specifically, refer to FIG. 9, which is a flow diagram of a model deregistration flow disclosed in an embodiment of the present application. As shown in FIG. 9, the model deregistration flow can include but is not limited to the following steps:

[0219] 901. The model provider sends a fifth request to the management service producer, the fifth request being used to request model deregistration, and the fifth request including identification information of the first model.

[0220] In an embodiment of the present application, the model provider can send a deregistration request to the management service producer to deregister the related model registered before. For example, it is assumed that the model provider needs to permanently offline the deployed first model, in which case the model provider can deregister the first model registered before, and the model provider sends a fifth request to the management service producer. Correspondingly, the management service producer can receive the fifth request from the model provider, the fifth request being used to request model deregistration, and the fifth request including identification information of the model to be deregistered, i.e., identification information of the first model. The identification information of the first model can include information that can identify the first model, such as an identifier of the first model, a name of the first model, etc., which are not limited herein.

[0221] It should be understood that the fifth request used to request model deregistration can also be understood as the fifth request used to request model instance deregistration.

[0222] It should be understood that the above description of the information carried in the fifth request is only an example, and embodiments of the present application are not limited thereto. For example, the model deregistration request can carry identification information of multiple models to be deregistered, such as identification information of a second model in the fifth request.

[0223] 902. The management service producer deregisters the first model according to the identification information of the first model in the fifth request.

[0224] After receiving the fifth request from the model provider, the management service producer can determine that the first model needs to be deregistered according to the identification information of the first model carried in the fifth request. For example, the management service producer can mark the first model as unavailable, and can delete the stored information of the first model.

[0225] In some possible implementation manners, after receiving the fifth request, the management service producer can perform authorization verification on the fifth request, and in a case where the authorization verification is passed, the management service producer can deregister the first model according to the identification information of the first model carried in the fifth request, and in a case where the authorization verification is not passed, the management service producer can not perform processing, and can discard the fifth request.

[0226] 903. The management service producer sends a response to the model provider for the fifth request, where the response to the fifth request is used to indicate that the model deregistration is successful.

[0227] After deregistering the first model according to the identification information of the first model carried in the fifth request, the management service producer can send a response to the fifth request to the model provider. Correspondingly, the model provider can receive the response to the fifth request from the management service producer, where the response to the fifth request is used to indicate that the model deregistration is successful, that is, information indicating that the first model is successfully deregistered or the first model is successfully deregistered. It should be understood that the response to the fifth request is also a model deregistration request response (ModelDeregister_response).

[0228] In the above processing flow, the management service producer can provide a model deregistration service to the outside, and the model provider can send a model deregistration request to the management service producer to deregister the information of the model currently stored (registered) by the management service producer, so as to ensure the accuracy of the information of the model stored by the management service producer.

[0229] Embodiment Five

[0230] Embodiment Five is a related processing flow of model subscription. In Embodiment Five, the management service consumer can send a model subscription request (ModelSubscribe_request) to the management service producer, and can carry information of a model to be subscribed in the model subscription request, such as identification of the model, name of the model, and the like. Specifically, refer to FIG. 10, which is a flow diagram of a model subscription according to an embodiment of the present application. As shown in FIG. 10, the model subscription flow can include but is not limited to the following steps:

[0231] 1001. The management service consumer sends a third request to the management service producer, where the third request is used to request to perform model subscription, and the third request includes identification information of a first model.

[0232] It can be understood that after the management service consumer sends the model discovery request to the management service producer to obtain the information of one or more models, the information of the one or more models can be updated, and the one or more models can also be unregistered. Therefore, in order to ensure the accuracy of the information of the model used by the management service consumer, the management service consumer can send a subscription request to the management service producer to subscribe to the information of the related model. For example, assuming that the management service consumer needs to subscribe to the information of the first model, the management service consumer can send a third request to the management service producer. Accordingly, the management service producer can receive the third request from the management service consumer, and the third request can be used to request model subscription. The third request can include identification information of the model that needs to be subscribed, i.e., identification information of the first model. The identification information of the first model can include information that can identify the first model, such as an identifier of the first model, a name of the first model, and the like, which are not limited herein.

[0233] In some possible implementation manners, the management service consumer can not want to subscribe to all the information of the first model. In this case, the third request can further include second indication information, and the second indication information can indicate the information of the first model that needs to be subscribed. For example, if the management service consumer only wants to subscribe to the key of the first model and the service address of the first model, the third request can include the second indication information, and the second indication information can be ("key" and "serviceURL").

[0234] It should be understood that the third request for requesting model subscription can also be understood as the third request for requesting model instance subscription.

[0235] It should be understood that the above description of the information carried in the third request is only an example, and embodiments of the present application are not limited thereto. For example, the model subscription request can carry identification information of multiple models that need to be subscribed, such as identification information of the second model in the fifth request.

[0236] 1002. In a case where the information of the first model changes, the management service producer sends the changed information of the first model to the management service consumer.

[0237] After the management service producer receives the third request from the management service consumer, the management service producer can determine that the first model needs to be subscribed according to the identification information of the first model carried in the third request. When the information of the first model changes later, the management service producer can send the changed information of the first model to the management service consumer. It should be understood that the changed information of the first model sent by the management service producer to the management service consumer can only include information of the first model that has changed, and information of the first model that has not changed can not be sent.

[0238] For example, assuming the second indication information is (‘key’ and ‘serviceURL’), indicating the key and the service address of the first model, in this case, the management service producer can send the changed key of the first model (only the key of the first model is changed) to the management service consumer, or send the changed service address of the first model (only the service address of the first model is changed) to the management service consumer, or send the changed key of the first model and the changed service address of the first model (both the key and the service address of the first model are changed) to the management service consumer, only when at least one of the two information (the key and the service address) of the first model is changed.

[0239] In some possible implementation, after receiving the third request, the management service producer can perform authorization verification on the third request, and if the authorization verification is passed, the management service producer can subscribe to the first model according to the identification information of the first model carried in the third request, and if the authorization verification is not passed, the management service producer can not perform processing, and can discard the third request.

[0240] It can be understood that in some cases, after receiving the third request from the management service consumer, the management service producer can send a response of the third request to the management service consumer. Accordingly, the management service producer can receive the response of the third request from the management service consumer, and the response of the third request can be used to indicate that the model subscription is successful, i.e., to indicate that the first model is successfully subscribed or the information of the first model is successfully subscribed. It should be understood that the response of the third request is also a model subscription request response (ModelSubscribe_response).

[0241] In the above processing flow, the management service producer can provide a model subscription service to the outside, and the management service consumer can send a model subscription request to the management service producer to subscribe to the information of the model stored (registered) by the management service producer, so as to ensure the accuracy of the information of the model obtained / used by the management service consumer.

[0242] It should be noted that the related information (i.e., the same information or similar information) and the related description in the above different embodiments can be mutually referred.

[0243] The above mainly introduces the communication method provided by the embodiments of the present application. It can be understood that the above management service producer, model provider and management service consumer can contain hardware structure and / or software module for performing the corresponding functions in order to realize the above corresponding functions. The units and steps of each example described in combination with the embodiments disclosed herein can be realized in the form of hardware or a combination of hardware and computer software. Whether a certain function is performed by hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods for each specific application to implement the described functions, but such implementation should not be considered beyond the scope of the embodiments of the present application.

[0244] The embodiments of the present application can divide the functional modules of the management service producer, model provider and management service consumer according to the above method examples. For example, each functional module can be divided according to each function, or two or more functions can be integrated in one module. The above integrated module can be realized in the form of hardware or software functional module. It should be noted that the division of the modules by the embodiments of the present application is illustrative, and is only a logical functional division. Actual implementation can have another division manner.

[0245] In the case of dividing each functional module according to each function, FIG. 11 shows a possible structural schematic diagram of a communication device 1100. The communication device 1100 includes a receiving unit 1101 and a storage unit 1102, and the communication device 1100 can also include a sending unit 1103. In a possible design, the communication device 1100 can be the above management service producer, or can be a chip in the management service producer, or can be a processing system, chip system, circuit or functional module in the management service producer, etc. Wherein:

[0246] The receiving unit 1101 is configured to receive a first request from a model provider, the first request being used to request model registration, and the first request including information of one or more models; and the storage unit 1102 is configured to store the information of the one or more models.

[0247] In a possible implementation, the receiving unit 1101 is further configured to receive a second request from a management service consumer, the second request being used to request model discovery, and the second request including model discovery information; and the device 1100 can further include a sending unit 1103, configured to send a response of the second request to the management service consumer according to the model discovery information and the stored model information, and the response of the second request including information of one or more models.

[0248] In a possible implementation, the receiving unit 1101 is further configured to receive a third request from the management service consumer, the third request being used to request model subscription, and the third request comprising identification information of a first model; and the sending unit 1103 is further configured to send, to the management service consumer, information of the first model after the information of the first model is changed.

[0249] In a possible implementation, the information of the model comprises one or more of the following: identification of the model, name of the model, download address of the model, service address of the model, type of the model, size of the model, key of the model, service state of the model, and function description information of the model.

[0250] In a possible implementation, the information of the model further comprises applicable scenarios of the model.

[0251] In a possible implementation, the information of the model further comprises information of one or more models associated with the applicable scenarios of the model.

[0252] In a possible implementation, the model discovery information comprises one or more of the following: identification of the model, name of the model, type of the model, size of the model, service state of the model, and applicable scenarios of the model.

[0253] In a possible implementation, the sending unit 1103 is further configured to send, to the model provider, a response to the first request, the response being used to indicate that model registration is successful.

[0254] The specific operations of each unit in the communication apparatus 1100 described above can be refer to the corresponding description of the management service producer in the method embodiments described above, such as the corresponding description of the management service producer in FIG. 6-10, and details are not described herein again.

[0255] FIG. 12 shows a possible structural schematic diagram of a communication apparatus 1200. The communication apparatus 1200 comprises a processing unit 1201 and a sending unit 1202, and the communication apparatus 1200 can further comprise a receiving unit 1203. In a possible design, the communication apparatus 1200 can be the model provider described above, or can be a chip in the model provider, or can be a processing system, a chip system, a circuit or a functional module in the model provider, and the like. Wherein:

[0256] The processing unit 1201 is configured to obtain information of one or more models; and the sending unit 1202 is configured to send, to a management service producer, a first request, the first request being used to request model registration, and the first request comprising the information of the one or more models.

[0257] In a possible implementation, the information of the model includes one or more of the following: an identifier of the model, a name of the model, a download address of the model, a service address of the model, a type of the model, a size of the model, a key of the model, a service state of the model, and function description information of the model.

[0258] In a possible implementation, the information of the model further includes a scenario applicable to the model.

[0259] In a possible implementation, the information of the model further includes information of one or more models associated with the applicable scenario of the model.

[0260] In a possible implementation, the apparatus 1200 further includes a receiving unit 1203 configured to receive a response to the first request from the management service producer, where the response to the first request is used to indicate that the model registration is successful.

[0261] The specific operations of each unit in the communication apparatus 1200 described above can refer to the corresponding description of the model provider in the method embodiments described above, such as the corresponding description of the model provider in FIGS. 6-10, and details are not described herein again.

[0262] FIG. 13 shows a possible structural schematic diagram of a communication apparatus 1300. The communication apparatus 1300 includes a sending unit 1301 and a receiving unit 1302. In a possible design, the communication apparatus 1300 can be the management service consumer described above, or can be a chip in the management service consumer, or can be a processing system, a chip system, a circuit, or a functional module in the management service consumer, and the like. Wherein:

[0263] The sending unit 1301 is configured to send a second request to a management service producer, where the second request is used to request model discovery, and the second request includes model discovery information; and the receiving unit 1302 is configured to receive a response to the second request from the management service producer, where the response to the second request includes information of one or more models associated with the model discovery information in information of models stored by the management service consumer.

[0264] In a possible implementation, the sending unit 1301 is further configured to send a third request to the management service producer, where the third request is used to request model subscription, and the third request includes identification information of a first model; and the receiving unit 1302 is further configured to receive information of the first model after a change in a case where the information of the first model changes.

[0265] In a possible implementation, the information of the model includes one or more of the following: an identifier of the model, a name of the model, a download address of the model, a service address of the model, a type of the model, a size of the model, a key of the model, a service state of the model, and function description information of the model.

[0266] In a possible implementation, the information of the model further includes a scenario applicable to the model.

[0267] In a possible implementation, the information of the model further includes information of one or more models associated with the applicable scenario of the model.

[0268] In a possible implementation, the model discovery information includes one or more of the following: an identifier of the model, a name of the model, a type of the model, a size of the model, a service state of the model, and a scenario applicable to the model.

[0269] The specific operations of each unit in the communication apparatus 1300 can be referred to the foregoing method embodiments or the corresponding description of managing a service consumer, such as the corresponding description of managing a service consumer in FIGS. 6-10, and details are not described herein.

[0270] In a possible implementation, the sending unit in the communication apparatus 1100, the communication apparatus 1200, and the communication apparatus 1300 can be a transmitter, and the receiving unit can be a receiver. The sending unit and the receiving unit can be integrated into one device, for example, a transceiver. In an example, the communication apparatus 1100, the communication apparatus 1200, and the communication apparatus 1300 can further include a processing unit, which can be one or more processors / logic circuits. In the process of executing the foregoing method, the process of sending information (for example, sending the first request) in the foregoing method can be understood as the process of outputting the foregoing information by the processor. When the foregoing information is output, the processor can output the foregoing information to the transceiver, so that the transceiver transmits the foregoing information. After the foregoing information is output by the processor, the foregoing information can need to be further processed before reaching the transceiver. Similarly, the process of receiving information (for example, receiving the first request) in the foregoing method can be understood as the process of receiving inputted foregoing information by the processor. When the processor receives the inputted information, the transceiver receives the foregoing information and inputs the foregoing information to the processor. Further, after the transceiver receives the foregoing information, the foregoing information can need to be further processed before being inputted to the processor. In the embodiments of the present application, the processor and the transceiver can be coupled, and the connection manner between the processor and the transceiver is not limited in the embodiments of the present application.

[0271] In another possible implementation, in the communication apparatus 1100, the communication apparatus 1200, and the communication apparatus 1300, the sending unit can be an output interface, the receiving unit can be an input interface, the sending unit and the receiving unit are integrated into one unit, for example, an input-output interface, or a communication interface, or an interface circuit, or an interface, etc.

[0272] Fig. 14 shows a possible hardware structure of the communication apparatus 1400 provided by the embodiment of the application. The communication apparatus 1400 can include a communication interface 1404 and at least one processor 1402. Optionally, it can also include a bus 1403. Further optionally, it can also include at least one memory 1401, wherein the memory 1401, the processor 1402, and the communication interface 1404 can be connected through the bus 1403.

[0273] The memory 1401 is configured to provide a storage space, in which an operating system and computer programs, etc. can be stored. The memory 1401 can be one or a combination of a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a compact disc read-only memory (CD-ROM), etc.

[0274] The processor 1402 is a module for arithmetic operation and / or logical operation, and can be one or a combination of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor unit (MPU), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a complex programmable logic device (CPLD), etc.

[0275] The communication interface 1404 is configured to receive data transmitted from outside and / or transmit data to outside, and can be a wired link interface including an Ethernet cable, or a wireless link (Wi-Fi, Bluetooth, general wireless transmission, etc.) interface. Optionally, the communication interface 1404 can further include a transmitter (such as a radio frequency transmitter, an antenna, etc.) coupled with the interface, or a receiver, etc.

[0276] In one design, the communication apparatus 1400 can be configured to perform the functions of the service producer in the foregoing embodiments. For details, refer to the related description in FIGS. 6-10, which will not be repeated here.

[0277] In another design, the communication apparatus 1400 can be configured to perform the functions of the model provider in the foregoing embodiments. For details, refer to the related description in FIGS. 6-10, which will not be repeated here.

[0278] In yet another design, the communication apparatus 1400 can be configured to perform the functions of the service consumer in the foregoing embodiments. For details, refer to the related description in FIGS. 6-10, which will not be repeated here.

[0279] In one possible design, the memory 1401 can store instructions, which can be a computer program, and the computer program can run on the processor 1402 to make the communication apparatus 1400 perform the operations performed by the service producer or the operations performed by the model provider or the operations performed by the service consumer in any of the method embodiments. For details, refer to the related description in FIGS. 6-10, which will not be repeated here.

[0280] It should be noted that the communication apparatus 1400 shown in FIG. 14 is only one implementation of the embodiments of the present application, and in actual applications, the communication apparatus 1400 can further include more or fewer components, which are not limited here.

[0281] It should be understood that the transmission in the embodiments of the present application can be direct transmission or indirect transmission. Direct transmission means that one device or module transmits information / data directly to the corresponding device or module, and indirect transmission means that one device or module transmits information / data to the corresponding device or module through other devices or modules.

[0282] It is apparent that the described embodiments are only some but not all of the embodiments of the present application. Referring to "an embodiment" herein means that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the application. The appearances of the phrase "in an embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are they necessarily mutually exclusive of one another. As used in this application, the term "exemplary" is intended to present one or more examples only, and is not intended to convey an indication of a preferred or important embodiment. As used in this application, the terms "identical" or "exactly identical" are intended to mean that the two sequences are identical in every position. As used in this application, the terms "first", "second", "third", etc. are used to distinguish different objects, not necessarily in order of importance. Furthermore, the terms "comprises", "comprising", "includes", "including", "has", "having" and their variants are intended to cover non-exclusive inclusions. For example, a process, method, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements, but can include additional steps or elements not expressly listed or inherent to such process, method, article, or apparatus. It is further understood that with respect to the architecture of a system or apparatus having multiple devices or modules, if one device or module generates an information and another device or module utilizes the information, the manner in which the other device obtains the information can be varied, such as, for example, the device or module that generates the information can send the information directly to the device or module that utilizes the information (direct sending), or the device or module that generates the information can send the information to the device or module that utilizes the information through other devices or modules (indirect sending).

[0283] It is understood that only those parts of the apparatus pertinent to the application are shown in the drawings. It should be understood that some of the exemplary embodiments are described as processes or methods depicted as flow diagrams. Although the processes are described in a particular sequential order, many of the processes can be performed concurrently, in parallel, or simultaneously. In addition, the order of the processes can be re-arranged. The processes can be terminated when their operations are completed, but the processes can also end in response to events that are external to some of the processes. The processes can correspond to methods, functions, procedures, subroutines, subprograms, etc.

[0284] The terms “component,” “module,” “system,” “unit,” etc., used in this specification are used to refer to computer-related entities, hardware, firmware, combinations of hardware and software, software, or software in execution. For example, a unit can be, but is not limited to, a process running on a processor, a processor, an object, an executable file, a thread of execution, a program, and / or distributed between two or more computers. Furthermore, these units can be executed from various computer-readable media on which various data structures are stored. For example, a unit can communicate via local and / or remote processes based on signals having one or more data packets (e.g., data from a second unit interacting with another unit between a local system, a distributed system, and / or a network; for example, the Internet interacting with other systems via signals).

[0285] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above description is only a specific embodiment of this application and is not intended to limit the scope of protection of this application. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of this application should be included within the scope of protection of this application.

Claims

A communication method characterized by comprising: The method applied to a management service producer comprises: receiving a first request from a model provider, the first request being used for requesting model registration, the first request comprising information of one or more models; storing the information of the one or more models. The method of claim 1, wherein The method further comprises: receiving a second request from a management service consumer, the second request being used for requesting model discovery, the second request comprising model discovery information; sending a response of the second request to the management service consumer according to the model discovery information and the stored information of models, the response of the second request comprising information of one or more models. The method according to claim 1 or 2, characterized in that The method further comprises: receiving a third request from the management service consumer, the third request being used for requesting model subscription, the third request comprising identification information of a first model; in a case where information of the first model changes, sending the changed information of the first model to the management service consumer. The method according to any one of claims 1 to 3, characterized in that The information of the model comprises one or more of the following: identification of the model, name of the model, download address of the model, service address of the model, type of the model, size of the model, key of the model, service state of the model, and function description information of the model. The method according to claim 4, characterized in that The information of the model further comprises applicable scenarios of the model. The method according to claim 5, characterized in that The information of the model further comprises information of one or more models associated with the applicable scenarios of the model. The method according to any one of claims 2-6, characterized in that The model discovery information comprises one or more of the following: identification of the model, name of the model, type of the model, size of the model, service state of the model, and applicable scenarios of the model. The method according to any one of claims 1 to 7, characterized in that The method further comprises: sending a response of the first request to the model provider, the response of the first request being used for indicating successful model registration. A communication method characterized by comprising: The method applied to a model provider comprises: obtaining information of one or more models; sending a first request to a management service producer, the first request being used for requesting model registration, the first request comprising the information of the one or more models. The method of claim 9, wherein The information of the model comprises one or more of the following: identification of the model, name of the model, download address of the model, service address of the model, type of the model, size of the model, key of the model, service state of the model, and function description information of the model. The method of claim 10, wherein The information of the model further comprises applicable scenarios of the model. The method of claim 11, wherein The information of the model further comprises information of one or more models associated with the applicable scenarios of the model. The method according to any one of claims 9-12, characterized in that The method further comprises: receiving a response of the first request from the management service producer, the response of the first request being used for indicating successful model registration. A communication method characterized by comprising: The method applied to a management service consumer comprises: sending a second request to a management service producer, the second request being used for requesting model discovery, the second request comprising model discovery information; receiving a response of the second request from the management service producer, the response of the second request comprising information of one or more models associated with the model discovery information in information of models stored by the management service consumer. The method of claim 14, wherein The method further comprises: sending a third request to the management service producer, the third request being used to request a model subscription, the third request comprising identification information of a first model; in a case where information of the first model changes, receiving the changed information of the first model from the management service producer. The method according to claim 14 or 15, characterized in that The information of the model comprises one or more of the following: identification of the model, name of the model, download address of the model, service address of the model, type of the model, size of the model, key of the model, service status of the model, function description information of the model. The method of claim 16, wherein The information of the model further comprises applicable scenarios of the model. The method of claim 17, wherein The information of the model further comprises information of one or more models associated with the applicable scenarios of the model. The method according to any one of claims 14-18, characterized in that The model discovery information comprises one or more of the following: identification of the model, name of the model, type of the model, size of the model, service status of the model, applicable scenarios of the model. A communication system characterized by The system comprises a management service producer, a model provider and a management service consumer, the management service producer is configured to implement the method of any one of claims 1-8; the model provider is configured to implement the method of any one of claims 9-13; the management service consumer is configured to implement the method of any one of claims 14-19. A communication device characterized by comprising: The system comprises a processor and a communication interface; the communication interface is configured to receive and / or send data; the processor invokes computer programs or computer instructions stored in a memory to implement the method of any one of claims 1-8, or to implement the method of any one of claims 9-13, or to implement the method of any one of claims 14-19. A computer-readable storage medium, characterized by The computer readable storage medium stores computer programs or computer instructions, which are executed by the processor to implement the method of any one of claims 1-8, or to implement the method of any one of claims 9-13, or to implement the method of any one of claims 14-19. A computer program product, characterized in that The computer program product comprises computer program codes or computer instructions, which, when executed, implement the method of any one of claims 1-8, or implement the method of any one of claims 9-13, or implement the method of any one of claims 14-19.

Images