Task processing method, automatic question answering method, and task processing system

Abstract

Embodiments of the present specification provide a task processing method, an automatic question answering method and a task processing system, wherein the task processing method comprises: receiving a task processing request, wherein the task processing request carries to-be-processed data; selecting a target model parameter corresponding to a target requestor from a plurality of candidate model parameters, wherein the candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requestor, and each candidate model parameter corresponds to a different requestor; constructing a task processing model according to the target model parameter and shared model parameters, wherein the shared model parameters are model parameters shared by each requestor in the pre-trained processing model, and the shared model parameters are fixed and unchanged in the training process of the pre-trained processing model; and inputting the to-be-processed data into the task processing model to obtain a task processing result. Without providing a task processing model for each requestor, the model deployment cost and the task processing resource consumption are reduced.

Description

Technical Field

[0001] This specification relates to the field of computer technology, and in particular to a task processing method. One or more embodiments of this specification also relate to an automatic question-answering method, a task processing system, a task processing apparatus, an automatic question-answering apparatus, a computing device, a computer-readable storage medium, and a computer program. Background Technology

[0002] With the development of computer technology, large-scale models have begun to shine, demonstrating extraordinary capabilities in language understanding, generation, interaction, and reasoning, and are widely used in natural language processing fields such as dialogue, translation, and code generation. However, due to the massive number of parameters in large-scale models and the varying needs of different enterprises, how to provide personalized model services to a large number of enterprises has gradually become a research focus.

[0003] Currently, enterprise-specific large-scale models are typically trained using supervised fine-tuning: during the training process, all parameters of the large model are adjusted according to the enterprise's needs to achieve personalized training. However, this approach suffers from significant resource consumption due to the large number of parameters in the large model and the deployment of a separate dedicated large model for each enterprise. This results in the deployment and storage costs of N large models for N enterprises, leading to substantial resource consumption. Therefore, a task processing solution with lower resource consumption is urgently needed. Summary of the Invention

[0004] In view of the above, embodiments of this specification provide a task processing method. One or more embodiments of this specification also relate to an automatic question-answering method, a task processing system, a task processing apparatus, an automatic question-answering apparatus, a computing device, a computer-readable storage medium, and a computer program, to address the technical deficiencies existing in the prior art.

[0005] According to a first aspect of the embodiments of this specification, a task processing method is provided, comprising:

[0006] Receive a task processing request, which carries data to be processed;

[0007] The target model parameters corresponding to the target requester are selected from multiple candidate model parameters. The candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester. Each candidate model parameter corresponds to a different requester.

[0008] Based on the target model parameters and the shared model parameters, a task processing model is constructed. The shared model parameters are the model parameters shared by each requester in the pre-trained processing model. The shared model parameters remain unchanged during the training process of the pre-trained processing model.

[0009] Input the data to be processed into the task processing model to obtain the task processing results.

[0010] According to a second aspect of the embodiments of this specification, an automatic question-answering method is provided, comprising:

[0011] Receive a problem handling request, which carries data to be answered;

[0012] The target model parameters corresponding to the target requester are selected from multiple candidate model parameters. The candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester. Each candidate model parameter corresponds to a different requester.

[0013] Based on the target model parameters and the shared model parameters, a task processing model is constructed. The shared model parameters are the model parameters shared by each requester in the pre-trained processing model. The shared model parameters remain unchanged during the training process of the pre-trained processing model.

[0014] Input the data to be answered into the task processing model to obtain the answer to the question.

[0015] According to a third aspect of the embodiments of this specification, a task processing system is provided, including a service unit and a candidate parameter storage unit;

[0016] Candidate parameter storage unit, used to store multiple candidate model parameters;

[0017] The service unit receives task processing requests, which carry data to be processed; it selects target model parameters corresponding to the requester from multiple candidate model parameters, wherein the candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester, and each candidate model parameter corresponds to a different requester; it constructs a task processing model based on the target model parameters and shared model parameters, wherein the shared model parameters are model parameters shared by all requesters in the pre-trained processing model, and the shared model parameters remain fixed during the training process of the pre-trained processing model; and it inputs the data to be processed into the task processing model to obtain the task processing result.

[0018] According to a fourth aspect of the embodiments of this specification, a task processing apparatus is provided, comprising:

[0019] The first receiving module is configured to receive task processing requests, wherein the task processing requests carry data to be processed.

[0020] The first filtering module is configured to filter out the target model parameters corresponding to the target requester from multiple candidate model parameters. The candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester, and each candidate model parameter corresponds to a different requester.

[0021] The first building module is configured to build a task processing model based on the target model parameters and the shared model parameters. The shared model parameters are the model parameters shared by each requester in the pre-trained processing model. The shared model parameters remain fixed during the training process of the pre-trained processing model.

[0022] The first input module is configured to input the data to be processed into the task processing model and obtain the task processing result.

[0023] According to a fifth aspect of the embodiments of this specification, an automatic question-answering device is provided, comprising:

[0024] The second receiving module is configured to receive problem handling requests, wherein the problem handling requests carry data to be answered;

[0025] The second filtering module is configured to filter out the target model parameters corresponding to the target requester from multiple candidate model parameters. The candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester, and each candidate model parameter corresponds to a different requester.

[0026] The second building module is configured to build a task processing model based on the target model parameters and the shared model parameters. The shared model parameters are the model parameters shared by each requester in the pre-trained processing model. The shared model parameters remain fixed during the training process of the pre-trained processing model.

[0027] The second input module is configured to input the data to be answered into the task processing model to obtain the answer to the question.

[0028] According to a sixth aspect of the embodiments of this specification, a computing device is provided, comprising:

[0029] Memory and processor;

[0030] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the method provided in the first or second aspect above.

[0031] According to a seventh aspect of the embodiments of this specification, a computer-readable storage medium is provided that stores computer-executable instructions, which, when executed by a processor, implement the steps of the methods provided in the first or second aspect described above.

[0032] According to an eighth aspect of the embodiments of this specification, a computer program is provided, wherein when the computer program is executed in a computer, it causes the computer to perform the steps of the method provided in the first or second aspect described above.

[0033] This specification provides a task processing method according to one embodiment, which receives a task processing request carrying data to be processed; selects target model parameters corresponding to the target requester from multiple candidate model parameters, wherein the candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester, and each candidate model parameter corresponds to a different requester; constructs a task processing model based on the target model parameters and shared model parameters, wherein the shared model parameters are model parameters shared by all requesters in the pre-trained processing model, and the shared model parameters remain fixed during the training of the pre-trained processing model; inputs the data to be processed into the task processing model to obtain the task processing result. By keeping the shared model parameters in the pre-trained processing model fixed during the training process using sample sets specified by different requesters, the method can efficiently obtain the candidate model parameters corresponding to each requester. This achieves the goal of using only one shared model parameter in conjunction with different candidate model parameters to serve different requesters, eliminating the need to provide a separate task processing model for each requester, thus reducing model deployment costs and task processing resource consumption. Attached Figure Description

[0034] Figure 1 This is an architecture diagram of a task processing system;

[0035] Figure 2 This is an architecture diagram of a first task processing system provided in one embodiment of this specification;

[0036] Figure 3 This is an architecture diagram of a second task processing system provided in one embodiment of this specification;

[0037] Figure 4 This is an architecture diagram of a third task processing system provided in one embodiment of this specification;

[0038] Figure 5 This is a flowchart illustrating a task processing method provided in one embodiment of this specification;

[0039] Figure 6 This is a flowchart illustrating an automatic question-and-answer method provided in one embodiment of this specification;

[0040] Figure 7 This is a flowchart illustrating the processing procedure of a task processing method provided in one embodiment of this specification.

[0041] Figure 8 This is a schematic diagram of an automatic question-and-answer interface provided in one embodiment of this specification;

[0042] Figure 9 This is a schematic diagram of the structure of a task processing device provided in one embodiment of this specification;

[0043] Figure 10 This is a schematic diagram of the structure of an automatic question-and-answer device provided in one embodiment of this specification;

[0044] Figure 11 This is a structural block diagram of a computing device provided in one embodiment of this specification. Detailed Implementation

[0045] Many specific details are set forth in the following description to provide a full understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar extensions without departing from the spirit of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0046] The terminology used in one or more embodiments of this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the one or more embodiments of this specification. The singular forms “a,” “described,” and “the” as used in one or more embodiments of this specification and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0047] It should be understood that although the terms first, second, etc., may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first may also be referred to as second without departing from the scope of one or more embodiments of this specification, and similarly, second may also be referred to as first. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0048] Furthermore, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties. Moreover, the collection, use and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0049] In one or more embodiments of this specification, a large model refers to a deep learning model with a large number of model parameters, typically containing hundreds of millions, tens of billions, hundreds of billions, trillions, or even tens of trillions of model parameters. A large model can also be called a foundation model. It is pre-trained using large-scale unlabeled corpora to produce a pre-trained model with hundreds of millions of parameters. Such models can adapt to a wide range of downstream tasks and have good generalization ability. Examples include Large Language Models (LLMs) and multi-modal pre-training models.

[0050] In practical applications, large models only require a small number of samples to fine-tune the pre-trained model before they can be applied to different tasks. Large models can be widely used in fields such as Natural Language Processing (NLP) and Computer Vision. Specifically, they can be applied to computer vision tasks such as Visual Question Answering (VQA), Image Captioning (IC), and Image Generation, as well as natural language processing tasks such as text-based sentiment classification, text summarization, and machine translation. The main application scenarios for large models include digital assistants, intelligent robots, search, online education, office software, e-commerce, and intelligent design.

[0051] First, the terms and concepts used in one or more embodiments of this specification will be explained.

[0052] Language model: A language model is an abstract mathematical model of language based on objective facts of language, and it is an important direction in the field of natural language research.

[0053] With the development of computer technology, large-scale models have begun to shine, demonstrating extraordinary capabilities in language understanding, generation, interaction, and reasoning, and are widely used in natural language processing fields such as dialogue, translation, and code generation. However, due to the massive number of parameters in large-scale models and the varying needs of different enterprises, how to provide personalized model services to a large number of enterprises has gradually become a research focus.

[0054] Currently, enterprise-specific large-scale models are typically trained using supervised fine-tuning (SFT): during training, all parameters of the large model are adjusted according to the enterprise's needs to achieve personalized training. However, this approach results in very high training costs due to the extremely large number of parameters in the large model. Furthermore, in practical deployment, deploying a separate dedicated large model for each enterprise would result in the deployment and storage costs of N large models for N enterprises, leading to enormous resource consumption. See also Figure 1 , Figure 1 An architecture diagram of a task processing system is shown, such as Figure 1 As shown, this task processing system is a multi-enterprise, multi-model task processing system. Enterprise A corresponds to task processing model A, enterprise B corresponds to task processing model B, enterprise C corresponds to task processing model C, and enterprise D corresponds to task processing model D.

[0055] In actual production, multiple enterprises often need to deploy multiple large models, resulting in enormous deployment costs. This specification's embodiments propose a training and deployment scheme that utilizes the same large model platform when deploying models for enterprise users. This supports the deployment of different enterprise models based on a unified large model platform, allowing different enterprises to fine-tune their respective model parameters, thereby reducing model deployment costs. See also... Figure 2 , Figure 2 An architecture diagram of a first task processing system provided in one embodiment of this specification is shown, such as... Figure 2 As shown in the embodiments of this specification, the task processing system proposed is a multi-enterprise unified shared model parameter and multiple candidate model parameters task processing system. Enterprise A, Enterprise B, Enterprise C and Enterprise D share the shared model parameters of a pre-trained processing model, and each enterprise corresponds to different candidate model parameters. The candidate model parameters are obtained by training the pre-trained processing model based on the sample set specified by the enterprise. Specifically, Enterprise A corresponds to candidate model parameter A, Enterprise B corresponds to candidate model parameter B, Enterprise C corresponds to candidate model parameter C, and Enterprise D corresponds to candidate model parameter D.

[0056] Specifically, this specification proposes a task processing method that involves receiving a task processing request, wherein the request carries data to be processed; selecting target model parameters corresponding to the target requester from multiple candidate model parameters, wherein the candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester, and each candidate model parameter corresponds to a different requester; constructing a task processing model based on the target model parameters and shared model parameters, wherein the shared model parameters are model parameters shared by all requesters in the pre-trained processing model, and the shared model parameters remain fixed during the training of the pre-trained processing model; and inputting the data to be processed into the task processing model to obtain the task processing result. By keeping the shared model parameters in the pre-trained processing model fixed during the training process using sample sets specified by different requesters, the method efficiently obtains the candidate model parameters corresponding to each requester. This achieves the goal of using only one shared model parameter in conjunction with different candidate model parameters to serve different requesters, eliminating the need to provide a separate task processing model for each requester, thus reducing model deployment costs and task processing resource consumption.

[0057] This specification provides a task processing method, and also relates to an automatic question-and-answer method, a task processing system, a task processing device, an automatic question-and-answer device, a computing device, a computer-readable storage medium, and a computer program, which will be described in detail in the following embodiments.

[0058] See Figure 3 , Figure 3 This specification illustrates an architecture diagram of a second task processing system provided in one embodiment of the present specification. The task processing system may include a candidate parameter storage unit 100 and a service unit 200.

[0059] Candidate parameter storage unit 100 is used to store multiple candidate model parameters;

[0060] Service unit 200 is used to receive task processing requests, wherein the task processing requests carry data to be processed; to select target model parameters corresponding to the target requester from multiple candidate model parameters, wherein the candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester, and the requesters corresponding to each candidate model parameter are different; to construct a task processing model based on the target model parameters and shared model parameters, wherein the shared model parameters are model parameters shared by all requesters in the pre-trained processing model, and the shared model parameters remain unchanged during the training of the pre-trained processing model; and to input the data to be processed into the task processing model to obtain the task processing result.

[0061] In one optional embodiment of this specification, the task processing system further includes a shared parameter storage unit 300; the shared parameter storage unit 300 is used to store shared model parameters; the service unit 200 is further used to obtain shared model parameters from the shared parameter storage unit 300.

[0062] By applying the scheme of the embodiments of this specification, in the process of training the pre-trained processing model using the sample sets specified by different requesters, the shared model parameters in the pre-trained processing model are kept unchanged, thereby efficiently obtaining the candidate model parameters corresponding to each requester. This realizes that only one shared model parameter is used in conjunction with different candidate model parameters to serve different requesters, without having to provide a separate task processing model for each requester, thus reducing the model deployment cost and task processing resource consumption.

[0063] See Figure 4 , Figure 4 This specification illustrates an architecture diagram of a third task processing system according to an embodiment. The task processing system may include multiple clients 400 and a service unit 200. The multiple clients 400 can establish communication connections through the service unit 200. In a task processing scenario, the service unit 200 is used to provide task processing services between the multiple clients 400. The multiple clients 400 can act as either senders or receivers, communicating through the service unit 200.

[0064] The target requester can interact with the service unit 200 through client 400 to receive data sent by other clients 400, or send data to other clients 400, etc. In a task processing scenario, the target requester can publish a data stream to the service unit 200 through client 400, and the service unit 200 can generate a task processing result based on the data stream and push the task processing result to other clients that have established communication.

[0065] In this system, the client 400 and the service unit 200 establish a connection via a network. The network provides a communication link between the client 400 and the service unit 200. The network can include various connection types, such as wired or wireless communication links or fiber optic cables. The data transmitted by the client 400 may need to undergo encoding, transcoding, compression, or other processing before being published to the service unit 200.

[0066] Client 400 can be a browser, an app (application), a web application such as an H5 (HyperText Markup Language 5) application, a lightweight application (also known as a mini-program), or a cloud application. Client 400 can be developed based on the software development kit (SDK) of the corresponding service provided by service unit 200, such as a real-time communication (RTC) SDK. Client 400 can be deployed in electronic devices and depends on the device or certain apps on the device to run. Electronic devices may have displays and support information browsing, such as personal mobile terminals like mobile phones, tablets, and personal computers. Various other types of applications can also be configured in electronic devices, such as human-computer interaction applications, model training applications, text processing applications, web browser applications, shopping applications, search applications, instant messaging tools, email clients, and social media platform software.

[0067] Service unit 200 may include servers providing various services, such as servers providing communication services to multiple clients, servers supporting backend training of models used on clients, and servers processing data sent by clients. It should be noted that service unit 200 can be implemented as a distributed server cluster composed of multiple servers, or as a single server. The server can also be a server in a distributed system, or a server integrated with blockchain. The server can also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.

[0068] It is worth noting that the task processing methods provided in the embodiments of this specification are generally executed by the service unit. However, in other embodiments of this specification, the client may also have similar functions to the service unit, thereby executing the task processing methods provided in the embodiments of this specification. In other embodiments, the task processing methods provided in the embodiments of this specification may also be executed jointly by the client and the service unit.

[0069] See Figure 5 , Figure 5 This specification shows a flowchart of a task processing method according to an embodiment, which specifically includes the following steps:

[0070] Step 502: Receive a task processing request, wherein the task processing request carries data to be processed.

[0071] In one or more embodiments of this specification, a task processing request can be received, and task processing can be performed based on the data to be processed carried in the task processing request to obtain the task processing result.

[0072] Specifically, the data to be processed is the object requested to be processed by the task processing request. The data to be processed can be data from different scenarios, such as data from a meeting scenario, data from an e-commerce scenario, etc. The data to be processed can also be different types of data, such as text data, audio data, image data, etc.

[0073] In practical applications, there are various ways to receive task processing requests, and the specific method should be selected according to the actual situation. This specification does not limit the methods used in this embodiment. In one possible implementation, this specification can receive task processing requests sent by the target requester through a client. In another possible implementation, this specification can receive task processing requests sent periodically by the client.

[0074] Step 504: Select the target model parameters corresponding to the target requester from multiple candidate model parameters. The candidate model parameters are obtained by training the pre-trained processing model based on the sample set specified by the requester, and each candidate model parameter corresponds to a different requester.

[0075] In one or more embodiments of this specification, after receiving a task processing request, the target model parameter corresponding to the target requester can be further selected from multiple candidate model parameters.

[0076] Specifically, the target requester refers to the object that sends the task processing request. The target requester can be an individual user or an enterprise user. The candidate model parameters are obtained by training the pre-trained processing model based on the sample set specified by the requester. The pre-trained processing model can be trained using Parameter-Efficient Fine-Tuning (PEFT). PEFT is a training technique that freezes most of the parameters in the pre-trained processing model, training only a small subset of parameters while ensuring that the training effect is essentially the same as full training. In other words, the candidate model parameters are obtained by training the pre-trained processing model with frozen shared model parameters based on the sample set specified by the requester. PEFT methods include, but are not limited to, LoRA, Adapter, P-tuning, and prefix-tuning. Preferably, in the embodiments of this specification, the candidate model parameters can be obtained by training the pre-trained processing model using the LoRA method, that is, by learning a low-rank matrix with small parameters to approximate the parameter updates of the model weight matrix, and only optimizing the low-rank matrix parameters during training.

[0077] The target model parameters refer to the model parameters obtained by training the pre-trained processing model on a sample set specified by the target requester. The pre-trained processing model is a deep learning model, which can be obtained by training a large model with all parameters based on the training set. Different requesters correspond to different candidate model parameters; for example, requester A corresponds to candidate model parameter A, requester B corresponds to candidate model parameter B, and requester C corresponds to candidate model parameter C.

[0078] It is worth noting that if the target model parameters corresponding to the target requester are not included among the multiple candidate model parameters, the sample set specified by the target requester can be obtained, and the target model parameters can be generated by training the pre-trained processing model using the sample set specified by the target requester.

[0079] In practical applications, there are various ways to select the target model parameters corresponding to the target requester from multiple candidate model parameters. The specific method should be selected according to the actual situation. This specification does not limit the specific methods used in this embodiment.

[0080] In one possible implementation of this specification, the target model parameter corresponding to the target requester can be directly selected from multiple candidate model parameters based on the requester information of the target requester. For example, if the requester information of the target requester is number A, then the requester information can be matched with the candidate model parameter number, and candidate model parameter A can be selected as the target model parameter from candidate model parameter A, candidate model parameter B, and candidate model parameter C.

[0081] In another possible implementation of this specification, a parameter filtering matrix can be generated based on the requester information of the target requester, and the target model parameter can be further filtered from multiple candidate model parameters using the parameter filtering matrix. That is, the above-mentioned filtering of the target model parameter corresponding to the target requester from multiple candidate model parameters may include the following steps:

[0082] Obtain the requester information of the target requester;

[0083] Construct a parameter filtering matrix based on the requester's information;

[0084] Based on the parameter selection matrix, the target model parameters are selected from multiple candidate model parameters.

[0085] Specifically, requester information refers to information that uniquely identifies the target requester, including but not limited to requester identification number, Internet Protocol address, etc. In the parameter filtering matrix, the value in the row corresponding to the target requester is 1, and the values ​​in the other rows are 0. Rows with values ​​of 1 in the parameter filtering matrix retain the target model parameters, while candidate model parameters corresponding to rows with values ​​of 0 are eliminated. The parameter filtering matrix is ​​used to filter the target model parameters from multiple candidate model parameters.

[0086] It should be noted that there are multiple ways to obtain the requester information of the target requester, and the specific method should be selected according to the actual situation. This specification does not impose any limitations on this method in the embodiments. In one possible implementation of this specification, the target requester information can be obtained from a database. In another possible implementation of this specification, the target requester information can be received from the client.

[0087] Furthermore, after obtaining the requester information of the target requester, the following steps are taken: First, an original parameter filtering matrix can be obtained, where each row corresponds to a requester. Then, based on the requester information, the target row corresponding to the target requester is determined from the original parameter filtering matrix, and the value in the target row is set to 1, while the values ​​in other rows are set to 0, thus obtaining the parameter filtering matrix. Finally, the parameter filtering matrix is ​​multiplied by multiple candidate model parameters to filter out the target model parameters corresponding to the target requester from the multiple candidate model parameters.

[0088] The scheme described in this specification involves obtaining the requester information of the target requester; constructing a parameter filtering matrix based on the requester information; and filtering the target model parameters from multiple candidate model parameters based on the parameter filtering matrix. By multiplying the parameter filtering matrix with multiple candidate model parameters, the target model parameters are accurately filtered from multiple candidate model parameters.

[0089] In one optional embodiment of this specification, multiple candidate model parameters can be pre-generated, and the target model parameter corresponding to the target requester can be further selected from the pre-generated multiple candidate model parameters. That is, before selecting the target model parameter corresponding to the target requester from the multiple candidate model parameters, the following steps may be included:

[0090] Retrieve sample sets specified by multiple requesters;

[0091] For the first requester, a pre-trained processing model is trained based on the sample set specified by the first requester to obtain the pre-trained model parameters corresponding to the first requester, wherein the first requester is any one of multiple requesters;

[0092] Candidate model parameters for each requester are selected from the pre-trained model parameters for each requester.

[0093] Specifically, the sample sets specified by multiple requesters can be the same or different. Pre-trained model parameters refer to the model parameters of the pre-trained processing model obtained after training the pre-trained processing model using the sample set specified by the first requester. Pre-trained model parameters include shared model parameters and candidate model parameters corresponding to the first requester. The shared model parameters remain fixed during the training of the pre-trained processing model.

[0094] In practical applications, there are various ways to obtain sample sets specified by multiple requesters, and the specific method should be selected according to the actual situation. This specification does not impose any limitations on these methods in its embodiments. In one possible implementation, the sample sets specified by each requester can be read from other databases or data acquisition devices. In another possible implementation, sample sets sent by each requester through a client can be received.

[0095] It should be noted that after obtaining the pre-trained model parameters corresponding to the first requester, the shared model parameters and candidate model parameters can be separated from the pre-trained model parameters through tensor-related functions, and then stored separately so that the corresponding candidate model parameters can be called later based on the requester information.

[0096] The scheme described in this specification involves obtaining sample sets specified by multiple requesters. For a first requester, a pre-trained processing model is trained based on the sample set specified by the first requester to obtain pre-trained model parameters corresponding to the first requester. The first requester can be any one of the multiple requesters. Candidate model parameters corresponding to each requester are then selected from the pre-trained model parameters. During the training of the pre-trained processing model using sample sets specified by different requesters, the shared model parameters in the pre-trained processing model remain unchanged, thereby efficiently obtaining the candidate model parameters corresponding to each requester. This achieves the goal of serving different requesters using only one shared model parameter combined with different candidate model parameters.

[0097] Furthermore, the above-mentioned process of training a pre-trained processing model based on the sample set specified by the first requester to obtain the pre-trained model parameters corresponding to the first requester may include the following steps:

[0098] Extract first sample data from the sample set specified by the first requester, wherein the first sample data is any sample data in the sample set and carries a first sample label;

[0099] Input the first sample data into the pre-trained processing model to obtain the prediction result of the first sample;

[0100] Based on the prediction result of the first sample and the label of the first sample, adjust the model parameters other than the shared model parameters in the pre-trained processing model, and return to the step of extracting the first sample data from the sample set specified by the first requester, until the first preset stopping condition is reached, and obtain the pre-trained model parameters corresponding to the first requester.

[0101] It should be noted that the pre-trained processing model is trained using supervised training. The first sample data, carrying the first sample label, serves as the true label for the pre-trained processing model during training; it can also be understood as the training objective, used to guide the training process. The sample set includes multiple sample data points, which can be data from different scenarios, such as data from a meeting or e-commerce scenario. The sample data can also be of different types, such as audio data or text data. It should be noted that if the sample data is of a type other than text, it can be converted to obtain text-based sample data.

[0102] In practical applications, after inputting the first sample data into the pre-trained processing model and obtaining the first sample prediction result, the first loss value can be calculated based on the first sample prediction result and the first sample label. The model parameters other than the shared model parameters in the pre-trained processing model can be adjusted based on the first loss value, and the process can return to the step of extracting the first sample data from the sample set specified by the first requester until the first preset stopping condition is reached, thereby obtaining the pre-trained model parameters corresponding to the first requester.

[0103] In one possible implementation of this specification, the first preset stopping condition includes a first loss value being less than or equal to a first preset threshold. After calculating the first loss value based on the first sample prediction result and the first sample label, the first loss value is compared with the first preset threshold.

[0104] Specifically, if the first loss value is greater than the first preset threshold, it indicates that the difference between the first sample prediction result and the first sample label is large, and the pre-trained processing model has poor prediction ability for the first sample prediction result. At this time, the model parameters other than the shared model parameters in the pre-trained processing model can be adjusted, and the step of extracting the first sample data from the sample set specified by the first requester can be returned to continue training the pre-trained processing model until the first loss value is less than or equal to the first preset threshold, indicating that the difference between the first sample prediction result and the first sample label is small, the first preset stopping condition is reached, and the pre-trained model parameters corresponding to the first requester are obtained.

[0105] In another possible implementation of this specification, in addition to comparing the magnitude of the first loss value and the first preset threshold, a preset number of iterations can also be used to determine whether the pre-trained model parameters corresponding to the first requester are obtained.

[0106] Specifically, if the first loss value is greater than the first preset threshold, the model parameters other than the shared model parameters in the pre-trained processing model are adjusted, and the step of extracting the first sample data from the sample set specified by the first requester is returned to be executed. The pre-trained processing model is then trained until the first preset number of iterations is reached, the iteration is stopped, and the pre-trained model parameters corresponding to the first requester are obtained. The first preset threshold and the first preset number of iterations are selected according to the actual situation, and the embodiments in this specification do not limit them in any way.

[0107] In practical applications, there are many functions for calculating the first loss value, such as the cross-entropy loss function, the L1 norm loss function, the maximum loss function, the mean squared error loss function, and the log loss function. The specific function to be selected depends on the actual situation, and the embodiments in this specification do not impose any limitations on this.

[0108] The scheme implemented in this specification involves calculating a first loss value based on the prediction result and label of the first sample. This first loss value is then compared to a first preset stopping condition. If the first preset stopping condition is not met, the pre-trained processing model continues to be trained until it is reached, thus completing the training and obtaining the pre-trained model parameters corresponding to the first requester. By continuously adjusting the model parameters other than the shared model parameters in the pre-trained processing model, the final pre-trained model parameters can be made more accurate.

[0109] It's worth noting that by fixing the shared model parameters in the pre-trained model, only about 1% of the parameters need to be trained to achieve good training results. Furthermore, due to the reduced number of parameters to be adjusted during training, the required GPU and storage space are significantly reduced. For example, training a 13B model requires at least two 80GB GPUs to train all parameters of the pre-trained model. Also, storing 10 checkpoints during training requires at least 10 * 20GB = 200GB of storage space, which is extremely resource-intensive. However, by efficiently fine-tuning the parameters using fixed shared model parameters, only one 80GB GPU is needed to train the pre-trained model. Moreover, storing 10 checkpoints during training only requires 10 * 20GB * 1% = 2GB of storage space, resulting in a significant reduction in storage space.

[0110] In one optional embodiment of this specification, after obtaining the pre-trained model parameters corresponding to the first requester, the pre-trained model parameters can be sent to the first requester. The first requester can build a model locally based on the pre-trained model parameters and determine whether the current pre-trained model parameters meet the requirements. If they do not meet the requirements, the first requester can send sample adjustment data to continue training the pre-trained model. That is, after obtaining the pre-trained model parameters corresponding to the first requester by training the pre-trained processing model based on the sample set specified by the first requester, the following steps may also be included:

[0111] Send the pre-trained model parameters to the first requester;

[0112] Receive sample adjustment data sent by the first requester based on the parameters of the pre-trained model, and adjust the parameters of the pre-trained processing model according to the sample adjustment data.

[0113] Specifically, sample adjustment data can be sample data outside the sample set, or it can be the number of model training iterations during parameter adjustment. Sample adjustment data is used to adjust model parameters other than the shared model parameters in the pre-trained model. If the sample adjustment data is sample data outside the sample set, it can be data from different scenarios, such as data from a meeting scenario or e-commerce scenario. Sample adjustment data can also be different types of data, such as audio data or text data. It should be noted that if the sample adjustment data is of a type other than text data, it can be converted to obtain text-based sample adjustment data.

[0114] In practical applications, when sending pre-trained model parameters to the first requester, the pre-trained model parameters can be sent to the first requester through a client. Furthermore, sample adjustment data sent by the first requester through a client can be received, wherein the sample adjustment data is obtained based on the pre-trained model parameters.

[0115] It should be noted that the method of adjusting the parameters of the pre-trained processing model based on the sample adjustment data is the same as the implementation method of "training the pre-trained processing model based on the sample set specified by the first requester" described above, and will not be repeated in the embodiments of this specification.

[0116] The scheme of the embodiments of this specification is used to send the pre-trained model parameters to the first requester; receive the sample adjustment data sent by the first requester based on the pre-trained model parameters, and adjust the parameters of the pre-trained processing model according to the sample adjustment data, so that the candidate model parameters corresponding to the first requester are more accurate, and further ensure the accuracy of the task processing model.

[0117] In one optional embodiment of this specification, the pre-trained model parameters include shared model parameters; after filtering out the candidate model parameters corresponding to each requester from the pre-trained model parameters corresponding to each requester, the following steps may also be included:

[0118] Store the candidate model parameters corresponding to each requester into the candidate parameter storage unit respectively;

[0119] Store the shared model parameters in the shared parameter storage unit.

[0120] Specifically, the candidate parameter storage unit and the shared parameter storage unit can be different databases or different storage sub-units in the same database.

[0121] It should be noted that when storing the candidate model parameters corresponding to each requester into the candidate parameter storage unit, the candidate model parameters can be stored in a columnar storage manner according to the requester information of each requester.

[0122] Applying the scheme of the embodiments in this specification, after selecting candidate model parameters for each requester from the pre-trained model parameters corresponding to each requester, the candidate model parameters for each requester are stored in the candidate parameter storage unit; the shared model parameters are stored in the shared parameter storage unit. This facilitates the subsequent direct extraction of the target model parameters corresponding to the target requester from the candidate parameter storage unit, and the extraction of the shared model parameters from the shared parameter storage unit to construct the task processing model.

[0123] In practical applications, shared model parameters can be directly stored in the shared parameter storage unit. Furthermore, since the shared model parameters are model parameters shared by all requesters and are fixed, if the shared model parameters are already stored in the shared parameter storage unit, there is no need to store them again. That is, before storing the shared model parameters in the shared parameter storage unit, the following steps may also be included:

[0124] Determine whether the shared parameter storage unit contains shared model parameters;

[0125] Storing shared model parameters into a shared parameter storage unit may include the following steps:

[0126] If the shared model parameters are not included in the shared parameter storage unit, the shared model parameters are stored in the shared parameter storage unit.

[0127] It should be noted that there are multiple ways to determine whether a shared parameter storage unit includes shared model parameters, and the appropriate method should be selected based on the actual situation. This specification does not impose any limitations on this method in the embodiments.

[0128] In one possible implementation of this specification, it can be identified whether the shared parameter storage unit is empty. If the shared parameter storage unit is empty, it means that the shared parameter storage unit does not include the shared model parameters, and the shared model parameters can be stored in the shared parameter storage unit. If the shared parameter storage unit is not empty, it means that the shared parameter storage unit includes the shared model parameters, and there is no need to store the shared model parameters in the shared parameter storage unit.

[0129] In another possible implementation of this specification, since the data stored in the shared parameter storage unit may not be the shared model parameters if the shared parameter storage unit is not empty, the stored data in the shared parameter storage unit can be obtained and compared with the shared model parameters. If the stored data and the shared model parameters are different, it means that the shared parameter storage unit does not include the shared model parameters, and the shared model parameters can be stored in the shared parameter storage unit. If the stored data and the shared model parameters are the same, it means that the shared parameter storage unit includes the shared model parameters, and there is no need to store the shared model parameters in the shared parameter storage unit.

[0130] The scheme implemented in the embodiments of this specification determines whether the shared parameter storage unit includes shared model parameters; if the shared parameter storage unit does not include shared model parameters, the shared model parameters are stored in the shared parameter storage unit. This avoids redundant storage of shared model parameters and reduces the occupancy rate of the shared parameter storage unit.

[0131] Step 506: Construct a task processing model based on the target model parameters and the shared model parameters. The shared model parameters are the model parameters shared by each requester in the pre-trained processing model. The shared model parameters remain fixed during the training process of the pre-trained processing model.

[0132] In one or more embodiments of this specification, after receiving a task processing request and filtering out the target model parameters corresponding to the target requester from multiple candidate model parameters, a task processing model can be further constructed based on the target model parameters and shared model parameters.

[0133] Specifically, shared model parameters refer to parameters that remain unchanged during the training of the pre-trained processing model based on the sample set specified by the requester; they can also be understood as frozen parameters. Since the target model parameters are obtained by training the pre-trained processing model based on the sample set specified by the target requester, the task processing model can be understood as a processing model exclusive to the target requester.

[0134] It should be noted that when constructing a task processing model based on the target model parameters and the shared model parameters, the target model parameters and the shared model parameters can be concatenated to obtain the model parameters of the task processing model.

[0135] Step 508: Input the data to be processed into the task processing model to obtain the task processing results.

[0136] In one or more embodiments of this specification, a task processing request is received, target model parameters corresponding to the target requester are selected from multiple candidate model parameters, and a task processing model is constructed based on the target model parameters and shared model parameters. Further, the data to be processed can be input into the task processing model to obtain the task processing result.

[0137] By applying the scheme of the embodiments of this specification, in the process of training the pre-trained processing model using the sample sets specified by different requesters, the shared model parameters in the pre-trained processing model are kept unchanged, thereby efficiently obtaining the candidate model parameters corresponding to each requester. This realizes that only one shared model parameter is used in conjunction with different candidate model parameters to serve different requesters, without having to provide a separate task processing model for each requester, thus reducing the model deployment cost and task processing resource consumption.

[0138] In one optional embodiment of this specification, after inputting the data to be processed into the task processing model and obtaining the task processing result, the following steps may be further included:

[0139] Send the task processing result to the client so that the client can display the task processing result to the target requester;

[0140] The system receives adjustment information sent by the target requester through the client and adjusts the target model parameters according to the adjustment information, which is obtained based on the task processing results.

[0141] It should be noted that after inputting the data to be processed into the task processing model and obtaining the task processing result, the task processing result can be sent to the client, and the model parameters of the task processing model can also be sent to the client.

[0142] In practical applications, there are various ways for a client to display the task processing result to the target requester, and the specific method chosen depends on the actual situation. This specification does not impose any limitations on this approach. In one possible implementation, the client may only display the task processing result to the target requester. In another possible implementation, the client may simultaneously display both the task processing result and the model parameters of the task processing model to the target requester.

[0143] It is worth noting that after obtaining the task processing result, the result can be quality checked. If the quality check passes, the result is sent to the client; if the quality check fails, the task processing is restarted.

[0144] Furthermore, after the client displays the task processing results to the target requester, the target requester can process the data independently based on the task processing results, or adjust the task processing results displayed by the client to generate adjustment information. The client then sends the adjustment information to the task processing system so that the task processing system can adjust the target model parameters based on the adjustment information. The adjustment information may include the data to be processed and the adjusted task processing results.

[0145] The solution implemented in this specification sends task processing results to the client so that the client can display the task processing results to the target requester; it receives adjustment information sent by the target requester through the client and adjusts the target model parameters according to the adjustment information, so that the target requester can accurately obtain the task processing results, increases the interaction between the client and the target requester, and improves the target requester's satisfaction.

[0146] In one optional embodiment of this specification, before filtering out the target model parameters corresponding to the target requester from multiple candidate model parameters, the following steps may be included:

[0147] Obtain the training set, which includes multiple training data sets, each carrying a training label.

[0148] Extract the first training data from the training set, where the first training data is any one of the data in the training set;

[0149] Input the first training data into the original processing model to obtain the first training prediction result;

[0150] Based on the first training prediction result and the first training label carried by the first training data, the original processing model is trained to obtain the pre-trained processing model.

[0151] It should be noted that the original processing model can be a deep learning model for natural language processing. The original processing model is trained using supervised training. The training labels carried by the training data are the true labels from the training process of the original processing model; these training labels can also be understood as training objectives, used to guide the training process. The training set includes multiple training data sets, which can be data from different scenarios, such as data from a meeting or e-commerce scenario. The training data can also be different types of data, such as audio data or text data. It should be noted that if the training data is not text-based, it can be converted to obtain text-based training data.

[0152] There are multiple ways to obtain the training set, and the specific method should be selected according to the actual situation. This specification does not limit the methods used in this embodiment. In one possible implementation, multiple training data sets can be read from other data acquisition devices or databases, and a training set can be constructed based on these multiple training data sets. In another possible implementation, a large amount of training data can be manually input, and a training set can be constructed based on this manually input training data.

[0153] In practical applications, after inputting the first training data into the original processing model and obtaining the first training prediction result, the second loss value can be calculated based on the first training prediction result and the first training label. The model parameters of the original processing model are adjusted based on the second loss value, and the process returns to the step of extracting the first training data from the training set until the second preset stopping condition is reached, thus obtaining the pre-trained processing model that has been trained.

[0154] In one possible implementation of this specification, the second preset stopping condition includes a second loss value being less than or equal to a second preset threshold. After calculating the second loss value based on the first training prediction result and the first training label, the second loss value is compared with the second preset threshold.

[0155] Specifically, if the second loss value is greater than the second preset threshold, it indicates that the difference between the first training prediction result and the first training label is large, and the original processing model has poor prediction ability for the first training prediction result. At this time, the model parameters of the original processing model can be adjusted, and the step of extracting the first training data from the training set can be returned to continue training the original processing model until the second loss value is less than or equal to the second preset threshold, indicating that the difference between the first training prediction result and the first training label is small, and the second preset stopping condition is reached, thus obtaining the pre-trained processing model that has been trained.

[0156] In another possible implementation of this specification, in addition to comparing the relationship between the second loss value and the second preset threshold, a preset number of iterations can also be used to determine whether the pre-trained processing model has been trained.

[0157] Specifically, if the second loss value is greater than the second preset threshold, the model parameters of the original processing model are adjusted, and the step of extracting the first training data from the training set is returned to continue training the original processing model until the second preset number of iterations is reached, the iteration is stopped, and the pre-trained processing model is obtained. The second preset threshold and the second preset number of iterations are selected according to the actual situation, and the embodiments in this specification do not limit them in any way.

[0158] In practical applications, there are many functions for calculating the second loss value, such as the cross-entropy loss function, the L1 norm loss function, the maximum loss function, the mean squared error loss function, the log loss function, etc. The specific function to be selected depends on the actual situation, and the embodiments in this specification do not impose any limitations on this.

[0159] The scheme implemented in this specification involves calculating a second loss value based on the first training prediction result and the first training label. This second loss value is then compared to a second preset stopping condition. If the second preset stopping condition is not met, training of the original processing model continues until the condition is met, resulting in a pre-trained processing model. By continuously adjusting the model parameters of the original processing model, the pre-trained processing model can be made more accurate.

[0160] In the embodiments of this specification, the above-described task processing method has various application scenarios, such as data recommendation scenarios, intelligent question answering scenarios, etc. The following description, in conjunction with the appendix... Figure 6 Taking the application of the task processing method provided in this specification in an intelligent question-answering scenario as an example, the task processing method will be further explained. See [link to documentation]. Figure 6 , Figure 6 This specification shows a flowchart of an automatic question-answering method according to an embodiment, which specifically includes the following steps:

[0161] Step 602: Receive a problem handling request, wherein the problem handling request carries data to be answered.

[0162] Step 604: Select the target model parameters corresponding to the target requester from multiple candidate model parameters. The candidate model parameters are obtained by training the pre-trained processing model based on the sample set specified by the requester. Each candidate model parameter corresponds to a different requester.

[0163] Step 606: Construct a task processing model based on the target model parameters and the shared model parameters. The shared model parameters are the model parameters shared by each requester in the pre-trained processing model. The shared model parameters remain fixed during the training process of the pre-trained processing model.

[0164] Step 608: Input the data to be answered into the task processing model to obtain the answer to the question.

[0165] It should be noted that the data to be answered is also the data to be processed, and the answer to the question is also the task processing result. The implementation methods of steps 602 to 608 are the same as those of steps 502 to 508 described above, and will not be repeated in the embodiments of this specification.

[0166] By applying the scheme of the embodiments of this specification, in the process of training the pre-trained processing model using the sample sets specified by different requesters, the shared model parameters in the pre-trained processing model are kept unchanged, thereby efficiently obtaining the candidate model parameters corresponding to each requester. This enables the use of only one shared model parameter in conjunction with different candidate model parameters to serve different requesters, without the need to provide a separate task processing model for each requester, thus reducing the model deployment cost and the consumption of automatic question answering resources.

[0167] See Figure 7 , Figure 7 This specification illustrates a flowchart of a task processing method according to an embodiment, which specifically includes the following steps:

[0168] Step 702: Obtain the sample sets specified by each of the multiple requesters.

[0169] Step 704: For the first requester, train a pre-trained processing model based on the sample set specified by the first requester to obtain the pre-trained model parameters corresponding to the first requester, wherein the first requester is any one of multiple requesters.

[0170] Step 706: Select candidate model parameters for each requester from the pre-trained model parameters corresponding to each requester.

[0171] Step 708: Store the candidate model parameters corresponding to each requester into the candidate parameter storage unit respectively.

[0172] Step 710: Store the shared model parameters from the pre-trained model parameters into the shared parameter storage unit.

[0173] Step 712: Receive a task processing request, wherein the task processing request carries the data to be processed.

[0174] Step 714: Select the target model parameters corresponding to the target requester from the multiple candidate model parameters stored in the candidate parameter storage unit. The candidate model parameters are obtained by training the pre-trained processing model based on the sample set specified by the requester, and the requesters corresponding to each candidate model parameter are different.

[0175] Step 716: Construct a task processing model based on the target model parameters and the shared model parameters in the shared parameter storage unit. The shared model parameters are the model parameters shared by each requester in the pre-trained processing model. The shared model parameters remain fixed during the training process of the pre-trained processing model.

[0176] Step 718: Input the data to be processed into the task processing model to obtain the task processing results.

[0177] It should be noted that the implementation methods of steps 702 to 718 are the same as those of steps 502 to 508 described above, and will not be repeated in the embodiments of this specification.

[0178] By applying the scheme of the embodiments of this specification, in the process of training the pre-trained processing model using the sample sets specified by different requesters, the shared model parameters in the pre-trained processing model are kept unchanged, thereby efficiently obtaining the candidate model parameters corresponding to each requester. This realizes that only one shared model parameter is used in conjunction with different candidate model parameters to serve different requesters, without having to provide a separate task processing model for each requester, thus reducing the model deployment cost and task processing resource consumption.

[0179] The task processing method proposed in the embodiments of this specification utilizes efficient parameter fine-tuning to train dedicated parameters for a specified enterprise. During the deployment phase, using the same pre-trained processing model as a base, it dynamically switches the corresponding dedicated parameters for different enterprises to achieve the goal of serving different enterprises. This achieves the goal of simultaneously serving multiple enterprises using only one large model with multiple dedicated parameters, significantly reducing deployment and storage costs. The task processing method mainly includes three parts: efficient training, parameter splitting, storage and assembly, and efficient deployment architecture. Efficient training includes full parameter training and efficient parameter fine-tuning.

[0180] Full parameter training: Obtain the training set, extract the first training data from the training set, input the first training data into the original processing model, obtain the first training prediction result, and train the original processing model based on the first training prediction result and the first training label carried by the first training data to obtain the pre-trained processing model.

[0181] It should be noted that during the training phase of the original processing model, after preparing the training data, supervised fine-tuning of all parameters of the original processing model can be performed. The advantage of this method is that by adjusting all parameters of the original processing model, it can better fit the training data and achieve better training results. However, due to the large number of parameters in the original processing model, full parameter training consumes a significant amount of GPU resources, and storing the trained checkpoint files also requires a large amount of storage space. Therefore, this specification proposes a training method for efficient parameter fine-tuning after the original processing model has been trained.

[0182] Efficient parameter fine-tuning: Obtain sample sets specified by multiple requesters. For the first requester, train a pre-trained processing model based on the sample set specified by the first requester to obtain the pre-trained model parameters corresponding to the first requester. Select candidate model parameters corresponding to each requester from the pre-trained model parameters corresponding to each requester.

[0183] Parameter Splitting, Storage, and Assembly: In actual production, to address the challenges of efficient training and deployment across multiple enterprises, this specification's embodiments, in addition to developing algorithms for efficient parameter fine-tuning, also include the separate storage of candidate model parameters and parameter assembly during model deployment. Specifically, candidate model parameters corresponding to each requester are stored separately in candidate parameter storage units, while shared model parameters are stored in shared parameter storage units.

[0184] It's worth noting that parameter splitting, storage, and assembly can be achieved through PyTorch's underlying tensor operations. During parameter assembly, shared model parameters and target model parameters can be loaded separately. Then, the shared and target model parameters are multiplied by the input matrix, followed by addition, to complete the task.

[0185] Efficient Deployment Architecture: For training tasks targeting N enterprises, a single pre-trained processing model can serve as the foundation, with N smaller candidate model parameters and the shared model parameters of the foundation model loaded into GPU memory. Compared to deploying N different task processing models, this approach effectively reduces GPU consumption by N times. When used online, the corresponding smaller candidate model parameters can be automatically switched based on requests from different enterprises, forming a temporary task processing model along with the shared model parameters to handle requests and generate answers.

[0186] The solution implemented in this specification effectively solves the resource consumption problem when training and deploying large models by multiple enterprises by using efficient parameter tuning. By simultaneously combining and deploying the shared model parameters in the pre-trained model of the base and the candidate model parameters of multiple enterprises, it is possible to deploy the model with only 1.x times the base model resources, which greatly reduces the resource consumption problem when deploying collaboratively by multiple enterprises.

[0187] See Figure 8 , Figure 8 This diagram illustrates an automatic question-and-answer interface according to an embodiment of this specification. The automatic question-and-answer interface is divided into a question processing request input interface and a response result display interface. The question processing request input interface includes a question processing request input box, an "OK" control, and a "Cancel" control. The response result display interface includes a response result display box.

[0188] The requesting party inputs a problem-solving request through the problem-solving request input box displayed on the client. This request includes the data to be answered. Upon clicking the "OK" button, the service unit receives the request and selects the target model parameters corresponding to the requesting party from multiple candidate model parameters. These candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requesting party, and each candidate model parameter corresponds to a different requesting party. Based on the target model parameters and shared model parameters, a task processing model is constructed. The shared model parameters are those shared by all requesting parties in the pre-trained processing model and remain fixed during training. The data to be answered is input into the task processing model to obtain the problem answer result, which is then sent to the client. The client displays the problem answer result in the answer result display box.

[0189] In practical applications, the target requester may operate the control in any way, including clicking, double-clicking, touching, hovering the mouse, swiping, long-pressing, voice control, or shaking, depending on the actual situation. This specification does not limit this in any way.

[0190] Corresponding to the above-described task processing method embodiments, this specification also provides embodiments of a task processing apparatus. Figure 9 A schematic diagram of a task processing apparatus according to one embodiment of this specification is shown. Figure 9 As shown, the device includes:

[0191] The first receiving module 902 is configured to receive a task processing request, wherein the task processing request carries data to be processed.

[0192] The first screening module 904 is configured to select the target model parameters corresponding to the target requester from multiple candidate model parameters. The candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester, and each candidate model parameter corresponds to a different requester.

[0193] The first construction module 906 is configured to construct a task processing model based on the target model parameters and the shared model parameters. The shared model parameters are the model parameters shared by each requester in the pre-trained processing model. The shared model parameters remain fixed during the training process of the pre-trained processing model.

[0194] The first input module 908 is configured to input the data to be processed into the task processing model to obtain the task processing result.

[0195] Optionally, the device further includes: a first training module configured to acquire sample sets specified by multiple requesters respectively; for the first requester, to train a pre-trained processing model based on the sample sets specified by the first requester to obtain pre-trained model parameters corresponding to the first requester, wherein the first requester is any one of the multiple requesters; and to select candidate model parameters corresponding to each requester from the pre-trained model parameters corresponding to each requester.

[0196] Optionally, the first training module is further configured to extract first sample data from a sample set specified by the first requester, wherein the first sample data is any sample data in the sample set and carries a first sample label; input the first sample data into a pre-trained processing model to obtain a first sample prediction result; adjust the model parameters other than the shared model parameters in the pre-trained processing model according to the first sample prediction result and the first sample label, and return to execute the step of extracting the first sample data from the sample set specified by the first requester until a first preset stopping condition is reached to obtain the pre-trained model parameters corresponding to the first requester.

[0197] Optionally, the first training module is further configured to send pre-trained model parameters to the first requester; receive sample adjustment data sent by the first requester based on the pre-trained model parameters; and adjust the parameters of the pre-trained processing model according to the sample adjustment data.

[0198] Optionally, the pre-trained model parameters include shared model parameters; the first training module is further configured to store the candidate model parameters corresponding to each requester into the candidate parameter storage unit respectively; and to store the shared model parameters into the shared parameter storage unit.

[0199] Optionally, the first training module is further configured to determine whether the shared parameter storage unit includes shared model parameters; if the shared parameter storage unit does not include shared model parameters, the shared model parameters are stored in the shared parameter storage unit.

[0200] Optionally, the first filtering module 904 is further configured to obtain the requester information of the target requester; construct a parameter filtering matrix based on the requester information; and filter the target model parameters from multiple candidate model parameters based on the parameter filtering matrix.

[0201] Optionally, the device further includes: a second training module configured to acquire a training set, wherein the training set includes multiple training data, and the training data carries training labels; extract first training data from the training set, wherein the first training data is any one of the training data; input the first training data into the original processing model to obtain a first training prediction result; and train the original processing model according to the first training prediction result and the first training label carried by the first training data to obtain a pre-trained processing model.

[0202] Optionally, the device further includes: a sending module configured to send task processing results to a client so that the client can display the task processing results to the target requester; receiving adjustment information sent by the target requester through the client, and adjusting the target model parameters according to the adjustment information, wherein the adjustment information is obtained based on the task processing results.

[0203] By applying the scheme of the embodiments of this specification, in the process of training the pre-trained processing model using the sample sets specified by different requesters, the shared model parameters in the pre-trained processing model are kept unchanged, thereby efficiently obtaining the candidate model parameters corresponding to each requester. This realizes that only one shared model parameter is used in conjunction with different candidate model parameters to serve different requesters, without having to provide a separate task processing model for each requester, thus reducing the model deployment cost and task processing resource consumption.

[0204] The above is an illustrative scheme of a task processing device according to this embodiment. It should be noted that the technical solution of this task processing device and the technical solution of the task processing method described above belong to the same concept. For details not described in detail in the technical solution of the task processing device, please refer to the description of the technical solution of the task processing method described above.

[0205] Corresponding to the above embodiments of the automatic question-answering method, this specification also provides embodiments of the automatic question-answering device. Figure 10 A schematic diagram of an automatic question-and-answer device according to one embodiment of this specification is shown. Figure 10 As shown, the device includes:

[0206] The second receiving module 1002 is configured to receive a problem handling request, wherein the problem handling request carries data to be answered;

[0207] The second filtering module 1004 is configured to filter out the target model parameters corresponding to the target requester from multiple candidate model parameters. The candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester, and each candidate model parameter corresponds to a different requester.

[0208] The second construction module 1006 is configured to construct a task processing model based on the target model parameters and the shared model parameters. The shared model parameters are the model parameters shared by each requester in the pre-trained processing model. The shared model parameters remain fixed during the training process of the pre-trained processing model.

[0209] The second input module 1008 is configured to input the data to be answered into the task processing model to obtain the answer to the question.

[0210] By applying the scheme of the embodiments of this specification, in the process of training the pre-trained processing model using the sample sets specified by different requesters, the shared model parameters in the pre-trained processing model are kept unchanged, thereby efficiently obtaining the candidate model parameters corresponding to each requester. This enables the use of only one shared model parameter in conjunction with different candidate model parameters to serve different requesters, without the need to provide a separate task processing model for each requester, thus reducing the model deployment cost and the consumption of automatic question answering resources.

[0211] The above is an illustrative scheme of an automatic question-and-answer device according to this embodiment. It should be noted that the technical solution of this automatic question-and-answer device and the technical solution of the automatic question-and-answer method described above belong to the same concept. For details not described in detail in the technical solution of the automatic question-and-answer device, please refer to the description of the technical solution of the automatic question-and-answer method described above.

[0212] Figure 11 A structural block diagram of a computing device according to one embodiment of this specification is shown. The components of the computing device 1100 include, but are not limited to, a memory 1110 and a processor 1120. The processor 1120 is connected to the memory 1110 via a bus 1130, and a database 1150 is used to store data.

[0213] The computing device 1100 also includes an access device 1140, which enables the computing device 1100 to communicate via one or more networks 1160. Examples of these networks include Public Switched Telephone Network (PSTN), Local Area Network (LAN), Wide Area Network (WAN), Personal Area Network (PAN), or combinations of communication networks such as the Internet. The access device 1140 may include one or more of any type of wired or wireless network interface (e.g., Network Interface Card (NIC)), such as an IEEE 802.11 Wireless Local Area Networks (WLAN) interface, a Wi-MAX (World Interoperability for Microwave Access) interface, an Ethernet interface, a Universal Serial Bus (USB) interface, a cellular network interface, a Bluetooth interface, a Near Field Communication (NFC) interface, and so on.

[0214] In one embodiment of this specification, the aforementioned components of the computing device 1100 and Figure 11 Other components, not shown, can also be connected to each other, for example, via a bus. It should be understood that... Figure 11 The block diagram of the computing device shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art can add or replace other components as needed.

[0215] The computing device 1100 can be any type of stationary or mobile computing device, including mobile computers or mobile computing devices (e.g., tablet computers, personal digital assistants, laptop computers, notebook computers, netbooks, etc.), mobile phones (e.g., smartphones), wearable computing devices (e.g., smartwatches, smart glasses, etc.) or other types of mobile devices, or stationary computing devices such as desktop computers or personal computers (PCs). The computing device 1100 can also be a mobile or stationary server.

[0216] The processor 1120 is configured to execute the following computer-executable instructions, which, when executed by the processor, implement the steps of the above-described task processing method or automatic question-and-answer method.

[0217] The above is an illustrative scheme of a computing device according to this embodiment. It should be noted that the technical solution of this computing device belongs to the same concept as the technical solutions of the task processing method and the automatic question answering method described above. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solutions of the task processing method or the automatic question answering method described above.

[0218] An embodiment of this specification also provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the above-described task processing method or automatic question-and-answer method.

[0219] The above is an illustrative scheme of a computer-readable storage medium according to this embodiment. It should be noted that the technical solution of this storage medium belongs to the same concept as the technical solutions of the task processing method and the automatic question answering method described above. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solutions of the task processing method or the automatic question answering method described above.

[0220] An embodiment of this specification also provides a computer program, wherein when the computer program is executed in a computer, it causes the computer to perform the steps of the above-described task processing method or automatic question-and-answer method.

[0221] The above is an illustrative scheme of a computer program according to this embodiment. It should be noted that the technical solution of this computer program belongs to the same concept as the technical solutions of the task processing method and the automatic question answering method described above. For details not described in detail in the technical solution of the computer program, please refer to the description of the technical solutions of the task processing method or the automatic question answering method described above.

[0222] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0223] The computer instructions include computer program code, which may be in the form of source code, object code, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0224] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the embodiments in this specification are not limited to the described order of actions, because according to the embodiments in this specification, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to the embodiments in this specification.

[0225] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0226] The preferred embodiments disclosed above are merely illustrative of this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the embodiments described herein. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the embodiments, thereby enabling those skilled in the art to better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A task processing method, comprising: Receive a task processing request, wherein the task processing request carries data to be processed; The target model parameters corresponding to the target requester are selected from multiple candidate model parameters. The candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester. Each candidate model parameter corresponds to a different requester. A task processing model is constructed based on the target model parameters and the shared model parameters, wherein the shared model parameters are model parameters shared by each requester in the pre-trained processing model, and the shared model parameters remain fixed during the training process of the pre-trained processing model; The data to be processed is input into the task processing model to obtain the task processing result.

2. The method according to claim 1, further comprising, before filtering out the target model parameters corresponding to the target requester from multiple candidate model parameters: Retrieve sample sets specified by multiple requesters; For the first requester, a pre-trained processing model is trained based on the sample set specified by the first requester to obtain the pre-trained model parameters corresponding to the first requester, wherein the first requester is any one of the plurality of requesters; Candidate model parameters for each requester are selected from the pre-trained model parameters for each requester.

3. The method according to claim 2, wherein training a pre-trained processing model based on the sample set specified by the first requester to obtain the pre-trained model parameters corresponding to the first requester includes: Extract first sample data from the sample set specified by the first requester, wherein the first sample data is any sample data in the sample set, and the first sample data carries a first sample label; The first sample data is input into the pre-trained processing model to obtain the prediction result of the first sample; Based on the prediction result of the first sample and the label of the first sample, adjust the model parameters other than the shared model parameters in the pre-trained processing model, and return to the step of extracting the first sample data from the sample set specified by the first requester until the first preset stopping condition is reached, and obtain the pre-trained model parameters corresponding to the first requester.

4. The method according to claim 2, after obtaining the pre-trained model parameters corresponding to the first requester by training the pre-trained processing model according to the sample set specified by the first requester, it further includes: Send the pre-trained model parameters to the first requester; The system receives sample adjustment data sent by the first requester based on the parameters of the pre-trained model, and adjusts the parameters of the pre-trained processing model according to the sample adjustment data.

5. The method according to claim 2, wherein the pre-trained model parameters include shared model parameters; After selecting candidate model parameters for each requester from the pre-trained model parameters corresponding to each requester, the process further includes: The candidate model parameters corresponding to each requester are stored in the candidate parameter storage unit respectively. The shared model parameters are stored in the shared parameter storage unit.

6. The method according to claim 5, further comprising, before storing the shared model parameters to the shared parameter storage unit: Determine whether the shared model parameters are included in the shared parameter storage unit; The step of storing the shared model parameters to the shared parameter storage unit includes: If the shared model parameters are not included in the shared parameter storage unit, the shared model parameters are stored in the shared parameter storage unit.

7. The method according to claim 1, wherein filtering the target model parameters corresponding to the target requester from a plurality of candidate model parameters includes: Obtain the requester information of the target requester; Construct a parameter filtering matrix based on the requester information; The target model parameters are selected from multiple candidate model parameters based on the parameter filtering matrix.

8. The method according to claim 1, further comprising, before filtering out the target model parameters corresponding to the target requester from multiple candidate model parameters: Obtain a training set, wherein the training set includes multiple training data, and the training data carries training labels; Extract first training data from the training set, wherein the first training data is any one of the training data in the training set; Input the first training data into the original processing model to obtain the first training prediction result; Based on the first training prediction result and the first training label carried by the first training data, the original processing model is trained to obtain a pre-trained processing model.

9. The method according to claim 1, further comprising, after inputting the data to be processed into the task processing model and obtaining the task processing result: The task processing result is sent to the client so that the client can display the task processing result to the target requester; The system receives adjustment information sent by the target requester through the client, and adjusts the target model parameters according to the adjustment information, wherein the adjustment information is obtained based on the task processing result.

10. An automatic question-answering method, comprising: Receive a problem handling request, wherein the problem handling request carries data to be answered; The target model parameters corresponding to the target requester are selected from multiple candidate model parameters. The candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester. Each candidate model parameter corresponds to a different requester. A task processing model is constructed based on the target model parameters and the shared model parameters, wherein the shared model parameters are model parameters shared by each requester in the pre-trained processing model, and the shared model parameters remain fixed during the training process of the pre-trained processing model; The data to be answered is input into the task processing model to obtain the answer to the question.

11. A task processing system, comprising a service unit and a candidate parameter storage unit; The candidate parameter storage unit is used to store multiple candidate model parameters; The service unit is used to receive task processing requests, wherein... The task processing request carries data to be processed; Target model parameters corresponding to the target requester are selected from the plurality of candidate model parameters. The candidate model parameters are obtained by training a pre-trained processing model based on a sample set specified by the requester, and each candidate model parameter corresponds to a different requester. A task processing model is constructed based on the target model parameters and shared model parameters. The shared model parameters are model parameters shared by all requesters in the pre-trained processing model, and the shared model parameters remain unchanged during the training of the pre-trained processing model. The data to be processed is input into the task processing model to obtain the task processing result.

12. The system according to claim 11 further includes a shared parameter storage unit; The shared parameter storage unit is used to store shared model parameters; The service unit is also used to obtain the shared model parameters from the shared parameter storage unit.

13. A computing device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of any one of claims 1 to 9 or the method of claim 10.

14. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of any one of claims 1 to 9 or the method of claim 10.

Images