Task execution method and system, and electronic device, storage medium and program product

Abstract

The present disclosure relates to the fields of large model technology and machine learning. Disclosed are a task execution method and system, and an electronic device, a storage medium and a program product. The method comprises: determining a task to be executed; identifying the current task type of said task; acquiring from an agent model set at least one current agent model, which matches the current task type, wherein the agent model set comprises different agent models, which match different task types, and any of the different agent models reuses the same pre-trained model on the basis of model parameters of a task processing model under a corresponding task type; and inputting said task into the current agent model, and using current model parameters under the current task type to control the pre-trained model to analyze said task, so as to obtain a task execution result.

Description

Task execution method, system, electronic device, storage medium and program product

[0001] Cross-reference

[0002] This disclosure claims priority to the Chinese patent application filed with the China Patent Office on January 23, 2024, with application number 202410096587.4 and invention name “Task execution method, system, electronic device, storage medium and program product”, the entire contents of which are incorporated by reference into this disclosure. Technical Field

[0003] The present disclosure relates to large model technology and the field of machine learning, and more specifically, to a task execution method, system, electronic device, storage medium, and program product. Background Art

[0004] Currently, multi-agent systems based on large models are a hot topic in the field of artificial intelligence. Multi-agent systems can automatically complete complex tasks such as code generation and solution research.

[0005] Related technologies typically define intelligent agent systems using a "universal large model + system instructions" approach. However, this approach relies heavily on the universal large model, making it expensive and time-consuming, leading to high model deployment costs.

[0006] To address the above-mentioned problems, no effective solutions have been proposed so far. Summary of the Invention

[0007] The embodiments of the present disclosure provide a task execution method, system, electronic device, storage medium, and program product to at least solve the technical problem of high model deployment cost.

[0008] According to one aspect of an embodiment of the present disclosure, a task execution method is provided. The method may include: determining a task to be executed; identifying a current task type of the task to be executed; obtaining at least one current agent model that matches the current task type from an agent model set, wherein the agent model set includes different agent models that match different task types, and any of the different agent models reuses the same pre-trained model by using model parameters of a task processing model under the corresponding task type; inputting the task to be executed into the current agent model, and controlling the pre-trained model to analyze the task to be executed using the current model parameters under the current task type to obtain a task execution result.

[0009] According to another aspect of the embodiments of the present disclosure, a model deployment method is also provided. The method may include: determining different task types; obtaining task processing models for different task types; combining the model parameters of the task processing models for different task types with the same pre-trained model to obtain intelligent agent models that match the different task types; and outputting the intelligent agent models that match the different task types.

[0010] According to another aspect of the embodiments of the present disclosure, another model generation method is also provided. The method may include: obtaining training samples for different task types; using the training samples for the different task types, training task processing models for the corresponding task types; wherein the model parameters of the task processing models for the different task types are used to combine with the same pre-trained model to obtain intelligent agent models that match the different task types; the intelligent agent model is used to control the pre-trained model to analyze the task to be executed input using the corresponding model parameters to obtain the task execution result.

[0011] According to another aspect of the embodiment of the present disclosure, another task execution method is also provided. The method can be applied to an intelligent agent system and may include: monitoring the tasks to be executed in the intelligent agent system by calling a first interface, wherein the first interface includes a first parameter, and the parameter value of the first parameter is the task to be executed; identifying the current task type of the task to be executed; obtaining at least one current intelligent agent model that matches the current task type in the intelligent agent model set, wherein the intelligent agent model set includes different intelligent agent models that match different task types, and any intelligent agent model in the different intelligent agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; inputting the task to be executed into the current intelligent agent model, using the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed, and obtaining the task execution result; outputting the task execution result by calling a second interface, wherein the second interface includes a second parameter, and the parameter value of the second parameter is the task execution result.

[0012] According to another aspect of the embodiment of the present disclosure, another task execution method is also provided. The method may include: responding to an inquiry message received in a dialogue interface, determining the task to be executed in the intelligent agent system in the inquiry message; displaying the current task type of the task to be executed on the dialogue interface; displaying the reply information corresponding to the current task type on the dialogue interface, wherein the reply information is used to indicate the task execution result of the task to be executed of the current task type, and the reply information is obtained by controlling the pre-trained model to analyze the task to be executed using the current model parameters under the current task type in the current intelligent agent model, the current intelligent agent model is obtained from the intelligent agent model set, the intelligent agent model set includes different intelligent agent models matching different task types, and any intelligent agent model in the different intelligent agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type.

[0013] According to another aspect of the embodiments of the present disclosure, an intelligent agent system is also provided. The system may include: a model training end configured to train task processing models for different task types; a model deployment end configured to combine model parameters of the task processing models for different task types with the same pre-trained model to obtain intelligent agent models that match the different task types; and output the intelligent agent models that match the different task types.

[0014] According to another aspect of an embodiment of the present disclosure, an electronic device is provided. The electronic device may include a memory and a processor; the memory is configured to store computer-executable instructions, and the processor is configured to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the method described above in the embodiment of the present disclosure is implemented.

[0015] According to another aspect of an embodiment of the present disclosure, a processor is further provided, wherein the processor is configured to run a program, wherein any one of the above methods is executed when the program is running.

[0016] According to another aspect of the present disclosure, a computer-readable storage medium is provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the storage medium is located is controlled to execute any one of the above methods.

[0017] According to another aspect of the present disclosure, a computer program product is provided, which may include computer instructions, and when the computer instructions are executed by a processor, any one of the above methods is implemented.

[0018] In an embodiment of the present disclosure, a task to be executed is determined; the current task type of the task to be executed is identified; and in the agent model set, at least one current agent model that matches the current task type is obtained, wherein the agent model set includes different agent models that match different task types, and any agent model among the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; the task to be executed is input into the current agent model, and the current model parameters under the current task type are used to control the pre-trained model to analyze the task to be executed, and obtain the task execution result. That is, in this embodiment, a corresponding task processing model can be trained for each task type, and when a task to be executed is obtained, at least one current agent model can be called based on the current task type of the task to be executed. Among them, the intelligent agent model can include the model parameters of the task processing model under the corresponding task type and the same pre-trained model. The current model parameters in the current intelligent agent model are used to control the pre-trained model to analyze the task to be executed and obtain the execution result. Therefore, a single pre-trained model (for example, a large model base) can be used to mount the model parameters of the task processing model under each task type. This method utilizes the deployment cost of a single large model to realize the deployment of different intelligent agent models, and can meet the needs of customized multi-agent systems at a lower cost, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high model deployment cost.

[0019] It is easy to note that the above general description and the following detailed description are only for the purpose of exemplifying and explaining the present disclosure, and do not constitute a limitation of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present disclosure and constitute a part of the present disclosure. The exemplary embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation of the present disclosure. In the drawings:

[0021] FIG1 is a schematic diagram of an application scenario of a task execution method according to an embodiment of the present disclosure;

[0022] FIG2 is a block diagram of a computing environment according to an embodiment of the present disclosure;

[0023] FIG3 is a flowchart of a task execution method according to an embodiment of the present disclosure;

[0024] FIG4 is a flow chart of generating a model according to an embodiment of the present disclosure;

[0025] FIG5 is a flowchart of another task execution method according to an embodiment of the present disclosure;

[0026] FIG6 is a flowchart of another task execution method according to an embodiment of the present disclosure;

[0027] FIG7( a ) is a flowchart of another task execution method according to an embodiment of the present disclosure;

[0028] FIG7( b ) is a schematic diagram of an intelligent agent system according to an embodiment of the present disclosure;

[0029] FIG8 is a flow chart of an efficient implementation scheme of a multi-agent system based on LoRA according to an embodiment of the present disclosure;

[0030] FIG9 is a schematic diagram of completing a task based on multi-agent interaction according to an embodiment of the present disclosure;

[0031] FIG10 is a schematic diagram of an agent model deployment according to an embodiment of the present disclosure;

[0032] FIG11 is a hardware structure block diagram of a computer terminal (or mobile device) according to a task execution method according to an embodiment of the present disclosure;

[0033] FIG12 is a schematic diagram of a task execution device according to an embodiment of the present disclosure;

[0034] FIG13 is a schematic diagram of a model generation device according to an embodiment of the present disclosure;

[0035] FIG14 is a schematic diagram of another task execution device according to an embodiment of the present disclosure;

[0036] FIG15 is a schematic diagram of another task execution device according to an embodiment of the present disclosure;

[0037] FIG16 is a schematic diagram of another task execution device according to an embodiment of the present disclosure;

[0038] FIG17 is a structural block diagram of a computer terminal according to an embodiment of the present disclosure;

[0039] FIG18 is a block diagram of an electronic device according to a task execution method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0040] In order to enable those skilled in the art to better understand the solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the embodiments described are only part of the embodiments of the present disclosure, not all of the embodiments. Based on the embodiments of the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present disclosure.

[0041] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way are interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or components is not necessarily limited to those steps or components clearly listed, but may include other steps or components that are not clearly listed or inherent to these processes, methods, products or devices.

[0042] The technical solution provided by the present disclosure can be implemented using large-scale model technology. The large model here refers to a deep learning model with large-scale model parameters, which can usually contain hundreds of millions, tens of billions, hundreds of billions, trillions, or even more than ten trillion model parameters. The large model can also be called a cornerstone model / foundation model (Foundation Model). The large model is pre-trained by large-scale unlabeled corpus to produce a pre-trained model with more than 100 million parameters. This model can adapt to a wide range of downstream tasks and has good generalization capabilities, such as large-scale language models (LLMs) and multi-modal pre-training models.

[0043] It should be noted that when the large model is actually applied, the pre-trained model can be fine-tuned through a small number of samples, so that the large model can be applied to different tasks. For example, the large model can be widely used in natural language processing (NLP), computer vision, speech processing and other fields. Specifically, it can be applied to computer vision tasks such as visual question answering (VQA), image description (IC), and image generation. It can also be widely used in natural language processing tasks such as text-based sentiment classification, text summary generation, and machine translation. Therefore, the main application scenarios of the large model include but are not limited to digital assistants, intelligent robots, search, online education, office software, e-commerce, intelligent design, etc. In the embodiment of the present disclosure, data processing through a machine learning model in a dialogue scenario is used as an example for explanation.

[0044] First, some nouns or terms that appear in the description of the embodiments of the present disclosure are subject to the following explanations:

[0045] Multi-agent, which can refer to a system consisting of multiple information processing and decision-making units existing in a shared environment, can be used to interact in the shared environment to achieve the same or conflicting goals;

[0046] System prompts are special prompts that can be used to guide the behavior of the model. By formulating system prompts, the style and tasks of large models can be specified within a certain range, making them more customizable and adaptable to various use cases.

[0047] Low-Rank Adaptation (LoRA) is an efficient training method for large models that can be used to perform adaptive adjustments between training data and test data.

[0048] According to a method of an embodiment of the present disclosure, a task execution method is provided. As an optional implementation, the above-mentioned task execution method may include but is not limited to being applied to an application scenario as shown in Figure 1. Figure 1 is a schematic diagram of an application scenario of a task execution method according to an embodiment of the present disclosure. As shown in Figure 1, in the application scenario, the terminal device 12 may, but is not limited to, communicate with the server 16 through the network 14. For example, it may be used to transmit tasks to be executed, task execution results, etc. The server 16 may, but is not limited to, perform operations on the database 18, such as write data operations or read data operations. The above-mentioned terminal device 12 may, but is not limited to, include a human-computer interaction screen, a processor, and a memory. The above-mentioned human-computer interaction screen may, but is not limited to, be used to display a set of candidate products and products to be recommended on the terminal device 12. The above-mentioned processor may include, but is not limited to, being used to respond to the above-mentioned human-computer interaction operation, perform corresponding operations, or generate corresponding instructions and send the generated instructions to the server 16. The above-mentioned memory is used to store relevant processing data, such as task processing models, pre-trained models, etc.

[0049] As an optional method, the following steps in the task execution method can be executed on the server 16: Step S102, determine the task to be executed; Step S104, identify the current task type of the task to be executed; Step S106, obtain at least one current agent model that matches the current task type in the agent model set, wherein the agent model set includes different agent models that match different task types, and any agent model in the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; Step S108, input the task to be executed into the current agent model, and use the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed to obtain the task execution result.

[0050] By adopting the above method, a corresponding task processing model is trained for each task type. When a task to be executed is obtained, at least one current agent model can be called based on the current task type of the task to be executed. Among them, the agent model can include the model parameters of the task processing model under the corresponding task type and the same pre-trained model. The current model parameters in the current agent model are used to control the pre-trained model to analyze the task to be executed and obtain the execution result. In this way, a single pre-trained model (for example, a large model base) can be used to mount the model parameters of the task processing model under each task type (also known as LoRA network parameters). Through this method, the deployment cost of a single large model can be used to realize the deployment of different agent models. The requirements of a customized multi-agent system can be met at a lower cost, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high model deployment cost.

[0051] In another optional embodiment, FIG2 shows in a block diagram an embodiment of using a computer terminal (or mobile device) as a computing node in a computing environment 201. FIG2 is a structural block diagram of a computing environment according to an embodiment of the present disclosure. As shown in FIG2 , the computing environment 201 includes multiple (shown in the figure as 210-1, 210-2, ...) computing nodes (such as servers) running on a distributed network. The computing nodes all contain local processing and memory resources, and the end user 202 can remotely run applications or store data in the computing environment 201. The application can be provided as multiple services 220-1, 220-2, 220-3 and 220-4 in the computing environment 201, representing services "A", "D", "E" and "H" respectively.

[0052] End user 202 can provide and access services through a web browser or other software application on a client. In some embodiments, the provisioning and / or request of end user 202 can be provided to the ingress gateway 230. The ingress gateway 230 may include a corresponding agent to handle the provisioning and / or request for services (one or more services provided in the computing environment 201).

[0053] Services are provided or deployed based on various virtualization technologies supported by the computing environment 201. In some embodiments, services can be provided based on virtual machine (VM)-based virtualization, container-based virtualization, and / or similar methods. Virtual machine-based virtualization can simulate a real computer by initializing a virtual machine, executing programs and applications without directly contacting any actual hardware resources. While the virtual machine virtualizes the machine, according to container-based virtualization, a container can be started to virtualize the entire operating system (OS) so that multiple workloads can run on a single operating system instance.

[0054] In one embodiment based on container virtualization, several containers of a service can be assembled into a computing component (e.g., a Kubernetes Pod). For example, as shown in Figure 2, service 220-2 can be equipped with one or more computing components (Pods) Pod240-1, 240-2, ..., 240-N (collectively referred to as Pods). The Pod may include a proxy 245 and one or more containers 242-1, 242-2, ..., 242-M (collectively referred to as containers). One or more containers in the Pod process requests related to one or more corresponding functions of the service, and the proxy 245 generally controls network functions related to the service, such as routing, load balancing, etc. Other services can also be equipped with Pods similar to Pods.

[0055] During operation, executing a user request from end user 202 may require invoking one or more services in computing environment 201. Executing one or more functions of one service may require invoking one or more functions of another service. As shown in FIG2 , service "A" 220-1 receives a user request from end user 202 from ingress gateway 230. Service "A" 220-1 may invoke service "D" 220-2, and service "D" 220-2 may request service "E" 220-3 to execute one or more functions.

[0056] This computing environment can be a cloud computing environment, where resource allocation is managed by the cloud service provider, allowing for feature development without having to worry about implementing, adjusting, or scaling servers. This computing environment allows developers to execute code in response to events without building or maintaining complex infrastructure. Services can be partitioned to perform a set of functions that can scale independently and automatically, rather than scaling a single hardware device to handle the potential load.

[0057] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this disclosure, such as weather forecast results and other data, are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0058] In the above operating environment, the present disclosure provides a task execution method, which can be applied to an intelligent agent system, or can be applied to situations such as multiple large models and multiple agents performing tasks. It should be noted that this is only an example and does not impose specific restrictions on the application scenarios of the task execution method. Among them, the intelligent agent system can be implemented by a generative pre-trained language model, and can be a system composed of multiple information processing components and decision-making components in a shared environment. Figure 3 is a flowchart of a task execution method according to an embodiment of the present disclosure.

[0059] As shown in FIG3 , the method may include the following steps:

[0060] Step S302: Determine the task to be executed.

[0061] In the technical solution provided in step S302 of the present disclosure, a task to be performed is determined. The task to be performed can be work that needs to be completed in the intelligent system, can be a complex task such as code generation or solution research, or can be a customized task such as an advertising and marketing task. It should be noted that there is no specific limitation on the type of task to be performed.

[0062] Optionally, the pending tasks in this embodiment can be tasks within the intelligent agent system. This requires first identifying the work that the intelligent agent system needs to accomplish. This could be a specific task, such as ensuring safe driving of an autonomous vehicle on city streets, or a more abstract goal, such as optimizing energy efficiency in a smart home system. The pending tasks within the intelligent agent system can be determined by monitoring user input, automated system monitoring, and other methods.

[0063] Step S304: Identify the current task type of the task to be executed.

[0064] In the technical solution provided in step S304 of the present disclosure, after determining the pending tasks in the agent system, the current task type of the pending tasks can be identified. The current task type may include a scheduling task type, a functional task type, an evaluation task type, etc. This is merely an example and does not impose any specific limitation on the current task type.

[0065] Optionally, after determining the tasks to be performed in the intelligent agent system, the tasks to be performed may be further identified to determine the current task type of the tasks to be performed.

[0066] For example, suppose there's a home service robot that helps with household chores. After receiving a pending task input by the user via voice or text, the robot can determine that the pending task is grading children's homework. The robot can then identify the pending task as a functional task.

[0067] Step S306: Obtain at least one current agent model that matches the current task type from the agent model set, wherein the agent model set includes different agent models that match different task types, and any agent model among the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type.

[0068] In the technical solution provided in the above step S306 of the present disclosure, after identifying the current task type of the task to be executed, at least one current agent model that matches the current task type can be obtained in the agent model set. Among them, the agent model set can include different agent models that match different task types. The agent model can reuse the same pre-trained model through the model parameters of the task processing model under the corresponding task type, and can include a scheduling model (also known as a scheduling agent), an evaluation model (also known as an evaluation agent) and a functional model (also known as a functional agent). It should be noted that this is only an example and does not specifically limit the type of agent model. The task processing model can be a low-rank adaptation (LoRA) model, also known as a LoRA network, or a LoRA model, which can be used to effectively and efficiently fine-tune a pre-trained large neural network model. The model parameters can include a low-rank matrix, which can be used to reduce the number of parameters of the task physical model, and can also be called LoRA network parameters. The pre-trained model is the base of the large model. It can be a basic model, a backbone model or a general large model. It can include a convolutional neural network (CNN), a recurrent neural network (RNN), a general language model, an image recognition model or other deep learning models. It should be noted that this is only an example and does not impose specific restrictions on the type of pre-trained model.

[0069] Optionally, after training a task processing model and a pre-trained model based on the same large model base, the task processing model and the pre-trained model can be combined in an additive manner to obtain an agent model. In this way, multiple different agent models matching different task types can be pre-built to obtain an agent model set. After identifying the current task type of the task to be executed, at least one current agent model matching the current task type can be obtained from the agent model set.

[0070] Optionally, the agent model reuses the same pre-trained model by using the model parameters of the task processing model under the corresponding task type. Therefore, different agent models can be distinguished according to system instructions or LoRA networks.

[0071] Optionally, the agent model can load N different LoRA networks simultaneously during deployment. During inference, the LoRA network to be used is determined based on the current task type of the task to be executed. The LoRA network to be used can be combined with the large model base in real time to obtain the current agent model that matches the current task type. After determining the current task type, the current agent model can be obtained from the agent model set.

[0072] Step S308: input the task to be executed into the current agent model, use the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed, and obtain the task execution result.

[0073] In the technical solution provided in the above step S308 of the present disclosure, after calling the intelligent agent model, the task to be executed can be input into the current intelligent agent model, and the current model parameters under the current task type can be used to control the pre-trained model to analyze the task to be executed to obtain the task execution result. Among them, the current model parameters can be determined based on the model parameters of the task processing model in the intelligent agent model, or they can be adjusted or trained model parameters. There is no specific restriction on the source of the current model parameters here. The task execution result can be used to determine the execution status of the task, which can include whether the task processing is completed or not. For example, when the task to be executed is a classification task, the task execution result can also be a classification result. It should be noted that this is only an example, and there is no specific restriction on the type of task execution result.

[0074] In this embodiment, the task processing model is utilized to inject trainable model parameters into each layer of the neural network model architecture. The large model based on LoRA is fine-tuned in a supervised manner. There is no need to change the parameters of the pre-trained model. Only a small number of model parameters of the LoRA network need to be adjusted, which can reduce the requirements for machine specifications. This solves the technical problem of being able to effectively execute different tasks and solves the technical effect of being unable to effectively execute different tasks.

[0075] Optionally, tasks to be performed in the intelligent agent system are determined. For example, the tasks to be performed may be text classification tasks, image recognition tasks, speech recognition tasks, etc. The tasks to be performed are identified and the current task type of the tasks to be performed is determined. For example, the type of text classification tasks may be sentiment analysis, topic classification, etc. Based on the current task type, a matching intelligent agent model is found in the intelligent agent model set to obtain at least one current intelligent agent model. The model may include pre-trained models matching different task types and model parameters of the task processing model. The tasks to be performed can be input into the current intelligent agent model, and the current model parameters are used to control the pre-trained model in the current intelligent agent model to analyze the tasks to be performed to obtain the task execution results.

[0076] For example, suppose the task to be performed is sentiment analysis of a text, and the current task type is determined to be sentiment analysis. At least one corresponding sentiment analysis model (i.e., the current agent model) is found in the agent model set. This model may be a pre-trained text classification model. The task to be performed (the text mentioned above) is input into the sentiment analysis model. The current model parameters of the sentiment analysis model are used to control the pre-trained model in the sentiment analysis model to analyze the text, and the sentiment analysis result (i.e., the task processing result) is obtained, for example, whether the sentiment tendency of the text is positive, negative, or neutral.

[0077] In an embodiment of the present disclosure, a task to be executed is determined; the current task type of the task to be executed is identified; and in the agent model set, at least one current agent model matching the current task type is obtained, wherein the agent model set includes different agent models matching different task types, and any agent model among the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; the task to be executed is input into the current agent model, and the current model parameters under the current task type are used to control the pre-trained model to analyze the task to be executed, and obtain the task execution result, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high deployment cost of the model.

[0078] The above method of this embodiment is further introduced below.

[0079] As an optional implementation, step S306, obtaining at least one current agent model that matches the current task type in the agent model set, includes: combining the current model parameters and the pre-trained model in the agent model set to obtain the current agent model.

[0080] In this embodiment, after determining the current task type, the model parameters of the task processing model corresponding to the current task type can be determined from the agent model set. Based on the model parameters of the task processing model, the current model parameters can be determined, and the current model parameters can be combined with the pre-trained model to obtain the current agent model.

[0081] Although the intelligent agent system with a general large model plus system instructions can facilitate users to quickly get started, this method has extremely high requirements for the general large model's ability to follow, logical reasoning, etc., and this method is highly dependent on the general large model, resulting in high costs, high time consumption, and high deployment costs. At the same time, for some customized tasks that rely on non-public data, it is almost impossible to use the general large model to complete them simply by rewriting the instructions, resulting in the technical problem of being unable to effectively execute different tasks. To solve the above technical problems, in this embodiment, a method of mounting multiple task processing models on a single large model base is adopted, and the model parameters of the task processing model and the pre-trained model are combined to obtain the current intelligent agent model. Through the above method, at least one current intelligent agent model can be obtained, thereby realizing the deployment of multiple intelligent agent models at the deployment cost of a single large model, thereby achieving the purpose of satisfying customers' needs to build highly customized multi-agent systems at a lower cost. That is, this method utilizes the efficient training capabilities of the LoRA network to train intelligent agent models with strong vertical capabilities for users at a low cost, thereby achieving the technical effect of being able to effectively execute different tasks and solving the technical problem of being unable to effectively execute different tasks.

[0082] Optionally, in this embodiment, the current model parameters and the pre-trained model can be combined by replacement, fusion, stacking, mixing, etc. to obtain the current agent model. It should be noted that the combination method can be selected according to the specific application scenario and requirements to achieve the required performance and effect, and no specific limitation is imposed on the selected combination method.

[0083] For example, a pre-trained model can be a large pre-trained language model that can be used for various natural language processing tasks such as text generation, question answering, and translation. Task processing models can be designed for different task types. These multiple task processing models can be independent or share model parameters. When processing a pending task, the model parameters of the task processing model for the current task type can be determined as the current model parameters within the agent model set.

[0084] For another example, after determining the current model parameters, the current model parameters and the pre-trained model can be combined by replacement to obtain the current agent model. The current model parameters (for example, the parameters G and H of the LoRA network) can be directly replaced with the corresponding parameters of the pre-trained model, so that the LoRA network can directly replace the large model base for reasoning and prediction. Alternatively, after determining the current model parameters, the current model parameters and the pre-trained model can be combined by fusion to obtain the current agent model. The parameters G and H of the LoRA network can be fused with the corresponding parameters of the pre-trained model. When the current agent model performs data processing, the prediction results of the two can be fused in a certain weighted manner to obtain the final output.

[0085] Alternatively, after determining the current model parameters, the current model parameters and the pre-trained model can be combined by stacking to obtain the current intelligent agent model. When the current intelligent agent model processes data, the LoRA network and the large model base can be inferred and predicted separately, and then the output results of the two can be stacked to obtain the final output.

[0086] Alternatively, the current model parameters and the pre-trained model can be combined in a mixed manner to obtain the current intelligent agent model. When the current intelligent agent model processes data, the output results of the LoRA network and the large model base are mixed, and they are mixed in a certain way to obtain the final output.

[0087] Through the above method, the same large model base can be used to correspond to multiple different tasks, while keeping the size of the intelligent model under control and reducing computational costs.

[0088] As an optional implementation, in the intelligent agent model set, the current model parameters and the pre-trained model are combined to obtain the current intelligent agent model, including: adjusting the current model parameters in the intelligent agent model set; superimposing the adjusted current model parameters on the pre-trained model to obtain the current intelligent agent model.

[0089] In this embodiment, in the process of combining the current model parameters and the pre-trained model to obtain the current intelligent agent model, the current model parameters can be adjusted in the intelligent agent model set, and the adjusted current model parameters can be superimposed on the pre-trained model to obtain the current intelligent agent model.

[0090] Optionally, a pre-trained model can be constructed in advance. Once the pending task is determined, at least one current agent model can be invoked based on the current task type of the pending task. The current agent model can be constructed based on the pre-trained model, and the current model parameters (especially the low-rank part) have been fine-tuned for the data in the respective domain.

[0091] As an optional implementation, the method may further include: in the process of adjusting the current model parameters, keeping the model parameters of the pre-trained model unchanged.

[0092] In this embodiment, during the process of adjusting the current model parameters, only the model parameters of the task processing model corresponding to the task processing type are adjusted, and the model parameters of the pre-trained model may remain unchanged.

[0093] Optionally, in the process of adjusting the current model parameters, only the model parameters of the LoRA network are adjusted, and the model parameters of the pre-trained model are not adjusted. Therefore, the model parameters of the pre-trained model remain unchanged.

[0094] For example, a pre-trained model, pre-trained on a large-scale dataset, can be selected and the LoRa network layers added to it. These layers can be used to update low-rank parameters during fine-tuning. Training data can be obtained and used to fine-tune the pre-trained model. During this process, only the low-rank parameters in the LoRa network layers are updated, while the majority of the pre-trained model parameters remain unchanged. Ultimately, intelligent agent models can be trained for different task types. These intelligent agent models can then be deployed in real-world applications to predict, generate, or analyze new data.

[0095] In this embodiment, LoRA-based training requires lower resource specifications and faster iteration speed than supervised fine-tuning (SFT) for all parameters. Moreover, compared with the SFT training method that mixes different task data for training, LoRA-based training can train a LoRA network for each task separately, and deploy the intelligent agent model by combining the current model parameters and the pre-trained model, thereby achieving the purpose of deploying all intelligent agent models at a cost close to that of a single large model, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high model deployment cost.

[0096] For example, a task-specific dataset can be used to fine-tune the model parameters of the LoRA network in the agent model set. In this process, only the model parameters of the LoRA network are updated, while most parameters of the large model base can remain unchanged. The adjusted current model parameters can be superimposed on the pre-trained model to obtain the current agent model, thereby improving the efficiency and accuracy of the current agent model in processing the task to be executed.

[0097] As an optional implementation, in step S306, before obtaining at least one current agent model that matches the current task type in the agent model set, the method may also include: obtaining a task processing model under the current task type, wherein the task processing model is trained based on training samples under the current task type; and determining the current model parameters based on the model parameters of the trained task processing model.

[0098] In this embodiment, task processing models corresponding to different task types can be pre-selected and trained based on training data of different tasks. After identifying the current task type of the task to be executed, a trained task processing model for the current task type can be obtained, and the current model parameters can be determined based on the model parameters of the trained task processing model.

[0099] In this embodiment, a model obtained based on LoRA targeted training is used as the intelligent agent model, replacing the "general large model + system instructions" intelligent agent method, thereby achieving the purpose of improving the ability of the multi-agent system to complete customized tasks.

[0100] As an optional implementation, the current model parameters are determined based on the model parameters of the trained task processing model, including: identifying low-rank model parameters from the model parameters of the trained task processing model, wherein the parameter amount of the low-rank model parameters is less than a parameter amount threshold; and determining the low-rank model parameters as the current model parameters in the intelligent agent model set.

[0101] In this embodiment, after calling the trained task processing model, low-rank model parameters can be identified from the model parameters of the trained task processing model, and the low-rank model parameters can be determined as the current model parameters in the agent model set. The low-rank model parameters can be a trainable low-rank matrix, and the parameter amount is less than the parameter amount threshold.

[0102] Optionally, the trained task processing model may include at least one low-rank matrix, for example, a low-rank matrix X and a low-rank matrix Y. The low-rank matrix may be determined as a current model parameter in the agent model set.

[0103] This embodiment utilizes the LoRa network to inject a trainable low-rank matrix into each layer of the agent model set. Supervised fine-tuning of large models based on the LoRa network requires no changes to the parameters of the underlying large model, only a small number of LoRa network parameters need to be adjusted, thus reducing machine specifications. The agent model set can be a Transformer architecture.

[0104] Optionally, the model parameters of the LoRA network can include low-rank matrices A and B. The LoRA network and the large model base can be combined by addition. Therefore, multiple different LoRA networks can be loaded simultaneously during deployment. During inference, the LoRA network to be used can be determined based on the input task to be executed, and the LoRA network can be combined with the large model base in real time. Among them, multiple LoRA networks are trained based on the same large model base. Therefore, they can be deployed in the form of a large model base + N LoRA networks. Since the model parameters of the LoRA network are all low-rank matrices, the number of parameters of the LoRA network is usually less than 1% of the number of parameters of the large model base. When N<100, the resources consumed by this efficient deployment method are similar to those consumed by deploying a single large model.

[0105] Because the number of low-rank model parameters is less than the parameter threshold, the number of parameters in the task processing model is typically less than 1% of the number of parameters in the large model base. When the number of task processing models called is less than 100, the resource consumption of this efficient deployment method is similar to that of deploying a single large model, thus achieving the technical effect of reducing model deployment costs.

[0106] As an optional implementation, obtaining a task processing model under the current task type includes: obtaining a task processing model under the current task type from different task processing models under different task types, wherein the different task processing models under different task types are trained based on training samples under the corresponding task types.

[0107] In this embodiment, different task processing models for different task types are pre-trained based on training samples for different task types. After identifying the current task model for the task to be executed, the task processing model for the current task type can be obtained from the trained different task processing models for different task types.

[0108] As an optional implementation, determining the task to be executed includes: obtaining multiple subtasks of the input task, wherein the task execution results of each of the multiple subtasks are used to determine the task execution result of the input task; and determining the task to be executed among the multiple subtasks.

[0109] In this embodiment, multiple subtasks of an input task of the agent system are determined, and tasks to be executed can be determined from these multiple subtasks. The execution results of each of the multiple subtasks can be used to determine the execution result of the input task. Subtasks can be scheduling tasks, functional tasks, evaluation tasks, etc. These are examples only and are not specifically limited to subtask types.

[0110] For example, when inputting a task into an agent system, the input task can be broken down into multiple subtasks. These subtasks can represent different stages or steps in the overall task execution process. For example, if the input task is "organize a meeting," subtasks might include sending invitations, reserving a meeting room, arranging catering, and so on. Because different tasks have different execution stages, tasks to be executed can be identified from multiple subtasks.

[0111] As an optional implementation manner, identifying the current task type of the task to be executed includes: determining the task stage of the task to be executed among the multiple subtasks as the current task type.

[0112] In this embodiment, a task to be executed can be determined from among multiple subtasks, and the task stage of the task to be executed among the multiple subtasks can be determined as the current task type. Task stages can include scheduling stages, function stages, evaluation stages, etc., which are merely examples and do not impose specific limitations on the types of task stages.

[0113] Optionally, when the task stage is the scheduling stage, the current task type may be a scheduling task type. When the task stage is the function stage, the current task type may be a function task type. When the task stage is the evaluation stage, the current task type may be an evaluation task type.

[0114] In this embodiment, in addition to determining the current task type based on the task phase, the current task type corresponding to the subtask can also be determined based on the subtask's function, required resource information, and application scenario content. It should be noted that this is merely an example and does not impose specific limitations on the method for determining the current task type. After determining the current task type, the required current agent model can be determined based on the current task type.

[0115] Optionally, this embodiment may also directly determine the current agent model corresponding to the subtask according to the task stage in which the subtask is located.

[0116] For example, the agent model may include a scheduling agent model, a functional agent model, and an evaluation agent model. Among them, the scheduling agent model can be used to understand the task to be executed, to determine the functional agent model that needs to be scheduled in order to solve the task to be executed, and to determine the conditions for the completion of the task to be executed. The functional agent model can be used to accept scheduling and output the task execution results to other agent models or the evaluation agent model. The evaluation agent model can be used to determine whether the task to be executed has been completed and return the execution status of the task to be executed to the scheduling agent model. When a subtask is obtained, the scheduling agent model can be used to schedule the subtask. At this time, the task stage of the subtask is the scheduling stage, and the current task type can be determined to be a scheduling task type. The scheduling agent model needs to transfer the analyzed subtask to the functional agent model. At this time, the task stage of the subtask is the functional stage, and the current task type can be determined to be a functional task type. The functional agent model processes the subtask and outputs the output results to the evaluation agent to perform the evaluation task. The subtask here can be an evaluation task, the stage of the subtask is the evaluation stage, and the current task type can be an evaluation task type.

[0117] In the disclosed embodiment, a corresponding task processing model can be trained for each task type. When a task to be executed is obtained, at least one current agent model can be called based on the current task type of the task to be executed. The agent model can include the model parameters of the task processing model under the corresponding task type and the same pre-trained model. The current model parameters in the current agent model are used to control the pre-trained model to analyze the task to be executed and obtain the execution result. Thus, a single pre-trained model (for example, a large model base) can be used to mount the model parameters of the task processing model under each task type (for example, LoRA network parameters). This method uses the deployment cost of a single large model to achieve the deployment of different agent models, and can meet the needs of a customized multi-agent system at a lower cost, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high model deployment cost.

[0118] An embodiment of the present disclosure also provides a method for generating a model. FIG4 is a flowchart of a method for deploying a model according to an embodiment of the present disclosure. As shown in FIG4 , the method may include the following steps.

[0119] Step S402: Determine different task types.

[0120] In the technical solution provided in step S402 of the present disclosure, different task types to be processed by the agent model can be determined. Task types may include scheduling task types, functional task types, evaluation task types, etc., which are merely examples and are not specifically limited to task types.

[0121] Step S404: Acquire task processing models under different task types.

[0122] In the technical solution provided in step S404 above, the model parameters of the task processing model under different task types can be pre-constructed to obtain the task processing model under different task types. The model parameters can be low-rank model parameters of the task processing model, can be trainable low-rank matrices, and the parameter amount is less than the parameter amount threshold. The task processing model can be a LoRA network, also known as a LoRA model.

[0123] Optionally, the model parameters of the task processing model under different task types may include at least one low-rank matrix, for example, an X low-rank matrix and a Y low-rank matrix.

[0124] In step S406, the model parameters of the task processing models under different task types are respectively combined with the same pre-trained model to obtain intelligent agent models that match the different task types.

[0125] In the technical solution provided in step S406 above, after obtaining task processing models for different task types, the model parameters can be combined with the same pre-trained model to obtain agent models that match the different task types. The pre-trained model can be a large model base, a basic model, or a backbone model. This is for illustrative purposes only and does not impose any specific restrictions on the type of pre-trained model.

[0126] In this embodiment, multiple task processing models can be trained based on the same large model base. Therefore, the large model can be deployed in the form of a large model base + N task processing models (for example, a LoRA network). Since the model parameters of the LoRA network are all low-rank matrices, the number of parameters of the LoRA network is usually less than 1% of the number of parameters of the large model base. When N<100, the resource consumption of this efficient deployment method is similar to that of deploying a single large model, thereby achieving the purpose of efficient model deployment.

[0127] Step S410: outputting agent models that match different task types.

[0128] In the technical solution provided in the above step S410 of the present disclosure, intelligent agent models matching different task types can be output for subsequent processing of the tasks to be executed.

[0129] Optionally, the output object of the intelligent agent model can be an object that provides data processing services, which can be a developer, enterprise user, etc. There is no specific restriction on the output object of the intelligent agent model here.

[0130] In the disclosed embodiment, different task types are determined; task processing models under different task types are obtained; model parameters of the task processing models under different task types are respectively combined with the same pre-trained model to obtain intelligent agent models that match different task types respectively; and intelligent agent models that match different task types are output, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high deployment cost of the model.

[0131] According to an embodiment of the present disclosure, a method for generating a model is also provided. FIG5 is a flow chart of a method for generating a model according to an embodiment of the present disclosure. As shown in FIG5 , the method may include the following steps:

[0132] Step S502: Obtain training samples for different task types.

[0133] In the technical solution provided in the above step S502 of the present disclosure, during model training, training samples under different task types can be obtained respectively to train task processing models corresponding to different task types.

[0134] Step S504: Use training samples under different task types to train task processing models under corresponding task types, wherein the model parameters of the task processing models under different task types are used to combine with the same pre-trained model to obtain intelligent agent models that match different task types respectively. The intelligent agent model is used to use the corresponding model parameters to control the pre-trained model to analyze the input tasks to be executed and obtain the task execution results.

[0135] In the technical solution provided in the above step S504 of the present disclosure, task processing models under different task types can be trained using training samples under different task types. The model parameters of the task processing models under different task types can be combined with the same pre-trained model to obtain intelligent agent models that match different task types. The intelligent agent model can use the corresponding model parameters to control the pre-trained model to analyze the input task to be executed to obtain the task execution result. Among them, the combination method may include replacement, fusion, mixing, stacking, etc., which is only for example illustration and does not impose specific restrictions on the combination method. The task execution result may include task execution results in different task stages.

[0136] In this embodiment, training samples under different task types are used to train task processing models under corresponding task types, and the task processing models obtained based on LoRA targeted training are constructed to obtain an intelligent agent model, replacing the intelligent agent method of "general large model + system instructions", thereby achieving the purpose of improving the multi-agent system's ability to complete customized tasks.

[0137] As an optional embodiment, training samples under different task types are used to train task processing models under corresponding task types, including: using training samples under different task types to train the pre-trained model to obtain the task processing model under the corresponding task type.

[0138] In this embodiment, a common model base can be used during model training to obtain training models for different task types. That is, pre-trained models can be trained using training samples from different task types to obtain task processing models for the corresponding task types. This embodiment utilizes a common model base + N large models deployment approach to achieve efficient model deployment, achieve the technical effect of reducing model deployment costs, and resolve the technical issue of high model deployment costs.

[0139] In the disclosed embodiments, training samples under different task types are obtained respectively; the training samples under different task types are used to train task processing models under corresponding task types, wherein the model parameters of the task processing models under different task types are used to be combined with the same pre-trained model to obtain intelligent agent models that match different task types respectively, and the intelligent agent model is used to control the pre-trained model to analyze the input tasks to be executed using the corresponding model parameters to obtain task execution results, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high deployment cost of the model.

[0140] According to an embodiment of the present disclosure, another task execution method is also provided, which can be applied to an intelligent agent system. Figure 6 is a flow chart of another task execution method according to an embodiment of the present disclosure. As shown in Figure 6, the method may include the following steps:

[0141] Step S602 : monitoring the tasks to be executed in the agent system by calling a first interface, wherein the first interface includes a first parameter, and the parameter value of the first parameter is the task to be executed.

[0142] In the technical solution provided in the above step S602 of the present disclosure, the tasks to be executed in the intelligent system can be monitored by calling the first interface, wherein the first interface may include a first parameter, and the parameter value of the first parameter may be the task to be executed. The task to be executed may be generated during the operation of the application in the cloud native scenario. The task to be executed may be the work that the intelligent system needs to complete, and may be a complex task such as code generation, solution research, etc., or a customized task, for example, an advertising and marketing task. It should be noted that there is no specific restriction on the type of task to be executed here.

[0143] For example, the terminal device (i.e., the user end) can pass the task to be executed as the parameter value of the first parameter through the Application Programming Interface (API) interface of the Software as a Service (SAAS) service provider, so as to process the task to be executed. The system of the SAAS service provider will receive and process the task to be executed. Among them, the first interface can be an API endpoint provided by the SAAS platform. The user can send the task to be executed by calling this interface, and the first parameter is used to pass the task to be executed. In this way, the user can use the functions provided by the SAAS platform to process customized tasks to be executed. The SAAS platform can be a platform that deploys an intelligent system. This is only an example and does not specifically limit the type of SAAS platform.

[0144] Step S604: Identify the current task type of the task to be executed.

[0145] Step S606: Obtain at least one current agent model that matches the current task type from the agent model set, wherein the agent model set includes different agent models that match different task types, and any agent model among the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type.

[0146] Step S608: input the task to be executed into the current agent model, use the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed, and obtain the task execution result.

[0147] Step S610: Output the task execution result by calling the second interface, wherein the second interface includes a second parameter, and the parameter value of the second parameter is the task execution result.

[0148] In the technical solution provided in the above step S610 of the present disclosure, the task execution result can be output by calling the second interface, wherein the second interface can include a second parameter, and the parameter value of the second parameter can be the task execution result.

[0149] For example, the task execution result can be transmitted as the parameter value of the second parameter through the application programming interface to provide the task execution result to the terminal device. The SAAS service provider's system will transmit the processed task execution result to the terminal device and / or platform device through the interface. The second interface can be an API endpoint provided by the SAAS platform, and the second parameter can be sent by calling this interface, and the second parameter is used to transmit the task execution result.

[0150] In an embodiment of the present disclosure, a task to be executed in an intelligent agent system is monitored by calling a first interface, wherein the first interface includes a first parameter, and the parameter value of the first parameter is the task to be executed; the current task type of the task to be executed is identified; at least one current intelligent agent model matching the current task type is obtained from the intelligent agent model set, wherein the intelligent agent model set includes different intelligent agent models matching different task types, and any intelligent agent model in the different intelligent agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; the task to be executed is input into the current intelligent agent model, and the current model parameters under the current task type are used to control the pre-trained model to analyze the task to be executed to obtain the task execution result; the task execution result is output by calling a second interface, wherein the second interface includes a second parameter, and the parameter value of the second parameter is the task execution result, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high deployment cost of the model.

[0151] According to an embodiment of the present disclosure, another task execution method is also provided, which can be applied to an intelligent agent system. Figure 7(a) is a flow chart of another task execution method according to an embodiment of the present disclosure. As shown in Figure 7(a), the method may include the following steps:

[0152] Step S702 , in response to the query information received in the dialogue interface, determine the tasks to be performed in the agent system in the query information.

[0153] In the technical solution provided in step S702 above, a query message received in a dialogue interface can be obtained, and tasks to be performed in the agent system can be determined from the query message. The dialogue interface can be a chat interface on a mobile terminal, such as a mobile phone or computer. The query message can be natural language information, voice information, or other content. This is merely an example, and the specific query message type is not shown.

[0154] For example, in response to receiving the inquiry message "I am writing a novel about the future city, are there any interesting settings you can recommend" in the dialogue interface, it can be determined in the inquiry message that the tasks to be performed in the intelligent agent system are "providing inspiration" and "story clues".

[0155] Step S704: Display the current task type of the task to be executed on the dialogue interface.

[0156] In this embodiment, the current task type of the task to be executed is identified, and the current task type of the task to be executed can be displayed on the dialogue interface.

[0157] Step S706, display the reply information corresponding to the current task type on the dialogue interface, wherein the reply information is used to indicate the task execution result of the task to be executed of the current task type, and the reply information is obtained by controlling the pre-trained model to analyze the task to be executed using the current model parameters under the current task type in the current agent model. The current agent model is obtained from the agent model set, and the agent model set includes different agent models matching different task types. Any agent model among different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type.

[0158] In this embodiment, the current model parameters of the task processing model for the current task type are obtained from the model parameters of the task processing models for different task types. The current model parameters are combined with the pre-trained model to obtain at least one current agent model that matches the current task type. The at least one current agent model is obtained and then processed for the pending task to obtain a response corresponding to the current task type. The response can be used to represent the task execution result of the pending task of the current task type, and can be the task execution result at different task stages.

[0159] For example, in response to receiving a query in the dialogue interface, "I'm writing a novel about a future city. What interesting settings can you recommend?", the query can be used to determine that the pending tasks in the agent system are "providing inspiration" and "providing story clues." The current task types of the two pending tasks can be identified, and the task execution results of the pending tasks of the current task types can be determined to be provided inspiration content and story clue content. The inspiration content and story clue content can be displayed in the dialogue interface.

[0160] As an optional embodiment, the inquiry information is multimodal information, and the type of the multimodal information includes at least one of the following: text information containing character information, video frame information containing frame image information, and audio information; the type of the reply information includes at least one of the following: text information, image information, video information, and voice information.

[0161] In this embodiment, the inquiry information can be multimodal information, including text information containing character information, video frame information containing frame image information, audio information, etc. The types of reply information can include at least text information, image information, video information, and voice information. It should be noted that this is for illustrative purposes only and does not impose specific limitations on the types of inquiry information and reply information.

[0162] In an embodiment of the present disclosure, in response to an inquiry message received in a dialogue interface, a task to be executed in the intelligent agent system is determined in the inquiry message; the current task type of the task to be executed is displayed on the dialogue interface; and reply information corresponding to the current task type is displayed on the dialogue interface, wherein the reply information is used to indicate the task execution result of the task to be executed of the current task type, and the reply information is obtained by controlling the pre-trained model to analyze the task to be executed using the current model parameters under the current task type in the current intelligent agent model. The current intelligent agent model is obtained from an intelligent agent model set, and the intelligent agent model set includes different intelligent agent models matching different task types. Any intelligent agent model in different intelligent agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type, thereby achieving the technical effect of effectively executing different tasks and solving the technical problem of high model deployment cost.

[0163] According to an embodiment of the present disclosure, an embodiment of an intelligent agent system is also provided. Figure 7(b) is a schematic diagram of an intelligent agent system according to an embodiment of the present disclosure. As shown in Figure 7(b), the intelligent agent system 70 may include: a model training end 72 and a model deployment end 74.

[0164] The model training terminal 72 is configured to train task processing models under different task types.

[0165] In this embodiment, the model training terminal 72 can obtain training data of different task types, such as training data of functional tasks, training data of scheduling tasks, training data of evaluation tasks, etc. Based on the training data of different task types, training is performed to obtain task processing models for different task types.

[0166] Optionally, training data under different task types are obtained, and the large model base can be trained based on the training data under different task types at the model training end 72 to obtain task processing models under different task types.

[0167] Optionally, during model training, the model training terminal 72 may determine user needs and the task processing model to be trained. The task processing model may include a scheduling model, an evaluation model, and a function model. It should be noted that this is merely an example and does not impose any specific restrictions on the type of task processing model. Corresponding training data is obtained, and task processing models for different persona types are trained.

[0168] The model deployment terminal 74 is configured to combine the model parameters of the task processing models under different task types with the same pre-trained model to obtain intelligent agent models that match different task types; and output intelligent agent models that match different task types.

[0169] In this embodiment, the model deployment terminal 74 can monitor pending tasks and identify the current task type of the pending tasks. Based on the current task type, the model parameters of the task processing models for different task types are combined with the same pre-trained model to obtain agent models that match the different task types, and the agent models that match the different task types are output.

[0170] In this embodiment, the intelligent agent system may include a model training end 702 and a model deployment end 704, so that the intelligent agent system can continuously perform model training and optimization, thereby achieving the purpose of continuously improving its learning ability and adaptability.

[0171] In this embodiment, task processing models under different task types are trained through the model training terminal 72; model parameters of task processing models under different task types are combined with the same pre-trained model through the model deployment terminal 74 to obtain intelligent agent models that match different task types respectively; intelligent agent models that match different task types are output, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high deployment cost of the model.

[0172] Currently, multi-agent systems based on large models are a hot topic in the field of artificial intelligence. For example, the AutoGen framework, based on generative pre-trained models (such as GPT-3.5 or GPT-4), allows developers to define the roles of agents and the interaction methods between agents in natural language and computer code. This framework can automatically complete complex tasks such as code generation and solution research. Although this agent system with a general large model plus system instructions can help users get started quickly, it has extremely high requirements for the large model's compliance, logical reasoning and other capabilities. In addition, this method is highly dependent on the general large model, resulting in high costs, high time consumption and high deployment costs. At the same time, for some customized tasks that rely on non-public data, it is almost impossible to complete them using the general large model by simply rewriting the instructions, resulting in the technical problem of high model deployment costs.

[0173] As an alternative embodiment, general-purpose large models, such as GPT-4, possess strong command-following and role-playing capabilities. This method assumes that the general-purpose large model can accurately understand and complete tasks according to role definitions. This method utilizes a general "large model + system instructions" approach, adding system instructions to the input text used to call the general-purpose large model. For example, to define an automated evaluation agent for an advertising and marketing task, the following system instruction could be submitted to the general-purpose large model along with the sample to be evaluated: "You are an expert in advertising and marketing. After receiving a product description and a slogan designed for it, you need to evaluate its suitability for the product based on accuracy, appeal, and fluency." However, this method relies on high-level capabilities such as the general-purpose large model's role-playing and world knowledge, making it unsuitable for customized tasks that rely on non-public data. For example, if a painter client wants to generate images consistent with their style, this method is not applicable. Furthermore, because large models with high parameter counts require more time to infer than models with low parameter counts, this method imposes high financial and time costs on repeated calls to the general-purpose large model, leading to technical issues such as high model deployment costs.

[0174] To solve the above problems, in this embodiment, a method for efficiently implementing a multi-agent system based on LoRA is proposed. In order to reduce the requirements for large model capabilities and meet the customization needs of customers, this embodiment uses LoRA's efficient training capabilities to train vertical capabilities for customers at a low cost; in order to avoid the rapid increase in model deployment costs as the number of agent types increases, a "single large model base mounted with multiple LoRA networks" solution is adopted to deploy multiple agent models at the deployment cost of a single large model, thereby achieving the purpose of meeting customers' needs to build highly customized multi-agent systems at a lower cost, thereby achieving the technical effect of reducing the model deployment cost and solving the technical problem of high model deployment costs.

[0175] The following is a further introduction to an efficient implementation method of a LoRA-based multi-agent system proposed in an embodiment of the present disclosure.

[0176] In this embodiment, a multi-agent system is built for customers with highly customized requirements based on a pre-trained large model with a small number of parameters, wherein the multi-agent system may include at least one agent model. Figure 8 is a flow diagram of an efficient implementation solution of a multi-agent system based on LoRA according to an embodiment of the present disclosure. As shown in Figure 8, the method may include:

[0177] Step S801: Determine the intelligent agent model to be trained for the customer and perform efficient training.

[0178] In this embodiment, the agent model to be trained is determined based on the client's needs. The trained agent models may include a scheduling agent, an evaluation agent, and a functional agent. Note that this is merely an example and does not impose any specific limitations on the agent models to be trained. The agent model may include the model parameters of the task processing model and a pre-trained model.

[0179] Optionally, training the agent model mainly involves adjusting the model parameters of the task processing model in the agent model, and the model parameters of the pre-trained model can remain unchanged.

[0180] For example, as shown in Figure 8, training data for functional task 81 can be obtained to train agent-1, ..., training data for functional task 8N can be obtained to train agent-N, training data for scheduling task 82 can be obtained to train the scheduling agent, and training data for evaluation task 83 can be obtained to train the evaluation agent. The resulting intelligent agent model after training can be an intelligent agent model that matches user needs.

[0181] Optionally, the agent model may include a scheduling agent model, a functional agent model, and an evaluation agent model. The scheduling agent model can be used to understand the pending task, determine the functional agent model that needs to be scheduled to solve the pending task, and determine the conditions for completing the pending task. The functional agent model can be used to accept the schedule and output the task execution results to other agent models or the evaluation agent model. The evaluation agent model can be used to determine whether the pending task has been completed and return the execution status of the pending task to the scheduling agent model.

[0182] For example, Figure 9 is a schematic diagram of completing a task based on multi-agent interaction according to an embodiment of the present disclosure. As shown in Figure 9, the scheduling agent model 902 obtains the input task to be executed 901 and determines the functional agent model 903 and the functional agent model 904 that need to be scheduled for the task to be executed 901. The functional agent model 903 and the functional agent model 904 process the task to be executed 901 and output the processing results to the functional agent model 905. The functional agent model 905 can further process the processing results and output the final results to the evaluation agent model 906. The evaluation agent model 906 can determine whether the task to be executed is completed. If the task to be executed is not completed, the status of the task to be executed can be returned to the scheduling agent model 902. If the task to be executed is completed, the task execution result 907 can be output.

[0183] In this embodiment, LoRA-based training requires lower resource specifications and faster iteration speed than supervised fine-tuning (SFT) for all parameters. Moreover, compared with the SFT training method that mixes different task data for training, LoRA-based training can train a LoRA network for each task separately, and deploy the intelligent agent model by combining the current model parameters and the pre-trained model, thereby achieving the purpose of deploying all intelligent agent models at a cost close to that of a single large model, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high model deployment cost.

[0184] For example, a pre-trained model, pre-trained on a large-scale dataset, can be selected and the LoRa network layers added to it. These layers can be used to update low-rank parameters during fine-tuning. Training data can be obtained and used to fine-tune the pre-trained model. During this process, only the low-rank parameters in the LoRa network layers are updated, while the majority of the pre-trained model parameters remain unchanged. Ultimately, intelligent agent models can be trained for different task types. These intelligent agent models can then be deployed in real-world applications to predict, generate, or analyze new data.

[0185] In this embodiment, in order to address the shortcoming that the "general large model + system instructions" solution is difficult to implement customized tasks that rely on non-public data, this embodiment proposes an efficient training method based on LoRA to meet the customization needs of customers. The model based on LoRA targeted training is used as the intelligent agent to replace the "general large model + system instructions" intelligent agent solution, thereby improving the multi-agent system's ability to complete customized tasks.

[0186] Step S802: Efficiently deploy the intelligent agent model based on the large model base.

[0187] In this embodiment, at least one current agent model that matches the current task type can be obtained from a set of trained agent models, and the selected at least one current agent model can be deployed. The current agent model can be a trained agent model.

[0188] Optionally, as shown in Figure 8, task processing model 84 (LoRA-agent-1), task processing model 85 (LoRA-agent-2), task processing model 86 (LoRA-agent-1), task processing model 87 (LoRA-evaluation agent) and task processing model 88 (scheduling agent) are selected, and the task processing model is combined with the large model base to obtain at least one current intelligent agent model.

[0189] For example, Figure 10 is a schematic diagram of an intelligent agent model deployment according to an embodiment of the present disclosure. As shown in Figure 10, the model parameters of the LoRA network can include low-rank matrices A and B. The LoRA network and the large model base 1001 can be combined through addition. Through this method, multiple different LoRA networks can be loaded simultaneously during deployment. During inference, the LoRA network to be used can be determined based on the input task to be executed, and the LoRA network can be combined with the large model base in real time to obtain at least one intelligent agent model. Multiple LoRA networks are trained based on the same large model base. Therefore, they can be deployed using a large model base + N LoRA networks. Since the model parameters of the LoRA network are all low-rank matrices, the number of LoRA network parameters is typically less than 1% of the number of large model base parameters. When N < 100, the resource consumption of this efficient deployment method is similar to that of deploying a single large model. The low-rank matrices A and B can be obtained through the LoRA router 1002.

[0190] Optionally, the deployed agent model 1003 processes the input task to be executed to obtain an output task execution result.

[0191] Because the number of low-rank model parameters is less than the parameter threshold, the number of parameters in the task processing model is typically less than 1% of the number of parameters in the large model base. When the number of task processing models called is less than 100, the resource consumption of this efficient deployment method is similar to that of deploying a single large model, thus achieving the technical effect of reducing model deployment costs.

[0192] In this embodiment, the proposed efficient training + efficient deployment solution realizes the deployment of the entire multi-agent system at a deployment cost close to that of a single large model, which can effectively reduce the calling cost of the large model to solve the problem that the "general large model + system instructions" method requires high financial and time costs.

[0193] In this embodiment, a corresponding task processing model can be trained for each task type. When a task to be executed is obtained, at least one current agent model can be called based on the current task type of the task to be executed. The agent model can include the model parameters of the task processing model under the corresponding task type and the same pre-trained model. The current model parameters in the current agent model are used to control the pre-trained model to analyze the task to be executed and obtain the execution result. Thus, a single pre-trained model (for example, a large model base) can be used to mount the model parameters of the task processing model under each task type (also known as LoRA network parameters). This method can realize the deployment of different agent models at the deployment cost of a single large model, and can meet the needs of a customized multi-agent system at a lower cost, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high deployment cost of the model.

[0194] The method embodiments provided in the above embodiments of the present disclosure can be executed in a mobile terminal, a computer terminal or a similar computing device. Figure 11 is a hardware structure block diagram of a computer terminal (or mobile device) according to a task execution method of an embodiment of the present disclosure. As shown in Figure 11, the computer terminal 110 (or mobile device) may include one or more (1102a, 1102b, ..., 1102n are used in the figure to illustrate) processors 1102 (the processor 1102 may include but is not limited to a microprocessor (Microcontroller Unit, referred to as MCU) or a programmable logic device (Field Programmable Gate Array, referred to as FPGA) and other processing devices), a memory 1104 for storing data, and a transmission device 1106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which can be included as one of the ports of the BUS bus), a network interface, a power supply and / or a camera. It will be understood by those skilled in the art that the structure shown in Figure 11 is only for illustration and does not limit the structure of the above-mentioned electronic device. For example, the computer terminal 110 may also include more or fewer components than shown in FIG. 11 , or have a configuration different from that shown in FIG. 11 .

[0195] The hardware structure block diagram shown in Figure 11 can not only serve as an exemplary block diagram of the above-mentioned computer terminal 110 (or mobile device), but also as an exemplary block diagram of the above-mentioned server. In an optional embodiment, Figure 2 shows in a block diagram an embodiment of using the computer terminal 110 (or mobile device) shown in Figure 11 as a computing node in the computing environment 201.

[0196] The memory 1104 can be used to store software programs and components of application software, such as the program instructions / data storage device corresponding to the data processing method in the embodiment of the present disclosure. The processor executes various functional applications and data processing by running the software programs and components stored in the memory 1104, that is, realizing the above-mentioned data processing method. The memory 1104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1104 may further include a memory remotely located relative to the processor, and these remote memories may be connected to the computer terminal 110 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0197] Transmission device 1106 is configured to receive or transmit data via a network. A specific example of the aforementioned network may include a wireless network provided by the communications provider of computer terminal 110. In one embodiment, transmission device 1106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, transmission device 1106 may be a radio frequency (RF) component configured to communicate with the Internet wirelessly.

[0198] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the computer terminal 110 (or mobile device).

[0199] It should be noted that the object information (including but not limited to object device information, object personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this disclosure are all information and data authorized by the object or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for the object to choose to authorize or refuse.

[0200] It should be noted that for the aforementioned method embodiments, for simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the present disclosure is not limited by the order of the actions described, because according to the present disclosure, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and components involved are not necessarily required by the present disclosure.

[0201] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course can also be implemented by hardware. Based on this understanding, the technical solution of the present disclosure, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of the various embodiments of the present disclosure.

[0202] According to an embodiment of the present disclosure, a task execution device for implementing the task execution method shown in FIG. 3 is also provided.

[0203] FIG12 is a schematic diagram of a task execution apparatus according to an embodiment of the present disclosure. As shown in FIG12 , the task execution apparatus 1200 may include: a first determination component 1202 , a first identification component 1204 , a first acquisition component 1206 and a first processing component 1208 .

[0204] The first determining component 1202 is configured to determine a task to be executed.

[0205] A first identification component 1204 is configured to identify the current task type of the task to be performed;

[0206] The first acquisition component 1206 is configured to obtain at least one current agent model matching the current task type from the agent model set, wherein the agent model set includes different agent models matching different task types, and any agent model among the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type.

[0207] The first processing component 1208 is configured to input the task to be executed into the current agent model, use the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed, and obtain the task execution result.

[0208] Here, the first determination component 1202, the first identification component 1204, the first acquisition component 1206, and the first processing component 1208 correspond to steps S302 to S308 in the above embodiment. The four components and the corresponding steps implement the same examples and application scenarios, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above components can be hardware components or software components stored in a memory (e.g., memory 1104) and processed by one or more processors (e.g., processors 1102a, 1102b..., 1102n). The above components can also be part of the device and can be run in the computer terminal 110 provided in Example 3.

[0209] According to an embodiment of the present disclosure, a model deployment device for implementing the model deployment method shown in FIG. 4 is also provided.

[0210] Figure 13 is a schematic diagram of a model deployment device according to an embodiment of the present disclosure. As shown in Figure 13, the model deployment device 1300 may include: a second determination component 1302, a first acquisition component 1304, a processing component 1306 and a first output component 1308.

[0211] The second determining component 1302 is configured to determine different task types.

[0212] The first acquisition component 1304 is configured to acquire task processing models under different task types.

[0213] The processing component 1306 is configured to combine the model parameters of the task processing models under different task types with the same pre-trained model to obtain intelligent agent models that match the different task types.

[0214] The first output component 1308 is configured to output agent models that match different task types.

[0215] It should be noted that the second determination component 1302, the first acquisition component 1304, the processing component 1306, and the first output component 1308 correspond to steps S402 to S408 in the above embodiment. The examples and application scenarios implemented by the five components and the corresponding steps are the same, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above components can be hardware components or software components stored in a memory (e.g., memory 1104) and processed by one or more processors (e.g., processors 1102a, 1102b..., 1102n). The above components can also be part of the device and can be run in the computer terminal 110 provided in Example 3.

[0216] According to an embodiment of the present disclosure, a model generation device for implementing the model generation method shown in FIG. 5 is also provided.

[0217] FIG14 is a schematic diagram of a device for generating a model according to an embodiment of the present disclosure. As shown in FIG14 , the device for generating a model 1400 may include: a second acquisition component 1402 and a training component 1404 .

[0218] The second acquisition component 1402 is configured to respectively acquire training samples under different task types;

[0219] The training component 1404 is configured to use training samples under different task types to train task processing models under corresponding task types, wherein the model parameters of the task processing models under different task types are used to combine with the same pre-trained model to obtain intelligent agent models that match different task types respectively. The intelligent agent model is used to use the corresponding model parameters to control the pre-trained model to analyze the input tasks to be executed and obtain the task execution results.

[0220] It should be noted that the second acquisition component 1402 and the training component 1404 correspond to steps S502 to S504 in the above embodiment. The two components and the corresponding steps implement the same examples and application scenarios, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above components can be hardware components or software components stored in a memory (e.g., memory 1104) and processed by one or more processors (e.g., processors 1102a, 1102b..., 1102n). The above components can also be run as part of the device in the computer terminal 110 provided in Example 3.

[0221] According to an embodiment of the present disclosure, a task execution device for implementing the task execution method shown in FIG6 is also provided, and the device can be applied to an intelligent agent system.

[0222] Figure 15 is a schematic diagram of another task execution device according to an embodiment of the present disclosure. As shown in Figure 15, the task execution device 1500 may include: a second monitoring component 1502, a third identification component 1504, a third acquisition component 1506, a second processing component 1508 and a second output component 1510.

[0223] The second monitoring component 1502 is configured to monitor the tasks to be executed in the agent system by calling the first interface, wherein the first interface includes a first parameter, and the parameter value of the first parameter is the task to be executed.

[0224] The third identification component 1504 is configured to identify the current task type of the task to be executed.

[0225] The third acquisition component 1506 is configured to obtain at least one current agent model matching the current task type from the agent model set, wherein the agent model set includes different agent models matching different task types, and any agent model among the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type.

[0226] The second processing component 1508 is configured to input the task to be executed into the current agent model, use the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed, and obtain the task execution result.

[0227] The second output component 1510 is configured to output the task execution result by calling a second interface, wherein the second interface includes a second parameter, and the parameter value of the second parameter is the task execution result.

[0228] It should be noted that the second monitoring component 1502, the third identification component 1504, the third acquisition component 1506, the second processing component 1508, and the second output component 1510 correspond to steps S602 to S610 in the above embodiment. The examples and application scenarios implemented by the five components and the corresponding steps are the same, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above components can be hardware components or software components stored in a memory (e.g., memory 1104) and processed by one or more processors (e.g., processors 1102a, 1102b..., 1102n). The above components can also be run as part of the device in the computer terminal 110 provided in Example 3.

[0229] According to an embodiment of the present disclosure, a task execution device for implementing the task execution method shown in FIG. 7( a ) is also provided, and the device can be applied to an intelligent agent system.

[0230] FIG16 is a schematic diagram of another task execution apparatus according to an embodiment of the present disclosure. As shown in FIG16 , the task execution apparatus 1600 may include: a third determination component 1602 , a first display component 1604 , and a second display component 1606 .

[0231] The third determining component 1602 is configured to respond to the query information received in the dialogue interface, and determine the tasks to be performed in the agent system in the query information.

[0232] The first display component 1604 is configured to display the current task type of the task to be executed on the dialogue interface.

[0233] The second display component 1606 is configured to display the reply information corresponding to the current task type on the dialogue interface, wherein the reply information is configured to represent the task execution result of the task to be executed of the current task type, and the reply information is obtained by controlling the pre-trained model to analyze the task to be executed using the current model parameters under the current task type in the current agent model. The current agent model is obtained from the agent model set, and the agent model set includes different agent models matching different task types. Any agent model among the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type.

[0234] It should be noted that the third determination component 1602, the first display component 1604, and the second display component 1606 correspond to steps S702 to S706 in the above embodiment. The three components and the corresponding steps implement the same examples and application scenarios, but are not limited to the contents disclosed in the above embodiment. It should be noted that the above components can be hardware components or software components stored in a memory (e.g., memory 1104) and processed by one or more processors (e.g., processors 1102a, 1102b..., 1102n). The above components can also be part of the device and can be run in the computer terminal 110 provided in Example 3.

[0235] In the task execution device, a corresponding task processing model can be trained for each task type. When a task to be executed is obtained, at least one current agent model can be called based on the current task type of the task to be executed. Among them, the agent model can include the model parameters of the task processing model under the corresponding task type and the same pre-trained model. The current model parameters in the current agent model are used to control the pre-trained model to analyze the task to be executed and obtain the execution result. In this way, a single pre-trained model (for example, a large model base) can be used to mount the model parameters of the task processing model under each task type (for example, LoRA network parameters). This method uses the deployment cost of a single large model to realize the deployment of different agent models, and can meet the needs of a customized multi-agent system at a lower cost, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high model deployment cost.

[0236] The embodiment of the present disclosure may provide a computer terminal, which may be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the computer terminal may also be replaced by a terminal device such as a mobile terminal.

[0237] Optionally, in this embodiment, the computer terminal may be located in at least one network device among a plurality of network devices of a computer network.

[0238] In this embodiment, the above-mentioned computer terminal can execute the program code of the following steps in the task execution method: determining the task to be executed; identifying the current task type of the task to be executed; obtaining at least one current agent model that matches the current task type in the agent model set, wherein the agent model set includes different agent models that match different task types, and any agent model in the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; inputting the task to be executed into the current agent model, and using the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed, and obtain the task execution result.

[0239] Optionally, Figure 17 is a structural block diagram of a computer terminal according to an embodiment of the present disclosure. As shown in Figure 17, the computer terminal A may include: one or more (only one is shown in the figure) processors 1702, a memory 1704 and a transmission device 1706.

[0240] Among them, the memory can be configured to store software programs and components, such as program instructions / components corresponding to the task execution method and device in the embodiment of the present disclosure. The processor executes various functional applications and data processing by running the software programs and components stored in the memory, that is, realizing the above-mentioned task execution method. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include a memory remotely located relative to the processor, and these remote memories may be connected to the computer terminal A via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0241] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: determine the task to be executed; identify the current task type of the task to be executed; obtain at least one current agent model that matches the current task type in the agent model set, wherein the agent model set includes different agent models that match different task types, and any agent model in the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; input the task to be executed into the current agent model, and use the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed to obtain the task execution result.

[0242] Optionally, the processor may further execute the program code of the following steps: combining the current model parameters and the pre-trained model in the agent model set to obtain the current agent model.

[0243] Optionally, the processor may further execute the program code of the following steps: adjusting the current model parameters in the agent model set; and superimposing the adjusted current model parameters onto the pre-trained model to obtain the current agent model.

[0244] Optionally, the processor may further execute program code of the following steps: in the process of adjusting the current model parameters, keeping the model parameters of the pre-trained model unchanged.

[0245] Optionally, the processor may also execute the following program code: obtaining a task processing model under the current task type, wherein the task processing model is trained based on training samples under the current task type; and determining current model parameters based on the model parameters of the trained task processing model.

[0246] Optionally, the above-mentioned processor can also execute the program code of the following steps: identifying low-rank model parameters from the model parameters of the trained task processing model, wherein the parameter quantity of the low-rank model parameters is less than the parameter quantity threshold; and determining the low-rank model parameters as the current model parameters in the intelligent agent model set.

[0247] Optionally, the processor may also execute the program code of the following steps: obtaining the task processing model under the current task type from different task processing models under different task types, wherein the different task processing models under different task types are trained based on training samples under the corresponding task types.

[0248] Optionally, the processor may further execute program code of the following steps: obtaining multiple subtasks of the input task, wherein task execution results of each of the multiple subtasks are used to determine the task execution result of the input task; and determining a task to be executed among the multiple subtasks.

[0249] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: determine different task types; obtain task processing models under different task types; combine the model parameters of the task processing models under different task types with the same pre-trained model to obtain intelligent agent models that match different task types respectively; and output intelligent agent models that match different task types respectively.

[0250] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: respectively obtain training samples under different task types; use the training samples under different task types to train task processing models under corresponding task types, wherein the model parameters of the task processing models under different task types are used to combine with the same pre-trained model to obtain intelligent agent models that match different task types respectively, and the intelligent agent model is used to control the pre-trained model to analyze the input tasks to be executed using the corresponding model parameters to obtain the task execution results.

[0251] Optionally, the processor may further execute program code of the following steps: using training samples under different task types to train the pre-trained model to obtain a task processing model under the corresponding task type.

[0252] The processor can call the information and application programs stored in the memory through the transmission device to perform the following steps: monitor the tasks to be executed in the intelligent agent system by calling the first interface, wherein the first interface includes a first parameter, and the parameter value of the first parameter is the task to be executed; identify the current task type of the task to be executed; obtain at least one current intelligent agent model that matches the current task type in the intelligent agent model set, wherein the intelligent agent model set includes different intelligent agent models that match different task types, and any intelligent agent model in the different intelligent agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; input the task to be executed into the current intelligent agent model, and use the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed to obtain the task execution result; output the task execution result by calling the second interface, wherein the second interface includes a second parameter, and the parameter value of the second parameter is the task execution result.

[0253] The processor can call the information and application stored in the memory through the transmission device to perform the following steps: responding to the query information received in the dialogue interface, determining the task to be executed in the intelligent agent system in the query information; displaying the current task type of the task to be executed on the dialogue interface; displaying the reply information corresponding to the current task type on the dialogue interface, wherein the reply information is used to represent the task execution result of the task to be executed of the current task type, and the reply information is obtained by controlling the pre-trained model to analyze the task to be executed using the current model parameters under the current task type in the current intelligent agent model. The current intelligent agent model is obtained from the intelligent agent model set, and the intelligent agent model set includes different intelligent agent models matching different task types. Any intelligent agent model in the different intelligent agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type.

[0254] By adopting the embodiment of the present disclosure, a corresponding task processing model can be trained for each task type. When a task to be executed is obtained, at least one current agent model can be called based on the current task type of the task to be executed. Among them, the agent model can include the model parameters of the task processing model under the corresponding task type and the same pre-trained model. The current model parameters in the current agent model are used to control the pre-trained model to analyze the task to be executed and obtain the execution result. Therefore, a single pre-trained model (for example, a large model base) can be used to mount the model parameters of the task processing model under each task type. This method uses the deployment cost of a single large model to realize the deployment of different agent models, and can meet the needs of a customized multi-agent system at a lower cost, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high model deployment cost.

[0255] Those skilled in the art will appreciate that the structure shown in FIG17 is merely illustrative, and that computer terminal A may also be a smartphone (e.g., an Android phone, an iOS phone, etc.), a tablet computer, a PDA, a mobile internet device (MID), a PAD, or other terminal device. FIG17 does not limit the structure of the aforementioned computer terminal A. For example, computer terminal A may include more or fewer components (e.g., a network interface, a display device, etc.) than those shown in FIG17 , or may have a configuration different from that shown in FIG17 .

[0256] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing the hardware related to the terminal device through a program, and the program can be stored in a computer-readable storage medium, which may include: a flash drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0257] The embodiment of the present disclosure further provides a computer-readable storage medium. Optionally, in this embodiment, the computer-readable storage medium can be used to store the program code executed by the task execution method provided in the first embodiment.

[0258] Optionally, in this embodiment, the computer-readable storage medium may be located in any computer terminal in a computer terminal group in a computer network, or in any mobile terminal in a mobile terminal group.

[0259] Optionally, in this embodiment, the computer-readable storage medium is configured to store program code for performing the following steps: determining the task to be performed; identifying the current task type of the task to be performed; obtaining at least one current agent model that matches the current task type in the agent model set, wherein the agent model set includes different agent models that match different task types, and any agent model in the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; inputting the task to be performed into the current agent model, and using the current model parameters under the current task type to control the pre-trained model to analyze the task to be performed, and obtain the task execution result.

[0260] Optionally, the computer-readable storage medium may also execute program code for the following steps: combining the current model parameters and the pre-trained model in the agent model set to obtain the current agent model.

[0261] Optionally, the computer-readable storage medium may also execute program code for the following steps: adjusting the current model parameters in the agent model set; and superimposing the adjusted current model parameters on the pre-trained model to obtain the current agent model.

[0262] Optionally, the computer-readable storage medium may further execute program code for the following steps: in the process of adjusting the current model parameters, keeping the model parameters of the pre-trained model unchanged.

[0263] Optionally, the computer-readable storage medium can also execute the program code of the following steps: obtaining a task processing model under the current task type, wherein the task processing model is trained based on training samples under the current task type; and determining the current model parameters based on the model parameters of the trained task processing model.

[0264] Optionally, the above-mentioned computer-readable storage medium can also execute the program code of the following steps: identifying low-rank model parameters from the model parameters of the trained task processing model, wherein the parameter quantity of the low-rank model parameters is less than the parameter quantity threshold; and determining the low-rank model parameters as the current model parameters in the intelligent agent model set.

[0265] Optionally, the computer-readable storage medium may also execute the program code of the following steps: obtaining the task processing model under the current task type from different task processing models under different task types, wherein the different task processing models under different task types are obtained by training based on training samples under the corresponding task types.

[0266] Optionally, the computer-readable storage medium can also execute the program code of the following steps: determining different task types; obtaining task processing models under different task types; combining the model parameters of the task processing models under different task types with the same pre-trained model to obtain intelligent agent models that match different task types respectively; and outputting intelligent agent models that match different task types respectively.

[0267] As an optional example, a computer-readable storage medium is configured to store program code for performing the following steps: determining different task types; obtaining task processing models under different task types; combining the model parameters of the task processing models under different task types with the same pre-trained model to obtain intelligent agent models that match different task types respectively; and outputting intelligent agent models that match different task types respectively.

[0268] As an optional example, a computer-readable storage medium is configured to store program code for performing the following steps: obtaining training samples under different task types respectively; using the training samples under different task types to train task processing models under corresponding task types, wherein the model parameters of the task processing models under different task types are used to combine with the same pre-trained model to obtain intelligent agent models that match different task types respectively, and the intelligent agent model is used to control the pre-trained model to analyze the input tasks to be executed using the corresponding model parameters to obtain task execution results.

[0269] Optionally, the computer-readable storage medium may further execute program code for the following steps: using training samples under different task types to train the pre-trained model to obtain a task processing model under the corresponding task type.

[0270] As an optional example, a computer-readable storage medium is configured to store program code for performing the following steps: monitoring the tasks to be executed in the intelligent agent system by calling a first interface, wherein the first interface includes a first parameter, and the parameter value of the first parameter is the task to be executed; identifying the current task type of the task to be executed; obtaining at least one current intelligent agent model that matches the current task type in the intelligent agent model set, wherein the intelligent agent model set includes different intelligent agent models that match different task types, and any intelligent agent model in the different intelligent agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; inputting the task to be executed into the current intelligent agent model, and using the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed to obtain the task execution result; outputting the task execution result by calling a second interface, wherein the second interface includes a second parameter, and the parameter value of the second parameter is the task execution result.

[0271] As an optional example, a computer-readable storage medium is configured to store program code for performing the following steps: in response to an inquiry message received in a dialogue interface, determining the task to be performed in the intelligent agent system in the inquiry message; displaying the current task type of the task to be performed on the dialogue interface; and displaying reply information corresponding to the current task type on the dialogue interface, wherein the reply information is used to represent the task execution result of the task to be performed of the current task type, and the reply information is obtained by controlling the pre-trained model to analyze the task to be performed using the current model parameters under the current task type in the current intelligent agent model, and the current intelligent agent model is obtained from the intelligent agent model set, and the intelligent agent model set includes different intelligent agent models matching different task types, and any intelligent agent model among the different intelligent agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type.

[0272] In the embodiment of the present disclosure, a corresponding task processing model can be trained for each task type. When a task to be executed is obtained, at least one current agent model can be called based on the current task type of the task to be executed. The agent model can include the model parameters of the task processing model under the corresponding task type and the same pre-trained model. The current model parameters in the current agent model are used to control the pre-trained model to analyze the task to be executed and obtain the execution result. Thus, a single pre-trained model (for example, a large model base) can be used to mount the model parameters of the task processing model under each task type. This method uses the deployment cost of a single large model to achieve the deployment of different agent models, and can meet the needs of a customized multi-agent system at a lower cost, thereby achieving the technical effect of reducing the deployment cost of the model and solving the technical problem of high model deployment cost.

[0273] An embodiment of the present disclosure may provide an electronic device, which may include a memory and a processor.

[0274] FIG18 is a block diagram of an electronic device for a task execution method according to an embodiment of the present disclosure. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0275] As shown in FIG18 , device 1800 includes a computing component 1801 that can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 1802 or a computer program loaded from a storage component 1808 into a random access memory (RAM) 1803. Various programs and data required for the operation of device 1800 may also be stored in RAM 1803. Computing component 1801, ROM 1802, and RAM 1803 are connected to each other via a bus 1804. An input / output (I / O) interface 1805 is also connected to bus 1804.

[0276] Various components in device 1800 are connected to I / O interface 1805, including: input component 1806, such as a keyboard, mouse, etc.; output component 1804, such as various types of displays, speakers, etc.; storage component 1808, such as a magnetic disk, optical disk, etc.; and communication component 1809, such as a network card, modem, wireless communication transceiver, etc. Communication component 1809 allows device 1800 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0277] The computing component 1801 can be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing component 1801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing components that run machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing component 1801 performs the various methods and processes described above, such as the data verification method. For example, in some embodiments, the data verification method can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as the storage component 1808. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 1800 via the ROM 1802 and / or the communication component 1809. When the computer program is loaded into the RAM 1803 and executed by the computing component 1801, one or more steps of the data verification method described above can be performed. Alternatively, in other embodiments, the computing component 1801 may be configured to execute the data verification method in any other appropriate manner (for example, by means of firmware).

[0278] According to an embodiment of the present disclosure, a task execution method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0279] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0280] Program code configured to implement the methods of the present disclosure may be written in any combination of one or more programming languages. Such program code may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program code is executed by the processor or controller, the functions / operations specified in the flow charts and / or block diagrams are implemented. The program code may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0281] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0282] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or an LCD (liquid crystal display, monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0283] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0284] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0285] An embodiment of the present disclosure further provides a computer program product, including computer instructions, which, when executed by a processor, implement the database product status detection method provided by the embodiment of the present disclosure.

[0286] In this embodiment, the above-mentioned computer instructions can be stored in a read-only memory (ROM), or loaded from a storage component into a random access memory (RAM) so that the processor can execute various appropriate actions and processes in the status detection method of the database product.

[0287] In some embodiments, some or all of the aforementioned computer instructions may be loaded and / or installed on an electronic device via a read-only memory and / or a communication component. When the computer instructions are loaded into the random access memory and executed by the computing component, one or more steps of the database product status detection method described above may be performed.

[0288] It should be noted that the serial numbers of the above-mentioned embodiments of the present disclosure are only for description and do not represent the advantages or disadvantages of the embodiments.

[0289] In the above embodiments of the present disclosure, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0290] In the several embodiments provided in this disclosure, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only exemplary. For example, the division of components is only a logical function division. In actual implementation, there may be other division methods, such as multiple components or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection of components or components can be electrical or other forms.

[0291] Components described as separate parts may or may not be physically separate, and components shown as components may or may not be physical components, that is, they may be located in one place or distributed across multiple network components. Some or all of these components may be selected based on actual needs to achieve the objectives of this embodiment.

[0292] In addition, the functional components in the various embodiments of the present disclosure may be integrated into a single processing component, each component may exist physically separately, or two or more components may be integrated into a single component. The aforementioned integrated components may be implemented in the form of hardware or software functional components.

[0293] If the integrated components are implemented in the form of software functional components and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present disclosure is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the steps of the various embodiments of the present disclosure. The aforementioned storage media include: U disk, read-only memory (ROM), random access memory (RAM), mobile hard disk, magnetic disk or optical disk, etc., various media that can store program codes.

[0294] The above is only a preferred embodiment of the present disclosure. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present disclosure. These improvements and modifications should also be regarded as within the scope of protection of the present disclosure. Industrial Applicability

[0295] The solution provided by the embodiments of the present disclosure can be applied to the process of machine learning to determine the task to be executed; identify the current task type of the task to be executed; obtain at least one current agent model that matches the current task type in the agent model set, wherein the agent model set includes different agent models that match different task types, and any agent model in the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; input the task to be executed into the current agent model, and use the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed, and obtain the task execution result, thereby solving the technical problem of high model deployment cost.

[0296] .

Claims

1. A task execution method, comprising: Determine a task to be executed; Identify the current task type of the task to be executed; In an agent model set, obtain at least one current agent model that matches the current task type, wherein the agent model set includes different agent models that match different task types, and any agent model in the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; Input the task to be executed into the current agent model, and use the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed, and obtain a task execution result.

2. The method according to claim 1, wherein, In the agent model set, obtaining at least one current agent model that matches the current task type includes: In the agent model set, combine the current model parameters and the pre-trained model to obtain the current agent model.

3. The method according to claim 2, wherein, In the agent model set, combining the current model parameters and the pre-trained model to obtain the current agent model includes: In the agent model set, adjust the current model parameters; Overlay the adjusted current model parameters on the pre-trained model to obtain the current agent model.

4. The method according to claim 3, wherein, The method further includes: During the process of adjusting the current model parameters, keep the model parameters of the pre-trained model unchanged.

5. The method according to claim 1, wherein Before obtaining at least one current agent model that matches the current task type in the agent model set, the method further includes: Obtain the task processing model under the current task type, wherein the task processing model is trained based on training samples under the current task type; Determine the current model parameters based on the model parameters of the trained task processing model.

6. The method according to claim 5, wherein, Determining the current model parameters based on the model parameters of the trained task processing model includes: From the model parameters of the trained task processing model, identify low-rank model parameters, wherein the number of parameters of the low-rank model parameters is less than a parameter number threshold; Determine the low-rank model parameters as the current model parameters in the agent model set.

7. The method according to claim 5, wherein, Obtaining the task processing model under the current task type includes: In the different task processing models under the different task types, obtain the task processing model under the current task type, wherein the different task processing models under the different task types are trained based on training samples under the corresponding task types.

8. The method according to any one of claims 1 to 7, wherein, Determining a task to be executed includes: Obtain multiple subtasks of the input task, wherein the task execution results of the multiple subtasks are respectively used to determine the task execution result of the input task; In the multiple subtasks, determine the task to be executed.

9. A model deployment method, comprising: Determine different task types; Obtain the task processing models under the different task types; Respectively combine the model parameters of the task processing models under the different task types with the same pre-trained model to obtain agent models that respectively match the different task types; Output agent models respectively matching the different task types.

10. A method for generating a model, comprising: Obtaining training samples under different task types respectively; Using the training samples under the different task types to train task processing models corresponding to the task types, wherein the model parameters of the task processing models under the different task types are used to be combined with the same pre-trained model to obtain agent models respectively matching the different task types, and the agent models are used to control the pre-trained model to analyze an input task to be executed by using the corresponding model parameters to obtain a task execution result.

11. The method according to claim 10, wherein, Using the training samples under the different task types to train task processing models corresponding to the task types, comprising: Using the training samples under the different task types to train the pre-trained model to obtain the task processing model corresponding to the task type.

12. A task execution method applied to an agent system, comprising: Monitoring a task to be executed in the agent system by calling a first interface, wherein the first interface includes a first parameter, and the parameter value of the first parameter is the task to be executed; Identifying the current task type of the task to be executed; Obtaining at least one current agent model matching the current task type from an agent model set, wherein the agent model set includes different agent models respectively matching different task types, and any agent model in the different agent models reuses the same pre-trained model through the model parameters of the task processing model under the corresponding task type; Inputting the task to be executed into the current agent model, and using the current model parameters under the current task type to control the pre-trained model to analyze the task to be executed to obtain a task execution result; Outputting the task execution result by calling a second interface, wherein the second interface includes a second parameter, and the parameter value of the second parameter is the task execution result.

13. An agent system, comprising: A model training end configured to train task processing models under different task types; A model deployment end configured to respectively combine the model parameters of the task processing models under the different task types with the same pre-trained model to obtain agent models respectively matching the different task types; output agent models respectively matching the different task types.

14. An electronic device, comprising: A memory and a processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, and when the computer-executable instructions are executed by the processor, the steps of the method according to any one of claims 1 to 12 are implemented.

15. A computer-readable storage medium, the computer-readable storage medium including a stored program, wherein, Controlling the device where the storage medium is located to execute the method according to any one of claims 1 to 12 when the program is running.

16. A computer program product, comprising computer instructions, and when the computer instructions are executed by a processor, the method according to any one of claims 1 to 12 is implemented.

Images