Model updating method, electronic equipment and storage medium
By generating and testing models in a production environment, the problem that the model cannot meet production needs is solved, and the real-time and adaptability of the model is improved.
Patent Information
- Application Number
- CN202510453304.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-11
- Publication Date
- 2025-08-19
AI Technical Summary
Models generated or updated in the prior art may not meet the actual production needs, resulting in poor results in actual applications.
By obtaining business data in a production environment, performing training, evaluation or optimization operations, generating a second model, and testing it in a test environment, deploying the model in a production environment based on the test information to ensure that the model meets production needs.
It improves the real-time performance of the model in the production environment and the ability to be close to actual needs, ensuring that the model can better meet production needs after deployment.
Smart Images
Figure CN120509477A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular to a model updating method, electronic device, and storage medium. Background Art
[0002] In large model generation or update scenarios, model generation or updates can be performed in a test environment using data from a production environment.
[0003] However, this method has the problem that the generated or updated model may not meet the actual production needs and may not meet expectations. Summary of the Invention
[0004] The purpose of this application is to provide a model updating method, electronic device and storage medium, aiming to solve the technical problem in related technologies that the model may not meet the actual production needs.
[0005] To achieve the above objectives, this application adopts the following technical solutions:
[0006] This application provides a model updating method, comprising:
[0007] Obtain business data generated in the production environment;
[0008] Performing a target operation on the first model in a production environment based on business data to obtain a second model; the target operation includes at least one of the following: training, evaluation, and optimization;
[0009] Synchronize the second model to the test environment;
[0010] Performing a test on the second model in the test environment to obtain first test information; the first test information is used to indicate whether the second model passes the test or fails the test;
[0011] The first test information is synchronized to the production environment, and if the first test information indicates that the second model has passed verification, the second model is deployed in the production environment.
[0012] In the embodiments of this application, the production environment and the test environment are isolated from each other to ensure data security. However, this will result in a significant delay when the test environment obtains data from the production environment. Because the production environment can obtain the latest production-related data in a timely manner, training, evaluating, or optimizing the first model in the production environment can ensure that the resulting second model is more closely aligned with actual production needs and has greater real-time performance.
[0013] In some embodiments, performing a target operation on the first model in a production environment based on business data to obtain a second model includes:
[0014] When the amount of business data exceeds a data amount threshold, performing a target operation on the first model in a production environment based on the business data to obtain a second model; or
[0015] When the model performance of the first model is less than or equal to the performance threshold, a target operation is performed on the first model in a production environment based on the business data to obtain a second model.
[0016] In some embodiments, the method further comprises:
[0017] Collecting, in a production environment, predicted data generated by the first model within a preset period of time and real data corresponding to the predicted data;
[0018] Determining performance parameters of the first model within a preset time period based on the predicted data and the actual data;
[0019] Based on the performance parameter, it is determined whether a model performance of the first model is less than or equal to a performance threshold.
[0020] In some embodiments, the performance parameters include at least one of the following: prediction accuracy, response delay, number of requests processed per unit time, tolerance to input data noise, number of model crashes due to abnormal inputs, and number of crashes due to environmental problems.
[0021] In some embodiments, when the performance parameter includes one parameter, determining whether the model performance of the first model is less than or equal to the performance threshold based on the performance parameter includes: determining that the model performance of the first model is less than or equal to the performance threshold if a numerical value of the performance parameter is less than or equal to a numerical threshold corresponding to the performance parameter;
[0022] In a case where the performance parameter includes a plurality of parameters, determining whether the model performance of the first model is less than or equal to a performance threshold based on the performance parameter includes:
[0023] For each parameter in the performance parameters, based on the value of the parameter and the numerical threshold corresponding to the parameter, a score corresponding to the parameter is determined to obtain a score for each parameter in the performance parameters;
[0024] Perform weighted fusion on the scores of each parameter to obtain the scores corresponding to the performance parameters;
[0025] When the score corresponding to the performance parameter is less than or equal to the score threshold, it is determined that the model performance of the first model is less than or equal to the performance threshold.
[0026] In some embodiments, when the first test information indicates that the second model fails the test, the first test information also includes abnormality information of the second model; and the method further includes:
[0027] Based on the abnormal information, the second model is updated to obtain a third model;
[0028] Synchronizing the third model to the test environment, executing a test on the third model in the test environment, and obtaining second test information; the second test information is used to indicate whether the third model passes the test;
[0029] Synchronize the second test information to the production environment;
[0030] If the second test information indicates that the third model passes the test, the third model is deployed in the production environment.
[0031] In some embodiments, performing a test on the model to be tested in the test environment to obtain test information is achieved by:
[0032] Obtain test data sets corresponding to multiple test indicators;
[0033] For a test data set corresponding to each test indicator in the multiple test indicators, a test is performed on the model to be tested based on the test data set corresponding to the test indicator to obtain test information.
[0034] In some embodiments, multiple test indicators have test priorities; the higher the test priority of a test indicator, the earlier the test indicator is tested; for a test data set corresponding to each test indicator in the multiple test indicators, based on the test data set corresponding to the test indicator, a test is performed on the to-be-tested model to obtain test information, including:
[0035] Among the untested test indicators, based on the test data set of the test indicator with the highest priority, the test model to be tested is tested to obtain the test results corresponding to the test indicators;
[0036] If the test result corresponding to the test indicator is a failure, the test is stopped and the generated test result is determined as the test information;
[0037] When the test result corresponding to the test indicator is a pass, continue to select the test data set of the test indicator with the highest priority from the untested test indicators to perform the test on the test model until the test result of the test indicator is a fail.
[0038] In some embodiments, the present application provides an electronic device comprising: a processor and a memory configured to store processor-executable instructions; wherein the processor is configured to execute the instructions to implement any one of the above optional methods.
[0039] In some embodiments, the present application provides a computer-readable storage medium having instructions stored thereon. When the instructions in the computer-readable storage medium are executed by an electronic device, the electronic device is enabled to execute any of the above-mentioned optional methods. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.
[0041] Figure 1 A schematic diagram of the architecture of a model updating system provided in this application;
[0042] Figure 2 A flowchart of a model updating method provided in this application;
[0043] Figure 3 This is a structural diagram of a model updating device provided in this application. DETAILED DESCRIPTION
[0044] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.
[0045] In the description of this application, it should be understood that the terms "upper," "lower," "left," "right," "front," "back," "inner," "outer," and the like, indicating directions or positional relationships, are based on the directions or relative positional relationships shown in the accompanying drawings and are intended solely to facilitate the description of this application and simplify the description. They do not indicate or imply that the devices or components referred to must have a specific direction, be constructed, or operate in a specific direction. Therefore, they should not be construed as limitations on this application. Unless otherwise specified, the above-mentioned directionality descriptions may be flexibly set in actual application, provided that the relative positional relationships shown in the accompanying drawings are met.
[0046] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature specified as "first" or "second" may explicitly or implicitly include one or more of such features. Throughout this application, unless otherwise specified, "plurality" means two or more.
[0047] In the description of this application, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," "connected," and "connected" should be understood broadly. For example, they may refer to fixed connections, detachable connections, or integral connections. They may be directly connected, indirectly connected through an intermediary, or internally connected between two components. Those skilled in the art will understand the specific meanings of these terms in this application based on the specific circumstances.
[0048] In the embodiments of the present application, the terms "comprises," "comprising," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, article, or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not preclude the presence of other identical elements in the process, article, or device comprising the element.
[0049] In the embodiments of this application, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of this application should not be interpreted as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.
[0050] Exemplarily, the electronic device that executes the model updating method provided in the embodiments of the present application may be a mobile phone, tablet computer, desktop computer, laptop computer, handheld computer, notebook computer, ultra-mobile personal computer (UMPC), netbook computer, cellular phone, personal digital assistant (PDA), augmented reality (AR) or virtual reality (VR) device. The embodiments of the present application do not impose any particular restrictions on the specific form of the electronic device. It can interact with the user through one or more methods such as a keyboard, touchpad, touch screen, remote control, voice interaction, or handwriting device.
[0051] For example, Figure 1 The figure shows an architecture diagram of a model update system provided by this application, including: a test environment and a production environment.
[0052] The test environment includes intelligent business applications, model storage, image library, and model services.
[0053] The agent business application is used to test the model to be tested (or referred to as the agent to be tested). Optionally, a developer dev can perform development operations in the agent business application.
[0054] Model storage, including Object Storage Service (OBS), is used to store model data or synchronize data with the Object Storage Service in the model storage in the production environment.
[0055] A mirror library is used to store model images or to synchronize models or model images with the mirror library in the production environment.
[0056] Model services are used to provide support for intelligent business applications or be called by intelligent business applications to perform tests on models.
[0057] The production environment includes intelligent business applications, model experiments, model storage, image libraries, and model services.
[0058] Intelligent business applications are used to use models in production processes.
[0059] Model experiments, including model training, model evaluation, and model optimization, are used to perform model training, evaluation, and optimization, respectively.
[0060] Model storage includes models and a warehouse (including object storage). Model storage is used to store models generated through model experiments, training, or optimization. The object storage service in the production environment synchronizes model data with the object storage service in the test environment.
[0061] A mirror library is used to store model images or to synchronize models or model images with the mirror library in the test environment.
[0062] The model service is used to provide support for intelligent business applications in the production environment or be called by intelligent business applications so that the model can be used in the production process.
[0063] In the description of this specification, specific features, structures, materials or characteristics may be combined in an appropriate manner in any one or more embodiments or examples.
[0064] like Figure 2 As shown, a model updating method provided in an embodiment of the present application includes S201-S205:
[0065] S201. Acquire business data generated in the production environment.
[0066] S202: Perform a target operation on the first model in a production environment based on the business data to obtain a second model.
[0067] The target operation includes at least one of the following: training, evaluation, and optimization.
[0068] S203: Synchronize the second model to the test environment.
[0069] In some embodiments, the production environment synchronizes the second model to the test environment through a synchronization operation between an image repository in the production environment and an image repository in the test environment.
[0070] S204: Execute a test on the second model in the test environment to obtain first test information.
[0071] The first test information is used to indicate whether the second model test passed or failed.
[0072] S205: Synchronize the first test information to the production environment, and deploy the second model in the production environment if the first test information indicates that the second model has passed verification.
[0073] It's important to note that the production and test environments are isolated to ensure data security, but this can result in significant latency when the test environment accesses data from the production environment. Because the production environment has immediate access to the latest production-related data, training, evaluating, or optimizing the first model in the production environment ensures that the resulting second model is more aligned with actual production needs and offers enhanced real-time performance.
[0074] In some embodiments, performing a target operation on the first model in a production environment based on business data to obtain a second model includes: when the amount of business data exceeds a data amount threshold, performing a target operation on the first model in a production environment based on the business data to obtain the second model; or, when the model performance of the first model is less than or equal to a performance threshold, performing a target operation on the first model in a production environment based on the business data to obtain the second model. It should be understood that when the amount of business data exceeds the data amount threshold, it means that the business data has accumulated to a certain extent and has reached the data amount required to support model updates. At this time, performing operations such as model update, training, or evaluation can have a better update, training, or evaluation effect. When the model performance of the first model is less than or equal to the performance threshold, it means that the first model can no longer adapt to the current production environment and the model needs to be updated, trained, or evaluated to ensure the reliability of the production process.
[0075] In some embodiments, the method further includes: collecting, in a production environment, predicted data generated by the first model within a preset time period and real data corresponding to the predicted data; determining the performance parameters of the first model within the preset time period based on the predicted data and the real data; and determining, based on the performance parameters, whether the model performance of the first model is less than or equal to a performance threshold. It should be understood that the predicted data and the real data can reflect the difference between the results predicted by the model and the real results, and can reflect the performance of the model. Therefore, the performance parameters can be determined, and the relationship between the model performance of the first model and the performance threshold can be determined based on the performance parameters. Exemplarily, the relationship between the model performance of the first model and the performance threshold can be determined based on the numerical value of the performance parameter and the numerical threshold corresponding to the performance parameter.
[0076] In some embodiments, the performance parameters include at least one of the following: prediction accuracy, response delay, number of requests processed per unit time, tolerance to input data noise, number of model crashes due to abnormal inputs, and number of crashes due to environmental problems.
[0077] In some embodiments, when the performance parameter includes one parameter, determining whether the model performance of the first model is less than or equal to the performance threshold based on the performance parameter includes: determining that the model performance of the first model is less than or equal to the performance threshold when the numerical value of the performance parameter is less than or equal to the numerical threshold corresponding to the performance parameter. In this way, when the performance parameter includes only one parameter, it is possible to determine the relationship between the performance threshold of the first model and the data threshold.
[0078] In some embodiments, when the performance parameters include multiple parameters, determining whether the model performance of the first model is less than or equal to a performance threshold based on the performance parameters includes: for each parameter in the performance parameters, determining a score corresponding to the parameter based on the parameter's value and the corresponding numerical threshold, thereby obtaining a score for each parameter in the performance parameters; performing weighted fusion on the scores of each parameter to obtain a score corresponding to the performance parameter; and determining that the model performance of the first model is less than or equal to the performance threshold if the score corresponding to the performance parameter is less than or equal to the score threshold. Through weighted fusion, the performance level in multiple dimensions can be comprehensively determined based on the scores of multiple parameters, thereby more accurately and reliably determining the model performance of the first model. For example, the weights for different parameters can be determined based on the actual needs of the production environment. For example, if the production environment has a higher requirement for prediction accuracy than for input data noise, the weight of prediction accuracy can be higher than the weight of input data noise. For example, the weights of the two parameters can be determined based on the number of production process anomalies caused by prediction accuracy and the number of production process anomalies caused by input noise in historical data within a preset time period.
[0079] In some embodiments, when the first test information indicates that the second model failed a test, the first test information also includes exception information about the second model. The method further includes: updating the second model based on the exception information to obtain a third model; synchronizing the third model to a test environment, performing a test on the third model in the test environment, and obtaining second test information; the second test information indicates whether the third model passed the test; synchronizing the second test information to a production environment; and deploying the third model in the production environment if the second test information indicates that the third model passed the test. This ensures that if the second model fails to meet production requirements, the second model can be updated to obtain a third model that does. Optionally, if the third model fails to meet requirements, updates can be continued until a model that does meet the requirements is obtained.
[0080] In some embodiments, testing is performed on the model to be tested in a test environment to obtain test information, which is achieved by: obtaining a test data set corresponding to each of a plurality of test indicators; for the test data set corresponding to each of the plurality of test indicators, based on the test data set corresponding to the test indicator, testing is performed on the model to be tested to obtain test information.
[0081] In some embodiments, the test data sets corresponding to the multiple test indicators can be the same or different. In the case where the test data sets corresponding to the multiple test indicators are the same, the test results corresponding to the multiple test indicators can be obtained based on the test data sets.
[0082] In some embodiments, multiple test indicators have test priorities; the higher the test priority of a test indicator, the higher the test order of the test indicator; for the test data set corresponding to each test indicator among the multiple test indicators, based on the test data set corresponding to the test indicator, the test model to be tested is tested to obtain test information, including: among the untested test indicators, based on the test data set of the test indicator with the highest priority, the test model to be tested is tested to obtain the test result corresponding to the test indicator; when the test result corresponding to the test indicator is a test failure, the test is stopped, and the generated test result is determined as the test information; when the test result corresponding to the test indicator is a test pass, the test data set of the test indicator with the highest priority is continued to be selected from the untested test indicators to perform the test on the test model until the test result of the test indicator is a test failure. In this way, the test can be stopped when the test result of the test indicator fails, and the test information is output so that the production environment can execute the model update. This saves test overhead and improves test speed.
[0083] The embodiments of the present application can divide the functional modules of electronic devices, etc. 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 into one processing module. The above integrated modules can be implemented in the form of hardware or in the form of software functional modules. It should be noted that the division of modules in the embodiments of the present application is schematic and is only a logical function division. In actual implementation, there may be other division methods.
[0084] In the case of dividing each functional module into corresponding functional modules, Figure 3 A possible structural diagram of the model updating device involved in the above embodiment is shown. Figure 3 As shown, the model updating device may include: an acquisition module 301 and a processing module 302 .
[0085] Acquisition module 301, used to acquire business data generated in the production environment;
[0086] A processing module 302 is configured to perform a target operation on the first model in a production environment based on the business data to obtain a second model; the target operation includes at least one of the following: training, evaluation, and optimization;
[0087] The processing module 302 is further configured to synchronize the second model to the test environment;
[0088] The processing module 302 is further configured to perform a test on the second model in the test environment to obtain first test information; the first test information is used to indicate whether the second model passes the test or fails the test;
[0089] The processing module 302 is further configured to synchronize the first test information to the production environment, and deploy the second model in the production environment if the first test information indicates that the second model has passed verification.
[0090] In some embodiments, the processing module 302 is further configured to, when the amount of business data exceeds a data amount threshold, perform a target operation on the first model in a production environment based on the business data to obtain a second model; or
[0091] The processing module 302 is further configured to, when the model performance of the first model is less than or equal to a performance threshold, perform a target operation on the first model in a production environment based on the business data to obtain a second model.
[0092] In some embodiments, the acquisition module 301 is further configured to collect, in a production environment, prediction data generated by the first model within a preset period of time and real data corresponding to the prediction data;
[0093] The processing module 302 is further configured to determine the performance parameters of the first model within a preset time period based on the predicted data and the real data;
[0094] The processing module 302 is further configured to determine, based on the performance parameter, whether the model performance of the first model is less than or equal to a performance threshold.
[0095] In some embodiments, the performance parameters include at least one of the following: prediction accuracy, response delay, number of requests processed per unit time, tolerance to input data noise, number of model crashes due to abnormal inputs, and number of crashes due to environmental problems.
[0096] In some embodiments, when the performance parameter includes one parameter, the processing module 302 is further configured to determine that the model performance of the first model is less than or equal to the performance threshold when the value of the performance parameter is less than or equal to the value threshold corresponding to the performance parameter;
[0097] The processing module 302 is further configured to determine, for each parameter in the performance parameters, a score corresponding to the parameter based on the parameter's value and a numerical threshold corresponding to the parameter, thereby obtaining a score for each parameter in the performance parameters;
[0098] The processing module 302 is further configured to perform weighted fusion on the scores of each parameter to obtain a score corresponding to the performance parameter;
[0099] The processing module 302 is further configured to determine that the model performance of the first model is less than or equal to the performance threshold when the score corresponding to the performance parameter is less than or equal to the score threshold.
[0100] In some embodiments, when the first test information is used to indicate that the second model fails the test, the first test information also includes abnormality information of the second model;
[0101] The processing module 302 is further configured to update the second model based on the abnormal information to obtain a third model;
[0102] The processing module 302 is further configured to synchronize the third model to the test environment, execute a test on the third model in the test environment, and obtain second test information; the second test information is used to indicate whether the third model has passed the test;
[0103] The processing module 302 is further configured to synchronize the second test information to the production environment;
[0104] The processing module 302 is further configured to deploy the third model in a production environment if the second test information indicates that the third model has passed the test.
[0105] In some embodiments, testing is performed on the model to be tested in a test environment to obtain test information, which is achieved by: obtaining a test data set corresponding to each of a plurality of test indicators; for the test data set corresponding to each of the plurality of test indicators, based on the test data set corresponding to the test indicator, testing is performed on the model to be tested to obtain test information.
[0106] In some embodiments, multiple test indicators have test priorities; the higher the test priority of a test indicator, the higher the test order of the test indicator;
[0107] The processing module 302 is further configured to perform a test on the to-be-tested model based on the test data set of the test indicator with the highest priority among the untested test indicators, and obtain a test result corresponding to the test indicator;
[0108] The processing module 302 is further configured to stop the test if the test result corresponding to the test indicator is a test failure, and determine the generated test result as test information;
[0109] The processing module 302 is also used to continue to select a test data set of the test indicator with the highest priority from the untested test indicators to perform tests on the test model when the test result corresponding to the test indicator is a pass, until the test result of the test indicator is a fail.
[0110] The above are only specific embodiments of the present application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.
Claims
1. A model updating method, characterized in that: The method comprises: Obtain business data generated in the production environment; Performing a target operation on the first model in a production environment based on the business data to obtain a second model; the target operation includes at least one of the following: training, evaluation, and optimization; synchronizing the second model to the test environment; Performing a test on the second model in a test environment to obtain first test information; the first test information is used to indicate whether the second model passes the test or fails the test; The first test information is synchronized to a production environment, and if the first test information indicates that the second model has passed verification, the second model is deployed in the production environment.
2. The method according to claim 1, characterized in that The performing a target operation on the first model in a production environment based on the business data to obtain a second model includes: When the amount of the business data exceeds a data amount threshold, performing the target operation on the first model in a production environment based on the business data to obtain the second model; or When the model performance of the first model is less than or equal to a performance threshold, the target operation is performed on the first model in a production environment based on the business data to obtain the second model.
3. The method according to claim 2, characterized in that The method further comprises: Collecting, in a production environment, prediction data generated by the first model within a preset period of time and real data corresponding to the prediction data; Determining performance parameters of the first model within a preset time period based on the predicted data and the real data; Based on the performance parameter, it is determined whether a model performance of the first model is less than or equal to a performance threshold.
4. The method according to claim 3, characterized in that The performance parameters include at least one of the following: prediction accuracy, response delay, number of requests processed per unit time, tolerance to input data noise, number of model crashes due to abnormal inputs, and number of crashes due to environmental problems.
5. The method according to claim 4, characterized in that In a case where the performance parameter includes one parameter, the determining, based on the performance parameter, whether the model performance of the first model is less than or equal to a performance threshold, includes: determining that the model performance of the first model is less than or equal to the performance threshold when a value of the performance parameter is less than or equal to a value threshold corresponding to the performance parameter; In a case where the performance parameter includes a plurality of parameters, determining whether the model performance of the first model is less than or equal to a performance threshold based on the performance parameter includes: For each parameter in the performance parameters, determining a score corresponding to the parameter based on the value of the parameter and a numerical threshold corresponding to the parameter, to obtain a score for each parameter in the performance parameters; Performing weighted fusion on the score of each parameter to obtain a score corresponding to the performance parameter; When the score corresponding to the performance parameter is less than or equal to a score threshold, it is determined that the model performance of the first model is less than or equal to the performance threshold.
6. The method according to claim 1, characterized in that When the first test information indicates that the second model fails the test, the first test information also includes abnormal information of the second model; and the method further includes: Based on the abnormal information, updating the second model to obtain a third model; Synchronizing the third model to the test environment, performing a test on the third model in the test environment, and obtaining second test information; the second test information is used to indicate whether the third model passes the test; Synchronizing the second test information to the production environment; If the second test information indicates that the third model passes the test, the third model is deployed in a production environment.
7. The method according to claim 1 or 6, characterized in that Execute the test model in the test environment and obtain the test information. This is achieved by: Obtain test data sets corresponding to multiple test indicators; For a test data set corresponding to each test indicator among the multiple test indicators, the model to be tested is tested based on the test data set corresponding to the test indicator to obtain test information.
8. The method according to claim 7, characterized in that The multiple test indicators have test priorities; the higher the test priority of a test indicator, the earlier the test order of the test indicator; for the test data set corresponding to each test indicator in the multiple test indicators, the test is performed on the to-be-tested model based on the test data set corresponding to the test indicator to obtain test information, including: Among the untested test indicators, based on the test data set of the test indicator with the highest priority, the test is performed on the model to be tested to obtain the test result corresponding to the test indicator; If the test result corresponding to the test indicator is a test failure, stopping the test and determining the generated test result as the test information; When the test result corresponding to the test indicator is a pass, continue to select a test data set of the test indicator with the highest priority from the untested test indicators to perform testing on the model to be tested until the test result of the test indicator is a fail.
9. An electronic device, characterized in that: The electronic device comprises: processor; a memory configured to store instructions executable by the processor; The processor is configured to execute the instructions to implement the method according to any one of claims 1 to 8.
10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions in the computer-readable storage medium are executed by an electronic device, the electronic device is enabled to perform the method according to any one of claims 1 to 8.