Task scheduling method of edge data service and language model training method
By using two language models for task scheduling and adjustment in edge service management devices, the problem of inefficient scheduling of edge data service is solved, and efficient resource utilization and data service stability are achieved.
Patent Information
- Application Number
- CN202411143468.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-20
- Publication Date
- 2024-11-29
AI Technical Summary
In the prior art, the scheduling efficiency of edge data services is mainly due to the high complexity of algorithm model deployment and computing, which leads to low resource utilization efficiency and large computing overhead, making it difficult to deal with resource constraints and environmental dynamics.
By applying two language models in edge service management devices, the first language model is used to local scheduling decisions to optimize local computing overhead; the second language model combines local and distributed environment parameters to make task adjustment decisions, enhancing system robustness and stability.
It realizes efficient scheduling of edge data services, improves resource utilization efficiency, ensures the continuity and stability of data services, and adapts to resource changes and environmental fluctuations.
Smart Images

Figure CN119046001A_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of edge data services, and specifically relates to a task scheduling method for edge data services and a language model training method, device, system, storage medium, equipment and computer program product. Background Art
[0002] In today's digital age, data services have become a key technology for enterprises and individuals to manage and utilize data resources. These services include functions such as data collection, storage, processing, analysis and transmission, and can be deployed in the cloud or local environment. Especially in edge computing scenarios, the flexibility and efficiency of data services are crucial for processing large amounts of data. Edge computing can process data close to where the data is generated, thereby reducing latency and improving response speed, and is particularly suitable for the Internet of Things, smart cities, autonomous driving and other fields.
[0003] At present, data service scheduling and offloading mainly rely on exact solution and approximate solution methods. Exact solution methods, such as mathematical programming, can provide optimal solutions, but the computational complexity is high and it is not suitable for large-scale problems. Approximate solution methods, including search methods, heuristic optimization, and intelligent optimization, can provide approximate optimal solutions within a reasonable time, but these methods often cannot effectively use historical data for analysis and judgment, and have poor adaptability to dynamically changing conditions.
[0004] However, existing methods have difficulty coping with resource constraints and environmental dynamics, resulting in inefficient deployment and application of data services. Summary of the invention
[0005] The present application aims to provide a task scheduling method for edge data services and a training method, apparatus, system, storage medium, device and computer program product for a language model, which at least solves the problem of low resource utilization efficiency and high computational overhead caused by high algorithm model deployment and computational complexity in the prior art, resulting in low data service scheduling efficiency.
[0006] In a first aspect, an embodiment of the present application discloses a task scheduling method for edge data services, which is applied to an edge service management device, including:
[0007] Obtain local operating environment parameters of the local edge service platform, and determine the local data service and distributed data service of the edge data task; the local data service is used to execute in the local edge service platform, and the distributed data service is used to execute in the distributed edge service platform;
[0008] Inputting the local data service and the local operating environment parameters into a first language model to obtain a local scheduling decision for the edge data task, and sending the distributed data service to the distributed edge service platform;
[0009] According to the local scheduling decision, control the local edge service platform to execute the local data service to update the local operating environment parameters, and obtain the distributed operating environment parameters of the distributed data service when executed on the distributed edge service platform;
[0010] Inputting the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into a second language model to obtain a task adjustment decision for the edge data task;
[0011] According to the task adjustment decision, the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform are adjusted.
[0012] In a second aspect, an embodiment of the present application further discloses a method for training a language model, the training method being used to train a first language model in the task scheduling method for the edge data service as described in the first aspect, the training method comprising:
[0013] Constructing a first training set; the first training set includes a local data service of a training task and local training operating environment parameters of a local training edge service platform running the local data service;
[0014] The local data service and the local training operating environment parameters are input into the first language model, and the first language model is trained according to a preset first loss function to obtain a trained first language model.
[0015] In a third aspect, an embodiment of the present application further discloses a method for training a language model, the training method being used to train a second language model in the task scheduling method for the edge data service as described in the first aspect, the training method comprising:
[0016] Constructing a second training set; the second training set includes a local training scheduling decision of the training task, a local training operating environment parameter of the local training edge service platform executed according to the local training scheduling decision, and a distributed training operating environment parameter of the distributed training edge service platform;
[0017] The local training scheduling decision, the local training operating environment parameters, and the distributed training operating environment parameters are input into the second language model, and the second language model is trained according to a preset second loss function to obtain a trained second language model.
[0018] In a fourth aspect, the embodiment of the present application further discloses a scheduling system for edge data services, including:
[0019] Edge service management equipment, local edge service platform, distributed edge service platform;
[0020] The local edge service platform is used to perform local data services;
[0021] The distributed edge service platform is used to perform distributed data services;
[0022] The edge service management device is used to obtain the local operating environment parameters of the local edge service platform, determine the local data service and the distributed data service of the edge data task, and input the local data service and the local operating environment parameters into the first language model to obtain the local scheduling decision of the edge data task, and send the distributed data service to the distributed edge service platform, and control the local edge service platform to execute the local data service according to the local scheduling decision to update the local operating environment parameters, and obtain the distributed operating environment parameters of the distributed data service when it is executed on the distributed edge service platform, and input the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into the second language model to obtain the task adjustment decision of the edge data task, and adjust the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform according to the task adjustment decision.
[0023] In a fifth aspect, an embodiment of the present application further discloses a scheduling device for edge data services, which is applied to an edge service management device, and the device includes:
[0024] A data set acquisition module, used to obtain local operating environment parameters of the local edge service platform, and determine the local data service and distributed data service of the edge data task; the local data service is used to execute in the local edge service platform, and the distributed data service is used to execute in the distributed edge service platform;
[0025] A local scheduling module, configured to input the local data service and the local operating environment parameters into a first language model to obtain a local scheduling decision for the edge data task, and send the distributed data service to the distributed edge service platform;
[0026] A local execution module, configured to control the local edge service platform to execute the local data service according to the local scheduling decision, so as to update the local operating environment parameters, and obtain the distributed operating environment parameters of the distributed data service when it is executed on the distributed edge service platform;
[0027] A global scheduling module, configured to input the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into a second language model to obtain a task adjustment decision for the edge data task;
[0028] The scheduling adjustment module is used to adjust the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform according to the task adjustment decision.
[0029] In a sixth aspect, an embodiment of the present application further discloses a language model training device, the training device is used to train the first language model in the task scheduling method for edge data services as described in the first aspect, the training device comprising:
[0030] A first training set module, used to construct a first training set; the first training set includes a local data service of a training task and local training operating environment parameters of a local training edge service platform running the local data service;
[0031] The first training module is used to input the local data service and the local training operating environment parameters into the first language model, and train the first language model according to a preset first loss function to obtain a trained first language model.
[0032] In a seventh aspect, an embodiment of the present application further discloses a language model training device, the training device is used to train the second language model in the task scheduling method of the edge data service as described in the first aspect, the training device comprising:
[0033] A second training set module, used to construct a second training set; the second training set includes a local training scheduling decision of the training task, a local training operating environment parameter of the local training edge service platform executed according to the local training scheduling decision, and a distributed training operating environment parameter of the distributed training edge service platform;
[0034] The second training set module is used to input the local training scheduling decision, the local training operating environment parameters, and the distributed training operating environment parameters into the second language model, and train the second language model according to a preset second loss function to obtain a trained second language model.
[0035] In an eighth aspect, an embodiment of the present application further discloses an electronic device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps described in the first aspect, the second aspect, or the third aspect.
[0036] In the ninth aspect, an embodiment of the present application further discloses a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps described in the first aspect, the second aspect or the third aspect are implemented.
[0037] In the tenth aspect, an embodiment of the present application further discloses a computer program product, on which a computer program is stored, and when the computer program is executed by a processor, the steps described in the first aspect, the second aspect or the third aspect are implemented.
[0038] In summary, in the embodiment of the present application, by inputting the local data service and its environmental parameters into the first language model to obtain the local scheduling decision, and then executing the selected data service on the local edge service platform, the local computing overhead is optimized and unnecessary resource consumption is avoided; then the second language model is used to process the operating environment parameters updated according to the execution of the local scheduling decision, as well as the operating environment parameters of the distributed platform, to obtain the task adjustment decision for the local and distributed data services, thereby enhancing the robustness of the system, so that the system can maintain the continuity and stability of the data service in the face of resource changes and environmental fluctuations, thereby collaboratively completing the specified data service. Therefore, based on the method of the embodiment of the present application, efficient scheduling of edge data services is achieved, solving the problems of low resource utilization efficiency and high computing overhead in the related technology due to the high complexity of algorithm model deployment and calculation in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS
[0039] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:
[0040] Figure 1 This is a task scheduling method for edge data services provided by this embodiment;
[0041] Figure 2 This is another task scheduling method for edge data services provided by this embodiment;
[0042] Figure 3 This is a language model training method provided by this embodiment;
[0043] Figure 4 This is another language model training method provided by this embodiment;
[0044] Figure 5 is a structural diagram of a task scheduling device for edge data services provided by this embodiment;
[0045] Figure 6 is a structural diagram of a language model training device provided in this embodiment;
[0046] Figure 7 is a schematic diagram of the structure of another language model training device provided in this embodiment;
[0047] Figure 8 is a structural diagram of a task scheduling system for edge data services provided in this embodiment;
[0048] Fig. 9 It is the data flow process under the task scheduling method of the edge data service of this application;
[0049] Fig.10 is a block diagram of an electronic device provided in an embodiment of the present application;
[0050] Fig.11 It is a block diagram of another electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0051] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.
[0052] Figure 1 This embodiment provides a task scheduling method for edge data services, which is applied to an edge service management device. The method may include the following steps:
[0053] Step 101, obtain the local operating environment parameters of the local edge service platform, and determine the local data service and distributed data service of the edge data task.
[0054] Among them, the local data service is used to be executed in the local edge service platform, and the distributed data service is used to be executed in the distributed edge service platform.
[0055] In some embodiments of the present application, it involves obtaining the operating environment parameters of the local edge service platform and determining the local data service and distributed data service of the edge data task. Local data service refers to the service executed on the local edge service platform, while distributed data service refers to the service executed on the distributed edge service platform. The purpose of this step is to collect enough information so that the subsequent language model can make effective scheduling decisions based on these parameters. Operating environment parameters may include but are not limited to CPU usage, memory usage, network bandwidth and latency, etc., which directly affect the execution efficiency and quality of data services. In some practical application embodiments of the present application, changes in the environment of the edge service platform (such as weather changes) and platform target information (such as conversion targets) can also be used as input values of operating environment parameters, and this application does not make special restrictions.
[0056] For example, suppose a local edge service platform is responsible for processing data from an intelligent transportation system. The operating environment parameters of the platform include the current CPU usage of 70%, memory usage of 4GB, and network bandwidth of 100Mbps. Local data services may include vehicle identification and traffic flow analysis, while distributed data services may include storage and backup of large-scale video data. In this case, by collecting these parameters and determining which services are executed locally and which services require distributed processing, it is possible to ensure that the local platform does not affect service quality due to resource overload, while also being able to use the resources of the distributed platform for efficient computing.
[0057] Step 102, input the local data service and the local operating environment parameters into the first language model to obtain the local scheduling decision of the edge data task, and send the distributed data service to the distributed edge service platform.
[0058] In some embodiments of the present application, in order to use the first language model to analyze the local data service and operating environment parameters to formulate a local scheduling decision for the edge data task. The first language model will derive the optimal local scheduling decision through algorithm processing based on the input data service and environment parameters, and this decision will guide the execution of the local data service. At the same time, the distributed data service will be sent to the distributed edge service platform for execution there.
[0059] For example, suppose the local edge service platform is running a video analysis service that requires a lot of computing resources. The operating environment parameters show that the CPU usage has reached 80%, the memory usage is 6GB, and the network bandwidth is 50Mbps. In this case, the first language model may select an algorithm that meets the local environment conditions to perform the video encoding task. At the same time, some other tasks in the edge data service are sent to the distributed edge service platform to ensure the complete operation of the task.
[0060] Step 103, according to the local scheduling decision, control the local edge service platform to execute the local data service to update the local operating environment parameters, and obtain the distributed operating environment parameters when the distributed data service is executed on the distributed edge service platform.
[0061] In some embodiments of the present application, the process of local data service will be executed on the local edge service platform, and the local operating environment parameters will be updated according to the execution results. At the same time, the operating environment parameters of the distributed data service when it is executed on the distributed edge service platform are obtained. The purpose of this step is to ensure that the local data service can be optimized and adjusted according to the latest environmental status, and at the same time collect the operating data of the distributed platform to provide necessary information for subsequent task adjustment decisions.
[0062] For example, consider a local edge service platform in a smart manufacturing system that processes sensor data and controls the production line. Local data services may include real-time monitoring and predictive maintenance. After executing these services, the platform may find that the CPU usage has increased to 60%, the memory usage has increased to 5GB, and the network bandwidth has stabilized at 100Mbps. These updated parameters will be recorded. At the same time, the distributed edge service platform may be performing large-scale data analysis, and its operating environment parameters such as processing speed, response time, etc. will also be collected for further data service scheduling and offloading decisions. In this way, it can be ensured that the data services of both local and distributed platforms can operate in the best state.
[0063] Step 104 , input the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into the second language model to obtain a task adjustment decision for the edge data task.
[0064] In some embodiments of the present application, local scheduling decisions, updated local operating environment parameters, and distributed operating environment parameters are input into the second language model in order to obtain task adjustment decisions for edge data tasks. The local scheduling decision refers to the decision previously generated by the first language model, and the updated local operating environment parameters are the latest parameters obtained after executing the local data service. Distributed operating environment parameters refer to the environmental parameters when executing distributed data services on a distributed edge service platform. The second language model will integrate this information and conduct in-depth analysis to generate optimized task adjustment decisions, which will help further improve the efficiency and robustness of data services.
[0065] For example, suppose that in a smart city management system, the local edge service platform is responsible for monitoring urban traffic flow and has executed data analysis services based on the decisions of the first language model. The updated local operating environment parameters show that due to the arrival of rush hour, the CPU usage has risen to 85% and the memory usage has increased to 7GB. At the same time, the distributed edge service platform is processing a large amount of video surveillance data, and its distributed operating environment parameters include the number of video streams processed per second and the response time. In this case, the second language model may decide to increase the data processing capacity of the local platform and adjust the video data processing strategy of the distributed platform to meet the challenges brought by rush hour. Such task adjustment decisions can ensure that the system can still operate efficiently under resource constraints.
[0066] Step 105 , adjusting the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform according to the task adjustment decision.
[0067] In some embodiments of the present application, the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform will be adjusted according to the task adjustment decision. This step is to optimize the execution of the data service based on the task adjustment decision provided by the second language model to improve the overall service efficiency and system robustness. The adjustment may include the unloading of local tasks and the reloading of unloaded tasks on distributed servers to adapt to changes in the operating environment and updates to task objectives.
[0068] For example, suppose a local edge service platform is responsible for data processing tasks for a smart home system, while a distributed edge service platform handles a large amount of data backup tasks from smart home devices. During peak hours, the CPU usage of the local platform soars and memory usage approaches the limit. Based on the task adjustment decision, the local data service may offload the data analysis task and send it to the distributed platform for processing to ensure the security of critical data and the continuity of services. Such adjustments help maintain the stable operation of the system and the quality of service under resource constraints.
[0069] In summary, in the embodiment of the present application, by inputting the local data service and its environmental parameters into the first language model to obtain the local scheduling decision, and then executing the selected data service on the local edge service platform, the local computing overhead is optimized and unnecessary resource consumption is avoided; then the second language model is used to process the operating environment parameters updated according to the execution of the local scheduling decision, as well as the operating environment parameters of the distributed platform, to obtain the task adjustment decision for the local and distributed data services, thereby enhancing the robustness of the system, so that the system can maintain the continuity and stability of the data service in the face of resource changes and environmental fluctuations, thereby collaboratively completing the specified data service. Therefore, based on the method of the embodiment of the present application, efficient scheduling of edge data services is achieved, solving the problems of low resource utilization efficiency and high computing overhead in the related technology due to the high complexity of algorithm model deployment and calculation in the prior art.
[0070] Figure 2 Another edge data service task scheduling method provided by this embodiment is applied to an edge service management device. The method may include the following steps:
[0071] Step 201, obtain local operating environment parameters of the local edge service platform, and determine the local data service and distributed data service of the edge data task.
[0072] Among them, the local data service is used to be executed in the local edge service platform, and the distributed data service is used to be executed in the distributed edge service platform.
[0073] The method shown in this step has been explained in step 101 and will not be repeated here.
[0074] Optionally, the edge data task includes multiple subtasks. In order to determine the local data service and the distributed data service of the edge data task, step 201 includes the following substeps:
[0075] Sub-step 2011, determining a local data service according to a local operating environment parameter and a target sub-task determined from a plurality of sub-tasks.
[0076] In some embodiments of the present application, the target subtask is determined from multiple subtasks according to the local operating environment parameters, and the local data service is determined accordingly. This process involves analyzing the operating environment parameters of the local edge service platform, such as CPU usage, memory usage, network bandwidth, etc., as well as the resource requirements and priority of each subtask, so as to select the subtask that is most suitable for local execution. This can ensure the efficient operation of local data services while avoiding waste of resources.
[0077] For example, suppose an edge data task contains three subtasks: video analysis, data encryption, and real-time monitoring. The operating environment parameters of the local edge service platform show that a large amount of computing resources and network bandwidth are currently available. In this case, real-time monitoring can be determined as the target subtask because it requires low latency and high bandwidth, which can be met by the local platform. Therefore, real-time monitoring is determined as a local data service, while other subtasks such as video analysis and data encryption may be arranged as distributed data services and executed on a distributed edge service platform with more abundant resources. Such decisions help optimize resource allocation and improve the efficiency and effectiveness of task execution.
[0078] Sub-step 2012, determining a distributed data service based on the remaining subtasks in the edge data task excluding the target subtask.
[0079] In some embodiments of the present application, after removing the target subtask from the edge data task, the remaining subtasks are determined as distributed data services. This process requires analyzing the other subtasks except the target subtask, and determining which subtasks are suitable for execution on the distributed edge service platform based on their characteristics and requirements. Such decisions help ensure that distributed data services can be efficiently executed on a platform with more sufficient resources.
[0080] For example, assume that an edge data task includes three subtasks: data collection, big data analysis, and long-term data storage. If real-time data collection is determined as a local data service, then the remaining big data analysis and long-term data storage subtasks can be analyzed. Since these two subtasks may require more storage space and computing power, they are determined as distributed data services and are scheduled to be executed on the distributed edge service platform. This allocation allows the local platform to process real-time data efficiently, while the distributed platform can use its more powerful resources to handle more complex analysis and storage tasks, thereby optimizing the performance and resource utilization of the entire system.
[0081] Step 202 , input the local data service and the local operating environment parameters into the first language model to obtain the local scheduling decision of the edge data task, and send the distributed data service to the distributed edge service platform.
[0082] The method shown in this step has been explained in step 102 and will not be repeated here.
[0083] Step 203, according to the local scheduling decision, control the local edge service platform to execute the local data service to update the local operating environment parameters, and obtain the distributed operating environment parameters when the distributed data service is executed on the distributed edge service platform.
[0084] The method shown in this step has been explained in step 103 and will not be repeated here.
[0085] Step 204 , input the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into the second language model to obtain the task adjustment decision of the edge data task.
[0086] The method shown in this step has been described in step 104 and will not be repeated here.
[0087] Step 205 , adjusting the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform according to the task adjustment decision.
[0088] The method shown in this step has been explained in step 105 and will not be repeated here.
[0089] Optionally, the local data service includes at least one execution task, and the task adjustment decision includes an uninstallation decision for a target data service in the local data service. Step 205 includes the following sub-steps:
[0090] Sub-step 2051 , uninstalling the target data service executed on the local edge service platform according to the uninstallation decision, and sending the uninstalled target data service to the distributed edge service platform, so that the distributed edge service platform executes the target data service.
[0091] In some embodiments of the present application, it involves unloading the target data service on the local edge service platform and sending it to the distributed edge service platform for execution according to the task adjustment decision. This process is part of the task adjustment strategy, which aims to optimize the location of task execution to improve the efficiency and responsiveness of the overall system. The unloading decision is based on the analysis of the current operating environment parameters and task requirements, and decides to transfer certain tasks from the local platform to a distributed platform with more sufficient resources or more suitable for performing specific tasks.
[0092] For example, suppose that in an intelligent transportation system, the local edge service platform is responsible for real-time processing of data from traffic cameras. When the local platform's resources are tight, the deep analysis tasks of video data can be offloaded to the distributed edge service platform. In this case, the local platform will continue to perform tasks with high real-time requirements, such as vehicle detection and traffic flow statistics, while the deep analysis tasks that require a lot of computing resources are transferred to the distributed platform to use its stronger processing power to execute, thus ensuring the smooth operation of real-time tasks and the efficient processing of deep analysis tasks. This offloading and task allocation strategy helps to balance the load of different platforms and improve the overall performance of the system.
[0093] Optionally, there are multiple distributed edge service platforms, and the target data service has a corresponding target distributed edge service platform. In order to send the unloaded target data service to the distributed edge service platform so that the distributed edge service platform executes the target data service, sub-step 2051 includes the following sub-steps:
[0094] Sub-step 20511, sending the target data service to the target distributed edge service platform, so that the target distributed edge service platform executes the target data service.
[0095] In some embodiments of the present application, in order to ensure that the target data service can be effectively sent to the corresponding target distributed edge service platform and executed by the platform, the target data service is sent to the target distributed edge service platform. Among multiple distributed edge service platforms, each platform may focus on different types of task processing, so choosing the right platform is crucial for the successful execution of the task. This step involves the allocation and transmission of tasks, ensuring that each task can be processed on the platform that best suits its needs.
[0096] For example, suppose that in an intelligent medical system, there are multiple distributed edge service platforms, each capable of processing different types of medical data. If the target data service is to analyze electrocardiogram (ECG) data, and one of the distributed platforms is specially configured with high-performance algorithms and hardware for ECG analysis, then this task can be sent to the dedicated distributed edge service platform. In this way, the ECG data can be analyzed quickly and accurately, while also optimizing the use of resources and the efficiency of task execution.
[0097] Step 206 , in response to the modification command for the local scheduling decision, execute the modified local scheduling decision, obtain the local operating environment parameters after executing the modified local scheduling decision, and return to step 203 .
[0098] In some embodiments of the present application, in order to implement a dynamic adjustment process, in response to a modification command to the local scheduling decision, the modified local scheduling decision will be executed and the local operating environment parameters will be updated. This step ensures that the edge data service can adapt to environmental changes and changes in task requirements, and improves the flexibility of the service and the overall performance of the system through continuous adjustment and optimization. After executing the modified scheduling decision, the system will collect the latest local operating environment parameters, which will be used for the next round of service scheduling and optimization.
[0099] For example, consider a local edge service platform in an intelligent transportation system that is responsible for processing vehicle flow data. If a major road is closed due to an accident during the morning rush hour, the system may receive a modification command to increase the monitoring intensity of surrounding roads. In this case, a modified scheduling decision can be executed, for example, increasing the resource allocation of data analysis services for backup roads. After executing this decision, the system will update the local operating environment parameters, such as CPU usage and memory usage, and return to step 203 based on these latest parameters, that is, according to the local scheduling decision, control the local edge service platform to execute the local data service to update the local operating environment parameters, and obtain the distributed operating environment parameters of the distributed data service when it is executed on the distributed edge service platform, and re-evaluate and adjust the scheduling of the data service to ensure the continuity and effectiveness of the traffic monitoring service. Such a dynamic adjustment mechanism enables the system to respond quickly to emergencies and maintain efficient operation.
[0100] In some embodiments of the present application, the local operating environment parameters include:
[0101] One or more of the resource occupancy of the local edge service platform, the configuration information of the software and / or hardware environment of the local edge service platform, and the data content corresponding to the running tasks of the current local edge service platform.
[0102] Distributed operating environment parameters include:
[0103] One or more of the resource occupancy of the distributed edge service platform, the configuration information of the software and / or hardware environment of the distributed edge service platform, and the data content corresponding to the running tasks of the current distributed edge service platform.
[0104] In various embodiments of the present application, the specific values of the operating environment parameters of the local edge service platform or the distributed edge service platform include but are not limited to the central processing unit (CPU) usage rate, memory usage, disk space, network bandwidth usage, etc. of the corresponding edge service platform. These values can accurately provide the operating status of the corresponding edge service platform, helping system administrators or automated management systems to make more reasonable resource allocation and task scheduling decisions. Obtaining these values can ensure that the prediction results for task scheduling are more accurate, especially in edge computing environments with limited resources, where reasonable resource management is crucial to maintaining service quality. In practical applications, these data can be obtained in real time through various monitoring tools, and the corresponding resource management policies can be triggered by preset thresholds. The accurate acquisition and reasonable application of these values are the key to ensuring the performance of the edge service platform.
[0105] For example, if a local edge service platform of an intelligent transportation system detects that its memory usage is approaching its limit, the system can take measures in advance to offload this task to the distributed edge service platform to avoid service interruption due to resource exhaustion of the local edge service platform.
[0106] In summary, in the embodiment of the present application, by inputting the local data service and its environmental parameters into the first language model to obtain the local scheduling decision, and then executing the selected data service on the local edge service platform, the local computing overhead is optimized and unnecessary resource consumption is avoided; then the second language model is used to process the operating environment parameters updated according to the execution of the local scheduling decision, as well as the operating environment parameters of the distributed platform, to obtain the task adjustment decision for the local and distributed data services, thereby enhancing the robustness of the system, so that the system can maintain the continuity and stability of the data service in the face of resource changes and environmental fluctuations, thereby collaboratively completing the specified data service. Therefore, based on the method of the embodiment of the present application, efficient scheduling of edge data services is achieved, solving the problems of low resource utilization efficiency and high computing overhead in the related technology due to the high complexity of algorithm model deployment and calculation in the prior art.
[0107] Figure 3 A language model training method provided in this embodiment is used to train the first language model in the task scheduling method for edge data services mentioned in the above embodiment. The training method includes:
[0108] Step 301: construct a first training set.
[0109] Among them, the first training set includes the local data service of the training task and the local training operating environment parameters of the local training edge service platform that runs the local data service.
[0110] In some embodiments of the present application, in order to train the first language model, a first training set needs to be constructed. The first training set contains relevant information about local data services and operating environment parameters of the local edge service platform that runs these services. These parameters are critical for training models because they provide the environment that the model needs to learn and the context of service execution. The construction of the training set usually involves collecting historical data, simulation data, or experimental data to ensure that the model can make accurate predictions and decisions in practical applications.
[0111] For example, suppose a local edge service platform can be used to process sensor data in a smart manufacturing factory. Building the first training set may include tasks for collecting different sensors in the factory, as well as platform operating environment parameters (such as CPU usage, memory usage, etc.). This data will be input into the first language model as part of the training set to train the model how to optimize the scheduling of local data services based on sensor data and the current state of the platform. Such a training set helps the model learn how to effectively manage and schedule data service tasks under resource constraints.
[0112] Step 302: input the local data service and the local training operating environment parameters into the first language model, and train the first language model according to a preset first loss function to obtain a trained first language model.
[0113] In some embodiments of the present application, a process of training a first language model is performed, with local data services and local training operating environment parameters as input. These input data reflect the actual situation of the edge service platform when performing data services, including service type, resource usage, network status, etc. The preset first loss function is a standard for evaluating the prediction accuracy of the model during the training process, which guides the model to adjust its parameters by minimizing the loss, thereby improving the accuracy and efficiency of decision-making.
[0114] For example, suppose that local data services include real-time video analysis and sensor data processing, and local training operating environment parameters include CPU usage, memory usage, network latency, etc. During the training process, these data services and environment parameters will be input into the first language model. In this way, the first language model can learn how to make the most appropriate local data service scheduling decisions based on different service requirements and platform status.
[0115] In summary, in the embodiment of the present application, by inputting the local data service and its environmental parameters into the first language model to obtain the local scheduling decision, and then executing the selected data service on the local edge service platform, the local computing overhead is optimized and unnecessary resource consumption is avoided; then the second language model is used to process the operating environment parameters updated according to the execution of the local scheduling decision, as well as the operating environment parameters of the distributed platform, to obtain the task adjustment decision for the local and distributed data services, thereby enhancing the robustness of the system, so that the system can maintain the continuity and stability of the data service in the face of resource changes and environmental fluctuations, thereby collaboratively completing the specified data service. Therefore, based on the method of the embodiment of the present application, efficient scheduling of edge data services is achieved, solving the problems of low resource utilization efficiency and high computing overhead in the related technology due to the high complexity of algorithm model deployment and calculation in the prior art.
[0116] Figure 4This embodiment provides a language model training method, which is used to train the second language model in the edge data service task scheduling method mentioned in the above embodiment. The training method includes:
[0117] Step 401: construct a second training set.
[0118] Among them, the second training set includes the local training scheduling decision of the training task, the local training operating environment parameters of the local training edge service platform executed according to the local training scheduling decision, and the distributed training operating environment parameters of the distributed training edge service platform.
[0119] In some embodiments of the present application, in order to train the second language model, a second training set is constructed. The second training set includes the local training scheduling decision, the operating environment parameters when the local training edge service platform executes the decision, and the operating environment parameters of the distributed training edge service platform. These data provide the model with the necessary information to learn how to make effective task adjustment decisions in a changing environment.
[0120] For example, suppose in a smart grid system, the local edge service platform is processing energy consumption data, while the distributed edge service platform is performing energy allocation optimization. Local training scheduling decisions may include increasing the processing frequency of energy consumption data during peak demand hours. Local training operating environment parameters may include CPU utilization and memory usage of the local edge service platform when processing energy data, while distributed training operating environment parameters may include response time and accuracy of the energy allocation optimization algorithm. These data will be used to construct a second training set to train the second language model to make accurate task adjustment decisions under different operating environments. Through such training, the model can learn how to effectively adjust the execution of data services when resource usage and task requirements change.
[0121] Step 402: input the local training scheduling decision, the local training operating environment parameters, and the distributed training operating environment parameters into the second language model, and train the second language model according to a preset second loss function to obtain a trained second language model.
[0122] In some embodiments of the present application, local training scheduling decisions, local training operating environment parameters, and distributed training operating environment parameters are input into a second language model, and the model is trained using a preset second loss function. The purpose of this step is to optimize the parameters of the model through the training process so that it can make more accurate task adjustment decisions in practical applications. The second loss function is a mathematical standard for evaluating the difference between the model prediction and the actual value, which guides the model to improve the accuracy of the prediction by minimizing the loss.
[0123] For example, consider a local edge service platform for an intelligent logistics system that needs to adjust logistics routes based on the real-time location and status of goods. Local training scheduling decisions may include priority transportation routes for specific goods, local training operating environment parameters may include CPU usage and memory usage for cargo processing, and distributed training operating environment parameters may include the processing capacity and response time of each logistics node. Inputting these parameters into the second language model and training according to a second loss function such as Mean Squared Error (MSE) can enable the model to learn how to optimize logistics scheduling decisions based on real-time data and environmental changes, thereby improving the efficiency and responsiveness of the entire logistics system. The trained second language model can quickly make adjustments when new operating environment parameters appear to ensure the continuity and efficiency of logistics services.
[0124] It should be noted that the first language model and the second language model in various embodiments of the present application can be selected according to actual conditions and are not specifically limited in the present application.
[0125] In some preferred embodiments, the first language model and / or the second language model may be selected as a large language model (LLM). LLM is capable of extracting meaningful information from complex data and generating useful outputs, which is critical for edge data service platforms that require fast response and high accuracy. The accuracy and adaptability of these models make them ideal for performing edge data service tasks. The advantage of selecting LLM as the first language model and / or the second language model lies in the powerful data processing and pattern recognition capabilities of LLM. Large language models are able to understand and generate complex texts by training large amounts of data, which enables them to handle complex data service tasks such as natural language processing, predictive analysis, and decision support. In the relevant technical solutions of this application, the use of LLM will further enhance the responsiveness of the solution and the accuracy of the prediction.
[0126] In summary, in the embodiment of the present application, by inputting the local data service and its environmental parameters into the first language model to obtain the local scheduling decision, and then executing the selected data service on the local edge service platform, the local computing overhead is optimized and unnecessary resource consumption is avoided; then the second language model is used to process the operating environment parameters updated according to the execution of the local scheduling decision, as well as the operating environment parameters of the distributed platform, to obtain the task adjustment decision for the local and distributed data services, thereby enhancing the robustness of the system, so that the system can maintain the continuity and stability of the data service in the face of resource changes and environmental fluctuations, thereby collaboratively completing the specified data service. Therefore, based on the method of the embodiment of the present application, efficient scheduling of edge data services is achieved, solving the problems of low resource utilization efficiency and high computing overhead in the related technology due to the high complexity of algorithm model deployment and calculation in the prior art.
[0127] like Figure 5 As shown, the embodiment of the present application also discloses a scheduling device 50 for edge data services, which is applied to an edge service management device, including:
[0128] The data set acquisition module 501 is used to obtain the local operating environment parameters of the local edge service platform and determine the local data service and distributed data service of the edge data task; the local data service is used to execute in the local edge service platform, and the distributed data service is used to execute in the distributed edge service platform;
[0129] A local scheduling module 502, for inputting local data services and local operating environment parameters into a first language model to obtain a local scheduling decision for edge data tasks, and sending distributed data services to a distributed edge service platform;
[0130] The local execution module 503 is used to control the local edge service platform to execute the local data service according to the local scheduling decision, so as to update the local operating environment parameters and obtain the distributed operating environment parameters when the distributed data service is executed on the distributed edge service platform;
[0131] A global scheduling module 504, configured to input the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into a second language model to obtain a task adjustment decision for the edge data task;
[0132] The scheduling adjustment module 505 is used to adjust the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform according to the task adjustment decision.
[0133] Optionally, the local data service includes at least one execution task, the task adjustment decision includes an uninstallation decision for a target data service in the local data service, and the scheduling adjustment module 505 includes:
[0134] The scheduling adjustment submodule is used to unload the target data service executed on the local edge service platform according to the unloading decision, and send the unloaded target data service to the distributed edge service platform so that the distributed edge service platform executes the target data service.
[0135] Optionally, there are multiple distributed edge service platforms, the target data service has a corresponding target distributed edge service platform, and the scheduling adjustment submodule includes:
[0136] The task sending unit is used to send the target data service to the target distributed edge service platform so that the target distributed edge service platform executes the target data service.
[0137] Optionally, the edge data task includes multiple subtasks, and the data set acquisition module 501 includes:
[0138] A local partitioning submodule, for determining a local data service according to a target subtask determined from a plurality of subtasks based on a local operating environment parameter;
[0139] The distribution partitioning submodule is used to determine the distributed data service according to the remaining subtasks in the edge data task excluding the target subtask.
[0140] Optionally, the scheduling device 50 for edge data service further includes:
[0141] The task refresh module is used to respond to the modification command of the local scheduling decision, execute the modified local scheduling decision, obtain the local operating environment parameters after executing the modified local scheduling decision, and return to execute local data services on the local edge service platform according to the local scheduling decision.
[0142] like Figure 6 As shown, the embodiment of the present application further discloses a language model training device 51, which is used to train the first language model in the task scheduling method of the edge data service in the above embodiment, including:
[0143] A first training set module 511 is used to construct a first training set; the first training set includes a local data service of a training task and local training operating environment parameters of a local training edge service platform running the local data service;
[0144] The first training module 512 is used to input the local data service and the local training operating environment parameters into the first language model, and train the first language model according to a preset first loss function to obtain a trained first language model.
[0145] like Figure 7As shown, the embodiment of the present application further discloses a language model training device 52, which is used to train the second language model in the task scheduling method of the edge data service in the above embodiment, including:
[0146] The second training set module 521 is used to construct a second training set; the second training set includes a local training scheduling decision of the training task, a local training operating environment parameter of the local training edge service platform executed according to the local training scheduling decision, and a distributed training operating environment parameter of the distributed training edge service platform;
[0147] The second training set module 522 is used to input the local training scheduling decision, the local training operating environment parameters, and the distributed training operating environment parameters into the second language model, and train the second language model according to a preset second loss function to obtain a trained second language model.
[0148] like Figure 8 As shown, the embodiment of the present application also discloses a scheduling system 60 for edge data services, including:
[0149] A local edge service platform 601, a distributed edge service platform 602, and an edge service management device 603;
[0150] The local edge service platform 601 is used to perform local data services;
[0151] The distributed edge service platform 602 is used to perform distributed data services;
[0152] The edge service management device 603 is used to obtain the local operating environment parameters of the local edge service platform, determine the local data service and the distributed data service of the edge data task, and input the local data service and the local operating environment parameters into the first language model to obtain the local scheduling decision of the edge data task, and send the distributed data service to the distributed edge service platform, and control the local edge service platform to execute the local data service according to the local scheduling decision to update the local operating environment parameters, and obtain the distributed operating environment parameters of the distributed data service when it is executed on the distributed edge service platform, and input the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into the second language model to obtain the task adjustment decision of the edge data task, and adjust the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform according to the task adjustment decision.
[0153] like Fig. 9 As shown, under the scheduling system of the edge data service provided in the embodiment of the present application, the flow process of relevant data is as follows:
[0154] S1: Determine the specific content of local data tasks and distributed data tasks based on edge data tasks.
[0155] S2: Data services executed on the local edge service platform.
[0156] S3: Data services executed on the distributed edge service platform.
[0157] S4: Use the first language model to make scheduling decisions for local data services. The model will analyze local operating environment parameters and optimize the execution of local data services.
[0158] S5: After executing the local data service, the local operating environment parameters will change. These updated parameters will be used for further decision-making and adjustments.
[0159] S6: The second language model receives the updated local operating environment parameters and distributed operating environment parameters and generates a task adjustment decision.
[0160] S7: According to the output of the second language model, the execution of the local data service and the distributed data service is adjusted, which mainly includes the unloading of the local task and sending the unloaded task to the distributed data platform for execution.
[0161] In summary, in the embodiment of the present application, by inputting the local data service and its environmental parameters into the first language model to obtain the local scheduling decision, and then executing the selected data service on the local edge service platform, the local computing overhead is optimized and unnecessary resource consumption is avoided; then the second language model is used to process the operating environment parameters updated according to the execution of the local scheduling decision, as well as the operating environment parameters of the distributed platform, to obtain the task adjustment decision for the local and distributed data services, thereby enhancing the robustness of the system, so that the system can maintain the continuity and stability of the data service in the face of resource changes and environmental fluctuations, thereby collaboratively completing the specified data service. Therefore, based on the method of the embodiment of the present application, efficient scheduling of edge data services is achieved, solving the problems of low resource utilization efficiency and high computing overhead in the related technology due to the high complexity of algorithm model deployment and calculation in the prior art.
[0162] The embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, each process of the training method of the semantic evaluation model and the machine translation method of the streaming text is implemented, and the same technical effect can be achieved. To avoid repetition, it is not repeated here. The computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0163] Fig.10 700 is a block diagram of an electronic device 700 provided in an embodiment of the present application. For example, the electronic device 700 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.
[0164] Reference Fig.10 , the electronic device 700 may include one or more of the following components: a processing component 702 , a memory 704 , a power component 706 , a multimedia component 708 , an audio component 710 , an input / output (I / O) interface 712 , a sensor component 714 , and a communication component 716 .
[0165] The processing component 702 generally controls the overall operation of the electronic device 700, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 702 may include one or more processors 720 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 702 may include one or more modules to facilitate the interaction between the processing component 702 and other components. For example, the processing component 702 may include a multimedia module to facilitate the interaction between the multimedia component 708 and the processing component 702.
[0166] The memory 704 is used to store various types of data to support operations on the electronic device 700. Examples of such data include instructions for any application or method operating on the electronic device 700, contact data, phone book data, messages, pictures, multimedia, etc. The memory 704 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.
[0167] The power supply component 706 provides power to the various components of the electronic device 700. The power supply component 706 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device 700.
[0168] The multimedia component 708 includes a screen that provides an output interface between the electronic device 700 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 708 includes a front camera and / or a rear camera. When the electronic device 700 is in an operating mode, such as a shooting mode or a multimedia mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.
[0169] The audio component 710 is used to output and / or input audio signals. For example, the audio component 710 includes a microphone (MIC), and when the electronic device 700 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is used to receive an external audio signal. The received audio signal can be further stored in the memory 704 or sent via the communication component 716. In some embodiments, the audio component 710 also includes a speaker for outputting audio signals.
[0170] I / O interface 712 provides an interface between processing component 702 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: a home button, a volume button, a start button, and a lock button.
[0171] The sensor assembly 714 includes one or more sensors for providing various aspects of status assessment for the electronic device 700. For example, the sensor assembly 714 can detect the open / closed state of the electronic device 700, the relative positioning of components, such as the display and keypad of the electronic device 700, and the sensor assembly 714 can also detect the position change of the electronic device 700 or a component of the electronic device 700, the presence or absence of user contact with the electronic device 700, the orientation or acceleration / deceleration of the electronic device 700, and the temperature change of the electronic device 700. The sensor assembly 714 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 714 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 714 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.
[0172] The communication component 716 is used to facilitate wired or wireless communication between the electronic device 700 and other devices. The electronic device 700 can access a wireless network based on a communication standard, such as WiFi, a carrier network (such as 2G, 3G, 4G or 7G), or a combination thereof. In an exemplary embodiment, the communication component 716 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 716 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.
[0173] In an exemplary embodiment, the electronic device 700 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to implement a speech synthesis model training method provided in an embodiment of the present application.
[0174] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 704 including instructions, and the instructions can be executed by a processor 720 of an electronic device 700 to perform the above method. For example, the non-transitory storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.
[0175] Fig.11 8 is a block diagram of an electronic device 800 according to an exemplary embodiment. For example, the electronic device 800 may be provided as a server. Fig.11 The electronic device 800 includes a processing component 822, which further includes one or more processors, and a memory resource represented by a memory 832 for storing instructions executable by the processing component 822, such as an application. The application stored in the memory 832 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 822 is configured to execute instructions to perform a speech synthesis model training method provided in an embodiment of the present application.
[0176] The electronic device 800 may also include a power supply component 826 configured to perform power management of the electronic device 800, a wired or wireless network interface 850 configured to connect the electronic device 800 to a network, and an input / output (I / O) interface 858. The electronic device 800 may operate based on an operating system stored in the memory 832, such as Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™ or the like.
[0177] An embodiment of the present application also provides a computer program product, including a computer program, and a speech synthesis model training method implemented when the computer program is executed by a processor.
[0178] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the application disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present disclosure. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.
[0179] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.
[0180] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.
[0181] It is easy for a person skilled in the art to think that any combination of the above embodiments is feasible, so any combination of the above embodiments is an implementation scheme of the present application. However, due to space limitations, this specification will not describe them in detail here.
[0182] The task scheduling method for edge data services and the training method for language models provided herein are not inherently related to any specific computer, virtual system, or other device. Various general systems can also be used with the teachings based thereon. According to the above description, it is obvious that the structure required to construct a system with the solution of the present application is obvious. In addition, the present application is not directed to any specific programming language. It should be understood that the contents of the present application described herein can be implemented using various programming languages, and the above description of specific languages is to disclose the best implementation mode of the present application.
[0183] In the description provided herein, a large number of specific details are described. However, it is understood that the embodiments of the present application can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.
[0184] Similarly, it should be understood that in order to streamline the present application and aid in understanding one or more of the various application aspects, in the above description of the exemplary embodiments of the present application, the various features of the present application are sometimes grouped together into a single embodiment, figure, or description thereof. However, the disclosed method should not be interpreted as reflecting the following intention: the claimed application requires more features than the features explicitly recited in each claim. More specifically, as reflected in the claims, the application aspects lie in less than all the features of the single embodiment disclosed above. Therefore, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, with each claim itself serving as a separate embodiment of the present application.
[0185] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and in addition they may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.
[0186] In addition, those skilled in the art will appreciate that, although some embodiments described herein include certain features included in other embodiments but not other features, the combination of features of different embodiments is meant to be within the scope of the present application and form different embodiments. For example, in the claims, any one of the claimed embodiments can be used in any combination.
[0187] The various component embodiments of the present application can be implemented in hardware, or implemented in software modules running on one or more processors, or implemented in a combination thereof. It should be understood by those skilled in the art that a microprocessor or digital signal processor (DSP) can be used in practice to implement some or all functions of some or all components in the training method of the semantic evaluation model according to the embodiment of the present application and the machine translation method of the streaming text. The application can also be implemented as a device or apparatus program (for example, a computer program and a computer program product) for executing a part or all of the methods described herein. Such a program implementing the present application can be stored on a computer-readable medium, or can have the form of one or more signals. Such a signal can be downloaded from an Internet website, or provided on a carrier signal, or provided in any other form.
[0188] It should be noted that the above embodiments illustrate the present application rather than limit the present application, and that those skilled in the art may design alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets should not be constructed as a limitation to the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of multiple such elements. The present application may be implemented by means of hardware including several different elements and by means of a suitably programmed computer. In a unit claim that lists several devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names.
Claims
1. A task scheduling method for edge data services, applied to edge service management equipment, It is characterized in that include: Obtain the local operating environment parameters of the local edge service platform and determine the local data service and distributed data service of the edge data task; The local data service is used to execute in the local edge service platform, and the distributed data service is used to execute in the distributed edge service platform; Inputting the local data service and the local operating environment parameters into a first language model to obtain a local scheduling decision for the edge data task, and sending the distributed data service to the distributed edge service platform; According to the local scheduling decision, control the local edge service platform to execute the local data service to update the local operating environment parameters, and obtain the distributed operating environment parameters of the distributed data service when executed on the distributed edge service platform; Inputting the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into a second language model to obtain a task adjustment decision for the edge data task; According to the task adjustment decision, the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform are adjusted.
2. The method according to claim 1, It is characterized in that The local data service includes at least one execution task, the task adjustment decision includes an uninstallation decision for a target data service in the local data service, and adjusting the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform according to the task adjustment decision includes: The target data service executed on the local edge service platform is uninstalled according to the uninstallation decision, and the uninstalled target data service is sent to the distributed edge service platform, so that the distributed edge service platform executes the target data service.
3. The method according to claim 2, It is characterized in that There are multiple distributed edge service platforms, the target data service has a corresponding target distributed edge service platform, and sending the uninstalled target data service to the distributed edge service platform so that the distributed edge service platform executes the target data service includes: The target data service is sent to the target distributed edge service platform, so that the target distributed edge service platform executes the target data service.
4. The method according to claim 1, It is characterized in that The edge data task includes a plurality of subtasks, and the local data service and the distributed data service of the edge data task are determined, including: Determining the local data service according to the local operating environment parameter and a target subtask determined from the plurality of subtasks; The distributed data service is determined according to the remaining subtasks in the edge data task excluding the target subtask.
5. The method according to claim 1, It is characterized in that The method further comprises: In response to a modification command for the local scheduling decision, the modified local scheduling decision is executed, and the local operating environment parameters after the modified local scheduling decision is obtained, and the local data service is executed on the local edge service platform according to the local scheduling decision.
6. A scheduling system for edge data services, It is characterized in that include: Edge service management equipment, local edge service platform, distributed edge service platform; The local edge service platform is used to perform local data services; The distributed edge service platform is used to perform distributed data services; The edge service management device is used to obtain the local operating environment parameters of the local edge service platform, determine the local data service and the distributed data service of the edge data task, and input the local data service and the local operating environment parameters into the first language model to obtain the local scheduling decision of the edge data task, and send the distributed data service to the distributed edge service platform, and control the local edge service platform to execute the local data service according to the local scheduling decision to update the local operating environment parameters, and obtain the distributed operating environment parameters of the distributed data service when it is executed on the distributed edge service platform, and input the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into the second language model to obtain the task adjustment decision of the edge data task, and adjust the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform according to the task adjustment decision.
7. A scheduling device for edge data services, applied to edge service management equipment, It is characterized in that The device comprises: A data set acquisition module, used to obtain local operating environment parameters of the local edge service platform, and determine the local data service and distributed data service of the edge data task; the local data service is used to execute in the local edge service platform, and the distributed data service is used to execute in the distributed edge service platform; A local scheduling module, configured to input the local data service and the local operating environment parameters into a first language model to obtain a local scheduling decision for the edge data task, and send the distributed data service to the distributed edge service platform; A local execution module, configured to control the local edge service platform to execute the local data service according to the local scheduling decision, so as to update the local operating environment parameters, and obtain the distributed operating environment parameters of the distributed data service when it is executed on the distributed edge service platform; A global scheduling module, configured to input the local scheduling decision, the updated local operating environment parameters, and the distributed operating environment parameters into a second language model to obtain a task adjustment decision for the edge data task; The scheduling adjustment module is used to adjust the execution of the local data service on the local edge service platform and the execution of the distributed data service on the distributed edge service platform according to the task adjustment decision.
8. An electronic device, It is characterized in that include: A processor, a memory for storing instructions executable by the processor; The processor is configured to execute the instructions to implement the method according to any one of claims 1 to 6.
9. A computer-readable storage medium, It is characterized in that When the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the method as claimed in any one of claims 1 to 6.
10. A computer program product, It is characterized in that The computer program product stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.
