Large model application publishing method and device, equipment and medium

Abstract

The embodiment of the invention discloses a large model application publishing method and device, equipment and a medium. The method comprises the steps of obtaining a to-be-updated execution code and code updating associated data; determining whether a target project name matched with the project update name exists in preset candidate project names or not; if the to-be-updated execution code exists, taking a candidate application category corresponding to the target project name as a target application category, taking a candidate large model file corresponding to the target project name as a target large model file, and determining an update version number of the to-be-updated execution code according to the project update time; generating an application update package, and determining a corresponding target test case according to the application update package; and testing the application update package according to the target test case, determining a target test level, and determining the release permission of the updated large model application according to the target test level. Automatic publishing of the large model application is realized, and the publishing efficiency and accuracy of the large model application are improved.

Description

Technical Field

[0001] Embodiments of the present invention relate to the technical field of large model applications, and in particular to a method, device, equipment and medium for publishing large model applications. Background Art

[0002] Large model applications are packaged based on manual scripts and then released by operations and maintenance. The process is long and involves many links. On the centralized release day, there is a tendency for version inconsistency and short-term abnormalities in the interface that calls the large model application. Therefore, it is crucial to improve the release efficiency and accuracy of large model applications. Summary of the invention

[0003] The present invention provides a method, device, equipment and medium for publishing a large model application, so as to realize automatic publishing of the large model application, improve the publishing efficiency of the large model application and the accuracy of the published large model application.

[0004] According to one aspect of the present invention, a method for publishing a large model application is provided, comprising:

[0005] Obtaining the execution code to be updated and code update associated data of the large model application to be updated; wherein the code update associated data includes a project update name and a project update time;

[0006] Determine whether there is a target project name matching the project update name among the preset candidate project names;

[0007] If it exists, the candidate application category corresponding to the target project name is used as the target application category, the candidate large model file corresponding to the target project name is used as the target large model file, and the update version number of the execution code to be updated is determined according to the project update time;

[0008] Generate an application update package including the target application category, the target project name, the update version number, the execution code to be updated and the target large model file, and determine a corresponding target test case according to the application update package;

[0009] The application update package is tested according to the target test case to determine the target test level, and the release authority of the updated large model application is determined according to the target test level.

[0010] According to another aspect of the present invention, there is provided a publishing device for a large model application, comprising:

[0011] A data acquisition module, used to acquire the execution code to be updated of the large model application to be updated and the code update associated data; wherein the code update associated data includes the project update name and the project update time;

[0012] A name determination module, used to determine whether there is a target project name matching the project update name among the preset candidate project names;

[0013] A version update module, for, if it exists, using the candidate application category corresponding to the target project name as the target application category, using the candidate large model file corresponding to the target project name as the target large model file, and determining the update version number of the execution code to be updated according to the project update time;

[0014] A use case determination module, used to generate an application update package including the target application category, the target project name, the update version number, the execution code to be updated and the target large model file, and determine a corresponding target test case according to the application update package;

[0015] The permission determination module is used to test the application update package according to the target test case, determine the target test level, and determine the release permission of the updated large model application according to the target test level.

[0016] According to another aspect of the present invention, there is provided an electronic device, comprising:

[0017] one or more processors;

[0018] A memory for storing one or more programs;

[0019] When one or more programs are executed by one or more processors, the one or more processors can execute any one of the large model application publishing methods provided by the embodiments of the present invention.

[0020] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions, and the computer instructions are used to enable a processor to implement any one of the large model application publishing methods provided by the embodiments of the present invention when executed.

[0021] An embodiment of the present invention provides a publishing scheme for a large model application, which obtains the execution code to be updated and code update associated data of the large model application to be updated; wherein the code update associated data includes a project update name and a project update time; determines whether there is a target project name matching the project update name in the preset candidate project names; if so, takes the candidate application category corresponding to the target project name as the target application category, takes the candidate large model file corresponding to the target project name as the target large model file, and determines the update version number of the execution code to be updated according to the project update time; generates an application update package including the target application category, the target project name, the update version number, the execution code to be updated and the target large model file, and determines the corresponding target test case according to the application update package; tests the application update package according to the target test case to determine the target test level, and determines the publishing authority of the updated large model application according to the target test level. The above scheme performs functional testing on the application update package of the big model application to be updated by updating associated data and target test cases according to the code, and determines the release authority of the big model application according to the test level, thereby realizing the automatic release of the big model application and improving the release efficiency of the big model application. At the same time, the update version number is determined according to the project update time in the embodiment of the present invention, thereby improving the accuracy of the released big model application.

[0022] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, 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 invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 This is a flow chart of a method for publishing a large model application provided in the first embodiment of the present invention;

[0025] Figure 2 This is a flow chart of a method for publishing a large model application provided in Embodiment 2 of the present invention;

[0026] Figure 3 It is a structural diagram of a publishing device for a large model application provided in Embodiment 3 of the present invention;

[0027] Figure 4 It is a structural diagram of an electronic device for implementing a publishing method for a large model application provided in a fourth embodiment of the present invention. DETAILED DESCRIPTION

[0028] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0029] Embodiment 1

[0030] Figure 1 It is a flowchart of a method for publishing a large model application provided in Example 1 of the present invention. This embodiment can be applicable to the situation of automatically updating and publishing a large model application. The method can be executed by a publishing device for a large model application. The device can be implemented in software and / or hardware and can be configured in an electronic device that carries the publishing function of the large model application.

[0031] See also Figure 1 The method for publishing a large model application shown is applied to an application management system and includes:

[0032] S110, obtaining the execution code to be updated of the large model application to be updated and code update associated data; wherein the code update associated data includes a project update name and a project update time.

[0033] The execution code to be updated refers to the execution code in the large model application that needs to be updated. The code update associated data refers to the data related to the execution code to be updated. The project update name refers to the name of the execution code to be updated. The project update time refers to the update time of the execution code to be updated.

[0034] It should be noted that the large model application includes execution code and large model files.

[0035] S120: Determine whether there is a target project name matching the project update name among the preset candidate project names.

[0036] The candidate project name refers to the name of the pre-set execution code. The embodiment of the present invention does not impose any restrictions on the setting of the candidate project name, which can be set by technicians based on experience. The target project name refers to the candidate project name that is the same as the project update name.

[0037] S130. If it exists, the candidate application category corresponding to the target project name is used as the target application category, the candidate large model file corresponding to the target project name is used as the target large model file, and the update version number of the execution code to be updated is determined according to the project update time.

[0038] The candidate application category refers to the functional category of the execution code corresponding to the candidate project name. The target application category refers to the candidate application category corresponding to the target project name, that is, the target application category is the code functional category of the execution code to be updated.

[0039] The candidate large model file refers to the large model file corresponding to the preset candidate project name. The target large model file refers to the candidate large model file corresponding to the target project name, that is, the target large model file is the candidate large model file that can be called by the execution code to be updated.

[0040] Among them, the update version number refers to the version number of the execution code to be updated. The embodiment of the present invention does not specifically limit how to determine the update version number of the execution code to be updated based on the project update time. Exemplarily, the project update time can be directly used as the update version number of the execution code to be updated; or, based on the preset version format, the update version number of the execution code to be updated can be determined according to the project update time. The embodiment of the present invention does not impose any limitation on the preset version format, which can be set by technical personnel based on experience or needs, and only needs to ensure the uniqueness of the update version number.

[0041] In an optional embodiment, if the target project name does not exist, it indicates that the application management platform cannot process the execution code to be updated.

[0042] S140, generating an application update package including a target application category, a target project name, an update version number, an execution code to be updated, and a target large model file, and determining a corresponding target test case according to the application update package.

[0043] The application update package refers to a data package obtained by packaging the target application category, target project name, update version number, execution code to be updated, and target large model file. The target test case refers to a sample for functional testing of the application update package.

[0044] S150: Testing the application update package according to the target test case, determining the target test level, and determining the release authority of the updated large model application according to the target test level.

[0045] The target test level can be used to quantify the test results of the application update package. Release authority refers to the qualification to release the updated large model application.

[0046] In an optional embodiment, the application update package is tested according to the target test case to determine the target test level, including: based on the execution code to be updated, calling the target large model file, executing the target test case, and obtaining the corresponding test sub-results; according to the test sub-results, determining the target test level; wherein the target test level includes a first-level normal level, a second-level adjustment level, and a third-level abnormal level.

[0047] Among them, the test sub-result refers to the test result corresponding to any target test case. The first-level normal level means that there is no abnormality between the execution code to be updated in the application update package and the target large model file; accordingly, the release permission is allowed to release.

[0048] Among them, the second-level adjustment level means that there are minor anomalies between the execution code to be updated in the application update package and the target large model file; accordingly, the release authority is to release after adjustment. It should be noted that the second-level adjustment level means that the R&D personnel need to adjust the execution code to be updated and retest the adjusted execution code to be updated.

[0049] Among them, the third level of abnormality means that there is an abnormality between the execution code to be updated in the application update package and the target large model file; accordingly, the publishing permission is prohibited. It should be noted that the third level of abnormality means that the R&D personnel need to rewrite the execution code to be updated and retest the rewritten execution code to be updated.

[0050] It can be understood that by determining the target test level based on the test sub-results, the accuracy of the determined target test level is improved.

[0051] In an optional embodiment, a target test level is determined based on the test sub-results, including: if there is a deviation sub-result in the test sub-results, determining the use case type of the target test case corresponding to the deviation sub-result; if the use case type is a use case deviation type, determining the result similarity between the deviation sub-result and the corresponding actual sub-result, and determining whether the deviation sub-result is an abnormal sub-result based on the result similarity; if so, determining the number of abnormalities of the abnormal sub-results and the number of tests of the test sub-results; and determining whether the target test level is a first-level normal level or a second-level adjustment level based on the number of abnormalities and the number of tests.

[0052] If there is a difference between the test sub-result corresponding to any target test case and the actual sub-result of the target test case, the test sub-result of the target test case is a deviation sub-result. The actual sub-result refers to the real result corresponding to the target test case.

[0053] The result similarity can be used to quantify the similarity between the deviation sub-result of any target test case and the actual sub-result of the target test case. An abnormal sub-result refers to an erroneous test sub-result.

[0054] Exemplarily, for any target test case, if the result similarity between the deviation sub-result of the target test case and the actual sub-result of the target test case is less than a preset result similarity threshold, the deviation sub-result of the target test case is determined to be an abnormal sub-result; if not, it is prohibited to treat the deviation sub-result of the target test case as an abnormal sub-result.

[0055] The embodiment of the present invention does not impose any limitation on the value of the preset result similarity threshold, which can be set by a technician based on experience or needs, or determined repeatedly through a large number of experiments.

[0056] The number of exceptions refers to the number of abnormal sub-results, that is, the number of exceptions can also represent the number of target test cases with abnormal sub-results. The number of tests refers to the number of test sub-results, that is, the number of tests can also represent the number of target test cases.

[0057] The use case deviation type refers to the type of test case that allows differences between the test sub-result and the actual sub-result.

[0058] Optionally, determining the target test level as a first-level normal level or a second-level adjustment level based on the number of anomalies and the number of tests includes: determining an anomaly ratio based on the number of anomalies and the number of tests; and determining the target test level as a first-level normal level or a second-level adjustment level based on the anomaly ratio and a preset percentage threshold. The anomaly ratio can be used to quantify the percentage of anomaly sub-results in the test sub-results, that is, the anomaly ratio can be used to quantify the percentage of target test cases with anomaly sub-results in all target test cases.

[0059] Exemplarily, based on the abnormality ratio and the preset proportion threshold, the target test level is determined to be the first-level normal level or the second-level adjustment level, including: if the abnormality ratio is less than the preset proportion threshold, the target test level is determined to be the first-level normal level; if the abnormality ratio is greater than or equal to the preset proportion threshold, the target test level is determined to be the second-level adjustment level.

[0060] Or optionally, determining the target test level as the first-level normal level or the second-level adjusted level based on the number of exceptions and the number of tests, including: determining the degree of abnormality based on preset abnormal case weights, preset test case weights, the number of exceptions and the number of tests; determining the target test level as the first-level normal level or the second-level adjusted level based on the degree of abnormality and a preset abnormality threshold.

[0061] Among them, the abnormality degree can be used to quantify the abnormality degree of testing the application update package. The embodiment of the present invention does not impose any restrictions on the size of the preset abnormal case weight and the preset test case weight, which can be set by a technician based on experience or needs, or repeatedly determined through a large number of experiments. The embodiment of the present invention does not impose any restrictions on the size of the preset abnormality threshold, which can be set by a technician based on experience or needs, or repeatedly determined through a large number of experiments.

[0062] Exemplarily, based on the abnormality degree and the preset abnormality degree threshold, the target test level is determined to be the first-level normal level or the second-level adjustment level, including: if the abnormality degree is less than the preset abnormality degree threshold, the target test level is determined to be the first-level normal level; if the abnormality degree is greater than or equal to the preset abnormality degree threshold, the target test level is determined to be the second-level adjustment level.

[0063] It can be understood that when the use case type is the use case deviation type, the target test level is determined as the first-level normal level or the second-level adjustment level based on the number of abnormalities and the number of tests, which improves the accuracy of the determined target test level.

[0064] In an optional embodiment, the method further includes: if the use case type is a fixed use case type, determining the target test level to be a third level abnormality level.

[0065] The fixed type of test case refers to a type of test case that prohibits differences between test sub-results and actual sub-results.

[0066] It can be understood that if the use case type is a fixed use case type, the target test level is determined to be the third abnormal level, which improves the accuracy of the determined target test level.

[0067] An embodiment of the present invention provides a publishing scheme for a large model application, which obtains the execution code to be updated and code update associated data of the large model application to be updated; wherein the code update associated data includes a project update name and a project update time; determines whether there is a target project name matching the project update name in the preset candidate project names; if so, takes the candidate application category corresponding to the target project name as the target application category, takes the candidate large model file corresponding to the target project name as the target large model file, and determines the update version number of the execution code to be updated according to the project update time; generates an application update package including the target application category, the target project name, the update version number, the execution code to be updated and the target large model file, and determines the corresponding target test case according to the application update package; tests the application update package according to the target test case to determine the target test level, and determines the publishing authority of the updated large model application according to the target test level. The above scheme performs functional testing on the application update package of the big model application to be updated by updating associated data and target test cases according to the code, and determines the release authority of the big model application according to the test level, thereby realizing the automatic release of the big model application and improving the release efficiency of the big model application. At the same time, the update version number is determined according to the project update time in the embodiment of the present invention, thereby improving the accuracy of the released big model application.

[0068] Embodiment 2

[0069] Figure 2It is a flowchart of a method for publishing a large model application provided in the second embodiment of the present invention. Based on the above embodiments, this embodiment further refines the operation of "determining the corresponding target test case according to the application update package" into "using the candidate test case corresponding to the target project name as the first reference test case; using the default test case in the model basic data set as the second reference test case; determining the target test case of the application update package according to the first reference test case and / or the second reference test case" to improve the target test case determination mechanism. It should be noted that for the parts not described in detail in the embodiments of the present invention, reference may be made to the descriptions of other embodiments.

[0070] See also Figure 2 The release scheme of the large model application shown includes:

[0071] S210, obtaining the execution code to be updated of the large model application to be updated and the code update associated data; wherein the code update associated data includes the project update name and the project update time.

[0072] S220: Determine whether there is a target project name matching the project update name among the preset candidate project names.

[0073] S230. If it exists, the candidate application category corresponding to the target project name is used as the target application category, the candidate large model file corresponding to the target project name is used as the target large model file, and the update version number of the execution code to be updated is determined according to the project update time.

[0074] S240, generating an application update package including a target application category, a target project name, an update version number, an execution code to be updated, and a target large model file.

[0075] S250: Use the candidate test case corresponding to the target project name as the first reference test case.

[0076] The first reference test case refers to a sample that may be used for functional testing and is determined based on the target project name. The first reference test case refers to a sample set in the application management system for functional testing of the trained large model.

[0077] S260: Use the default test case in the model basic data set as the second reference test case.

[0078] Among them, the model basic data set refers to a general data set pre-stored in the application management system. Exemplarily, the model basic data set may include a training data set, a training simulation environment, and a default test case. The training data set refers to a data set stored in the application management system for training the large model. The training simulation environment refers to a simulation environment for training the large model in the application management system. The default test case refers to a test case set when training the large model.

[0079] It should be noted that the application management system pre-stores common data sets corresponding to each major model. When packaging the application update package, the common data sets corresponding to the large model can be directly obtained as the model basic data set.

[0080] S270: Determine a target test case for the application update package according to the first reference test case and / or the second reference test case.

[0081] In an optional embodiment, the target test case may be determined based on the first reference test case and the number of test cases.

[0082] In an optional embodiment, target test cases for an application update package are determined based on first reference test cases and second reference test cases, including: determining the first case quantity of first reference test cases in the target test cases, and determining the second case quantity of second reference test cases in the target test cases, based on a preset case selection ratio and the number of case studies of the target test cases; selecting the first number of first target test cases from the first reference test cases, and selecting the second number of second target test cases from the second reference test cases; generating target test cases including the first target test cases and the second target test cases.

[0083] The number of use cases refers to the number of pre-set target test cases. The embodiment of the present invention does not impose any limitation on the size of the use case selection ratio, which can be set by technicians based on experience or needs, or determined repeatedly through a large number of experiments.

[0084] The first number of test cases refers to the number of first reference test cases in the target test cases, and the second number of test cases refers to the number of second reference test cases in the target test cases.

[0085] The first target test case refers to the first reference test case in the target test case, and the second target test case refers to the second reference test case in the target test case.

[0086] Exemplarily, determine the use case similarity between each first reference test case and each second reference test case; determine whether there are similar test cases in the first reference test case and the second reference test case based on the use case similarity and a preset use case similarity threshold; if so, use the similar test cases in the first reference test case as part of the target test case; determine the number of similar use cases of similar test cases; determine other target test cases from other first reference test cases and other second reference test cases based on the number of similar use cases, the number of use cases, and the use case selection ratio; generate target test cases including part of the target test cases and other target test cases.

[0087] The use case similarity can be used to quantify the similarity between the first reference test case and the second reference test case. The embodiment of the present invention does not impose any limitation on the value of the preset use case similarity threshold, which can be set by technicians based on experience or needs, or determined repeatedly through a large number of experiments.

[0088] Among them, the first reference test case and the second reference test case whose case similarity is greater than or equal to a preset case similarity threshold are used as similar test cases.

[0089] Among them, part of the target test cases refers to part of the target test cases, and the number of similar test cases refers to the number of similar test cases.

[0090] Among them, other first reference test cases refer to other test cases in the first reference test cases except similar test cases. Other second reference test cases refer to other test cases in the second reference test cases except similar test cases. Other target test cases refer to the remaining test cases in the target test cases.

[0091] It should be noted that the target test cases at this time include similar test cases, first target test cases and second target test cases.

[0092] It can be understood that by determining the first number of use cases and the second number of use cases based on a preset case selection ratio and the number of use cases, and determining the first target test cases and the second target test cases based on the first number of use cases and the second number of use cases, and then obtaining the target test cases, the accuracy and diversity of the determined target test cases are improved.

[0093] S280: Testing the application update package according to the target test case, determining the target test level, and determining the release authority of the updated large model application according to the target test level.

[0094] The embodiment of the present invention provides a release scheme for a large model application, which refines the operation of determining the corresponding target test case according to the application update package into taking the candidate test case corresponding to the target project name as the first reference test case; taking the default test case in the model basic data set as the second reference test case; and determining the target test case of the application update package according to the first reference test case and / or the second reference test case, thereby improving the target test case determination mechanism. The above scheme improves the comprehensiveness of the determined target test case by determining the target test case of the application update package according to the first reference test case and / or the second reference test case.

[0095] Based on the above technical solution, an embodiment of the present invention further provides an application management system for automatically updating and publishing large model applications. Exemplarily, the application management system may include an application publishing subsystem and an application development subsystem, and the application publishing subsystem and the application development subsystem are connected in communication.

[0096] The application publishing subsystem can be used to perform functional tests on large model applications before release in an online publishing environment, and release large model applications that have passed the test. The application publishing subsystem can monitor data changes in the database connected to the application development subsystem in real time; if changes are found, it will start to determine the release authority. The database can store data fed back by the application development subsystem, such as the execution code to be updated and the code update related data.

[0097] The application development subsystem can monitor the status of the development application in real time and automatically build an online release package. R&D personnel can write the execution code to be updated in the development environment provided by the application development subsystem.

[0098] In the existing technology, the iteration cycle of large model applications is short. For example, the use of new parameters needs to be put online urgently to apply the latest effects, or the business urgently requires appropriate simplification of the output results, etc., which all require large model applications to provide quasi-real-time feedback; the environment of large model applications is changeable. For example, different GPU (Graphics Processing Unit) hardware requires different underlying dependency packages, which may lead to serious efficiency problems (CPU (Central Processing Unit) instead of GPU is used for operation).

[0099] The application management system provided by the embodiment of the present invention can automatically perceive the version number information file of the large model application; automatically build the image after perception, and then automatically deploy it based on the AI ​​platform (i.e., the application management system); after deployment, run the automated test script for testing and output the test sub-results; judge whether the output test sub-results pass and whether they meet the release conditions; according to version management, switch the new version to the current release version; and provide the ability to roll back the old version on the AI ​​platform.

[0100] Among them, the image refers to the container image, which is a deployment package based on the underlying virtual structure such as runc, and contains all the dependencies and environment required to run the application to ensure consistency and portability. According to the installation information during development, the image is packaged according to the target application category-target project name-update version number during packaging to obtain the application update package. Each change in the large model application corresponds to a main file (i.e., the target large model file) and code (i.e., the execution code to be updated) packaged in the image. The main file can be similar to the Dockfile file, and the code can be all the code of the large model application and the database script for this upgrade or rollback, etc.

[0101] Among them, the AI ​​platform relies on distributed deployment middleware. The middleware calls Linux commands to execute deployment commands. The output is the test sub-result, and if the functional test fails, it outputs error information and execution logs.

[0102] Exemplarily, the application built-in detection interface / info is called to determine whether the deployment is successful, and the interface configured by the automated test and the specification returned by the interface are called to determine whether the interface is normal. If the above is successful and normal, the functional test passes.

[0103] The application management system provided by this solution can realize automatic perception. Through subscription, when a new large model application is released, the successful latest large model application (with updated version number) is automatically added to the available list and the currently used list according to the event of successful container creation; it can realize automatic release. After successful development, it can be audited through automatic construction, automatic packaging, automatic release and recording of detailed automated logs, which improves efficiency (manual review requires a certain amount of inspection and approval time); with updated version numbers, by retaining historical large model applications for a period of time (such as 30 days), it can roll back and use historically released large model applications in real time, and can use GPU features and non-GPU resources to better manage versions and risk safety.

[0104] Embodiment 3

[0105] Figure 3This is a schematic diagram of the structure of a large model application publishing device provided in Embodiment 3 of the present invention. This embodiment is applicable to the case of automatically updating and publishing a large model application. The method can be executed by a large model application publishing device, which can be implemented in software and / or hardware and can be configured in an electronic device that carries the publishing function of the large model application.

[0106] like Figure 3 As shown, the device includes: a data acquisition module 310, a name determination module 320, a version update module 330, a use case determination module 340 and a permission determination module 350. Among them,

[0107] The data acquisition module 310 is used to acquire the execution code to be updated of the large model application to be updated and the code update associated data; wherein the code update associated data includes the project update name and the project update time;

[0108] A name determination module 320 is used to determine whether there is a target project name matching the project update name among the preset candidate project names;

[0109] A version update module 330, for, if it exists, using the candidate application category corresponding to the target project name as the target application category, using the candidate large model file corresponding to the target project name as the target large model file, and determining the update version number of the execution code to be updated according to the project update time;

[0110] A use case determination module 340 is used to generate an application update package including the target application category, the target project name, the update version number, the execution code to be updated and the target large model file, and determine a corresponding target test case according to the application update package;

[0111] The permission determination module 350 is used to test the application update package according to the target test case, determine the target test level, and determine the release permission of the updated large model application according to the target test level.

[0112] An embodiment of the present invention provides a publishing scheme for a large model application, which obtains the execution code to be updated and code update associated data of the large model application to be updated; wherein the code update associated data includes a project update name and a project update time; determines whether there is a target project name matching the project update name in the preset candidate project names; if so, takes the candidate application category corresponding to the target project name as the target application category, takes the candidate large model file corresponding to the target project name as the target large model file, and determines the update version number of the execution code to be updated according to the project update time; generates an application update package including the target application category, the target project name, the update version number, the execution code to be updated and the target large model file, and determines the corresponding target test case according to the application update package; tests the application update package according to the target test case to determine the target test level, and determines the publishing authority of the updated large model application according to the target test level. The above scheme performs functional testing on the application update package of the big model application to be updated by updating associated data and target test cases according to the code, and determines the release authority of the big model application according to the test level, thereby realizing the automatic release of the big model application and improving the release efficiency of the big model application. At the same time, the update version number is determined according to the project update time in the embodiment of the present invention, thereby improving the accuracy of the released big model application.

[0113] Optionally, if the application update package also includes a model basic data set, the use case determination module 340 includes:

[0114] A first test case determination unit, configured to use a candidate test case corresponding to the target project name as a first reference test case;

[0115] A second test case determination unit, configured to use a default test case in the model basic data set as a second reference test case;

[0116] A target case determination unit is used to determine a target test case of the application update package according to the first reference test case and / or the second reference test case.

[0117] Optional, target use case determination unit, specifically used to:

[0118] Determine the first number of first reference test cases in the target test cases and determine the second number of second reference test cases in the target test cases according to a preset case selection ratio and the number of case studies of the target test cases;

[0119] Selecting a first number of first target test cases from the first reference test cases, and selecting a second number of second target test cases from the second reference test cases;

[0120] A target test case including the first target test case and the second target test case is generated.

[0121] Optionally, the authority determination module 350 includes:

[0122] A sub-result determination unit, configured to call the target large model file based on the execution code to be updated, execute the target test case, and obtain a corresponding test sub-result;

[0123] The test level determination unit is used to determine a target test level according to the test sub-results; wherein the target test level includes a first-level normal level, a second-level adjustment level and a third-level abnormal level.

[0124] Optionally, the test level determination unit is specifically used to:

[0125] If there is a deviation sub-result in the test sub-result, determining the use case type of the target test case corresponding to the deviation sub-result;

[0126] If the use case type is a use case deviation type, determining the result similarity between the deviation sub-result and the corresponding actual sub-result, and determining whether the deviation sub-result is an abnormal sub-result based on the result similarity;

[0127] If so, determining the abnormal number of the abnormal sub-result and the test number of the test sub-result;

[0128] According to the abnormal number and the test number, the target test level is determined to be a first-level normal level or a second-level adjustment level.

[0129] Optionally, the device further comprises:

[0130] The third-level abnormality level determination unit is used to determine that the target test level is a third-level abnormality level if the use case type is a fixed use case type.

[0131] The publishing device for large model applications provided in the embodiment of the present invention can execute the publishing method for large model applications provided in any embodiment of the present invention, and has functional modules and beneficial effects corresponding to executing the publishing methods for large model applications.

[0132] In the technical solution of the present invention, the collection, storage, use, processing, transmission, provision and disclosure of the execution code to be updated and the code update-related data, etc., all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0133] Embodiment 4

[0134] Figure 4It is a structural diagram of an electronic device for implementing a publishing method for a large model application provided by Embodiment 4 of the present invention. Electronic device 410 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0135] like Figure 4 As shown, the electronic device 410 includes at least one processor 411, and a memory connected to the at least one processor 411 in communication, such as a read-only memory (ROM) 412, a random access memory (RAM) 413, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 411 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 412 or the computer program loaded from the storage unit 418 to the random access memory (RAM) 413. In the RAM 413, various programs and data required for the operation of the electronic device 410 can also be stored. The processor 411, the ROM 412, and the RAM 413 are connected to each other via a bus 414. An input / output (I / O) interface 415 is also connected to the bus 414.

[0136] Multiple components in the electronic device 410 are connected to the I / O interface 415, including: an input unit 416, such as a keyboard, a mouse, etc.; an output unit 417, such as various types of displays, speakers, etc.; a storage unit 418, such as a disk, an optical disk, etc.; and a communication unit 419, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 419 allows the electronic device 410 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0137] Processor 411 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of processor 411 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. Processor 411 performs the various methods and processes described above, such as a publishing method for a large model application.

[0138] In some embodiments, the publishing method of the large model application may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 418. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 410 via the ROM 412 and / or the communication unit 419. When the computer program is loaded into the RAM 413 and executed by the processor 411, one or more steps of the publishing method of the large model application described above may be performed. Alternatively, in other embodiments, the processor 411 may be configured to execute the publishing method of the large model application in any other appropriate manner (e.g., by means of firmware).

[0139] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0140] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0141] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0142] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0143] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0144] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0145] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0146] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for publishing a large model application, characterized in that: include: Obtaining the execution code to be updated and code update associated data of the large model application to be updated; wherein the code update associated data includes a project update name and a project update time; Determine whether there is a target project name matching the project update name among the preset candidate project names; If it exists, the candidate application category corresponding to the target project name is used as the target application category, the candidate large model file corresponding to the target project name is used as the target large model file, and the update version number of the execution code to be updated is determined according to the project update time; Generate an application update package including the target application category, the target project name, the update version number, the execution code to be updated and the target large model file, and determine a corresponding target test case according to the application update package; The application update package is tested according to the target test case to determine the target test level, and the release authority of the updated large model application is determined according to the target test level.

2. The method according to claim 1, characterized in that: If the application update package also includes a model basic data set, then determining a corresponding target test case according to the application update package includes: Taking the candidate test case corresponding to the target project name as the first reference test case; Using the default test case in the model basic data set as the second reference test case; A target test case for the application update package is determined according to the first reference test case and / or the second reference test case.

3. The method according to claim 2, characterized in that The determining, according to the first reference test case and the second reference test case, a target test case of the application update package includes: Determine the first number of first reference test cases in the target test cases and determine the second number of second reference test cases in the target test cases according to a preset case selection ratio and the number of case studies of the target test cases; Selecting a first number of first target test cases from the first reference test cases, and selecting a second number of second target test cases from the second reference test cases; A target test case including the first target test case and the second target test case is generated.

4. The method according to claim 1, characterized in that The step of testing the application update package according to the target test case to determine the target test level includes: Based on the execution code to be updated, calling the target large model file, executing the target test case, and obtaining corresponding test sub-results; According to the test sub-results, a target test level is determined; wherein the target test level includes a first-level normal level, a second-level adjustment level, and a third-level abnormal level.

5. The method according to claim 4, characterized in that Determining a target test level according to the test sub-results includes: If there is a deviation sub-result in the test sub-result, determining the use case type of the target test case corresponding to the deviation sub-result; If the use case type is a use case deviation type, determining the result similarity between the deviation sub-result and the corresponding actual sub-result, and determining whether the deviation sub-result is an abnormal sub-result based on the result similarity; If so, determining the abnormal number of the abnormal sub-result and the test number of the test sub-result; According to the abnormal number and the test number, the target test level is determined to be a first-level normal level or a second-level adjustment level.

6. The method according to claim 5, characterized in that The method further comprises: If the use case type is a fixed use case type, the target test level is determined to be a third-level abnormal level.

7. A publishing device for a large model application, characterized in that: include: A data acquisition module, used to acquire the execution code to be updated of the large model application to be updated and the code update associated data; wherein the code update associated data includes the project update name and the project update time; A name determination module, used to determine whether there is a target project name matching the project update name among the preset candidate project names; A version update module, for, if it exists, using the candidate application category corresponding to the target project name as the target application category, using the candidate large model file corresponding to the target project name as the target large model file, and determining the update version number of the execution code to be updated according to the project update time; A use case determination module, used to generate an application update package including the target application category, the target project name, the update version number, the execution code to be updated and the target large model file, and determine a corresponding target test case according to the application update package; The permission determination module is used to test the application update package according to the target test case, determine the target test level, and determine the release permission of the updated large model application according to the target test level.

8. The device according to claim 7, characterized in that If the application update package also includes a model basic data set, the use case determination module includes: A first test case determination unit, configured to use a candidate test case corresponding to the target project name as a first reference test case; A second test case determination unit, configured to use a default test case in the model basic data set as a second reference test case; A target case determination unit is used to determine a target test case of the application update package according to the first reference test case and / or the second reference test case.

9. An electronic device, characterized in that: include: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement a method for publishing a large model application as described in any one of claims 1-6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, a method for publishing a large model application as described in any one of claims 1 to 6 is implemented.

Images