Task processing method, apparatus, device, storage medium, and computer program product

Abstract

The application discloses a task processing method and device, equipment, a storage medium and a computer program product, and relates to the technical field of artificial intelligence. The method comprises the following steps: receiving a model configuration request sent by a user through an AI browser based on a special data channel; performing model configuration according to the model configuration request, and deploying a private large model after the configuration is completed; when a to-be-processed task sent by the user through the AI browser is received based on the special data channel, performing task splitting and task execution on the to-be-processed task based on the deployed private large model. In the application, the private large model can be configured and deployed through the model configuration request received through the special data channel between the AI browser, and the task processing is performed based on the deployed private large model, so that the technical problem that the cost of deploying a large model in an enterprise internal environment is high and the task processing efficiency is low due to the limited computing power of the deployed large model is solved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to task processing methods, apparatus, devices, storage media, and computer program products. Background Technology

[0002] With the rapid development of artificial intelligence technology, large-scale pre-trained models (such as the GPT series models and BERT models) have achieved remarkable results in fields such as Natural Language Processing (NLP) and Computer Vision (CV). These models are typically deployed in the cloud for easy access and updates. However, due to concerns about data privacy, security, and compliance, enterprises tend to deploy private AI models locally. Furthermore, enterprises have an increasing demand for customized, high-performance large models, which often contain the enterprise's core knowledge, business logic, and data assets. Therefore, achieving private deployment of large models has become crucial for enterprises to ensure data security and enhance business competitiveness.

[0003] Currently, deploying large models within enterprise environments typically employs containerization technology to ensure independent operation and resource isolation, while leveraging cloud-native technologies to achieve elastic scaling and automated operation and maintenance. However, this approach results in a complex task processing workflow, leading to high deployment costs and low efficiency. Furthermore, the computing power of large models deployed within enterprises is limited, resulting in low efficiency in task completion for the deployed large models according to existing deployment schemes. Summary of the Invention

[0004] The main objective of this application is to provide a task processing method, apparatus, device, storage medium, and computer program product, which aims to solve the technical problems of high cost of deploying large models in enterprise internal environments and low task processing efficiency due to the limited computing power of the deployed large models in the prior art.

[0005] To achieve the above objectives, this application proposes a task processing method. This method is applied to a private large-scale model, which establishes a dedicated data channel with an AI browser. The method includes:

[0006] Based on the dedicated data channel, the system receives model configuration requests sent by users through the AI ​​browser;

[0007] Configure the model according to the model configuration request, and deploy the configured private large model;

[0008] When a pending task is received from the user through the AI ​​browser via the dedicated data channel, the pending task is split and executed based on the deployed private big model.

[0009] In one embodiment, the step of splitting and executing the task to be processed based on the deployed private large model includes:

[0010] Based on the aforementioned privateized large model, the task to be processed is split into several sub-tasks to be processed.

[0011] Based on the aforementioned privatization model, the sub-tasks to be processed are assigned to the corresponding target computing devices, so that the target computing devices can perform task execution on the sub-tasks to be processed.

[0012] In one embodiment, the step of allocating the subtasks to be processed to the corresponding target computing devices based on the privatized large model includes:

[0013] The task inputs and task objectives corresponding to all computing devices are obtained through the aforementioned private large model;

[0014] The private large model determines the target computing device corresponding to the subtask to be processed from the computing device based on the task input and the task objective;

[0015] The subtask to be processed is assigned to the target computing device.

[0016] In one embodiment, the step of determining the target computing device corresponding to the subtask to be processed from the computing devices based on the task input and the task objective using the privatized large model includes:

[0017] The degree of association between the subtask to be processed and the computing device is determined by the privatized large model based on the task input and the task objective;

[0018] Determine the task type and task quantity corresponding to the subtask to be processed;

[0019] The target computing device corresponding to the subtask to be processed is determined from the computing devices based on the degree of association, the task type, and the task volume.

[0020] In one embodiment, the step of determining the target computing device corresponding to the subtask to be processed from the computing devices based on the degree of association, the task type, and the task volume includes:

[0021] The computing devices are sorted according to the degree of correlation to obtain a computing device sequence;

[0022] Based on the task type, the computing device sequence is filtered to obtain the target computing device sequence;

[0023] The target computing device corresponding to the subtask to be processed is determined from the target computing device sequence based on the task volume.

[0024] In one embodiment, the step of determining the target computing device corresponding to the subtask to be processed from the target computing device sequence based on the task volume includes:

[0025] Determine whether there are candidate computing devices in the target computing device sequence whose device task processing volume is greater than the task volume;

[0026] If it exists, the candidate computing device is determined as the target computing device corresponding to the subtask to be processed;

[0027] Alternatively, if none exist, then determine the computing devices to be merged in the target computing device sequence based on the amount of work;

[0028] The computing devices to be merged are merged to obtain the target computing device corresponding to the subtask to be processed.

[0029] In one embodiment, after the step of assigning the subtask to be processed to a corresponding target computing device so that the target computing device can process the subtask, the method further includes:

[0030] The user feedback document is split according to a preset text splitting method to obtain the document splitting results;

[0031] The user's question and the answer are determined based on the document splitting results;

[0032] Output a predicted answer based on the user's question;

[0033] The model parameters are adjusted based on the answer to the question and the predicted answer.

[0034] In one embodiment, the step of assigning the subtask to be processed to the corresponding target computing device includes:

[0035] The subtask to be processed is encrypted using a preset data encryption method to obtain the encrypted subtask to be processed.

[0036] The encrypted subtasks to be processed are assigned to the corresponding target computing devices.

[0037] In one embodiment, before the step of receiving the model configuration request sent by the user through the AI ​​browser based on the dedicated data channel, the method further includes:

[0038] Determine the data transmission interface and data transmission protocol between the AI ​​browser and the AI ​​browser;

[0039] A dedicated data channel is established between the AI ​​browser and the data transmission interface and the data transmission protocol.

[0040] Furthermore, to achieve the above objectives, this application also proposes a task processing device, wherein the device is equipped with a private large model, and a dedicated data channel is established between the private large model and the AI ​​browser, the device comprising:

[0041] The request receiving module is used to receive model configuration requests sent by users through the AI ​​browser based on the dedicated data channel;

[0042] The model deployment module is used to configure the model according to the model configuration request and deploy the configured private large model.

[0043] The task processing module is used to split and execute the pending task based on the deployed private big model when it receives the pending task sent by the user through the AI ​​browser based on the dedicated data channel.

[0044] In one embodiment, the task processing module is further configured to split the task to be processed based on the private large model to obtain several sub-tasks to be processed; and to allocate the sub-tasks to be processed to corresponding target computing devices based on the private large model, so that the target computing devices can perform task execution on the sub-tasks to be processed.

[0045] In one embodiment, the task processing module is further configured to obtain the task inputs and task targets corresponding to all computing devices through the private large model; determine the target computing device corresponding to the sub-task to be processed from the computing devices based on the task inputs and task targets through the private large model; and allocate the sub-task to be processed to the target computing device.

[0046] In one embodiment, the task processing module is further configured to determine the degree of association between the subtask to be processed and the computing device based on the task input and the task target using the private big model; determine the task type and task quantity corresponding to the subtask to be processed; and determine the target computing device corresponding to the subtask to be processed from the computing devices based on the degree of association, the task type and the task quantity.

[0047] In one embodiment, the task processing module is further configured to sort the computing devices according to the degree of association to obtain a computing device sequence; filter the computing device sequence based on the task type to obtain a target computing device sequence; and determine the target computing device corresponding to the subtask to be processed from the target computing device sequence according to the task volume.

[0048] In one embodiment, the task processing module is further configured to determine whether there are candidate computing devices in the target computing device sequence whose device task processing volume is greater than the task volume; if there are, the candidate computing device is determined as the target computing device corresponding to the sub-task to be processed; or, if there are no, the computing devices to be merged in the target computing device sequence are determined according to the task volume; the computing devices to be merged are merged to obtain the target computing device corresponding to the sub-task to be processed.

[0049] In one embodiment, the apparatus further includes: a model update module;

[0050] The model update module is used to split the user feedback document according to a preset text splitting method to obtain the document splitting result; determine the user question and the question answer based on the document splitting result; output a predicted answer based on the user question; and adjust the model parameters based on the question answer and the predicted answer.

[0051] In one embodiment, the device further includes: a data encryption module;

[0052] The data encryption module is used to encrypt the subtask to be processed using a preset data encryption method to obtain the encrypted subtask to be processed; and to allocate the encrypted subtask to the corresponding target computing device.

[0053] In addition, to achieve the above objectives, this application also proposes a task processing device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the task processing method as described above.

[0054] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and which, when executed by a processor, implements the steps of the task processing method described above.

[0055] In addition, to achieve the above objectives, this application also provides a computer program product, which includes a computer program that, when executed by a processor, implements the steps of the task processing method described above.

[0056] This application provides a task processing method, which discloses receiving model configuration requests sent by users through an AI browser via a dedicated data channel; configuring the model according to the configuration request and deploying the configured private large model; when receiving pending tasks sent by users through the AI ​​browser via the dedicated data channel, splitting and executing the pending tasks based on the deployed private large model. Compared with the prior art of deploying large models in an enterprise internal environment using containerization technology, which is complex and costly, and whose efficiency in solving tasks is low due to the limited computing power of the deployed large model, this invention solves the technical problems of high cost and low efficiency in task processing caused by the limited computing power of the deployed large model in an enterprise internal environment by using model configuration requests received through a dedicated data channel between the private large model and the AI ​​browser. Attached Figure Description

[0057] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0058] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0059] Figure 1 This is a flowchart illustrating an embodiment of the task processing method of this application.

[0060] Figure 2 This is a flowchart illustrating Embodiment 2 of the task processing method of this application;

[0061] Figure 3 This is a diagram illustrating the overall architecture of the task processing method in this application.

[0062] Figure 4 This is a flowchart illustrating Embodiment 3 of the task processing method of this application;

[0063] Figure 5 This is a schematic diagram of the module structure of the task processing device according to an embodiment of this application;

[0064] Figure 6 This is a schematic diagram of the device structure of the hardware operating environment involved in the task processing method in the embodiments of this application.

[0065] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0066] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0067] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0068] The main solution of this application embodiment is: receiving a model configuration request sent by the user through the AI ​​browser based on the dedicated data channel; configuring the model according to the model configuration request, and deploying the configured private large model; when receiving a task to be processed sent by the user through the AI ​​browser based on the dedicated data channel, splitting and executing the task to be processed based on the deployed private large model.

[0069] In current technologies, deploying large models within an enterprise environment typically employs containerization to ensure independent operation and resource isolation, while leveraging cloud-native technologies to achieve elastic scaling and automated operation and maintenance. However, this approach results in a complex task processing workflow, leading to high deployment costs and low efficiency. Furthermore, the computing power of large models deployed within enterprises is limited, resulting in low efficiency in task completion for the deployed large models according to existing deployment schemes.

[0070] This application provides a solution that enables private large models to be configured and deployed through model configuration requests received via a dedicated data channel between the private large model and the AI ​​browser, and to perform task processing based on the deployed private large model. This solves the technical problems in the prior art, such as the high cost of deploying large models in an enterprise internal environment and the low task processing efficiency due to the limited computing power of the deployed large model.

[0071] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or a private large-scale model capable of implementing the above functions, or a device integrating a private large-scale model, etc. The following description uses a private large-scale model as an example to illustrate this embodiment and the subsequent embodiments.

[0072] Based on this, the embodiments of this application provide a task processing method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the task processing method of this application.

[0073] In this embodiment, the task processing method is applied to a private large model, which establishes a dedicated data channel with the AI ​​browser. The method includes steps S10 to S30:

[0074] Step S10: Receive the model configuration request sent by the user through the AI ​​browser based on the dedicated data channel.

[0075] It should be noted that the aforementioned private large-scale model can be a large model deployed locally by the enterprise, such as the GPT model, BERT model, etc. This embodiment does not limit the type of private large-scale model. The aforementioned AI browser can be a web browser that integrates artificial intelligence technology. In this embodiment, the AI ​​browser can be responsible for user interface display and user interaction, and integrates AI algorithms for intelligent analysis and decision-making.

[0076] It should be noted that the aforementioned dedicated data channel can be used for data transmission between the private large model and the AI ​​browser.

[0077] It should be understood that the aforementioned model configuration request can be a request used to instruct the private large model to perform model configuration. In practical applications, a corresponding data connection mechanism can be set up between the private large model and the AI ​​browser. This data connection mechanism can include, but is not limited to, API gateways, message queues, asynchronous processing frameworks, etc., so that the AI ​​browser can quickly forward the user's input model configuration request to the private large model through these mechanisms. In addition, the response result of the private large model can also be returned to the user in real time, ensuring the security and integrity of data transmission.

[0078] In this implementation, users can configure the private large-scale model through the AI ​​browser interface and generate a model configuration request based on this configuration information. Upon detecting the model configuration request, the AI ​​browser can send it to the private large-scale model via a dedicated data channel. In this embodiment, since users can directly configure and manage the private large-scale model through the AI ​​browser interface without requiring specialized AI knowledge, the deployment and maintenance difficulty of the enterprise's private large-scale model is reduced. Enterprises do not need to rely on external cloud service providers, improving the system's operability and maintainability. Furthermore, enterprises can configure the type and parameters of the large-scale model according to their own needs through the AI ​​browser interface, thereby achieving customized deployment of the large-scale model and meeting the AI ​​service needs of different business scenarios.

[0079] Furthermore, prior to step S10, the method further includes: determining the data transmission interface and data transmission protocol between the AI ​​browser and the AI ​​browser; and establishing a dedicated data channel between the AI ​​browser and the AI ​​browser based on the data transmission interface and the data transmission protocol.

[0080] It is understood that the aforementioned data transmission interface can be the interface in the AI ​​browser used for data transmission with the private large model; the aforementioned data transmission protocol can be the protocol for data transmission between the AI ​​browser and the private large model. In this embodiment, the AI ​​browser can have a built-in dedicated data transmission protocol and data transmission interface for communication with the large model. Simultaneously, a dedicated data channel for data transmission with the private large model can be established based on this data transmission protocol and interface, enabling efficient and secure interaction with the enterprise's private large model through this dedicated data channel. Furthermore, the AI ​​browser in this embodiment supports multiple input methods (such as text, images, and voice) and possesses powerful data processing and display capabilities.

[0081] Step S20: Configure the model according to the model configuration request, and deploy the configured private large model.

[0082] In this embodiment, after receiving a model configuration request, the private large model can parse the model configuration request to obtain the model type and model parameters to be configured, configure the model according to the model type and model parameters, and deploy the configured private large model.

[0083] It should be understood that the aforementioned model type to be configured can be the type of the model deployed in this instance, such as a language model, a question-answering model, etc., and this embodiment does not impose any restrictions on this. The aforementioned model parameters to be configured are the parameters in the model deployed in this instance. In this embodiment, since the model configuration request is generated by the user configuring the model on the AI ​​browser interface according to the enterprise's needs, the private large model can parse the model configuration request after receiving it to obtain the model type and model parameters configured by the user, and determine the user-configured model type and model parameters as the model type and model parameters to be configured, respectively. After determining the model type and model parameters of the large model to be deployed in this instance, the private large model can select the corresponding model type and set the model parameters to achieve model configuration, and finally deploy the configured private large model.

[0084] Step S30: When a pending task is received from the user through the AI ​​browser based on the dedicated data channel, the pending task is split and executed based on the deployed private big model.

[0085] It is understandable that the above-mentioned tasks to be processed can be computational tasks for a private large model.

[0086] In this embodiment, the user can input a task that needs to be computed by the private large model through the interface of the AI ​​browser. At this time, the AI ​​browser can send the task to be processed to the private large model through a dedicated data channel. Then, the deployed private large model can split and execute the task.

[0087] This embodiment provides a task processing method. The method discloses receiving model configuration requests sent by users through an AI browser via a dedicated data channel; configuring the model according to the configuration request; and deploying the configured private large model. When receiving pending tasks sent by users through the AI ​​browser via the dedicated data channel, the method splits and executes the pending tasks based on the deployed private large model. Compared to the prior art using containerization technology to deploy large models in an enterprise's internal environment, which is complex and costly, and whose efficiency in handling tasks is low due to limited computing power, this embodiment solves the technical problems of high cost and low efficiency in task processing caused by limited computing power of deployed large models in an enterprise's internal environment by using model configuration requests received via a dedicated data channel with the AI ​​browser.

[0088] Based on the first embodiment of this application, in the second embodiment of this application, the content that is the same as or similar to that in the first embodiment described above can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 2 , Figure 2 This is a flowchart illustrating Embodiment 2 of the task processing method of this application.

[0089] In this embodiment, step S30 includes steps S31 to S33:

[0090] Step S31: Based on the privatized large model, the task to be processed is split into several sub-tasks to be processed.

[0091] It should be understood that a subtask to be processed is a subtask obtained by splitting the task to be processed. In this embodiment, a distributed computing module can be set up in the private large model. This distributed computing module can split the task to be processed according to the computing power of the model or the task type of the task to be processed to obtain multiple subtasks to be processed.

[0092] Step S32: Based on the privatized large model, the sub-tasks to be processed are assigned to the corresponding target computing devices, so that the target computing devices can perform task execution on the sub-tasks to be processed.

[0093] It should be noted that the aforementioned target computing device can be a device used for processing and computing tasks. In this embodiment, the distributed computing module can break down the computing tasks of the private large model into multiple subtasks, and allocate these subtasks to the corresponding target computing devices for processing. This allows the advantages of cluster computing to be utilized to execute the split subtasks in parallel, improving computing efficiency. Furthermore, this embodiment can dynamically adjust the task allocation strategy based on the performance of the computing devices and network conditions to ensure the efficient execution of computing tasks.

[0094] It should be noted that, in response to the performance bottleneck in the inference process of the private large model, this embodiment can use advanced optimization algorithms (such as model pruning, quantization, distillation, etc.) and acceleration technologies (such as GPU acceleration, TPU acceleration, etc.) to improve the inference speed and efficiency of the private large model, thereby ensuring that users can enjoy a smooth AI service experience.

[0095] Further, step S32 includes:

[0096] Step S321: Obtain the task inputs and task objectives corresponding to all computing devices through the privatized large model.

[0097] It is understandable that the above task input is the input of the computing device; the above task objective is the computing objective of the computing device.

[0098] Step S322: Based on the task input and the task objective, determine the target computing device corresponding to the subtask to be processed from the computing devices using the privatized large model.

[0099] Specifically, step S322 includes:

[0100] Step S322a: Determine the degree of association between the subtask to be processed and the computing device based on the task input and the task objective using the privatized large model.

[0101] It should be noted that the aforementioned correlation degree can be a feature used to characterize the reliability of the subtask being processed using the computing device. In this embodiment, the privatized large model can first determine the input and output of the subtask to be processed, and then compare the input and output of the subtask to be processed with the task input and task target of the computing device, respectively, to determine the first similarity between the input of the subtask to be processed and the task input, and the second similarity between the output of the subtask to be processed and the task target. Finally, the correlation degree between the subtask to be processed and the computing device is determined based on the first similarity and the second similarity.

[0102] Step S322b: Determine the task type and task quantity corresponding to the subtask to be processed.

[0103] It should be understood that the above task type can be the type corresponding to the subtask to be processed, such as question-and-answer tasks, language generation tasks, etc., and this embodiment does not limit this.

[0104] It is understood that the aforementioned workload can refer to the time and resources required to complete the sub-tasks to be processed. In this embodiment, the workload of the sub-tasks to be processed can be determined based on their complexity and urgency.

[0105] Step S322c: Based on the degree of association, the task type, and the task volume, determine the target computing device corresponding to the sub-task to be processed from the computing devices.

[0106] Specifically, step S322c includes: sorting the computing devices according to the degree of association to obtain a computing device sequence; filtering the computing device sequence based on the task type to obtain a target computing device sequence; and determining the target computing device corresponding to the subtask to be processed from the target computing device sequence according to the task volume.

[0107] It should be noted that in this embodiment, the computing devices can be sorted from high to low according to the degree of association between the subtask to be processed and the computing device, thereby obtaining the above-mentioned computing device sequence. Then, a search can be performed in the computing device sequence according to the task type corresponding to the subtask to be processed to determine all first computing devices in the computing device sequence that can process the task type corresponding to the subtask to be processed. Then, all computing devices in the computing device sequence other than the first computing devices are filtered to obtain the target computing device sequence. At this time, all computing devices stored in the target computing device sequence can process the subtask to be processed.

[0108] In this embodiment, the privatized large model can determine whether there are candidate computing devices in the target computing device sequence whose device task processing volume is greater than the task volume; if there are, the candidate computing device is determined as the target computing device corresponding to the sub-task to be processed; or, if there are no, the computing devices to be merged in the target computing device sequence are determined according to the task volume; the computing devices to be merged are merged to obtain the target computing device corresponding to the sub-task to be processed.

[0109] It is understood that the aforementioned task processing capacity can be the amount of tasks that a computing device can handle. The aforementioned candidate computing devices can be those in the target computing device sequence whose task processing capacity is greater than the task capacity of the subtasks to be processed.

[0110] Step S323: Assign the subtask to be processed to the target computing device.

[0111] In practical applications, if there are candidate computing devices in the target computing device sequence that can handle a larger amount of tasks than the subtasks to be processed, then these candidate devices can be identified as target computing devices, and the subtasks to be processed can be assigned to them for computation. Alternatively, if there are no candidate computing devices in the target computing device sequence that can handle a larger amount of tasks than the subtasks to be processed, then two or more computing devices can be merged based on the amount of tasks to be processed and the processing capacity of each computing device in the target computing device sequence. In this case, the processing capacity of the merged computing device will be greater than the amount of tasks to be processed, and the merged computing device can then be identified as the target computing device, and the subtasks to be processed can be assigned to it for computation.

[0112] In this embodiment, in order to ensure secure data transmission, step S32 further includes: encrypting the subtask to be processed using a preset data encryption method to obtain the encrypted subtask to be processed; and assigning the encrypted subtask to the corresponding target computing device.

[0113] It should be understood that the above-mentioned preset data encryption methods can be used to encrypt data in a private large model, such as the RSA algorithm, SM2 algorithm, etc. Among them, the RSA algorithm is an asymmetric encryption algorithm, and the SM2 algorithm is an elliptic curve public key cryptography algorithm.

[0114] In practical applications, enterprise data is highly sensitive and confidential. Uploading data to a private large-scale model platform for model training and inference carries the risk of data leakage. Therefore, in this embodiment, a data encryption and security module can be deployed in the private large-scale model. This module can encrypt enterprise and user data to ensure the security of data transmission and storage. Furthermore, this module can perform regular security checks and vulnerability patching to ensure the security of enterprise data and the confidentiality of user privacy.

[0115] In the specific implementation, refer to Figure 3 , Figure 3 This is a diagram illustrating the overall architecture of the task processing method in this application. Figure 3The AI ​​Browser's large model module can communicate with enterprise-owned private large models efficiently and securely through dedicated protocols and interfaces built into the AI ​​Browser. To ensure enterprise data security, data transmitted between the AI ​​Browser module and the enterprise-owned private large model is encrypted using RSA or SM2 algorithms. In this embodiment, the enterprise private large model can be deployed with a model loading and management module, a distributed computing module, a model update and maintenance module, and a data encryption and security module. The model loading and management module is responsible for loading and managing the enterprise private large model, supporting multiple model formats and protocols such as TensorFlow and PyTorch. This module can automatically optimize the model storage and loading process, reducing resource consumption. The distributed computing module utilizes distributed computing technology to decompose and distribute the computational tasks of the large model across multiple devices to improve computational efficiency. The data encryption and security module encrypts enterprise and user data, ensuring the security of data transmission and storage. The model update and maintenance module supports online updates and maintenance of the enterprise private large model, ensuring the model's accuracy and timeliness. Specifically, users can retrain and optimize the model through the AI ​​browser module. In this case, the model update and maintenance module can automatically record model update logs and provide detailed update reports to help users better manage and maintain the model.

[0116] In this embodiment, a method is disclosed to split the task to be processed into several sub-tasks based on a private large model; and to allocate the sub-tasks to be processed to the corresponding target computing devices based on the private large model, so that the target computing devices can execute the sub-tasks. Since this embodiment can split the task to be processed into multiple sub-tasks based on the private large model and allocate them to the corresponding computing devices for computation, the task processing efficiency can be improved.

[0117] Based on the first and / or second embodiments of this application, in the third embodiment of this application, the content that is the same as or similar to the above embodiments can be referred to the above description, and will not be repeated hereafter. Based on this, please refer to... Figure 4 , Figure 4 This is a flowchart illustrating Embodiment 3 of the task processing method of this application.

[0118] In this embodiment, after step S33, the method further includes steps S41 to S44:

[0119] Step S41: Split the user feedback document according to the preset text splitting method to obtain the document splitting result.

[0120] It is understood that the above-mentioned preset text splitting method can be a way of splitting text content. In this embodiment, the preset text splitting method may include, but is not limited to, splitting by semantics, splitting by punctuation marks, splitting by paragraphs, etc.

[0121] It should be understood that the aforementioned user feedback document can be a document storing user feedback information, such as user-submitted questions and answers. In this embodiment, the user can input feedback information through an AI browser, which can then generate a user feedback document based on the user's feedback information and send the user feedback document to a private large-scale model. Upon receiving the user feedback document, the private large-scale model can split it according to a preset text splitting method to obtain the document splitting result.

[0122] Step S42: Determine the user's question and the answer to the question based on the document splitting results.

[0123] It is understandable that the user questions and answers mentioned above can be the user's question in the user feedback and the actual answer to that question, respectively.

[0124] Step S43: Output a predicted answer based on the user's question.

[0125] It should be understood that the above predicted answer can be the answer predicted by the private large model based on the user's question.

[0126] Step S44: Adjust the model parameters based on the question answer and the predicted answer.

[0127] In practical applications, to improve the prediction accuracy of the private large-scale model, users can input questions and answers via the AI ​​browser. The AI ​​browser then generates a user feedback document based on the user's feedback and sends it to the private large-scale model. The private large-scale model can perform text segmentation on the user feedback document to identify the user's question and the true answer. It then predicts the answer to the user's question and compares the predicted answer with the true answer to determine the discrepancy. Based on this difference, the model parameters are adjusted to improve prediction accuracy. Furthermore, users can provide feedback on the private large-scale model's operation through the AI ​​browser. The model update and maintenance module can then continuously optimize the AI ​​algorithm and parameter configuration based on user feedback to periodically update the AI ​​model, thereby improving the large-scale model's intelligent analysis and decision-making capabilities.

[0128] This embodiment discloses the process of splitting user feedback documents according to a preset text splitting method to obtain document splitting results; determining user questions and answers based on the document splitting results; outputting predicted answers based on user questions; and adjusting model parameters based on question answers and predicted answers. Since this embodiment can adjust the parameters of the private large model according to the user feedback documents to update the private large model, it can improve the intelligent analysis and decision-making capabilities of the large model.

[0129] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the task processing method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0130] This application also provides a task processing device, please refer to... Figure 5 The task processing device includes a private large model, and a dedicated data channel is established between the private large model and the AI ​​browser. The device comprises:

[0131] The request receiving module 10 is used to receive model configuration requests sent by users through the AI ​​browser based on the dedicated data channel;

[0132] The model deployment module 20 is used to configure the model according to the model configuration request and deploy the configured private large model.

[0133] The task processing module 30 is used to split and execute the pending task based on the deployed private big model when it receives the pending task sent by the user through the AI ​​browser based on the dedicated data channel.

[0134] The task processing apparatus provided in this application, employing the task processing method described in the above embodiments, can solve the technical problems in the prior art where the cost of deploying large models in an enterprise internal environment is high, and the task processing efficiency is low due to the limited computing power of the deployed large models. Compared with the prior art, the beneficial effects of the task processing apparatus provided in this application are the same as those of the task processing method provided in the above embodiments, and other technical features in the task processing apparatus are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0135] This application provides a task processing device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the task processing method in Embodiment 1 above.

[0136] The following is for reference. Figure 6The diagram illustrates a structural schematic of a task processing device suitable for implementing embodiments of this application. The task processing device in the embodiments of this application may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 6 The task processing device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0137] like Figure 6 As shown, the task processing device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the task processing device. The processing unit 1001, ROM 1002, and RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the task processing device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows task processing devices with various systems, it should be understood that it is not required to implement or possess all of the systems shown. More or fewer systems may be implemented alternatively.

[0138] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from ROM 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0139] The task processing device provided in this application, employing the task processing method described in the above embodiments, can solve the technical problems of task processing. Compared with the prior art, the beneficial effects of the task processing device provided in this application are the same as those of the task processing method described in the above embodiments, and other technical features of the task processing device are the same as those disclosed in the method of the previous embodiment, and will not be repeated here.

[0140] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0141] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0142] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the task processing method described in the above embodiments.

[0143] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination thereof.

[0144] The aforementioned computer-readable storage medium may be included in the task processing device or may exist independently and not assembled into the task processing device.

[0145] The aforementioned computer-readable storage medium carries one or more programs. When the one or more programs are executed by the task processing device, the task processing device: receives a model configuration request sent by a user through the AI ​​browser via the dedicated data channel; configures the model according to the model configuration request and deploys the configured private large model; and when it receives a task to be processed sent by the user through the AI ​​browser via the dedicated data channel, it splits and executes the task to be processed based on the deployed private large model.

[0146] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0147] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0148] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0149] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described task processing method. This solves the technical problems in the prior art, such as the high cost of deploying large models in an enterprise internal environment and the low task processing efficiency due to the limited computing power of the deployed large models. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the task processing method provided in the above embodiments, and will not be repeated here.

[0150] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the task processing method described above.

[0151] The computer program product provided in this application can solve the technical problems in the prior art, such as high cost of deploying large models in enterprise internal environments and low task processing efficiency due to the limited computing power of the deployed large models. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as the beneficial effects of the task processing method provided in the above embodiments, and will not be repeated here.

[0152] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

[0153] This invention discloses A1, a task processing method, which is applied to a private large model, wherein a dedicated data channel is established between the private large model and the AI ​​browser, and the method includes:

[0154] Based on the dedicated data channel, the system receives model configuration requests sent by users through the AI ​​browser;

[0155] Configure the model according to the model configuration request, and deploy the configured private large model;

[0156] When a pending task is received from the user through the AI ​​browser via the dedicated data channel, the pending task is split and executed based on the deployed private big model.

[0157] A2. As described in A1, the steps of splitting and executing the task to be processed based on the deployed private large model include:

[0158] Based on the aforementioned privateized large model, the task to be processed is split into several sub-tasks to be processed.

[0159] Based on the aforementioned privatization model, the sub-tasks to be processed are assigned to the corresponding target computing devices, so that the target computing devices can perform task execution on the sub-tasks to be processed.

[0160] A3. As described in A2, the step of allocating the subtasks to be processed to the corresponding target computing devices based on the privatized large model includes:

[0161] The task inputs and task objectives corresponding to all computing devices are obtained through the aforementioned private large model;

[0162] The private large model determines the target computing device corresponding to the subtask to be processed from the computing device based on the task input and the task objective;

[0163] The subtask to be processed is assigned to the target computing device.

[0164] A4. As described in A3, the step of determining the target computing device corresponding to the subtask to be processed from the computing devices based on the task input and the task target using the privatized large model includes:

[0165] The degree of association between the subtask to be processed and the computing device is determined by the privatized large model based on the task input and the task objective;

[0166] Determine the task type and task quantity corresponding to the subtask to be processed;

[0167] The target computing device corresponding to the subtask to be processed is determined from the computing devices based on the degree of association, the task type, and the task volume.

[0168] A5. As described in A4, the step of determining the target computing device corresponding to the subtask to be processed from the computing devices based on the degree of association, the task type, and the task volume includes:

[0169] The computing devices are sorted according to the degree of correlation to obtain a computing device sequence;

[0170] Based on the task type, the computing device sequence is filtered to obtain the target computing device sequence;

[0171] The target computing device corresponding to the subtask to be processed is determined from the target computing device sequence based on the task volume.

[0172] A6. As described in A5, the step of determining the target computing device corresponding to the subtask to be processed from the target computing device sequence based on the task volume includes:

[0173] Determine whether there are candidate computing devices in the target computing device sequence whose device task processing volume is greater than the task volume;

[0174] If it exists, the candidate computing device is determined as the target computing device corresponding to the subtask to be processed;

[0175] Alternatively, if none exist, then determine the computing devices to be merged in the target computing device sequence based on the amount of work;

[0176] The computing devices to be merged are merged to obtain the target computing device corresponding to the subtask to be processed.

[0177] A7. The method as described in any one of A2 to A6, further comprising, after the step of assigning the subtask to be processed to a corresponding target computing device so that the target computing device performs task processing on the subtask:

[0178] The user feedback document is split according to a preset text splitting method to obtain the document splitting results;

[0179] The user's question and the answer are determined based on the document splitting results;

[0180] Output a predicted answer based on the user's question;

[0181] The model parameters are adjusted based on the answer to the question and the predicted answer.

[0182] A8. As described in A2, the step of assigning the subtask to be processed to the corresponding target computing device includes:

[0183] The subtask to be processed is encrypted using a preset data encryption method to obtain the encrypted subtask to be processed.

[0184] The encrypted subtasks to be processed are assigned to the corresponding target computing devices.

[0185] A9. The method described in A1, prior to the step of receiving the model configuration request sent by the user through the AI ​​browser based on the dedicated data channel, further includes:

[0186] Determine the data transmission interface and data transmission protocol between the AI ​​browser and the AI ​​browser;

[0187] A dedicated data channel is established between the AI ​​browser and the data transmission interface and the data transmission protocol.

[0188] This invention also discloses B10 and a task processing device, wherein the device is equipped with a private large model, and a dedicated data channel is established between the private large model and the AI ​​browser, and the device includes:

[0189] The request receiving module is used to receive model configuration requests sent by users through the AI ​​browser based on the dedicated data channel;

[0190] The model deployment module is used to configure the model according to the model configuration request and deploy the configured private large model.

[0191] The task processing module is used to split and execute the pending task based on the deployed private big model when it receives the pending task sent by the user through the AI ​​browser based on the dedicated data channel.

[0192] B11. In the apparatus described in B10, the task processing module is further configured to split the task to be processed into several sub-tasks based on the private large model; and to allocate the sub-tasks to be processed to corresponding target computing devices based on the private large model, so that the target computing devices can perform task execution on the sub-tasks to be processed.

[0193] B12. In the apparatus described in B11, the task processing module is further configured to obtain the task inputs and task targets corresponding to all computing devices through the private large model; determine the target computing device corresponding to the sub-task to be processed from the computing devices based on the task inputs and task targets through the private large model; and assign the sub-task to be processed to the target computing device.

[0194] B13. The apparatus as described in B12, wherein the task processing module is further configured to determine the degree of association between the subtask to be processed and the computing device based on the task input and the task target using the privatized large model; determine the task type and task quantity corresponding to the subtask to be processed; and determine the target computing device corresponding to the subtask to be processed from the computing devices based on the degree of association, the task type and the task quantity.

[0195] B14. The apparatus as described in B13, wherein the task processing module is further configured to sort the computing devices according to the degree of association to obtain a computing device sequence; filter the computing device sequence based on the task type to obtain a target computing device sequence; and determine the target computing device corresponding to the subtask to be processed from the target computing device sequence according to the task volume.

[0196] B15. The apparatus as described in B14, wherein the task processing module is further configured to determine whether there are candidate computing devices in the target computing device sequence whose device task processing volume is greater than the task volume; if so, determine the candidate computing device as the target computing device corresponding to the sub-task to be processed; or, if not, determine the computing devices to be merged in the target computing device sequence according to the task volume; merge the computing devices to be merged to obtain the target computing device corresponding to the sub-task to be processed.

[0197] B16. The apparatus of any one of B11 to B15, wherein the apparatus further comprises: a model update module;

[0198] The model update module is used to split the user feedback document according to a preset text splitting method to obtain the document splitting result; determine the user question and the question answer based on the document splitting result; output a predicted answer based on the user question; and adjust the model parameters based on the question answer and the predicted answer.

[0199] B17. The apparatus as described in B11, further comprising: a data encryption module;

[0200] The data encryption module is used to encrypt the subtask to be processed using a preset data encryption method to obtain the encrypted subtask to be processed; and to allocate the encrypted subtask to the corresponding target computing device.

[0201] The present invention also discloses C18, a task processing device, the device comprising: a memory, a processor, and a task processing program stored in the memory and executable on the processor, the task processing program being configured to implement the steps of the task processing method described above.

[0202] The present invention also discloses D19, a storage medium storing a task processing program, wherein the task processing program, when executed by a processor, implements the steps of the task processing method described above.

[0203] The present invention also discloses E20, a computer program product comprising a computer program that, when executed by a processor, implements the steps of the task processing method described above.

Claims

1. A task processing method, characterized in that, The method described is applied to a private large-scale model, which establishes a dedicated data channel with the AI ​​browser. The method includes: Based on the dedicated data channel, the system receives model configuration requests sent by users through the AI ​​browser; Configure the model according to the model configuration request, and deploy the configured private large model; When a pending task is received from the user through the AI ​​browser via the dedicated data channel, the pending task is split and executed based on the deployed private big model.

2. The method as described in claim 1, characterized in that, The steps of splitting and executing the task to be processed based on the deployed private large model include: Based on the aforementioned privateized large model, the task to be processed is split into several sub-tasks to be processed. Based on the aforementioned privatization model, the sub-tasks to be processed are assigned to the corresponding target computing devices, so that the target computing devices can perform task execution on the sub-tasks to be processed.

3. The method as described in claim 2, characterized in that, The step of allocating the subtasks to be processed to the corresponding target computing devices based on the privatized large model includes: The task inputs and task objectives corresponding to all computing devices are obtained through the aforementioned private large model; The private large model determines the target computing device corresponding to the subtask to be processed from the computing device based on the task input and the task objective; The subtask to be processed is assigned to the target computing device.

4. The method as described in claim 3, characterized in that, The step of determining the target computing device corresponding to the subtask to be processed from the computing devices based on the task input and the task objective using the privatized large model includes: The degree of association between the subtask to be processed and the computing device is determined by the privatized large model based on the task input and the task objective; Determine the task type and task quantity corresponding to the subtask to be processed; The target computing device corresponding to the subtask to be processed is determined from the computing devices based on the degree of association, the task type, and the task volume.

5. The method as described in claim 4, characterized in that, The step of determining the target computing device corresponding to the subtask to be processed from the computing devices based on the correlation degree, the task type, and the task volume includes: The computing devices are sorted according to the degree of correlation to obtain a computing device sequence; Based on the task type, the computing device sequence is filtered to obtain the target computing device sequence; The target computing device corresponding to the subtask to be processed is determined from the target computing device sequence based on the task volume.

6. The method according to any one of claims 2 to 5, characterized in that, After the step of assigning the subtask to be processed to the corresponding target computing device so that the target computing device can process the subtask, the method further includes: The user feedback document is split according to a preset text splitting method to obtain the document splitting results; The user's question and the answer are determined based on the document splitting results; Output a predicted answer based on the user's question; The model parameters are adjusted based on the answer to the question and the predicted answer.

7. A task processing device, characterized in that, The device includes a private large model, and a dedicated data channel is established between the private large model and the AI ​​browser. The device comprises: The request receiving module is used to receive model configuration requests sent by users through the AI ​​browser based on the dedicated data channel; The model deployment module is used to configure the model according to the model configuration request and deploy the configured private large model. The task processing module is used to split and execute the pending task based on the deployed private big model when it receives the pending task sent by the user through the AI ​​browser based on the dedicated data channel.

8. A task processing device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the task processing method as described in any one of claims 1 to 6.

9. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the task processing method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the task processing method as described in any one of claims 1 to 6.

Images