Data processing method, device, storage medium and electronic device

Abstract

The present application discloses a data processing method, device, storage medium and electronic device. Relating to the field of artificial intelligence, the method includes: obtaining resource usage information of a target application during operation in a current application scenario to obtain target resource usage information; judging whether there is a reference application scenario in multiple application scenarios based on the target resource usage information and reference resource usage information in the target application scenario in multiple application scenarios to obtain a judgment result; selecting a target model from a first model and multiple second models based on the judgment result, and determining parameter tuning information of the target application through the target model, wherein the first model is a large language model, and the second model is a neural network model other than the large language model. The present application solves the technical problem that a fixed neural network model is used in the related art to determine parameter tuning information, and there is low flexibility in determining parameter tuning information.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence, and more specifically, to a data processing method, device, storage medium and electronic device. Background Art

[0002] When tuning parameters, it is often necessary to cover parameters related to CPU (Central Processing Unit), memory, network, and storage at multiple levels, such as the operating system kernel, compiler, and database. There are a large number of these parameters, and the value space of each parameter is not small. As a result, parameter tuning is often a high-dimensional space search problem. Whether it is a Bayesian series of algorithms based on Gaussian distribution or a series of algorithms based on proxy functions, a large number of rounds of execution are required to obtain relatively effective tuning results (often 2.5 times the number of rounds of parameters are required, and more rounds are required when the parameter value space is large).

[0003] Currently, in the related technology, parameter tuning information is determined through a neural network model to reduce the time required to determine the parameter tuning information compared to Bayesian and proxy functions. However, when a fixed neural network model is used to determine the parameter tuning information, there is a problem of low flexibility in determining the parameter tuning information.

[0004] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0005] The embodiments of the present application provide a data processing method, device, storage medium and electronic device to at least solve the technical problem that a fixed neural network model is used in the related art to determine parameter tuning information, and there is low flexibility in determining the parameter tuning information.

[0006] According to one aspect of an embodiment of the present application, a data processing method is provided, including: obtaining resource usage information of a target application during operation in a current application scenario to obtain target resource usage information, wherein the target application is an application to be parameter tuned; judging whether there is a reference application scenario in multiple application scenarios based on the target resource usage information and reference resource usage information in a target application scenario in multiple application scenarios to obtain a judgment result, wherein a scene similarity between the reference application scenario and the current application scenario is greater than a target threshold; selecting a target model from a first model and multiple second models based on the judgment result, and determining parameter tuning information of the target application through the target model, wherein the first model is a large language model, and the second model is a neural network model other than the large language model, the first model is trained based on historical tuning data information in multiple application scenarios, and different second models are trained based on historical tuning data information in different application scenarios.

[0007] Furthermore, based on the target resource usage information and the reference resource usage information in the target application scenario in multiple application scenarios, it is judged whether there is a reference application scenario in the multiple application scenarios, and the judgment result obtained includes: based on the target resource usage information and the reference resource usage information in the target application scenario in multiple application scenarios, it is judged whether there is a first reference application scenario in the multiple reference resource usage information, and the similarity between the reference resource usage information and the target resource usage information is greater than a first threshold, to obtain a first judgment sub-result; determining a first sensitive parameter set for the target application in the current application scenario, wherein the sensitive parameters in the first sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the current application scenario is greater than a first preset degree; based on the first sensitive parameter set and the first reference sensitive parameter set in the target application scenario in multiple application scenarios, it is judged whether there is a first reference sensitive parameter set including the first sensitive parameter set in multiple first reference sensitive parameter sets, to obtain a second judgment sub-result; determining the judgment result based on the first judgment sub-result and the second judgment sub-result.

[0008] Furthermore, in the case where the target model is the second model, determining the parameter tuning information of the target application through the target model includes: obtaining tuning evaluation results corresponding to multiple groups of first parameter combinations of the target application; for a first parameter combination in the multiple groups of first parameter combinations, combining the first parameter combination and the tuning evaluation results corresponding to the first parameter combination to obtain first tuning data information; inputting the first tuning data information into the target model, and obtaining the parameter tuning information of the target application through the target model.

[0009] Furthermore, obtaining tuning evaluation results corresponding to multiple groups of first parameter combinations for the target application includes: obtaining a first reference sensitive parameter set for a reference application scenario, wherein the sensitive parameters in the first reference sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the reference application scenario is greater than a first preset degree; determining multiple groups of first parameter combinations corresponding to the target application according to the first reference sensitive parameter set; performing parameter tuning on the target application based on the first parameter combination in the multiple groups of first parameter combinations, and performing tuning evaluation on the target application after parameter tuning, to obtain tuning evaluation results corresponding to the multiple groups of first parameter combinations.

[0010] Furthermore, the second model is obtained in the following manner: obtaining historical tuning data information under the application scenario corresponding to the second model to be generated, wherein the historical tuning data information includes parameters, parameter values ​​and historical tuning evaluation results used when the application is historically tuned in the application scenario; using the historical tuning data information under the application scenario as training samples, and the application type and scenario type associated with the application scenario as true labels to construct a first training sample set; training the initial neural network model based on the first training sample set to obtain the second model.

[0011] Furthermore, the first model learns the second reference sensitive parameter set under multiple application scenarios based on the historical tuning data information. When the target model is the first model, determining the parameter tuning information of the target application through the target model includes: determining the second tuning data information according to multiple groups of second parameter combinations corresponding to the target application; inputting the second tuning data information into the target model, and determining the second sensitive parameter set of the target application in the current application scenario through the target model according to the second tuning data information and the second reference sensitive parameter set; determining multiple groups of target parameter combinations based on the second sensitive parameter set, and determining the tuning evaluation results corresponding to the multiple groups of target parameter combinations of the target application; combining the target parameter combinations in the multiple groups of target parameter combinations with the tuning evaluation results corresponding to the target parameter combinations to obtain the target tuning data information, and constructing a second training sample set according to the target tuning data information corresponding to the multiple groups of target parameter combinations; training the target model based on the second training sample set to obtain the trained target model, and determining the parameter tuning information through the trained target model.

[0012] Furthermore, the first model is obtained in the following manner: obtaining historical tuning data information under multiple application scenarios, wherein the historical tuning data information includes parameters, parameter values, and historical tuning evaluation results used when the application performs historical parameter tuning in the application scenarios; using the historical tuning data information under multiple application scenarios as training samples, and the application types and scenario types associated with the application scenarios as true labels to construct a third training sample set; training an initial large language model based on the third training sample set to obtain the first model.

[0013] Furthermore, selecting a target model from the first model and multiple second models based on the judgment result includes: detecting whether the first model has been deployed in the target server; if the first model has been deployed in the target server, detecting whether there is a target identifier in a target configuration file associated with the target application, wherein the target identifier represents that the first model is determined as the target model; if the target identifier does not exist in the target configuration file, selecting the target model from the first model and multiple second models based on the judgment result.

[0014] Furthermore, selecting a target model from the first model and multiple second models based on the judgment result includes: when the judgment result indicates the existence of a reference application scenario, determining a second model associated with the reference application scenario among the multiple second models as the target model; when the judgment result indicates the absence of a reference application scenario, determining the first model as the target model.

[0015] According to another aspect of an embodiment of the present application, a data processing method is also provided, including: obtaining resource usage information of a target application uploaded by a client during operation in a current application scenario, and obtaining target resource usage information, wherein the target application is an application to be parameter tuned; judging in a cloud server whether there is a reference application scenario in multiple application scenarios based on the target resource usage information and reference resource usage information in a target application scenario in multiple application scenarios, and obtaining a judgment result, selecting a target model from a first model and multiple second models based on the judgment result, and determining parameter tuning information of the target application through the target model, wherein the scenario similarity between the reference application scenario and the current application scenario is greater than a target threshold, the first model is a large language model, and the second model is a neural network model other than the large language model, the first model is trained based on historical tuning data information in multiple application scenarios, and different second models are trained based on historical tuning data information in different application scenarios; and feeding back the parameter tuning information of the target application to the client.

[0016] According to another aspect of an embodiment of the present application, a data processing device is also provided, including: a first acquisition unit, used to acquire resource usage information of a target application during operation in a current application scenario, and obtain target resource usage information, wherein the target application is an application to be parameter tuned; a judgment unit, used to judge whether there is a reference application scenario in multiple application scenarios based on the target resource usage information and reference resource usage information in the target application scenario in multiple application scenarios, and obtain a judgment result, wherein the scene similarity between the reference application scenario and the current application scenario is greater than a target threshold; a determination unit, used to select a target model from a first model and multiple second models based on the judgment result, and determine the parameter tuning information of the target application through the target model, wherein the first model is a large language model, and the second model is a neural network model other than the large language model, and the first model is trained based on historical tuning data information in multiple application scenarios, and different second models are trained based on historical tuning data information in different application scenarios.

[0017] Further, the judgment unit includes: a first judgment sub-unit, used to judge whether there is a first reference resource usage information in multiple reference resource usage information whose similarity with the target resource usage information is greater than a first threshold value based on the target resource usage information and the reference resource usage information in the target application scenario in multiple application scenarios, and obtain a first judgment sub-result; a first determination sub-unit, used to determine a first sensitive parameter set of the target application in the current application scenario, wherein the sensitive parameters in the first sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the current application scenario is greater than a first preset degree; a second judgment sub-unit, used to judge whether there is a first reference sensitive parameter set including the first sensitive parameter set in multiple first reference sensitive parameter sets based on the first sensitive parameter set and the first reference sensitive parameter set in the target application scenario in multiple application scenarios, and obtain a second judgment sub-result; a second determination sub-unit, used to determine the judgment result based on the first judgment sub-result and the second judgment sub-result.

[0018] Furthermore, in the case that the target model is the second model, the determination unit includes: an acquisition subunit, used to obtain tuning evaluation results corresponding to multiple groups of first parameter combinations of the target application; a first processing subunit, used to combine the first parameter combination and the tuning evaluation results corresponding to the first parameter combination for the first parameter combination in the multiple groups of first parameter combinations to obtain first tuning data information; and a second processing subunit, used to input the first tuning data information into the target model, and obtain parameter tuning information of the target application through the target model.

[0019] Furthermore, the acquisition subunit includes: an acquisition module, used to acquire a first reference sensitive parameter set for a reference application scenario, wherein the sensitive parameters in the first reference sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the reference application scenario is greater than a first preset degree; a determination module, used to determine multiple groups of first parameter combinations corresponding to the target application according to the first reference sensitive parameter set; and a processing module, used to perform parameter tuning on the target application based on a first parameter combination in the multiple groups of first parameter combinations, and to perform tuning evaluation on the target application after parameter tuning to obtain tuning evaluation results corresponding to the multiple groups of first parameter combinations.

[0020] Furthermore, the data processing device also includes: a second acquisition unit, used to obtain historical tuning data information in an application scenario corresponding to the second model to be generated, wherein the historical tuning data information includes parameters, parameter values ​​and historical tuning evaluation results used when performing historical parameter tuning in the application scenario; a first processing unit, used to use the historical tuning data information in the application scenario as a training sample, and the application type and scenario type associated with the application scenario as true labels to construct a first training sample set; and a second processing unit, used to train an initial neural network model based on the first training sample set to obtain a second model.

[0021] Furthermore, the first model learns the second reference sensitive parameter sets under multiple application scenarios based on the historical tuning data information. When the target model is the first model, the determination unit includes: a third determination subunit, which is used to determine the second tuning data information according to the multiple groups of second parameter combinations corresponding to the target application; a third processing subunit, which is used to input the second tuning data information into the target model, and determine the second sensitive parameter set of the target application in the current application scenario through the target model according to the second tuning data information and the second reference sensitive parameter set; a fourth determination subunit, which is used to determine multiple groups of target parameter combinations based on the second sensitive parameter set, and determine the tuning evaluation results corresponding to the multiple groups of target parameter combinations of the target application; a fourth processing subunit, which is used to combine the target parameter combinations in the multiple groups of target parameter combinations with the tuning evaluation results corresponding to the target parameter combinations to obtain the target tuning data information, and construct a second training sample set according to the target tuning data information corresponding to the multiple groups of target parameter combinations; a fifth processing subunit, which is used to train the target model based on the second training sample set to obtain the trained target model, and determine the parameter tuning information through the trained target model.

[0022] Furthermore, the data processing device also includes: a third acquisition unit, used to obtain historical tuning data information under multiple application scenarios, wherein the historical tuning data information includes parameters, parameter values ​​and historical tuning evaluation results used when the application performs historical parameter tuning in the application scenario; a third processing unit, used to use the historical tuning data information under multiple application scenarios as training samples, and the application type and scenario type associated with the application scenario as true labels to construct a third training sample set; a fourth processing unit, used to train an initial large language model based on the third training sample set to obtain a first model.

[0023] Furthermore, the determination unit includes: a first detection subunit, used to detect whether the first model has been deployed in the target server; a second detection subunit, used to detect whether there is a target identifier in the target configuration file associated with the target application when the first model has been deployed in the target server, wherein the target identifier represents that the first model is determined as the target model; and a selection subunit, used to select the target model from the first model and multiple second models based on the judgment result when the target identifier does not exist in the target configuration file.

[0024] Furthermore, the determination unit includes: a fifth determination subunit, which is used to determine a second model associated with a reference application scenario among multiple second models as a target model when the judgment result represents the existence of a reference application scenario; and a sixth determination subunit, which is used to determine the first model as the target model when the judgment result represents the absence of a reference application scenario.

[0025] According to another aspect of an embodiment of the present invention, there is further provided an electronic device, comprising: a memory storing an executable program; and a processor for running the program, wherein any one of the above-mentioned data processing methods is executed when the program is running.

[0026] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is provided, wherein the storage medium stores a program, wherein when the program is running, the device where the storage medium is located is controlled to execute any one of the above-mentioned data processing methods.

[0027] According to another aspect of an embodiment of the present invention, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the computer program implements any one of the above data processing methods.

[0028] In an embodiment of the present application, resource usage information of a target application during operation in a current application scenario is obtained to obtain target resource usage information, wherein the target application is an application to be parameter tuned; based on the target resource usage information and reference resource usage information in a target application scenario in multiple application scenarios, it is determined whether there is a reference application scenario in multiple application scenarios to obtain a judgment result, wherein a scene similarity between the reference application scenario and the current application scenario is greater than a target threshold; based on the judgment result, a target model is selected from a first model and multiple second models, and parameter tuning information of the target application is determined through the target model, wherein the first model is a large language model, and the second model is a neural network model other than the large language model, the first model is trained based on historical tuning data information in multiple application scenarios, and different second models are trained based on historical tuning data information in different application scenarios.

[0029] In the above process, by acquiring the resource usage information of the target application during the operation process in the current application scenario, the target resource usage information is obtained, and the relevant feature data that can characterize the current application scenario of the target application is effectively obtained. According to the target resource usage information and the reference resource usage information of the target application scenario in multiple application scenarios, it is judged whether there is a reference application scenario in multiple application scenarios, and it is judged whether there is an application scenario similar to the current application scenario in multiple application scenarios according to the feature data corresponding to the multiple application scenarios, that is, it is judged whether the type of the current application scenario can be accurately identified. By setting the first model to be trained based on the historical tuning data information in multiple application scenarios, and different second models to be trained based on the historical tuning data information in different application scenarios, the first model and the second model can learn the relevant knowledge in the corresponding application scenario, so that in the process of selecting the target model from the first model and the multiple second models according to the judgment result, it can be achieved that the target model is determined according to the knowledge of the application scenario learned by the first model and the second model and the recognition situation of the current application scenario, and then the target model is dynamically determined between the first model and the second model according to the actual application situation, so as to further determine the parameter tuning information, thereby effectively improving the flexibility of determining the parameter tuning information. Thereby, the purpose of dynamically selecting a model from a large language model and a neural network model other than the large language model to determine parameter tuning information is achieved, and the technical effect of effectively improving the flexibility of determining parameter tuning information is realized, thereby solving the technical problem of low flexibility in determining parameter tuning information in related technologies using a fixed neural network model to determine parameter tuning information. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0031] Figure 1 is a schematic diagram of a computer terminal provided according to Embodiment 1 of the present application;

[0032] Figure 2 This is the process of the data processing method provided in Example 1 of this application Figure 1 ;

[0033] Figure 3 is a schematic diagram of an optional target selection model provided according to Embodiment 1 of the present application;

[0034] Figure 4 This is the process of the optional data processing method provided in Example 1 of the present application. Figure 2 ;

[0035] Figure 5 is a flow chart of a data processing method provided according to Embodiment 2 of the present application;

[0036] Figure 6 is a schematic diagram of a data processing device provided according to Embodiment 3 of the present application;

[0037] Figure 7 It is a structural block diagram of an electronic device provided according to Embodiment 4 of the present application. DETAILED DESCRIPTION

[0038] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0039] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0040] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards in the relevant regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0041] First, some nouns or terms that appear in the description of the embodiments of the present application are subject to the following explanations:

[0042] Parameter tuning: The process of continuously improving the system's score on a benchmark program by adjusting system parameters.

[0043] System parameters: refers to configuration items in the operating system that can be dynamically set and affect application performance, such as kernel configuration parameters and application configuration parameters.

[0044] Benchmark program: An evaluation program used to estimate the performance of a tuning application and give a performance score.

[0045] Parameter tuning algorithm: A program that implements a certain algorithm strategy and continuously gives system parameter combinations to find parameter configurations that improve application performance.

[0046] Proxy model: A mathematical model that constructs a mapping relationship between system parameter combinations and benchmark program scores, and uses this to guide the operation of the parameter tuning algorithm.

[0047] LLM: Large Language Model, large language model.

[0048] Example 1

[0049] According to an embodiment of the present application, a data processing method is also provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0050] The method embodiment provided in the first embodiment of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 1 FIG. 1 shows a hardware structure block diagram of a computer terminal (or mobile device) for implementing a data processing method. Figure 1 As shown, the computer terminal (or mobile device) 10 may include a processor set 102 (the processor set 102 may include but is not limited to a processing device such as a microprocessor (Microcontroller Unit, referred to as MCU) or a programmable logic device (Field Programmable Gate Array, referred to as FPGA), and the processor set 102 may include a processor set, Figure 1 102a, 102b, ..., 102n are used to illustrate), a memory 104 for storing data, and a transmission device 106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a Universal Serial Bus (USB) port (which may be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. Those skilled in the art can understand that Figure 1The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0051] It should be noted that the one or more processors 102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuits". The data processing circuits may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuits may be a single independent processing module, or may be incorporated in whole or in part into any of the other components in the computer terminal 10 (or mobile device).

[0052] The memory 104 can be used to store software programs and modules of application software, such as program instructions / data storage devices corresponding to the data processing method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, realizing the above-mentioned data processing method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the computer terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0053] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the computer terminal 10. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0054] The display may be a touch screen type liquid crystal display, which may enable a user to interact with a user interface of the computer terminal 10 (or mobile device).

[0055] When performing parameter tuning, it is often necessary to cover parameters related to CPU, memory, network, and storage at multiple levels such as the operating system kernel, compiler, and database. There are a large number of these parameters, and the value space of each parameter is not small. As a result, parameter tuning is often a high-dimensional space search problem. Whether it is the Bayesian series of algorithms based on Gaussian distribution or a series of algorithms based on proxy functions, a large number of rounds of execution are required to obtain relatively effective tuning results (often 2.5 times the number of rounds of parameters are required, and more rounds are required when the parameter value space is large). In the actual tuning process, the execution time of each round of the benchmark program is generally relatively long. Taking the stress test of the analytical query capability of the database MySQL as an example, it takes more than 2.5 hours to execute one round. Then, if 200 parameters are tuned, at least 500 rounds of execution are required, which takes more than 50 days, which is difficult for users to accept.

[0056] Currently, in the related technology, parameter tuning information is determined through a neural network model to reduce the time required to determine the parameter tuning information compared to Bayesian and proxy functions. However, when a fixed neural network model is used to determine the parameter tuning information, there is a problem of low flexibility in determining the parameter tuning information.

[0057] In the above technical background, the present application provides Figure 2 The data processing method shown. Figure 2 This is the process of the data processing method provided in Example 1 of this application Figure 1 The method comprises:

[0058] Step S201, obtaining resource usage information of a target application during operation in a current application scenario, and obtaining target resource usage information, wherein the target application is an application to be parameter-optimized.

[0059] Optionally, electronic devices, application systems, servers and other devices can be used as the execution subject of the present application. In this embodiment, the target processing system is used as the execution subject to execute the data processing method. Among them, the target application includes but is not limited to database applications, server applications, big data applications, etc. Database applications include but are not limited to relational databases MySQL, Redis (Remote Dictionary Server, remote dictionary service), etc. Server applications include but are not limited to reverse proxy Web (World Wide Web, World Wide Web) servers, and big data applications include but are not limited to applications based on distributed storage and computing frameworks, or open source computing frameworks for large-scale data processing and other frameworks.

[0060] Optionally, the application scenario of the application includes two parts of information: application type and scenario type, where the application type can be MySQL, Redis, etc., and the scenario type can be a read scenario, a write scenario, etc. For example, an optional application scenario can be expressed as "MySQL read scenario". Depending on the actual application requirements, the same application may adopt different scenarios in different time ranges. For example, MySQL is a read scenario during the day, responsible for querying data according to user needs, and a write scenario at night, responsible for updating data in MySQL. Therefore, the same application may correspond to multiple application scenarios. Depending on the actual application requirements, the application type and scenario type are also different, and can be divided at a finer granularity. Therefore, in this embodiment, the content of the application type and scenario type is not specifically limited.

[0061] Optionally, the resource usage information is used to characterize the resource usage of the application. The resource usage information includes, but is not limited to, the usage information of resources such as CPU, memory, process, disk, network, etc. during the running of the application. For example, the CPU usage information may be the CPU usage rate, the memory usage information may be the remaining memory, and the process usage information may be the frequency of process context switching.

[0062] Optionally, the target processing system may obtain resource usage information of the target application during operation in the current application scenario through relevant detection tools, or a task manager, or by executing system commands, so as to obtain the target resource usage information.

[0063] Step S202, based on the target resource usage information and the reference resource usage information under the target application scenario in the multiple application scenarios, determine whether there is a reference application scenario in the multiple application scenarios, and obtain a judgment result, wherein the scene similarity between the reference application scenario and the current application scenario is greater than the target threshold.

[0064] When using an algorithm model for parameter tuning, if you want to use historical data to extract valid information from it, you need to normalize the historical data offline according to certain requirements and "label" it; then run the algorithm model based on the labeled data to identify parameter sensitivity, correlation, and characteristics of the candidate regions with better values, which is "training". When performing actual tuning online, the real-time data obtained is used to determine the actual better parameter candidate value combination (equivalent to parameter tuning information) based on the trained model, which is "inference".

[0065] Therefore, in an optional embodiment, the target processing system can pre-train two models using the marked historical tuning data information (equivalent to the aforementioned historical data), namely a large model and a small model. Among them, the large model is also a large language model. This type of model can be trained based on the historical tuning data information in multiple application scenarios, so that a single large model can learn the relevant knowledge of multiple application scenarios during the training process. The small model refers to a neural network model other than the large language model. Different small models can be trained based on the historical tuning data information in different application scenarios, so that a single small model can learn the relevant knowledge of a single application scenario during the training process. The small model can be a DNN (Deep Neural Networks). Considering the problem of historical information attenuation in traditional RNN (Recurrent Neural Network) and CNN (Convolutional Neural Networks), DNN can adopt an LSTM (Long Short-Term Memory) algorithm based on RNN improvement.

[0066] Optionally, the historical tuning data information for a certain application scenario includes the parameters, parameter values, and historical tuning evaluation results used when the application performs historical parameter tuning in the application scenario. The real label added to the historical tuning data information for the application scenario may be the application type and scenario type associated with the application scenario, that is, the application scenario is used as the real label of the historical tuning data information. For example, an optional real label may be "MySQL_read_only_full_load", where "MySQL" is equivalent to the application type, and "read_only_full_load" is equivalent to the scenario type of "read-only, full data loading".

[0067] When using the labeled historical tuning data information for model training and reasoning, the large model and the small model have the following characteristics: For the small model, one application scenario of an application corresponds to a small model, and the final result is a model library including multiple small models. The size of the small model is from several MB to more than ten MB, and the computing, memory and other resources consumed during operation are relatively small. Therefore, it can be deployed in the application's operating environment. Its advantage is that the model is small and can be deployed in the application's operating environment. The deployment is simple and the operation speed is fast. Its disadvantage is that it is necessary to determine which small model to use before obtaining the result during operation; and because the small model only has the knowledge of training data extraction, if the data input to the small model during the reasoning process is different from the data used in the training of the small model in the application scenario, it will lead to poor tuning effect. For the large model, its model size is from more than ten GB to hundreds of GB, and the computing, memory and other resources consumed during operation are very large. Therefore, it cannot be deployed in the application's operating environment, but needs to be deployed in an environment that can be connected to the application's operating environment network, and the reasoning task is processed through network requests when it is used. Its advantage is that it can identify the application scenario of the application based on the model itself, and there is no need for the target processing system to add an operation to determine the application scenario. Moreover, since the large model is fine-tuned based on the pre-trained LLM, the pre-trained LLM itself has certain knowledge. Combined with the historical tuning data information used in the fine-tuning process of the large model, the large model can have more preset knowledge in terms of sensitive parameters, value space, parameter correlation, etc. Its disadvantage is that the model is large, can only be deployed in an independent environment, and the deployment is complex and the running speed is slow.

[0068] In actual application, compared with large models, small models consume less computing, memory and other resources when running. Therefore, if small models are used to replace large models to determine parameter tuning information under certain circumstances, the efficiency of determining parameter tuning information can be effectively improved. Furthermore, since different small models are trained based on historical tuning data information in different application scenarios, and small models only have knowledge extracted from training data, it is difficult to accurately process knowledge in all application scenarios during the reasoning process. Therefore, in order to minimize the impact on the accuracy of parameter tuning information when using small models to determine parameter tuning information, it is possible to first determine whether a small model that has learned the knowledge of the current application scenario of the target application can be matched from multiple pre-trained small models, so as to determine whether to use the small model to determine the parameter tuning information.

[0069] Optionally, the target processing system may first obtain reference resource usage information in a target application scenario among multiple application scenarios. The multiple application scenarios may be understood as application scenarios based on which multiple small models are trained, that is, the pre-trained multiple small models have learned the knowledge in these multiple application scenarios, and the target application scenario may be any one of the multiple application scenarios.

[0070] Optionally, reference resource usage information corresponding to multiple application scenarios can be pre-stored in a preset storage area. The reference resource usage information corresponding to the target application scenario can be determined by the target processing system based on the resource usage accumulated during the historical operation of the application in the target application scenario, or it can be preset by the user. The reference resource usage information can be understood as representing the theoretical resource usage of the application. The content form of the reference resource usage information is the same as that of the target resource usage information, so it will not be repeated here.

[0071] Optionally, scenario similarity can be understood as the similarity between two application scenarios in the application working method (or application working mode). The target processing system can determine whether there is a reference application scenario in multiple application scenarios based on the similarity between the target resource usage information and the reference resource usage information under the target application scenario in multiple application scenarios to obtain a judgment result.

[0072] In an optional embodiment, the target processing system may determine the scene similarity between the reference application scene and the current application scene based on the similarity between the resource usage information, and then compare the scene similarity with the target threshold to obtain a judgment result. For example, the target threshold is a preset value, and the similarity between the target resource usage information and the reference resource usage information is determined as the scene similarity between the current application scene and the target application scene. If there is reference resource usage information that is relatively similar to the target resource usage information (i.e., reference resource usage information with a similarity greater than the target threshold), it is determined that there is a reference application scene in the multiple application scenes, and the reference application scene is the application scene corresponding to the aforementioned relatively similar reference resource usage information. If there is no reference resource usage information that is relatively similar to the target resource usage information, it is determined that there is no reference application scene in the multiple application scenes. For another example, the target threshold is a preset value, and the target processing system can determine the first sensitive parameter set of the target application in the current application scenario, and then determine the similarity between the first sensitive parameter set and the first reference sensitive parameter set in the target application scenario based on the inclusion relationship between the first sensitive parameter set and the first reference sensitive parameter set. For example, if a first reference sensitive parameter set includes the first sensitive parameter set, then the similarity between the first reference sensitive parameter set and the first sensitive parameter set is determined to be the first similarity; if a first reference sensitive parameter set does not include the first sensitive parameter set, then the similarity between the first reference sensitive parameter set and the first sensitive parameter set is determined to be the second similarity, and the first similarity is greater than the second similarity, so that the scene similarity between the current application scenario and the target application scenario is determined based on the aforementioned determined similarity and the similarity between the target resource usage information and the reference resource usage information, and then the scene similarity is compared with the target threshold to obtain a judgment result.

[0073] In an optional embodiment, the target processing system may also determine whether there is an application scenario that meets the preset conditions in multiple application scenarios based on the similarity between the resource usage information, thereby determining whether there is a reference application scenario in multiple application scenarios based on whether there is an application scenario that meets the preset conditions in multiple application scenarios. For example, when there is an application scenario that meets the preset conditions in multiple application scenarios, it is determined that there is a reference application scenario in multiple application scenarios, and the application scenario that meets the preset conditions is the reference application scenario. When there is no application scenario that meets the preset conditions in multiple application scenarios, it is determined that there is no reference application scenario in multiple application scenarios, that is, the preset condition can be understood as the condition that the application scenario needs to meet to be the reference application scenario. For example, it is determined whether there is a reference resource usage information that is similar to the target resource usage information in multiple reference resource usage information to obtain a first judgment sub-result, it is determined whether there is a first reference sensitive parameter set including a first sensitive parameter set in multiple first reference sensitive parameter sets to obtain a second judgment sub-result, and then it is determined whether there is an application scenario that meets the preset conditions in multiple application scenarios based on the first judgment sub-result and the second judgment sub-result, thereby facilitating obtaining a judgment result.

[0074] Step S203: Select a target model from the first model and multiple second models based on the judgment result, and determine the parameter tuning information of the target application through the target model, wherein the first model is a large language model, and the second model is a neural network model other than the large language model. The first model is trained based on historical tuning data information under multiple application scenarios, and different second models are trained based on historical tuning data information under different application scenarios.

[0075] In an optional embodiment, the first model is equivalent to the large model mentioned above, and the second model is equivalent to the small model mentioned above.

[0076] In an optional embodiment, the current application scenario of the target application may belong to multiple application scenarios. However, affected by the accuracy of the target resource usage information, the target processing system may be unable to accurately identify which of the multiple application scenarios the current application scenario of the target application belongs to, that is, it is unable to determine the reference application scenario. In this case, if a second model is selected from multiple second models to determine the parameter tuning information of the target application, the probability of selecting the wrong second model is relatively large, which will affect the accuracy of the parameter tuning information.

[0077] Therefore, the target processing system can determine that there is an application scenario that is relatively similar to the current application scenario among multiple application scenarios when determining that there is a reference application scenario, that is, it can be determined that the reference application scenario can be determined as the current application scenario, thereby determining that a small model that has learned the knowledge of the current application scenario can be matched from multiple pre-trained small models, and then determining to select a target model from multiple second models to improve the efficiency of determining parameter tuning information. For example, based on the association relationship between multiple second models and multiple application scenarios, a second model associated with the reference application scenario among multiple second models is determined as the target model. Optionally, the second model is trained based on historical tuning data information in the application scenario associated with the second model.

[0078] Optionally, the target processing system may determine that there is no application scenario that is relatively similar to the current application scenario among the multiple application scenarios when it is determined that there is no reference application scenario, that is, it is impossible to relatively accurately determine which application scenario among the multiple application scenarios the current application scenario belongs to, and there is a possibility of recognition error, thereby determining that it is impossible to match the small model that has learned the knowledge of the current application scenario of the target application from the pre-trained multiple small models, and then determining the first model as the target model to improve the accuracy of determining the parameter tuning information.

[0079] Optionally, after determining the target model, the target processing system can determine the parameter tuning information of the target application through the target model. The parameter tuning information of the target application can be understood as the information required for parameter tuning of the target application. The parameter tuning information may include multiple parameters required for parameter tuning of the target application and parameter values ​​corresponding to the parameters. Since both the first model and the second model are trained using labeled historical tuning data information, the first model and the second model can learn the correlation between the application tuning data information and the application scenario of the application, as well as the correlation between the tuning evaluation results and the tuning parameters and parameter values ​​during the training process. Therefore, the target processing system can determine the tuning data information of the target application in the current application scenario, that is, obtain the real-time data of the target application, and then process the tuning data information of the target application in the current application scenario through the target model to obtain the parameter tuning information of the target application.

[0080] Optionally, after obtaining the parameter tuning information, the target processing system may send the parameter tuning information to the user, so that when receiving the user's application tuning request, the target processing system may perform parameter tuning on the target application based on the parameter tuning information. Optionally, after obtaining the parameter tuning information, the target processing system may also directly perform parameter tuning on the target application based on the parameter tuning information.

[0081] It should be noted that by dynamically determining the target model between the first model and the second model according to actual application conditions to further determine the parameter tuning information, not only the efficiency of determining the parameter tuning information can be effectively improved, but also the flexibility of determining the parameter tuning information can be improved.

[0082] In this solution, resource usage information of the target application during operation in the current application scenario is obtained to obtain the target resource usage information, thereby realizing effective acquisition of relevant feature data that can characterize the current application scenario of the target application. According to the target resource usage information and the reference resource usage information of the target application scenario in multiple application scenarios, it is judged whether there is a reference application scenario in multiple application scenarios, thereby realizing judging whether there is an application scenario similar to the current application scenario in multiple application scenarios according to the feature data corresponding to the multiple application scenarios, that is, judging whether the type of the current application scenario can be accurately identified. By setting the first model to be trained based on the historical tuning data information in multiple application scenarios, and different second models to be trained based on the historical tuning data information in different application scenarios, the first model and the second model can learn the relevant knowledge in the corresponding application scenario, thereby in the process of selecting the target model from the first model and the multiple second models according to the judgment result, it is possible to realize determining the target model according to the knowledge of the application scenario learned by the first model and the second model and the recognition situation of the current application scenario, thereby realizing dynamically determining the target model between the first model and the second model according to the actual application situation, so as to further determine the parameter tuning information, thereby effectively improving the flexibility of determining the parameter tuning information. Thereby, the purpose of dynamically selecting a model from a large language model and a neural network model other than the large language model to determine parameter tuning information is achieved, and the technical effect of effectively improving the flexibility of determining parameter tuning information is realized, thereby solving the technical problem of low flexibility in determining parameter tuning information in related technologies using a fixed neural network model to determine parameter tuning information.

[0083] How to judge whether there is an application scenario that is relatively similar to the current application scenario in multiple application scenarios is crucial. Therefore, in the data processing method provided in the first embodiment of the present application, based on the target resource usage information and the reference resource usage information in the target application scenario in multiple application scenarios, it is judged whether there is a reference application scenario in multiple application scenarios, and the judgment result obtained includes: based on the target resource usage information and the reference resource usage information in the target application scenario in multiple application scenarios, it is judged whether there is a reference resource usage information in multiple reference resource usage information and the target resource usage information has a similarity greater than a first threshold, and a first judgment sub-result is obtained; determine the first sensitive parameter set of the target application in the current application scenario, wherein the sensitive parameters in the first sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the current application scenario is greater than a first preset degree; based on the first sensitive parameter set and the first reference sensitive parameter set in the target application scenario in multiple application scenarios, it is judged whether there is a first reference sensitive parameter set including the first sensitive parameter set in multiple first reference sensitive parameter sets, and a second judgment sub-result is obtained; and the judgment result is determined based on the first judgment sub-result and the second judgment sub-result.

[0084] Optionally, the target application scenario may be any one of a plurality of application scenarios, and the plurality of reference resource usage information corresponds to the plurality of application scenarios.

[0085] In the process of determining whether there is any reference resource usage information among multiple reference resource usage information and the target resource usage information whose similarity is greater than a first threshold, optionally, the target processing system may input multiple reference resource usage information and the target resource usage information into a pre-trained discriminant model, and determine a first judgment sub-result through the discriminant model.

[0086] Optionally, the target processing system may also compare the resource usage information in the reference resource usage information with the resource usage information in the target resource usage information, thereby determining the similarity relationship between the reference resource usage information and the target resource usage information based on the comparison results corresponding to the resources, and further determining the first judgment sub-result based on the similarity relationship between multiple reference resource usage information and the target resource usage information.

[0087] For example, for the same resource in a certain reference resource usage information and the target resource usage information, when the difference between the values ​​of the resource in the reference resource usage information and the target resource usage information is less than the preset difference corresponding to the resource, it is determined that the comparison result corresponding to the resource indicates that the similarity of the usage information of the resource in the reference resource usage information and the target resource usage information is greater than the first threshold value, otherwise, it is determined that the comparison result corresponding to the resource indicates that the similarity of the usage information of the resource in the reference resource usage information and the target resource usage information is less than or equal to the first threshold value. Further, the target processing system can determine the similarity between the reference resource usage information and the target resource usage information according to the comparison result corresponding to the resource. For example, when there is no comparison result indicating that the similarity corresponding to the resource is less than or equal to the first threshold value in the comparison result corresponding to the reference resource usage information, it is determined that the similarity between the reference resource usage information and the target resource usage information is greater than the first threshold value, otherwise, when there is a comparison result indicating that the similarity corresponding to the resource is less than or equal to the first threshold value in the comparison result corresponding to the reference resource usage information, it is determined that the similarity between the reference resource usage information and the target resource usage information is less than or equal to the first threshold value. Thus, the first judgment sub-result can be determined according to the similarity relationship between multiple reference resource usage information and the target resource usage information.

[0088] In order to further improve the accuracy of determining whether there is a reference application scenario in multiple application scenarios, the determination result may be determined by combining the sensitive parameter set of the current application scenario with the sensitive parameter sets of multiple application scenarios.

[0089] In an optional embodiment, the first sensitive parameter set may be determined by the target processing system through a model interpreter. In the process of determining the first sensitive parameter set of the target application in the current application scenario, the target processing system may first determine the multiple sets of second parameter combinations corresponding to the target application in the current application scenario, wherein the second parameter combination includes multiple parameters and parameter values ​​corresponding to the multiple parameters, and the parameters in the second parameter combination may be the full amount of adjustable parameters of the target application. Afterwards, the target processing system may determine the second tuning data information of the target application in the current application scenario according to the multiple sets of second parameter combinations corresponding to the target application. For example, the target application is parameter tuned based on the second parameter combination in the multiple sets of second parameter combinations, and the tuning evaluation of the target application after parameter tuning is performed to obtain the tuning evaluation results corresponding to the multiple sets of second parameter combinations. Then, for the second parameter combination in the multiple sets of second parameter combinations, the tuning evaluation results corresponding to the second parameter combination and the second parameter combination are combined to obtain the second tuning data information. Among them, the tuning evaluation result may include multiple evaluation indicators and indicator values ​​corresponding to the evaluation indicators, and the evaluation indicators include but are not limited to response time, throughput, throughput rate, number of concurrency, throughput bandwidth, etc. Tuning evaluation can be achieved through benchmark programs or other system detection tools.

[0090] Optionally, after determining the second tuning data information corresponding to the target application, the target processing system may input the second tuning data information into the model interpreter, and then use the parameter interpretability algorithm of the model interpreter to determine the influence degree value of the parameters in the second parameter combination on the tuning effect of the target application in the current application scenario according to the second tuning data information. Thereafter, based on the influence degree values ​​corresponding to the parameters in the second parameter combination, the first sensitive parameter set is determined according to the parameters in the second parameter combination. Optionally, the parameter interpretability algorithm may be a Lasso (Least Absolute Shrinkage and Selection Operator) algorithm, or an XGboost (eXtreme Gradient Boosting) algorithm, or other algorithms.

[0091] In an optional embodiment, the first reference sensitive parameter set can be pre-stored in a sensitive information library, and the first reference sensitive parameter set is determined by the target processing system through a model interpreter. Optionally, for a target application scenario among multiple application scenarios, the target processing system can input the historical tuning data information under the target application scenario into the model interpreter, thereby obtaining the influence degree value of the parameters in the historical tuning data information of the target application scenario on the tuning effect applied in the target application scenario, and then the parameters in the historical tuning data information of the target application scenario can be screened based on the influence degree value corresponding to the parameters in the historical tuning data information of the target application scenario to obtain the first reference sensitive parameter set corresponding to the target application scenario. For example, the parameters in the historical tuning data information of the target application scenario are sorted in order from high to low according to the influence degree value, and the sorted parameters are obtained, and then the first N parameters in the sorted parameters are determined as sensitive parameters in the first reference sensitive parameter set. For another example, the parameters in the historical tuning data information of the target application scenario whose influence degree value is greater than the preset degree value are determined as sensitive parameters in the first reference sensitive parameter set.

[0092] Optionally, after determining the first sensitive parameter set and the first reference sensitive parameter sets of multiple application scenarios, the target processing system can determine whether there is a first reference sensitive parameter set including the first sensitive parameter set in the multiple first reference sensitive parameter sets to obtain a second judgment sub-result. Optionally, the multiple first reference sensitive parameter sets correspond to multiple application scenarios.

[0093] For example, for a first reference sensitive parameter set among multiple first reference sensitive parameter sets, the target processing system can determine whether the first reference sensitive parameter set includes the first sensitive parameter set to obtain a third judgment sub-result. For example, if a first reference sensitive parameter set includes A parameters, B parameters, and C parameters, and the first sensitive parameter set includes A parameters and B parameters, then it is determined that the first reference sensitive parameter set includes the first sensitive parameter set. Optionally, in the case where there is a third judgment sub-result representing that the first reference sensitive parameter set includes the first sensitive parameter set, it is determined that there is a first reference sensitive parameter set including the first sensitive parameter set among the multiple first reference sensitive parameter sets; conversely, in the case where there is no third judgment sub-result representing that the first reference sensitive parameter set includes the first sensitive parameter set, it is determined that there is no first reference sensitive parameter set including the first sensitive parameter set among the multiple first reference sensitive parameter sets.

[0094] After determining the first judgment sub-result and the second judgment sub-result, the target processing system may determine a judgment result according to the first judgment sub-result and the second judgment sub-result.

[0095] Optionally, the target processing system may determine the judgment result based on OR logic. For example, when the first judgment sub-result indicates that there is a similarity between the reference resource usage information and the target resource usage information greater than the first threshold, or when the second judgment sub-result indicates that there is a first reference sensitive parameter set including the first sensitive parameter set, it is determined that there is an application scenario that meets the preset condition in the multiple application scenarios, that is, it is determined that the judgment result indicates that there is a reference application scenario in the multiple application scenarios, and the application scenario corresponding to the reference resource usage information whose similarity with the target resource usage information is greater than the first threshold is determined as the reference application scenario, or the application scenario corresponding to the first reference sensitive parameter set including the first sensitive parameter set is determined as the reference application scenario, that is, the preset condition can be understood as "the similarity between the reference resource usage information corresponding to the application scenario and the target resource usage information is greater than the first threshold", or "the first reference sensitive parameter set corresponding to the application scenario includes the first sensitive parameter set". When the first judgment sub-result indicates that there is no similarity between the reference resource usage information and the target resource usage information greater than the first threshold, and the second judgment sub-result indicates that there is no first reference sensitive parameter set including the first sensitive parameter set, it is determined that there is no application scenario that meets the preset condition in the multiple application scenarios, that is, it is determined that the judgment result indicates that there is no reference application scenario in the multiple application scenarios.

[0096] Optionally, the target processing system may also determine the judgment result based on the AND logic. For example, when the first judgment sub-result indicates that the similarity between the reference resource usage information and the target resource usage information is greater than the first threshold, and the second judgment sub-result indicates that the first reference sensitive parameter set including the first sensitive parameter set exists, it is determined whether the application scenarios corresponding to the reference resource usage information and the first reference sensitive parameter set are the same. If they are the same, it is determined that there is an application scenario that meets the preset conditions among the multiple application scenarios, that is, it is determined that the judgment result indicates that there is a reference application scenario among the multiple application scenarios, and the application scenario corresponding to the reference resource usage information and the first reference sensitive parameter set is determined as the reference scenario, that is, the preset condition can be understood as "the similarity between the reference resource usage information corresponding to the application scenario and the target resource usage information is greater than the first threshold", and "the first reference sensitive parameter set corresponding to the application scenario includes the first sensitive parameter set", if the reference resource usage information and the application scenario corresponding to the first reference sensitive parameter set are not the same, it is determined that there is no application scenario that meets the preset conditions among the multiple application scenarios, that is, it is determined that the judgment result indicates that there is no reference application scenario among the multiple application scenarios. When the first judgment sub-result indicates that there is no reference resource usage information and the similarity between the target resource usage information is greater than the first threshold, or the second judgment sub-result indicates that there is no first reference sensitive parameter set including the first sensitive parameter set, it is determined that there is no application scenario that meets the preset conditions among the multiple application scenarios, that is, it is determined that the judgment result indicates that there is no reference application scenario among the multiple application scenarios.

[0097] It should be noted that by combining sensitive parameters to determine whether there is a reference application scenario in multiple application scenarios based on resource usage information, it is possible to evaluate the similarity between the current application scenario and multiple application scenarios based on multi-dimensional information, thereby effectively improving the accuracy of the determined judgment results.

[0098] In order to improve the accuracy of selecting a target model, in the data processing method provided in the first embodiment of the present application, selecting a target model from a first model and multiple second models based on a judgment result includes: detecting whether the first model has been deployed in the target server; if the first model has been deployed in the target server, detecting whether there is a target identifier in a target configuration file associated with the target application, wherein the target identifier represents that the first model is determined as the target model; if the target identifier does not exist in the target configuration file, selecting the target model from the first model and multiple second models based on the judgment result.

[0099] Optionally, the target server refers to a server that supports deployment of the first model. The target server may be preset, and the target server is connected to the operating environment network of the target application.

[0100] Optionally, the target configuration file associated with the target application can be stored in the target application's operating environment or in a storage area outside the target application's operating environment. The target configuration file is at least used to record a target identifier, which can be set by a user according to actual application requirements, and the target identifier indicates that the first model is determined as the target model.

[0101] In an optional embodiment, it is possible to use Figure 3 The schematic diagram shown in Figure 1 is used to select the target model. Figure 3 is a schematic diagram of an optional target model selection provided according to the first embodiment of the present application, such as Figure 3 As shown, the target processing system can first detect whether the first model has been deployed in the target server. If not, the target processing system can select a second model from multiple second models as the target model. For example, based on the similarity between the target resource usage information and the reference resource usage information under the target application scenario in multiple application scenarios, an application scenario that is more similar to the current application scenario can be determined from multiple application scenarios, and the second model associated with the application scenario can be determined as the target model. If the first model has been deployed in the target server, the target processing system can detect whether there is a target identifier (i.e., Figure 3 In the step of determining whether to use the first model, if there is a target identifier, the first model is directly determined as the target model. If there is no target identifier, the target model is selected from the first model and the plurality of second models according to the determination result.

[0102] Optionally, in the process of detecting whether the first model has been deployed in the target server, the target processing system may use a detection tool to detect whether the first model has been deployed in the target server. For example, the target processing system may detect whether the first model is recorded in the configuration file of the target server, or detect whether there is a configuration file associated with the first model in the target server, so as to detect whether the first model has been deployed in the target server.

[0103] It should be noted that by dynamically selecting whether to use the first model or the second model based on the deployment situation, whether to manually select to use LLM, and whether the real-time data of the application matches a certain small model (i.e., the process of selecting the target model based on the judgment result), on the one hand, the flexibility and accuracy of the model selection are improved, that is, the accuracy of the determined parameter tuning information is improved, and on the other hand, the efficiency of determining the parameter tuning information is effectively improved. The above method can effectively adapt to the differences in environment deployment, scenarios, etc. when the actual application parameters are tuned, and provide accurate and effective parameter tuning information as much as possible while using as few resources as possible.

[0104] In order to improve the accuracy of selecting a target model, in the data processing method provided in Example 1 of the present application, selecting a target model from a first model and multiple second models based on a judgment result includes: when the judgment result indicates that there is a reference application scenario, determining a second model associated with the reference application scenario among multiple second models as the target model; when the judgment result indicates that there is no reference application scenario, determining the first model as the target model.

[0105] Alternatively, if Figure 3 As shown, the target processing system can determine that there is an application scenario that is relatively similar to the current application scenario among multiple application scenarios when it is determined that there is a reference application scenario, that is, it can be determined that the reference application scenario can be determined as the current application scenario, thereby determining a second model that can match the knowledge of the current application scenario of the target application that has been learned from multiple pre-trained second models, and then based on the association relationship between the multiple second models and the multiple application scenarios, the second model associated with the reference application scenario among the multiple second models is determined as the target model.

[0106] Alternatively, if Figure 3 As shown, the target processing system can determine that there is no application scenario that is relatively similar to the current application scenario among multiple application scenarios when it is determined that there is no reference application scenario, and it is impossible to relatively accurately determine which application scenario among the multiple application scenarios the current application scenario belongs to, and there is a possibility of recognition error, thereby determining that it is impossible to match the second model that has learned the knowledge of the current application scenario of the target application from the pre-trained multiple second models, and then determine the first model as the target model.

[0107] It should be noted that by selecting a target model from multiple second models when the judgment result representation exists a reference application scenario, the efficiency of determining the parameter tuning information can be effectively improved. By determining the first model as the target model when the judgment result representation does not exist a reference application scenario, it is avoided that when the second model is selected as the target model when there is no application scenario similar to the current application scenario among multiple application scenarios, there may be a large difference between the knowledge learned by the second model during the training process and the knowledge input into the second model during the reasoning process, thereby affecting the accuracy of determining the parameter tuning information.

[0108] In order to improve the accuracy of determining parameter tuning information, in the data processing method provided in Example 1 of the present application, when the target model is the second model, determining the parameter tuning information of the target application through the target model includes: obtaining tuning evaluation results corresponding to multiple groups of first parameter combinations of the target application; for a first parameter combination in the multiple groups of first parameter combinations, combining the first parameter combination and the tuning evaluation results corresponding to the first parameter combination to obtain first tuning data information; inputting the first tuning data information into the target model, and obtaining the parameter tuning information of the target application through the target model.

[0109] Optionally, the target processing system may first determine multiple groups of first parameter combinations of the target application in the current application scenario. The first parameter combination includes multiple parameters and parameter values ​​corresponding to the multiple parameters, and the parameters in the first parameter combination may be determined according to the first reference sensitive parameter set corresponding to the reference application scenario.

[0110] Optionally, after determining multiple groups of first parameter combinations for the target application in the current application scenario, the target processing system may obtain tuning evaluation results corresponding to the multiple groups of first parameter combinations for the target application in the current application scenario. For example, for a first parameter combination among multiple groups of first parameter combinations, the target processing system may perform parameter tuning on the target application according to the first parameter combination in the current application scenario of the target application, and perform tuning evaluation on the target application after parameter tuning to obtain a tuning evaluation result corresponding to the first parameter combination. For another example, for a first parameter combination among multiple groups of first parameter combinations, the target processing system may input the first parameter combination into a target prediction model, and predict the tuning evaluation result corresponding to the first parameter combination through the target prediction model.

[0111] Optionally, for a first parameter combination in the multiple groups of first parameter combinations, the target processing system may combine the first parameter combination and a tuning evaluation result corresponding to the first parameter combination to obtain first tuning data information corresponding to the first parameter combination.

[0112] Afterwards, the target processing system can input the first tuning data information into the target model, and obtain the parameter tuning information of the target application through the target model. Since different second models are trained based on historical tuning data information under different application scenarios, the second model can learn the association between the application tuning data information and the application scenario of the application according to the historical tuning data information during the training process, and learn the association between the parameter combination and the tuning evaluation result according to the historical tuning data information during the training process. Therefore, in the actual reasoning process, when the target model receives the first tuning data information, it can infer the parameter tuning information of the target application according to the first tuning data information.

[0113] It should be noted that by obtaining the tuning evaluation results corresponding to multiple groups of first parameter combinations of the target application and combining them to obtain the first tuning data information, the real-time data of the target application in the current application scenario is effectively acquired. By using the target model to process the real-time data of the target application to determine the parameter tuning information, the accuracy of the determined parameter tuning information is improved.

[0114] In order to further improve the efficiency of determining parameter tuning information, in the data processing method provided in Example 1 of the present application, obtaining tuning evaluation results corresponding to multiple groups of first parameter combinations for the target application includes: obtaining a first reference sensitive parameter set for a reference application scenario, wherein the sensitive parameters in the first reference sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the reference application scenario is greater than a first preset degree; determining multiple groups of first parameter combinations corresponding to the target application based on the first reference sensitive parameter set; performing parameter tuning on the target application based on the first parameter combination in the multiple groups of first parameter combinations, and performing tuning evaluation on the target application after parameter tuning, to obtain tuning evaluation results corresponding to the multiple groups of first parameter combinations.

[0115] Optionally, the first reference sensitive parameter set of the reference application scenario may be pre-stored in the sensitive information library, and the first reference sensitive parameter set of the reference application scenario is determined by the target processing system through the model interpreter.

[0116] In an optional embodiment, in the process of determining multiple groups of first parameter combinations corresponding to the target application according to the first reference sensitive parameter set, the target processing system selects any one value from the parameter value range corresponding to the sensitive parameter as the parameter value of the sensitive parameter for the sensitive parameter in the first reference sensitive parameter set, thereby generating a group of first parameter combinations according to the sensitive parameters in the first reference sensitive parameter set and their corresponding parameter values. Thus, multiple groups of first parameter combinations are obtained by random multiple times. Optionally, the parameter value range corresponding to the sensitive parameter in the first reference sensitive parameter set can be preset by the user.

[0117] Optionally, the target processing system may also use a Bayesian algorithm or a proxy model to determine parameter values ​​of sensitive parameters in the first reference sensitive parameter set, thereby obtaining multiple groups of first parameter combinations.

[0118] After determining multiple groups of first parameter combinations, for a first parameter combination in the multiple groups of first parameter combinations, the target processing system can perform parameter tuning on the target application in the current application scenario of the target application based on the first parameter combination, that is, update the parameter value of the parameter in the target application based on the first parameter combination, thereby performing tuning evaluation on the target application after parameter tuning, and obtaining a tuning evaluation result corresponding to the first parameter combination. The target processing system can use a benchmark program or other methods to perform tuning evaluation on the target application after parameter tuning.

[0119] It should be noted that by determining multiple groups of first parameter combinations corresponding to the target application based on the first reference sensitive parameter set, the amount of data in the input data of the target model is effectively reduced, avoiding the impact of the working efficiency of the target model due to the large amount of parameters when the full amount of parameters is input into the target model. Furthermore, by tuning the parameters of the target application based on the first parameter combination and then performing a tuning evaluation to determine the tuning evaluation result, the effectiveness of the obtained tuning evaluation result is improved, thereby improving the accuracy of the parameter tuning information determined by the target model.

[0120] In order to obtain a second model with better performance, in the data processing method provided in Example 1 of the present application, the second model is obtained in the following manner: obtaining historical tuning data information under the application scenario corresponding to the second model to be generated, wherein the historical tuning data information includes parameters, parameter values ​​and historical tuning evaluation results used when the application is historically tuned in the application scenario; using the historical tuning data information under the application scenario as training samples, and the application type and scenario type associated with the application scenario as true labels to construct a first training sample set; training the initial neural network model based on the first training sample set to obtain the second model.

[0121] Optionally, the correspondence between the second model and the application scenario can be preset, and the target processing system can determine the application scenario corresponding to the second model to be generated based on the correspondence. The historical tuning data information under the application scenario corresponding to the second model includes the parameters, parameter values ​​and historical tuning evaluation results used when the application performs historical parameter tuning in the application scenario. The historical tuning evaluation results can be obtained by evaluating the application after the application is historically tuned according to the parameters and parameter values ​​used by the application in historical parameter tuning. The historical tuning evaluation results include multiple evaluation indicators and indicator values ​​corresponding to the evaluation indicators. The evaluation indicators include but are not limited to response time, throughput, throughput rate, concurrency, throughput bandwidth, etc. The tuning evaluation can be implemented through a benchmark program or through other system detection tools.

[0122] After determining the historical tuning data information under the application scenario corresponding to the second model to be generated, the historical tuning data information under the application scenario can be used as training samples, and the application type and scenario type associated with the application scenario can be used as true labels to construct a first training sample set, thereby training the initial neural network model based on the first training sample set to obtain the second model.

[0123] In an optional embodiment, it is possible to use Figure 4 The schematic diagram shown in FIG. 1 is used to implement the training of the second model. Figure 4 This is the process of the optional data processing method provided in Example 1 of the present application. Figure 2 ,like Figure 4 As shown, a first training sample set is constructed based on the historical tuning data information in the application scenario corresponding to the second model to be generated, and the first training sample set and the full adjustable parameter library are used as inputs of the initial neural network model to train the initial neural network model to obtain a second model. Among them, the full adjustable parameter library contains multiple levels and domain parameters of the kernel, kernel driver, compiler, runtime, and application, and the number of parameters can be in the range of 2000-100000. By using the full adjustable parameter library as the input of the initial neural network model, the second model can learn information about all parameters. Optionally, the full adjustable parameter library can also include parameter value ranges corresponding to the parameters in the full adjustable parameters. The parameter value ranges corresponding to the above-mentioned sensitive parameters can be determined from the full adjustable parameter library. As Figure 4 As shown, the second model is deployed in the operating environment of the target application. After the second model is trained, the target processing system can select the target model from the first model and multiple second models according to the real-time data of the target application, thereby determining the parameter tuning information.

[0124] It should be noted that by setting the historical tuning data information to include the parameters, parameter values ​​and historical tuning evaluation results used in the historical parameter tuning, the richness of the historical tuning data information is improved. By constructing the first training sample set in the above manner and training the initial neural network model based on the first training sample set, the trained second model can learn the relevant knowledge when applying parameter tuning based on the rich historical data, thereby improving the accuracy of the parameter tuning information determined by the second model during the reasoning process.

[0125] In order to improve the accuracy of determining parameter tuning information, in the data processing method provided in Example 1 of the present application, the first model learns the second reference sensitive parameter set under multiple application scenarios based on the historical tuning data information. When the target model is the first model, determining the parameter tuning information of the target application through the target model includes: determining the second tuning data information according to multiple groups of second parameter combinations corresponding to the target application; inputting the second tuning data information into the target model, and determining the second sensitive parameter set of the target application in the current application scenario through the target model according to the second tuning data information and the second reference sensitive parameter set; determining multiple groups of target parameter combinations based on the second sensitive parameter set, and determining the tuning evaluation results corresponding to the multiple groups of target parameter combinations of the target application; combining the target parameter combinations in the multiple groups of target parameter combinations with the tuning evaluation results corresponding to the target parameter combinations to obtain the target tuning data information, and constructing a second training sample set according to the target tuning data information corresponding to the multiple groups of target parameter combinations; training the target model based on the second training sample set to obtain the trained target model, and determining the parameter tuning information through the trained target model.

[0126] In an optional embodiment, in the process of training the first model, the first model can learn the second reference sensitive parameter set in different application scenarios based on the historical tuning data information of multiple application scenarios. The sensitive parameters in the second reference sensitive parameter set corresponding to a certain application scenario refer to the parameters whose influence on the tuning effect of the application in the application scenario is greater than the second preset degree. In layman's terms, the sensitive parameters in the second reference sensitive parameter set can be understood as the parameters required for parameter tuning of the application determined by the first model.

[0127] In an optional embodiment, the target processing system may first determine multiple sets of second parameter combinations corresponding to the target application in the current application scenario, wherein the second parameter combinations include multiple parameters and parameter values ​​corresponding to the multiple parameters, and the parameters in the multiple sets of second parameter combinations may be all adjustable parameters of the target application. Afterwards, the target processing system may determine the second tuning data information of the target application in the current application scenario based on the multiple sets of second parameter combinations corresponding to the target application. For example, the target application is parameter tuned based on the second parameter combination in the multiple sets of second parameter combinations, and the tuning evaluation of the target application after parameter tuning is performed to obtain the tuning evaluation results corresponding to the multiple sets of second parameter combinations. Then, for the second parameter combination in the multiple sets of second parameter combinations, the second parameter combination and the tuning evaluation results corresponding to the second parameter combination are combined to obtain the second tuning data information.

[0128] Since the first model is trained based on historical tuning data information in multiple application scenarios, the first model can learn relevant knowledge of multiple application scenarios during the training process, and the current application scenario of the target application has not been accurately identified. Therefore, when the first model is used as the target model, the second sensitive parameter set of the target application in the current application scenario can be first determined through the target model, so as to determine the data to be input into the target model for the current application scenario.

[0129] Optionally, relying on the billion-level parameters corresponding to the hidden layer of the large language model, the large language model has a certain fuzzy tolerance, and the large language model has strong learning and generalization capabilities. Therefore, when the first model has learned the second reference sensitive parameter set for multiple application scenarios, the target processing system can input the above-mentioned second tuning data information into the target model, and determine the second sensitive parameter set of the target application in the current application scenario through the target model according to the second tuning data information and the second reference sensitive parameter set.

[0130] In an optional embodiment, the sensitive parameters in the second sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the current application scenario is greater than the second preset degree. The sensitive parameters in the second sensitive parameter set and the second reference sensitive parameter set are determined based on the first model, while the sensitive parameters in the first sensitive parameter set and the first reference sensitive parameter set are determined based on the model interpreter. The aforementioned first preset degree and the second preset degree may be the same or different.

[0131] In an optional embodiment, the set of parameters in the second parameter combination is greater than or equal to the union of the first reference sensitive parameter set and the second reference sensitive parameter set.

[0132] Optionally, after determining the second sensitive parameter set, the target processing system can determine multiple sets of target parameter combinations for the target application in the current application scenario based on the second sensitive parameter set, and determine the tuning evaluation results corresponding to the multiple sets of target parameter combinations for the target application in the current application scenario.

[0133] In an optional embodiment, in the process of determining multiple groups of target parameter combinations corresponding to the target application according to the second sensitive parameter set, for a sensitive parameter in the second reference sensitive parameter set, any value can be selected from the parameter value range corresponding to the sensitive parameter as the parameter value of the sensitive parameter, thereby generating a group of target parameter combinations according to the sensitive parameters in the second sensitive parameter set and their corresponding parameter values. Thus, multiple groups of target parameter combinations are obtained by random multiple times.

[0134] Optionally, the parameter value range corresponding to the sensitive parameters in the second reference sensitive parameter set may be determined by the target processing system according to the above-mentioned fully adjustable parameter library.

[0135] Optionally, the parameter value range corresponding to the sensitive parameters in the second reference sensitive parameter set may also be determined by the target processing system through the first model. For example, for a target application scenario in multiple application scenarios, there are multiple historical parameter tunings for the application in the target application scenario, that is, there are historical tuning data information corresponding to multiple historical parameter tunings for the application in the target application scenario. For the same target application scenario, the operating environment of the application in different historical parameter tunings in the target application scenario may be different. Therefore, the first model can learn the parameters required for parameter tuning in different application scenarios (that is, the second reference sensitive parameter set in different application scenarios) and the parameter value range of the parameters in the same application scenario in different operating environments based on the historical tuning data information. Environmental information includes, but is not limited to, CPU information, memory information, information of the operating system where the application is located, disk space, graphics card information, etc. in the current operating environment of the application in the application scenario. Among them, CPU information may refer to the number of CPU cores, memory information may refer to memory capacity, information of the operating system where the application is located may refer to the system type and version number of the operating system where the application is located, and graphics card information may refer to the graphics card model or performance.

[0136] Therefore, in the process of inputting the second tuning data information into the target model, the target processing system can also input the environmental information of the current operating environment of the target application into the target model, so that the target model processes the environmental information to obtain the parameter value range corresponding to the sensitive parameters in the second sensitive parameter set.

[0137] Optionally, after determining multiple groups of target parameter combinations, the target processing system may determine the tuning evaluation results corresponding to the multiple groups of target parameter combinations in accordance with the manner of determining the tuning evaluation results corresponding to the multiple groups of first parameter combinations, which will not be described in detail here.

[0138] Optionally, the target processing system can combine the target parameter combinations in the multiple sets of target parameter combinations with the tuning evaluation results corresponding to the target parameter combinations to obtain target tuning data information, and construct a second training sample set based on the target tuning data information corresponding to the multiple sets of target parameter combinations. For example, while outputting the second sensitive parameter set, the first model can also output the application type and scenario type associated with the current application scenario determined by the first model. In this case, the target processing system can use the target tuning data information corresponding to the multiple sets of target parameter combinations as training samples, and the application type and scenario type associated with the current application scenario as true labels to construct a second training sample set. Thereby, the second training sample set is input into the target model to train the target model based on the second training sample set. For another example, the target processing system can use the target tuning data information corresponding to the multiple sets of target parameter combinations as training samples, obtain a second training sample set based on the training samples, and generate a target prompt statement, thereby inputting the target prompt statement and the second training sample set into the target model to train the target model based on the target prompt statement and the second training sample set. Among them, the target prompt statement is used to guide the target model to use the application type and scenario type associated with the current application scenario determined in the process of processing the second tuning data information as the true label, and label the received data according to the true label, and then process based on the labeled data.

[0139] Optionally, when a trained target model is obtained, the trained target model can automatically output parameter tuning information. For example, when training the target model based on the second training sample set, the target processing system can input a prompt statement to the target model to guide the target model to output parameter tuning information when the trained target model is obtained through the prompt statement.

[0140] It should be noted that, by inputting the second tuning data information into the target model, the target model can infer the parameters (that is, the second sensitive parameter set) required for parameter tuning of the target application in the current application scenario, and by determining the second training sample set based on the parameters required for parameter tuning of the target application in the current application scenario, it is possible to construct training samples based on parameter combinations in the search space determined by the target model, and obtain a trained target model through training with the second training sample set to determine parameter tuning information, thereby achieving learning of multiple parameter combinations in the search space and optimizing parameter information through the trained target model, thereby improving the efficiency and accuracy of determining parameter tuning information.

[0141] In order to obtain a first model with better performance, in the data processing method provided in Example 1 of the present application, the first model is obtained in the following manner: obtaining historical tuning data information under multiple application scenarios, wherein the historical tuning data information includes parameters, parameter values ​​and historical tuning evaluation results used when the application is historically tuned in the application scenarios; using the historical tuning data information under multiple application scenarios as training samples, and the application types and scenario types associated with the application scenarios as true labels, to construct a third training sample set; training an initial large language model based on the third training sample set to obtain the first model.

[0142] In some embodiments, the initial large language model can adopt the Transformer algorithm, which is a neural network model based on the self-attention mechanism. The initial large language model can be understood as a pre-trained large language model. The process of obtaining the first model based on the third training sample set and the process of obtaining the trained target model based on the second training sample set can be understood as a fine-tuning process.

[0143] In an optional embodiment, it is possible to use Figure 4 The schematic diagram shown in FIG. 1 is used to train the second model. Figure 4 As shown, a third training sample set is constructed based on historical tuning data information in multiple application scenarios, and the third training sample set and the full adjustable parameter library are used as inputs of the initial large language model to train the initial large language model to obtain the first model. By using the full adjustable parameter library as input of the initial large language model, the first model can learn information about all parameters. Figure 4 As shown, the first model is deployed outside the running environment of the target application.

[0144] Optionally, in the process of training the first model based on the third training sample set, the first model can learn relevant knowledge in multiple application scenarios according to historical tuning data information and real labels.

[0145] In an embodiment of the present application, resource usage information of a target application during operation in a current application scenario is obtained to obtain target resource usage information, wherein the target application is an application to be parameter tuned; based on the target resource usage information and reference resource usage information in a target application scenario in multiple application scenarios, it is determined whether there is a reference application scenario in multiple application scenarios to obtain a judgment result, wherein a scene similarity between the reference application scenario and the current application scenario is greater than a target threshold; based on the judgment result, a target model is selected from a first model and multiple second models, and parameter tuning information of the target application is determined through the target model, wherein the first model is a large language model, and the second model is a neural network model other than the large language model, the first model is trained based on historical tuning data information in multiple application scenarios, and different second models are trained based on historical tuning data information in different application scenarios.

[0146] In the above process, by acquiring the resource usage information of the target application during the operation process in the current application scenario, the target resource usage information is obtained, and the relevant feature data that can characterize the current application scenario of the target application is effectively obtained. According to the target resource usage information and the reference resource usage information of the target application scenario in multiple application scenarios, it is judged whether there is a reference application scenario in multiple application scenarios, and it is judged whether there is an application scenario similar to the current application scenario in multiple application scenarios according to the feature data corresponding to the multiple application scenarios, that is, it is judged whether the type of the current application scenario can be accurately identified. By setting the first model to be trained based on the historical tuning data information in multiple application scenarios, and different second models to be trained based on the historical tuning data information in different application scenarios, the first model and the second model can learn the relevant knowledge in the corresponding application scenario, so that in the process of selecting the target model from the first model and the multiple second models according to the judgment result, it can be achieved that the target model is determined according to the knowledge of the application scenario learned by the first model and the second model and the recognition situation of the current application scenario, and then the target model is dynamically determined between the first model and the second model according to the actual application situation, so as to further determine the parameter tuning information, thereby effectively improving the flexibility of determining the parameter tuning information. Thereby, the purpose of dynamically selecting a model from a large language model and a neural network model other than the large language model to determine parameter tuning information is achieved, and the technical effect of effectively improving the flexibility of determining parameter tuning information is realized, thereby solving the technical problem of low flexibility in determining parameter tuning information in related technologies using a fixed neural network model to determine parameter tuning information.

[0147] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0148] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0149] Example 2

[0150] According to an embodiment of the present application, a data processing method is also provided, such as Figure 5 As shown, the method includes:

[0151] Step S501, obtaining resource usage information of a target application uploaded by a client during operation in a current application scenario, and obtaining target resource usage information, wherein the target application is an application to be parameter-optimized.

[0152] Step S502: In the cloud server, based on the target resource usage information and the reference resource usage information in the target application scenario in the multiple application scenarios, determine whether there is a reference application scenario in the multiple application scenarios, obtain a judgment result, select a target model from the first model and the multiple second models based on the judgment result, and determine the parameter tuning information of the target application through the target model, wherein the scenario similarity between the reference application scenario and the current application scenario is greater than the target threshold, the first model is a large language model, and the second model is a neural network model other than the large language model, the first model is trained based on historical tuning data information in multiple application scenarios, and different second models are trained based on historical tuning data information in different application scenarios.

[0153] Step S503: Feedback the parameter tuning information of the target application to the client.

[0154] Through the above scheme, the resource usage information of the target application in the running process in the current application scenario is obtained, and the target resource usage information is obtained, so that the relevant feature data that can characterize the current application scenario of the target application is effectively obtained. According to the target resource usage information and the reference resource usage information of the target application scenario in multiple application scenarios, it is judged whether there is a reference application scenario in multiple application scenarios, and it is judged whether there is an application scenario similar to the current application scenario in multiple application scenarios according to the feature data corresponding to the multiple application scenarios, that is, it is judged whether the type of the current application scenario can be accurately identified. By setting the first model to be trained based on the historical tuning data information in multiple application scenarios, and the different second models to be trained based on the historical tuning data information in different application scenarios, the first model and the second model can learn the relevant knowledge in the corresponding application scenario, so that in the process of selecting the target model from the first model and the multiple second models according to the judgment result, it can be realized that the target model is determined according to the knowledge of the application scenario learned by the first model and the second model and the recognition situation of the current application scenario, and then the target model is dynamically determined between the first model and the second model according to the actual application situation, so as to further determine the parameter tuning information, thereby effectively improving the flexibility of determining the parameter tuning information. Thereby, the purpose of dynamically selecting a model from a large language model and a neural network model other than the large language model to determine parameter tuning information is achieved, and the technical effect of effectively improving the flexibility of determining parameter tuning information is realized, thereby solving the technical problem of low flexibility in determining parameter tuning information in related technologies using a fixed neural network model to determine parameter tuning information.

[0155] In the cloud server, the specific method of the data processing method is the same as that in the first embodiment and will not be repeated here.

[0156] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the described order of actions, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the present application.

[0157] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, disk, CD), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0158] Example 3

[0159] According to an embodiment of the present application, a data processing device for implementing the above data processing method is also provided. Figure 6 As shown, the device includes: a first acquisition unit 601, a judgment unit 602, and a determination unit 603.

[0160] The first acquisition unit 601 is used to acquire resource usage information of a target application during operation in a current application scenario, and obtain target resource usage information, wherein the target application is an application to be parameter-tuned;

[0161] A judging unit 602 is used to judge whether there is a reference application scenario in the multiple application scenarios according to the target resource usage information and the reference resource usage information under the target application scenario in the multiple application scenarios, and obtain a judgment result, wherein the scene similarity between the reference application scenario and the current application scenario is greater than a target threshold;

[0162] The determination unit 603 is used to select a target model from the first model and multiple second models based on the judgment result, and determine the parameter tuning information of the target application through the target model, wherein the first model is a large language model, and the second model is a neural network model other than the large language model. The first model is trained based on historical tuning data information under multiple application scenarios, and different second models are trained based on historical tuning data information under different application scenarios.

[0163] In the data processing device provided in Example 3 of the present application, resource usage information of the target application during operation in the current application scenario is acquired through the first acquisition unit 601 to obtain target resource usage information, wherein the target application is an application to be parameter tuned; the judgment unit 602 judges whether there is a reference application scenario in multiple application scenarios based on the target resource usage information and the reference resource usage information in the target application scenario in multiple application scenarios, and obtains a judgment result, wherein the scene similarity between the reference application scenario and the current application scenario is greater than the target threshold; the determination unit 603 selects a target model from the first model and multiple second models based on the judgment result, and determines the parameter tuning information of the target application through the target model, wherein the first model is a large language model, and the second model is a neural network model other than the large language model, and the first model is trained based on historical tuning data information in multiple application scenarios, and different second models are trained based on historical tuning data information in different application scenarios. In this solution, by acquiring the resource usage information of the target application during the operation in the current application scenario, the target resource usage information is obtained, and the relevant feature data that can characterize the current application scenario of the target application is effectively obtained. According to the target resource usage information and the reference resource usage information of the target application scenario in multiple application scenarios, it is judged whether there is a reference application scenario in multiple application scenarios, and it is judged whether there is an application scenario similar to the current application scenario in multiple application scenarios according to the feature data corresponding to the multiple application scenarios, that is, it is judged whether the type of the current application scenario can be accurately identified. By setting the first model to be trained based on the historical tuning data information in multiple application scenarios, and the different second models to be trained based on the historical tuning data information in different application scenarios, the first model and the second model can learn the relevant knowledge in the corresponding application scenario, so that in the process of selecting the target model from the first model and the multiple second models according to the judgment result, it can be achieved that the target model is determined according to the knowledge of the application scenario learned by the first model and the second model and the recognition situation of the current application scenario, and then the target model is dynamically determined between the first model and the second model according to the actual application situation, so as to further determine the parameter tuning information, thereby effectively improving the flexibility of determining the parameter tuning information. Thereby, the purpose of dynamically selecting a model from a large language model and a neural network model other than the large language model to determine parameter tuning information is achieved, and the technical effect of effectively improving the flexibility of determining parameter tuning information is realized, thereby solving the technical problem of low flexibility in determining parameter tuning information in related technologies using a fixed neural network model to determine parameter tuning information.

[0164] Optionally, in the data processing device provided in Example 3 of the present application, the judgment unit 602 includes: a first judgment sub-unit, used to judge whether there is a first reference resource usage information in multiple reference resource usage information, based on the target resource usage information and the reference resource usage information in the target application scenario in multiple application scenarios, and whether the similarity between the reference resource usage information and the target resource usage information is greater than a first threshold, to obtain a first judgment sub-result; a first determination sub-unit, used to determine a first sensitive parameter set of the target application in the current application scenario, wherein the sensitive parameters in the first sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the current application scenario is greater than a first preset degree; a second judgment sub-unit, used to judge whether there is a first reference sensitive parameter set including the first sensitive parameter set in multiple first reference sensitive parameter sets, based on the first sensitive parameter set and the first reference sensitive parameter set in the target application scenario in multiple application scenarios, and obtain a second judgment sub-result; a second determination sub-unit, used to determine the judgment result based on the first judgment sub-result and the second judgment sub-result.

[0165] Optionally, in the data processing device provided in Example 3 of the present application, when the target model is the second model, the determination unit 603 includes: an acquisition subunit, used to obtain the tuning evaluation results corresponding to the target application in multiple groups of first parameter combinations; a first processing subunit, used to combine the first parameter combination and the tuning evaluation results corresponding to the first parameter combination for the first parameter combination in the multiple groups of first parameter combinations to obtain first tuning data information; and a second processing subunit, used to input the first tuning data information into the target model, and obtain parameter tuning information of the target application through the target model.

[0166] Optionally, in the data processing device provided in Example 3 of the present application, the acquisition subunit includes: an acquisition module, used to acquire a first reference sensitive parameter set for a reference application scenario, wherein the sensitive parameters in the first reference sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the reference application scenario is greater than a first preset degree; a determination module, used to determine multiple groups of first parameter combinations corresponding to the target application according to the first reference sensitive parameter set; and a processing module, used to perform parameter tuning on the target application based on the first parameter combination in the multiple groups of first parameter combinations, and to perform tuning evaluation on the target application after parameter tuning to obtain tuning evaluation results corresponding to the multiple groups of first parameter combinations.

[0167] Optionally, in the data processing device provided in Example 3 of the present application, the data processing device also includes: a second acquisition unit, used to obtain historical tuning data information in an application scenario corresponding to the second model to be generated, wherein the historical tuning data information includes parameters, parameter values ​​and historical tuning evaluation results used when performing historical parameter tuning in the application scenario; a first processing unit, used to use the historical tuning data information in the application scenario as a training sample, and the application type and scenario type associated with the application scenario as true labels to construct a first training sample set; and a second processing unit, used to train an initial neural network model based on the first training sample set to obtain a second model.

[0168] Optionally, in the data processing device provided in Example 3 of the present application, the first model learns the second reference sensitive parameter set under multiple application scenarios based on the historical tuning data information. When the target model is the first model, the determination unit 603 includes: a third determination subunit, which is used to determine the second tuning data information according to the multiple groups of second parameter combinations corresponding to the target application; a third processing subunit, which is used to input the second tuning data information into the target model, and determine the second sensitive parameter set of the target application in the current application scenario through the target model according to the second tuning data information and the second reference sensitive parameter set; a fourth determination subunit, which is used to determine multiple groups of target parameter combinations based on the second sensitive parameter set, and determine the tuning evaluation results corresponding to the multiple groups of target parameter combinations of the target application; a fourth processing subunit, which is used to combine the target parameter combinations in the multiple groups of target parameter combinations with the tuning evaluation results corresponding to the target parameter combinations to obtain the target tuning data information, and construct a second training sample set according to the target tuning data information corresponding to the multiple groups of target parameter combinations; a fifth processing subunit, which is used to train the target model based on the second training sample set to obtain the trained target model, and determine the parameter tuning information through the trained target model.

[0169] Optionally, in the data processing device provided in Example 3 of the present application, the data processing device also includes: a third acquisition unit, used to acquire historical tuning data information under multiple application scenarios, wherein the historical tuning data information includes parameters, parameter values ​​and historical tuning evaluation results used when the application performs historical parameter tuning in the application scenario; a third processing unit, used to use the historical tuning data information under multiple application scenarios as training samples, and the application type and scenario type associated with the application scenario as true labels to construct a third training sample set; and a fourth processing unit, used to train an initial large language model based on the third training sample set to obtain a first model.

[0170] Optionally, in the data processing device provided in Example 3 of the present application, the determination unit 603 includes: a first detection subunit, used to detect whether the first model has been deployed in the target server; a second detection subunit, used to detect whether there is a target identifier in the target configuration file associated with the target application when the first model has been deployed in the target server, wherein the target identifier represents that the first model is determined as the target model; and a selection subunit, used to select a target model from the first model and multiple second models based on the judgment result when the target identifier does not exist in the target configuration file.

[0171] Optionally, in the data processing device provided in Example 3 of the present application, the determination unit 603 includes: a fifth determination subunit, used to determine a second model associated with a reference application scenario among multiple second models as a target model when the judgment result represents the existence of a reference application scenario; and a sixth determination subunit, used to determine the first model as the target model when the judgment result represents the absence of a reference application scenario.

[0172] It should be noted that the first acquisition unit 601, the judgment unit 602, and the determination unit 603 described above correspond to steps S201 to S203 in Example 1, and the examples and application scenarios implemented by the above units and corresponding steps are the same, but are not limited to the contents disclosed in the above Example 1. It should be noted that the above modules, as part of the device, can be run in the computer terminal 10 provided in Example 1.

[0173] It should be noted that the preferred implementation scheme involved in the above embodiments of the present application is the same as the scheme provided in Example 1, as well as the application scenario and implementation process, but is not limited to the scheme provided in Example 1.

[0174] Example 4

[0175] The embodiment of the present application may provide an electronic device, which may be any electronic device in a group of electronic devices. Optionally, in this embodiment, the electronic device may also be replaced by a terminal device such as a mobile terminal.

[0176] Optionally, in this embodiment, the electronic device may be located in at least one network device among a plurality of network devices of a computer network.

[0177] In this embodiment, the electronic device can execute the program code corresponding to the steps in the data processing method provided in any one of the method embodiments.

[0178] Optionally, Figure 7 is a structural block diagram of an electronic device provided according to Embodiment 4 of the present application. Figure 7 As shown, the electronic device 70 may include: one or more ( Figure 7 Only one is shown in the figure) processor 702, memory 704. The electronic device 70 may also include a storage controller, through which the memory 704 is controlled and managed; the electronic device 70 may also include a peripheral interface, through which the radio frequency module, audio module and display screen are connected.

[0179] Among them, the memory can be used to store software programs and modules, such as program instructions / modules corresponding to the data processing method and device in the embodiment of the present application. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, that is, realizing the above-mentioned data processing method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the terminal 10 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0180] The processor can call the information and application programs stored in the memory through the transmission device to execute the program codes corresponding to the steps in the data processing method provided in any one of the above method embodiments.

[0181] It can be understood by those skilled in the art that Figure 7 The structure shown is for illustration only, and the electronic device may also be a smart phone (such as an Android phone, an iOS phone), a tablet computer, a PDA, a mobile Internet device (MID), a PAD, or other terminal devices. Figure 7 The structure of the electronic device is not limited. Figure 7 More or fewer components (such as network interfaces, display devices, etc.) shown in, or with Figure 7 Different configurations are shown.

[0182] A person of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, and the storage medium may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0183] Example 5

[0184] The embodiment of the present application further provides a computer-readable storage medium. Optionally, in this embodiment, the storage medium can be used to store the program code executed by the data processing method provided in the first embodiment.

[0185] Optionally, in this embodiment, the above storage medium may be located in any electronic device in a group of electronic devices in a computer network, or in any mobile terminal in a group of mobile terminals.

[0186] Example 6

[0187] The embodiment of the present application further provides a computer program product. Optionally, in this embodiment, the computer program product may include a computer program, and when the computer program is executed by a processor, the data processing method provided in the first embodiment is implemented.

[0188] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0189] In the above embodiments of the present application, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, please refer to the relevant description of other embodiments.

[0190] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0191] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0192] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0193] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk and other media that can store program codes.

[0194] The above are only preferred implementations of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A data processing method, characterized in that: include: Obtaining resource usage information of a target application during operation in a current application scenario to obtain target resource usage information, wherein the target application is an application to be parameter-tuned; According to the target resource usage information and the reference resource usage information under the target application scenario in the multiple application scenarios, it is determined whether there is a reference application scenario in the multiple application scenarios, and a determination result is obtained, wherein the scene similarity between the reference application scenario and the current application scenario is greater than a target threshold; A target model is selected from the first model and the plurality of second models according to the judgment result, and parameter tuning information of the target application is determined by the target model, wherein the first model is a large language model, the second model is a neural network model other than the large language model, the first model is trained based on the historical tuning data information under the plurality of application scenarios, and different second models are trained based on the historical tuning data information under different application scenarios; According to the target resource usage information and the reference resource usage information under the target application scenario in the multiple application scenarios, it is determined whether there is a reference application scenario in the multiple application scenarios, and the determination result includes: According to the target resource usage information and the reference resource usage information in the target application scenario in the multiple application scenarios, it is determined whether there is any reference resource usage information in the multiple reference resource usage information whose similarity with the target resource usage information is greater than a first threshold, and a first determination sub-result is obtained; Determine a first sensitive parameter set of the target application in the current application scenario, wherein the sensitive parameters in the first sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the current application scenario is greater than a first preset degree; According to the first sensitive parameter set and a first reference sensitive parameter set in a target application scenario among the multiple application scenarios, determining whether there is a first reference sensitive parameter set including the first sensitive parameter set among the multiple first reference sensitive parameter sets, to obtain a second determination sub-result; The judgment result is determined according to the first judgment sub-result and the second judgment sub-result.

2. The method according to claim 1, characterized in that When the target model is the second model, determining the parameter tuning information of the target application through the target model includes: Obtaining tuning evaluation results corresponding to multiple groups of first parameter combinations of the target application; For a first parameter combination in the plurality of first parameter combinations, combining the first parameter combination and a tuning evaluation result corresponding to the first parameter combination to obtain first tuning data information; The first tuning data information is input into the target model, and parameter tuning information of the target application is obtained through the target model.

3. The method according to claim 2, characterized in that Obtaining the tuning evaluation results corresponding to the target application in the plurality of first parameter combinations includes: Acquire a first reference sensitive parameter set of the reference application scenario, wherein the sensitive parameters in the first reference sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the reference application scenario is greater than a first preset degree; Determine a plurality of first parameter combinations corresponding to the target application according to the first reference sensitive parameter set; Parameters of the target application are tuned based on a first parameter combination in the multiple groups of first parameter combinations, and tuning evaluation is performed on the target application after parameter tuning to obtain tuning evaluation results corresponding to the multiple groups of first parameter combinations.

4. The method according to claim 1, characterized in that: The second model is obtained by: Obtaining historical tuning data information in an application scenario corresponding to the second model to be generated, wherein the historical tuning data information includes parameters, parameter values, and historical tuning evaluation results used when historical parameter tuning is performed in the application scenario; The historical tuning data information in the application scenario is used as a training sample, and the application type and scenario type associated with the application scenario are used as true labels to construct a first training sample set; An initial neural network model is trained based on the first training sample set to obtain the second model.

5. The method according to claim 1, characterized in that The first model learns to obtain the second reference sensitive parameter set under the multiple application scenarios according to the historical tuning data information. When the target model is the first model, determining the parameter tuning information of the target application through the target model includes: Determining second tuning data information according to a plurality of second parameter combinations corresponding to the target application; Inputting the second tuning data information into the target model, and determining the second sensitive parameter set of the target application in the current application scenario according to the second tuning data information and the second reference sensitive parameter set through the target model; Determine multiple groups of target parameter combinations according to the second sensitive parameter set, and determine the tuning evaluation results corresponding to the multiple groups of target parameter combinations of the target application; Combining the target parameter combinations in the multiple groups of target parameter combinations with the tuning evaluation results corresponding to the target parameter combinations to obtain target tuning data information, and constructing a second training sample set according to the target tuning data information corresponding to the multiple groups of target parameter combinations; The target model is trained based on the second training sample set to obtain a trained target model, and the parameter tuning information is determined through the trained target model.

6. The method according to claim 1, characterized in that The first model is obtained by: Acquire historical tuning data information under the multiple application scenarios, wherein the historical tuning data information includes parameters, parameter values, and historical tuning evaluation results used when the application is used to perform historical parameter tuning under the application scenarios; Using the historical tuning data information under the multiple application scenarios as training samples, and the application types and scenario types associated with the application scenarios as true labels, to construct a third training sample set; An initial large language model is trained based on the third training sample set to obtain the first model.

7. The method according to claim 1, characterized in that Selecting a target model from the first model and the plurality of second models according to the judgment result includes: Detecting whether the first model has been deployed in the target server; In the case where the first model has been deployed in the target server, detecting whether a target identifier exists in a target configuration file associated with the target application, wherein the target identifier indicates that the first model is determined as the target model; When the target identifier does not exist in the target configuration file, the target model is selected from the first model and the plurality of second models according to the judgment result.

8. The method according to claim 1 or 7, characterized in that: Selecting a target model from the first model and the plurality of second models according to the judgment result includes: In a case where the judgment result indicates that the reference application scenario exists, determining a second model associated with the reference application scenario among the plurality of second models as the target model; When the judgment result indicates that the reference application scenario does not exist, the first model is determined as the target model.

9. A data processing method, characterized in that: include: Obtain resource usage information of a target application uploaded by a client during operation in a current application scenario, and obtain target resource usage information, wherein the target application is an application to be parameter-tuned; In the cloud server, based on the target resource usage information and the reference resource usage information in the target application scenario in the multiple application scenarios, it is determined whether there is a reference application scenario in the multiple application scenarios, and a judgment result is obtained; a target model is selected from the first model and the multiple second models according to the judgment result, and parameter tuning information of the target application is determined by the target model, wherein the scenario similarity between the reference application scenario and the current application scenario is greater than a target threshold, the first model is a large language model, the second model is a neural network model other than the large language model, the first model is trained based on the historical tuning data information in the multiple application scenarios, and different second models are trained based on the historical tuning data information in different application scenarios; Feedback parameter tuning information of the target application to the client; Wherein, judging whether there is a reference application scenario in the multiple application scenarios according to the target resource usage information and the reference resource usage information in the target application scenario in the multiple application scenarios, and obtaining the judgment result includes: According to the target resource usage information and the reference resource usage information in the target application scenario in the multiple application scenarios, it is determined whether there is any reference resource usage information in the multiple reference resource usage information whose similarity with the target resource usage information is greater than a first threshold, and a first determination sub-result is obtained; Determine a first sensitive parameter set of the target application in the current application scenario, wherein the sensitive parameters in the first sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the current application scenario is greater than a first preset degree; According to the first sensitive parameter set and a first reference sensitive parameter set in a target application scenario among the multiple application scenarios, determining whether there is a first reference sensitive parameter set including the first sensitive parameter set among the multiple first reference sensitive parameter sets, to obtain a second determination sub-result; The judgment result is determined according to the first judgment sub-result and the second judgment sub-result.

10. A data processing device, characterized in that: include: A first acquisition unit is used to acquire resource usage information of a target application during operation in a current application scenario to obtain target resource usage information, wherein the target application is an application to be parameter-tuned; a judgment unit, configured to judge whether there is a reference application scenario in the multiple application scenarios according to the target resource usage information and the reference resource usage information under the target application scenario in the multiple application scenarios, and obtain a judgment result, wherein the scene similarity between the reference application scenario and the current application scenario is greater than a target threshold; a determination unit, configured to select a target model from the first model and the plurality of second models according to the judgment result, and determine the parameter tuning information of the target application through the target model, wherein the first model is a large language model, the second model is a neural network model other than the large language model, the first model is trained based on the historical tuning data information under the plurality of application scenarios, and different second models are trained based on the historical tuning data information under different application scenarios; The judging unit includes: A first judgment subunit is used to judge whether there is a similarity between the reference resource usage information and the target resource usage information in the target application scenario in the multiple application scenarios, based on the target resource usage information and the reference resource usage information in the target application scenario, and obtain a first judgment subresult; A first determining subunit is used to determine a first sensitive parameter set of the target application in the current application scenario, wherein the sensitive parameters in the first sensitive parameter set refer to parameters whose influence on the tuning effect of the target application in the current application scenario is greater than a first preset degree; a second judgment subunit, configured to judge, based on the first sensitive parameter set and a first reference sensitive parameter set in a target application scenario among the multiple application scenarios, whether there is a first reference sensitive parameter set including the first sensitive parameter set among the multiple first reference sensitive parameter sets, and obtain a second judgment subresult; The second determining subunit is configured to determine the judgment result according to the first judgment subresult and the second judgment subresult.

11. An electronic device, characterized in that: include: A memory storing an executable program; A processor, used to run the program, wherein the program executes the data processing method of any one of claims 1 to 9 when running.

12. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored executable program, wherein when the executable program is run, the device where the storage medium is located is controlled to execute the data processing method of any one of claims 1 to 9.

13. A computer program product, characterized in that The method comprises a computer program or instructions, which, when executed by a processor, implements the data processing method according to any one of claims 1 to 9.

Images