CN-122020108-A - Data processing system for optimizing model operation

CN122020108ACN 122020108 ACN122020108 ACN 122020108ACN-122020108-A

Abstract

The application discloses a data processing system for optimizing model operation, which is used for carrying out analysis decision processing on first model information and operation environment information to obtain target operation result information.

Inventors

DANG JIE
HONG YU

Assignees

深圳市TCL高新技术开发有限公司

Dates

Publication Date: 20260512
Application Date: 20241111

Claims (16)

1. A method, comprising: acquiring first model information and running environment information; And analyzing and deciding the first model information and the operation environment information to obtain target operation result information.
2. The method of claim 1, wherein the performing an analysis decision process on the first model information and the operating environment information to obtain target operating result information comprises: Extracting features of the operation environment information to obtain first operation feature information, wherein the first operation feature information characterizes the number of a plurality of first target models currently in operation; And determining target operation result information based on the first model information and the first operation characteristic information.
3. The method of claim 2, wherein the determining target operation result information based on the first model information and the first operation feature information comprises: If the feature value corresponding to the first operation feature information is smaller than or equal to a first threshold value, determining that the target operation result information is the first operation result information, and/or, And if the characteristic value corresponding to the first operation characteristic information is larger than the first threshold value, determining target operation result information based on the first model information.
4. A method according to claim 3, wherein the first model information includes resource occupancy information and second operational characteristic information, the resource occupancy information characterizing a resource occupancy condition of a second target model at the time of operation; The determining, based on the first model information, target operation result information includes: If the resource occupation value corresponding to the resource occupation information is larger than or equal to a second threshold value, determining that the target operation result information is second operation result information, and/or, And if the resource occupation value corresponding to the resource occupation information is smaller than the second threshold value, determining target operation result information based on the first target characteristic information corresponding to the second operation characteristic information.
5. The method of claim 4, wherein determining target operation result information based on the first target feature information corresponding to the second operation feature information comprises: determining second target feature information based on first target feature information and third operation feature information corresponding to the second operation feature information; And determining target operation result information based on the second target characteristic information.
6. The method of claim 4, wherein the second operational characteristic information includes a plurality of first characteristic information and second characteristic information corresponding to the second target model, the second target model includes a plurality of first models, the first characteristic information characterizes inference times of the respective first models, and the second characteristic information characterizes operational times of the second target model in a first time; the first target characteristic information is determined based on: Carrying out weighted summation processing on a plurality of first characteristic information to obtain third characteristic information; and multiplying the third characteristic information and the second characteristic information to obtain first target characteristic information.
7. The method of claim 5, wherein determining second target feature information based on the first target feature information and third operating feature information corresponding to the second operating feature information comprises: determining third target feature information based on the third operation feature information; and subtracting the first target characteristic information corresponding to the third target characteristic information and the second operation characteristic information to obtain second target characteristic information.
8. The method of claim 7, wherein the third operation feature information includes fourth feature information and fifth feature information corresponding to each of the first object models, the fourth feature information characterizing a time required for each of the first object models to operate once, and the fifth feature information characterizing a number of operations of each of the first object models in a first time; The determining, based on the third operation feature information, third target feature information includes: Multiplying the fourth characteristic information and the fifth characteristic information corresponding to each first target model to obtain sixth characteristic information corresponding to each first target model; Adding the sixth characteristic information corresponding to the plurality of first target models to obtain seventh characteristic information; And subtracting the first time and the seventh characteristic information to obtain third target characteristic information.
9. The method of claim 8, wherein the fifth characteristic information for each of the first object models is determined based on: the method comprises the steps of obtaining eighth characteristic information of each first target model, wherein the eighth characteristic information characterizes the running times of each first target model in a second time; And dividing the eighth characteristic information and the second time to obtain fifth characteristic information corresponding to each first target model.
10. The method of claim 5, wherein determining target operation result information based on the second target feature information comprises: If the target feature value corresponding to the second target feature information is greater than a third threshold value, determining that the target operation result information is the first operation result information, and/or, And if the target characteristic value is smaller than or equal to the third threshold value, determining target operation result information based on the first priority information corresponding to the second target model and the second priority information corresponding to the plurality of first target models.
11. The method according to claim 10, wherein determining the target operation result information based on the first priority information corresponding to the second target model and the second priority information corresponding to the plurality of first target models includes: Determining whether a model with a priority lower than that of the second target model exists in the plurality of first target models or not based on first priority information corresponding to the second target model and second priority information corresponding to the plurality of first target models; when a model with lower priority than the second target model exists in the plurality of first target models, determining that the target operation result information is third operation result information, and/or, And when the models with the priority lower than that of the second target model do not exist in the plurality of first target models, determining the target operation result information as fourth operation result information.
12. The method according to any one of claims 1 to 11, wherein the performing an analysis decision process on the first model information and the operation environment information to obtain target operation result information includes: acquiring fourth operation characteristic information of a plurality of third target models which are currently operated and fifth operation characteristic information of a plurality of third target models, wherein the fourth operation characteristic information represents actual operation times of each third target model in a first time, and the fifth operation characteristic information represents minimum operation times of each third target model in the first time; and optimizing the operation environments of a plurality of third target models based on the fourth operation characteristic information and the fifth operation characteristic information.
13. The method of claim 12, wherein optimizing the operating environments of the plurality of third object models based on the fourth operating characteristic information and the fifth operating characteristic information comprises: Comparing the fourth operational characteristic information of the third object model with the fifth operational characteristic information of the third object model for any one of a plurality of the third object models; If the number of times of operation corresponding to the fourth operation characteristic information of the third target model is smaller than the number of times of operation corresponding to the fifth operation characteristic information of the third target model, determining a fourth target model from a plurality of third target models based on third priority information of the plurality of third target models, and stopping operating the fourth target model, wherein the fourth target model is a model with priority meeting a first condition in the plurality of third target models.
14. A system, comprising: the information acquisition module is used for acquiring the first model information and the operation environment information; the analysis processing module is used for carrying out analysis decision processing on the first model information and the operation environment information to obtain target operation result information; Optionally, the analysis processing module performs analysis decision processing on the first model information and the running environment information to obtain target running result information, including: Extracting features of the operation environment information to obtain first operation feature information, wherein the first operation feature information characterizes the number of a plurality of first target models currently in operation; Determining target operation result information based on the first model information and the first operation characteristic information; Optionally, the analysis processing module determines target operation result information based on the first model information and the first operation feature information, including: If the feature value corresponding to the first operation feature information is smaller than or equal to a first threshold value, determining that the target operation result information is the first operation result information, and/or, If the characteristic value corresponding to the first operation characteristic information is larger than the first threshold value, determining target operation result information based on the first model information; optionally, the first model information includes resource occupation information and second operation feature information, the resource occupation information characterizes a resource occupation condition of the second target model when in operation, and the analysis processing module determines target operation result information based on the first model information, including: If the resource occupation value corresponding to the resource occupation information is larger than or equal to a second threshold value, determining that the target operation result information is second operation result information, and/or, If the resource occupation value corresponding to the resource occupation information is smaller than the second threshold value, determining target operation result information based on first target characteristic information corresponding to the second operation characteristic information; Optionally, the analysis processing module determines target operation result information based on the first target feature information corresponding to the second operation feature information, including: determining second target feature information based on first target feature information and third operation feature information corresponding to the second operation feature information; Determining target operation result information based on the second target feature information; Optionally, the second operation feature information includes a plurality of first feature information and second feature information corresponding to the second target model, the second target model includes a plurality of first models, the first feature information characterizes inference time of each first model, and the second feature information characterizes operation times of the second target model in the first time; The first target characteristic information is obtained by the analysis processing module executing the following steps: Carrying out weighted summation processing on a plurality of first characteristic information to obtain third characteristic information; multiplying the third characteristic information and the second characteristic information to obtain first target characteristic information; optionally, the analysis processing module determines second target feature information based on the first target feature information and third operation feature information corresponding to the second operation feature information, including: determining third target feature information based on the third operation feature information; subtracting the first target characteristic information corresponding to the third target characteristic information and the second operation characteristic information to obtain second target characteristic information; Optionally, the third operation feature information includes fourth feature information and fifth feature information corresponding to each first target model, where the fourth feature information characterizes time required for each first target model to operate once, and the fifth feature information characterizes operation times of each first target model in a first time; the analysis processing module determines third target feature information based on the third operation feature information, and the analysis processing module comprises: Multiplying the fourth characteristic information and the fifth characteristic information corresponding to each first target model to obtain sixth characteristic information corresponding to each first target model; Adding the sixth characteristic information corresponding to the plurality of first target models to obtain seventh characteristic information; Subtracting the first time and the seventh characteristic information to obtain third target characteristic information; optionally, fifth feature information corresponding to each first target model is obtained by executing the following steps by the analysis processing module: the method comprises the steps of obtaining eighth characteristic information of each first target model, wherein the eighth characteristic information characterizes the running times of each first target model in a second time; dividing the eighth characteristic information and the second time to obtain fifth characteristic information corresponding to each first target model; Optionally, the analysis processing module determines target operation result information based on the second target feature information, including: If the target feature value corresponding to the second target feature information is greater than a third threshold value, determining that the target operation result information is the first operation result information, and/or, If the target characteristic value is smaller than or equal to the third threshold value, determining target operation result information based on first priority information corresponding to the second target model and second priority information corresponding to a plurality of first target models; Optionally, the analysis processing module determines target operation result information based on first priority information corresponding to the second target model and second priority information corresponding to a plurality of first target models, including: Determining whether a model with a priority lower than that of the second target model exists in the plurality of first target models or not based on first priority information corresponding to the second target model and second priority information corresponding to the plurality of first target models; when a model with lower priority than the second target model exists in the plurality of first target models, determining that the target operation result information is third operation result information, and/or, When a model with a priority lower than that of the second target model does not exist in the plurality of first target models, determining that the target operation result information is fourth operation result information; Optionally, the analysis processing module performs analysis decision processing on the first model information and the operation environment information, and after obtaining the target operation result information, the analysis processing module is further configured to: acquiring fourth operation characteristic information of a plurality of third target models which are currently operated and fifth operation characteristic information of a plurality of third target models, wherein the fourth operation characteristic information represents actual operation times of each third target model in a first time, and the fifth operation characteristic information represents minimum operation times of each third target model in the first time; Optimizing the operation environments of a plurality of third target models based on the fourth operation characteristic information and the fifth operation characteristic information; Optionally, the analysis processing module performs optimization processing on the operating environments of the plurality of third target models based on the fourth operating characteristic information and the fifth operating characteristic information, and includes: Comparing the fourth operational characteristic information of the third object model with the fifth operational characteristic information of the third object model for any one of a plurality of the third object models; If the number of times of operation corresponding to the fourth operation characteristic information of the third target model is smaller than the number of times of operation corresponding to the fifth operation characteristic information of the third target model, determining a fourth target model from a plurality of third target models based on third priority information of the plurality of third target models, and stopping operating the fourth target model, wherein the fourth target model is a model with priority meeting a first condition in the plurality of third target models.
15. A computer device, the computer device comprising: one or more processors; Memory, and One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the processor to implement the method of any of claims 1 to 13.
16. A computer readable storage medium, having stored thereon a computer program, the computer program being loaded by a processor to perform the steps of the method of any of claims 1 to 13.

Description

Data processing system for optimizing model operation Technical Field The application relates to the technical field of computers, in particular to a data processing system for optimizing model operation. Background With the rapid development of artificial intelligence (ARTIFICIAL INTELLIGENCE, AI) technology, the number of AI models running on a terminal is increasing, and the number of scenes in which a plurality of AI models need to be run simultaneously is also increasing. However, the AI computing power and the system resources on the terminal are limited, and the reasoning time consumption of the AI model can have larger fluctuation along with the shortage degree of the system resources, so that the actual operation effect of the AI algorithm is poor when a plurality of AI algorithms are simultaneously operated on the terminal. Disclosure of Invention The embodiment of the application provides a data processing system for optimizing model operation. In a first aspect, the present application provides a method comprising: acquiring first model information and running environment information; and analyzing and deciding the first model information and the operation environment information to obtain target operation result information. In a second aspect, the present application provides a system comprising: the information acquisition module is used for acquiring the first model information and the operation environment information; And the analysis processing module is used for carrying out analysis decision processing on the first model information and the operation environment information to obtain target operation result information. In a third aspect, the present application also provides a computer device comprising: one or more processors; Memory, and One or more applications, wherein the one or more applications are stored in memory and configured to be executed by a processor to implement the method of any of the first aspects. In a fourth aspect, embodiments of the present application provide a computer readable storage medium having stored thereon a computer program to be loaded by a processor for performing the steps of the method of any of the first aspects. Drawings In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments described in the present invention, and other drawings may be obtained according to the drawings without inventive effort to those skilled in the art. FIG. 1 is a schematic diagram of a scenario of a data processing system provided by an embodiment of the present invention; FIG. 2 is a flow chart of one embodiment of a data processing method provided by an embodiment of the present invention; FIG. 3 is a flowchart of an embodiment of an analysis decision process for first model information and operating environment information according to an embodiment of the present invention; FIG. 4 is a flow chart of another embodiment of a data processing method provided by an embodiment of the present invention; FIG. 5 is a functional block diagram of a data processing system provided by an embodiment of the present invention; Fig. 6 is a schematic structural diagram of an embodiment of a computer device according to an embodiment of the present invention. Detailed Description The following description of the embodiments of the present application will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present application, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to fall within the scope of the application. In the description of the present application, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. indicate orientations or positional relationships based on the drawings are merely for convenience in describing the present application and simplifying the description, and do not indicate or imply that the apparatus or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and thus should not be construed as limiting the present application. Furthermore, the terms "first," "second," "third," and the like are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, features defining "first," "seco