Task execution method and device, electronic equipment and storage medium

Abstract

The invention provides a task execution method and device, electronic equipment and a storage medium, and relates to the technical field of computers, in particular to the technical fields of artificial intelligence, agents, large models, man-machine collaboration and the like. The specific implementation scheme comprises the steps of obtaining a sub-task chain based on a target task; after a current subtask is determined from the subtask chain, executing the current subtask by calling a current tool corresponding to the current subtask to obtain an initial intermediate result; sending the initial intermediate result to the terminal equipment; wherein the terminal equipment is used for displaying the initial intermediate result; in response to the result determination request, obtaining a target intermediate result corresponding to the initial intermediate result; wherein the result determination request is generated by the terminal equipment in response to the result determination operation and is sent to the service equipment; and obtaining an execution result of the target task based on the target intermediate result. According to the invention, the adaptation degree between the execution result of the target task and the actual demand of the user can be improved.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to the fields of artificial intelligence, intelligent agents, large models, and human-computer collaboration, specifically to a task execution method, apparatus, electronic device, and storage medium. Background Technology

[0002] When using intelligent agents to perform target tasks, it is often necessary to call multiple tools to leverage their collaboration in order to obtain the execution result of the target task. Summary of the Invention

[0003] This disclosure provides a task execution method, apparatus, electronic device, and storage medium.

[0004] According to a first aspect of this disclosure, a task execution method is provided, comprising: Based on the target task, a chain of subtasks is obtained; After determining the current subtask from the subtask chain, the current subtask is executed by calling the current tool corresponding to the current subtask in order to obtain the initial intermediate result; The initial intermediate results are sent to the terminal device; the terminal device is used to display the initial intermediate results. In response to the result determination request, a target intermediate result corresponding to the initial intermediate result is obtained; wherein, the result determination request is generated by the terminal device in response to the result determination operation and sent to the service device; Based on the intermediate results of the target, the execution result of the target task is obtained.

[0005] According to a second aspect of this disclosure, a task execution method is provided, comprising: Obtain initial intermediate results; wherein, the initial intermediate results are obtained by the service device based on the target task, obtaining a sub-task chain, and after determining the current sub-task from the sub-task chain, executing the current sub-task by calling the current tool corresponding to the current sub-task; Display the initial intermediate results; In response to the result confirmation operation, a result confirmation request is generated; The result determination request is sent to the service device; wherein, in response to the result determination request, the service device obtains the target intermediate result corresponding to the initial intermediate result, and obtains the execution result of the target task based on the target intermediate result.

[0006] According to a third aspect of this disclosure, a task execution apparatus is provided, comprising: The call chain determination unit is used to obtain the subtask chain based on the target task; The first intermediate result acquisition unit is used to obtain the initial intermediate result by calling the current tool corresponding to the current subtask after determining the current subtask from the subtask chain; An intermediate result sending unit is used to send initial intermediate results to a terminal device; wherein, the terminal device is used to display the initial intermediate results; The second intermediate result acquisition unit is used to obtain the target intermediate result corresponding to the initial intermediate result in response to the result determination request; wherein, the result determination request is generated by the terminal device in response to the result determination operation and sent to the service device. The execution result acquisition unit is used to obtain the execution result of the target task based on the target intermediate results.

[0007] According to a fourth aspect of this disclosure, a task execution apparatus is provided, comprising: The third intermediate result acquisition unit is used to acquire the initial intermediate result; wherein, the initial intermediate result is obtained by the service device based on the target task, obtaining the sub-task chain, and after determining the current sub-task from the sub-task chain, executing the current sub-task by calling the current tool corresponding to the current sub-task; Intermediate result display unit, used to display initial intermediate results; The request generation unit is used to generate a result determination request in response to the result determination operation; The request sending unit is used to send a result determination request to the service device; wherein the service device is used to respond to the result determination request, obtain the target intermediate result corresponding to the initial intermediate result, and obtain the execution result of the target task based on the target intermediate result.

[0008] According to a fifth aspect of this disclosure, an electronic device is provided, comprising: At least one processor; Memory that is communicatively connected to at least one processor; The memory stores instructions that can be executed by at least one processor, which are executed by at least one processor to enable the at least one processor to perform the method provided in the first aspect of this disclosure.

[0009] According to a sixth aspect of this disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided; wherein the computer instructions are used to cause a computer to perform the method provided in the first aspect of this disclosure.

[0010] According to a seventh aspect of this disclosure, a computer program product is provided, including a computer program; wherein, when executed by a processor, the computer program is capable of implementing the method provided in the first aspect of this disclosure.

[0011] Using this disclosure can improve the fit between the execution results of the target task and the actual needs of the user.

[0012] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0013] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein: Figure 1 A flowchart illustrating a task execution method provided in an embodiment of this disclosure; Figure 2 A schematic diagram of a sub-task chain provided in an embodiment of this disclosure; Figure 3 A flowchart illustrating a task execution method provided in an embodiment of this disclosure; Figure 4 This is a schematic diagram of the system architecture of a task execution system provided in an embodiment of the present disclosure; Figure 5 This is a schematic diagram of the functional architecture of a task execution system provided in an embodiment of the present disclosure; Figure 6 This is a schematic diagram illustrating an application scenario of a task execution method provided in an embodiment of the present disclosure; Figure 7 A schematic structural block diagram of a task execution device provided in an embodiment of this disclosure; Figure 8 A schematic structural block diagram of a task execution device provided in an embodiment of this disclosure; Figure 9 This is a schematic structural block diagram of an electronic device provided in an embodiment of the present disclosure. Detailed Implementation

[0014] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0015] In traditional approaches, when utilizing the collaboration of multiple tools to obtain the execution result of a target task, the target task can be decomposed into multiple sub-tasks, and multiple tools can be invoked to collaboratively execute these sub-tasks to obtain the target task's execution result. However, the inventors' research has revealed that in traditional approaches, achieving a high degree of automation in task execution by the intelligent agent often leads to other problems, such as: When an intelligent agent invokes a tool to execute its corresponding subtask, the intermediate results output by that tool are directly treated as fixed inputs for the next tool. This means the user cannot intervene or adjust the intermediate results during the execution of the target task. This "black box" forward pass method makes the execution process of the target task an uncontrollable, unguided, and unadjustable "black box" for the user. This severely reduces the user's controllability over the execution process, thereby decreasing the fit between the target task's execution result and the user's actual needs.

[0016] To address the above problems, this disclosure provides a task execution method that can be applied to a service device. The service device can be a server, workbench, mainframe computer, or other similar computing device. The following will be combined with... Figure 1 The flowchart shown illustrates a task execution method provided by an embodiment of this disclosure. It should be noted that although a logical order is shown in the flowchart, in some cases, the steps shown or described in the flowchart may be executed in a different order.

[0017] Step S101: Based on the target task, obtain the sub-task chain.

[0018] The target task can be any task that can be implemented based on the content service platform, such as content acquisition tasks, content generation tasks, and content processing tasks. Here, content acquisition tasks can include document download tasks and information retrieval tasks; content generation tasks can include document writing tasks and report generation tasks; and content processing tasks can include summary extraction tasks, formatting tasks, and multilingual translation tasks.

[0019] Furthermore, in this embodiment of the disclosure, the subtask chain may include only one first subtask, or it may include multiple first subtasks with an execution order. These multiple first subtasks may include at least some first subtasks that need to be executed serially, or at least some first subtasks that need to be executed in parallel; this embodiment of the disclosure does not impose any limitations on this.

[0020] Step S102: After determining the current subtask from the subtask chain, execute the current subtask by calling the current tool corresponding to the current subtask to obtain the initial intermediate result.

[0021] In this embodiment of the disclosure, when the subtask chain includes only one first subtask, the first subtask can be determined as the current subtask; when the subtask chain includes multiple first subtasks with an execution order, the current subtask can be determined from the multiple first subtasks according to the execution order.

[0022] Furthermore, in this embodiment of the disclosure, the current tool may be included among multiple candidate tools. These candidate tools may include application programming interfaces (APIs), scripts, executable services, artificial intelligence (AI) tools, etc. Here, the AI ​​tool may be obtained by training a target large model with specific vertical capabilities; the target large model may be a large language model (LLM), and the LLM may be a pre-trained neural network model (e.g., an autoregressive generative model with a Transformer architecture), which possesses general language knowledge, world knowledge, and domain-specific expertise (e.g., expertise in the field of computer technology).

[0023] Step S103: Send the initial intermediate results to the terminal device.

[0024] The terminal device can be a mainframe computer, a conventional computer (e.g., a desktop computer, a laptop computer, a tablet computer), a portable terminal, a smart wearable device, or other similar computing device, used to display the initial intermediate results so that users can promptly learn about and review them.

[0025] Step S104: In response to the result determination request, obtain the target intermediate result corresponding to the initial intermediate result.

[0026] The result determination request can be generated by the terminal device in response to the result determination operation and sent to the service device. Here, the result determination operation can be triggered by the user, and it can be one of the result input operation and the result selection operation, or at least one of the result input operation, the result selection operation and the result modification operation. This disclosure does not limit this.

[0027] Step S105: Based on the intermediate results of the target, obtain the execution result of the target task.

[0028] In this embodiment of the disclosure, when the subtask chain includes only one first subtask and the first subtask is determined as the current subtask, the intermediate result of the target can be used as the execution result of the target task.

[0029] In this embodiment of the disclosure, when the subtask chain includes multiple first subtasks with an execution order, and the current subtask is not the last subtask in the subtask chain, a new current subtask can be determined from the subtask chain, and the target intermediate result can be used as the input of the new current subtask to enter the subsequent execution flow of the target task; or, when the subtask chain includes multiple first subtasks with an execution order, and the current subtask is the last subtask in the subtask chain, the target intermediate result can be used as the execution result of the target task.

[0030] The task execution method provided in this disclosure allows for the generation of a sub-task chain based on a target task. After determining the current sub-task from the sub-task chain, the current sub-task is executed by calling the corresponding tool to obtain an initial intermediate result. This initial intermediate result is then sent to a terminal device for display. Subsequently, in response to a result confirmation request, a target intermediate result corresponding to the initial intermediate result is obtained, and the execution result of the target task is obtained based on the target intermediate result. The result confirmation request can be generated by the terminal device in response to a result confirmation operation and sent to a service device. Compared to traditional solutions, the task execution method provided in this disclosure introduces an "initial intermediate result display process" and a "process for generating a result confirmation request in response to a result confirmation operation." This allows users to intervene in and / or adjust the initial intermediate result during the execution of the target task, obtaining a target intermediate result corresponding to the initial intermediate result. This enhances the user's controllability over the target task's execution process, improves the adaptability between the target task's execution result and the user's actual needs, and reduces the utilization of computing resources and the complexity of user operations.

[0031] In some optional implementations, step S101, namely, "obtaining a sub-task chain based on the target task", may include: The task planning model, obtained by training the target large model, is used to analyze the target task and obtain the sub-task chain.

[0032] The target task can include a task description, or a task description and reference data. For example, the task description could be, "Please help generate a 10-page, illustrated presentation in both Chinese and English based on the reference information I provide for the XX coffee cup." Here, the reference information refers to the reference data for the target task, which can include the basic attributes of the XX coffee cup (e.g., structural parameters, processing materials), images, market research reports, etc., to provide comprehensive background support for the execution of the target task, making the result (i.e., the generated presentation in both Chinese and English (Microsoft Office PowerPoint, PPT)) more professional and detailed.

[0033] Furthermore, as mentioned above, in the embodiments of this disclosure, the subtask chain may include only one first subtask, or it may include multiple first subtasks with an execution order. Based on this, it can be understood that in the embodiments of this disclosure, the aforementioned at least one first subtask is included in the output of the task planning model. In addition, the output of the task planning model may also include task parameters corresponding to at least a portion of the at least one first subtask.

[0034] For example, the task description of the target task is "Please help generate a 10-page presentation in both Chinese and English, with illustrations, based on the reference content I provided for the XX coffee cup," and the reference content serves as the reference data for the target task, including the basic attributes of the XX coffee cup, images, market research reports, etc. Please combine this with... Figure 2 After analyzing the target task using a task planning model trained on the target large model, the resulting subtask chain includes parallel first subtasks A1 and A2, first subtask B (executed after first subtasks A1 and A2), parallel first subtasks C1 and C2 (executed after first subtask B), and first subtask D (executed after first subtasks C1 and C2). Wherein: Subtask A1: Analyze the reference data of the target task to obtain the reference data analysis results; Subtask A2: Retrieve relevant external knowledge about the target task and obtain knowledge retrieval results; Subtask B: Perform operations such as summarizing and fusing the reference data parsing results and knowledge retrieval results to obtain data fusion results, which must include Chinese text and image parts; Subtask C1: Based on the data fusion results, design and generate a 10-page Chinese PPT for the XX coffee cup; First subtask C2: Translate the Chinese text portion included in the data fusion result to obtain the corresponding English text portion; Subtask D: Combine the Chinese PPT and English text to create a 10-page PPT in both Chinese and English, tailored to the XX coffee cup.

[0035] The first subtask C1 has corresponding task parameters, which can be "10 pages"; the task C2 has corresponding task parameters, which can be "English".

[0036] Furthermore, it should be noted that in this embodiment of the disclosure, the task planning model can be obtained by training the target large model with its vertical task planning capabilities. In one example, "training the target large model with its vertical task planning capabilities to obtain the task planning model" may include: Obtain the sample task and the corresponding reference task chain; Using the target large model, the sample tasks are analyzed to obtain the task chain output results; Based on the differences in task planning between the task chain output and the reference task chain, the vertical task planning capabilities of the target large model are trained to obtain the task planning model.

[0037] The sample task can be any task that can be implemented based on the content service platform, such as a content acquisition task, a content generation task, or a content processing task. The reference task chain can be obtained through manual annotation, and this embodiment of the disclosure does not impose any restrictions on this. The task planning difference between the task chain output result and the reference task chain can be obtained based on a preset first difference function. Here, the first difference function can be set according to application requirements, and this embodiment of the disclosure does not impose any restrictions on this.

[0038] In this embodiment of the present disclosure, a task planning model trained on a large target model can be used to analyze the target task and obtain a sub-task chain. Since the task planning model is trained with vertical capabilities specific to task planning, it possesses a more proficient and in-depth analytical and planning ability for the target task compared to a general target model. Therefore, using the task planning model to analyze the target task can more efficiently and accurately identify key elements and potential needs within the target task, thereby quickly generating a sub-task chain that better meets the user's actual needs. In this way, while improving the efficiency of obtaining the target task's execution results, it further enhances the fit between the target task's execution results and the user's actual needs.

[0039] Furthermore, it should be noted that in this embodiment of the disclosure, after using the task planning model obtained by training the target large model to parse the target task and obtain the sub-task chain, the sub-task chain can also be sent to the terminal device so that the terminal device can display the sub-task chain.

[0040] Subsequently, the terminal device can generate a planning determination request in response to the planning determination operation for the sub-task chain, and send the planning determination request to the service device. The service device then responds to the planning determination request by adjusting and / or adding instructions to the sub-task chain. The planning determination operation can be triggered by the user and may include sub-task deletion and / or sub-task modification operations, or tool specification operations. Correspondingly, adjusting the sub-task chain may include deleting and / or modifying some of the first sub-tasks included in the sub-task chain, or adding a first sub-task to the sub-task chain. Adding instructions to the sub-task chain may include: determining a first target sub-task from the sub-task chain (i.e., from at least one first sub-task included in the sub-task chain), and determining a first target tool specified for the first target sub-task from a plurality of candidate tools, to establish a first association between the first target sub-task and the first target tool (in this process, the first association between the first target sub-task and the first target tool is generated after the target task begins execution).

[0041] Furthermore, it should be noted that in this embodiment, the terminal device can also respond to a specified information input operation to obtain tool specification information and send the tool specification information to the service device. The service device is then used to determine a second target subtask from the subtask chain (i.e., from at least one first subtask included in the subtask chain) based on the tool specification information, and to determine a second target tool from multiple candidate tools, and to establish a first association between the second target subtask and the second target tool (in this process, the first association between the second target subtask and the second target tool is specified before the target task begins execution). The specified information input operation can be triggered by the user, for example, when inputting the target task, and uploaded to the service device along with the target task to specify the invocation of the specified second target tool when facing a second target subtask of a specific task type. For example, the specified information input operation can specify the unified invocation of the PPT generation tool when facing a document output type subtask.

[0042] Through the above methods, in this embodiment of the disclosure, the association between target subtasks and target tools (e.g., a first target subtask and a first target tool; or, for example, a second target subtask and a second target tool) in a subtask chain can be established by user specification. That is, when faced with a target subtask, the user can specify that the target tool should be directly invoked for execution. Since the user can specify the association between target subtasks and target tools according to actual needs, the tool planning model no longer needs to spend more time and computing resources dynamically matching tools during the execution of the target task. This avoids potential problems with inaccurate tool matching, thereby ensuring that the target task can be completed smoothly in the way and with the desired effect. At the same time, this method of specifying tools also helps to optimize the overall execution flow of the target task, making the execution flow of the target task more stable.

[0043] It should also be noted that, in this embodiment of the disclosure, multiple fixed workflows may be pre-stored in the service device. Each of the multiple fixed workflows may include multiple second subtasks having an execution order, and each of the multiple second subtasks has a candidate tool designated for it from multiple candidate tools and having established a second association with it.

[0044] Based on this, in this embodiment of the disclosure, step S102, "by calling the current tool corresponding to the current subtask to execute the current subtask in order to obtain the initial intermediate result," may include: If the current subtask is a self-planning task, use the tool planning model to determine the current tool corresponding to the current subtask from multiple candidate tools; Alternatively, if the current subtask does not belong to the autonomous planning category, select the candidate tool that has an association with the current subtask from among multiple candidate tools and use it as the current tool corresponding to the current subtask. Invoke the current tool to execute the current subtask in order to obtain initial intermediate results.

[0045] Among them, the self-planning task is the first sub-task that requires the use of planning tools and models for planning.

[0046] In this embodiment of the disclosure, if no candidate tool among multiple candidate tools is associated with the current subtask, it can be determined that the current subtask belongs to the autonomous planning type; or, if a candidate tool among multiple candidate tools is associated with the current subtask, it can be determined that the current subtask does not belong to the autonomous planning type. Here, "association" can be either the aforementioned first association or the second association.

[0047] Furthermore, in this embodiment of the disclosure, if it is determined that there is no candidate tool among multiple candidate tools that has a first association relationship with the current subtask, and simultaneously, it is determined that there is no fixed workflow among multiple fixed workflows that is the same as the subtask chain, then the current subtask is determined to be a self-planning task. Correspondingly, if it is determined that there is a candidate tool among multiple candidate tools that has a first association relationship with the current subtask, and / or it is determined that there is a fixed workflow among multiple fixed workflows that is the same as the subtask chain, then the current subtask is determined not to be a self-planning task. Here, "determining that there is no fixed workflow among multiple fixed workflows that is the same as the subtask chain" is equivalent to determining that there is no candidate tool among multiple candidate tools that has a second association relationship with the current subtask; "determining that there is a fixed workflow among multiple fixed workflows that is the same as the subtask chain" is equivalent to determining that there is a candidate tool among multiple candidate tools that has a second association relationship with the current subtask.

[0048] Furthermore, it should be noted that in this embodiment of the disclosure, when it is determined that among multiple candidate tools there is a candidate tool that has established a first association relationship with the current subtask, and at the same time, it is determined that among multiple fixed workflows there is a fixed workflow that is the same as the subtask chain, the candidate tool that has established a first association relationship with the current subtask can be taken as the candidate tool that has established an association relationship with the current subtask among multiple candidate tools, and then the candidate tool is taken as the current tool corresponding to the current subtask (that is, the current tool is designated by the user as the highest priority).

[0049] It should also be noted that, in this embodiment of the disclosure, the tool planning model can be obtained by training the vertical category capability of tool planning on the target large model. In one example, "training the vertical category capability of tool planning on the target large model to obtain the tool planning model" may include: Obtain sample subtasks and corresponding reference tools; Using the target large model, the tool selection result corresponding to the sample subtask is determined from multiple candidate tools; Based on the differences in tool planning between the tool selection results and the reference tools, the vertical category capability of tool planning is trained on the target large model to obtain the tool planning model.

[0050] The sample subtask can be any third subtask obtained by parsing any task that can be implemented based on the content service platform; the reference tool can be obtained through manual annotation, and this embodiment of the disclosure does not limit this; the difference between the tool selection result and the tool planning of the reference tool can be obtained based on a preset second difference function. Here, the second difference function can be set according to application requirements, and this embodiment of the disclosure does not limit this.

[0051] In this embodiment, by means of the above methods, when the current subtask belongs to the autonomous planning type, a tool planning model can be used to determine the current tool corresponding to the current subtask from multiple candidate tools; or, when the current subtask does not belong to the autonomous planning type, candidate tools that have established a relationship with the current subtask from multiple candidate tools can be used as the current tool corresponding to the current subtask. Afterwards, the current tool can be invoked to execute the current subtask to obtain initial intermediate results. Thus, on the one hand, for autonomous planning type tasks, the tool planning model can be used to quickly determine the current tool corresponding to the current subtask from multiple candidate tools, effectively improving the selection efficiency of the current tool, thereby improving the execution efficiency of the target task, that is, improving the efficiency of obtaining the execution result of the target task; on the other hand, for non-autonomous planning type tasks, candidate tools that have established a relationship with the current subtask from multiple candidate tools can be used as the current tool corresponding to the current subtask, thereby simplifying the tool determination process. This can also improve the efficiency of obtaining the execution result of the target task and reduce the resource consumption cost of service equipment.

[0052] Furthermore, in this embodiment, if no candidate tool among multiple candidate tools is associated with the current subtask, it can be determined that the current subtask belongs to the autonomous planning type; or, if a candidate tool among multiple candidate tools is associated with the current subtask, it can be determined that the current subtask does not belong to the autonomous planning type. In other words, this embodiment provides a clear, explicit, and concise judgment mechanism for classifying the current subtask, enabling rapid and accurate determination of whether the current subtask belongs to the autonomous planning type when facing complex and diverse task scenarios. This avoids execution confusion caused by ambiguous classification, thereby further improving the accuracy of the target task's execution results.

[0053] Furthermore, in some optional implementations, the result determination request can be used to request that the initial intermediate result be used as the target intermediate result, and may also include a result input request or a result modification request. Based on this, in this embodiment of the disclosure, step S104, namely, "in response to the result determination request, obtaining the target intermediate result corresponding to the initial intermediate result," may include one of the following: (1) In response to the result input request, obtain the pending result, and obtain the target intermediate result based on the pending result.

[0054] The result input request can be generated by the terminal device in response to the result input operation and sent to the service device. Here, the result input operation can be triggered by the user.

[0055] In other words, in this embodiment of the disclosure, the user can discard the initial intermediate result and obtain the result to be used through independent input, so as to obtain the target intermediate result based on the result to be used.

[0056] (2) In response to the result modification request, modify the initial intermediate result to obtain the pending result, and obtain the target intermediate result based on the pending result.

[0057] In this context, the result modification request can be generated by the terminal device in response to the result modification operation and sent to the service device. Here, the result modification operation can be triggered by the user.

[0058] In other words, in this embodiment of the disclosure, the user can obtain a pending result by modifying the initial intermediate result, and then obtain the target intermediate result based on the pending result.

[0059] In this embodiment of the present disclosure, in response to a result input request, a pending result can be obtained, and based on the pending result, a target intermediate result can be obtained. Alternatively, in response to a result modification request, the initial intermediate result can be modified to obtain a pending result, and based on the pending result, a target intermediate result can be obtained. Thus, on the one hand, users can completely discard the initial intermediate result according to their own needs and judgment, and obtain the target intermediate result by independently inputting a completely new pending result, which can then be used to generate the execution result of the target task, thereby further improving the fit between the execution result of the target task and the user's actual needs. On the other hand, when the user believes that the initial intermediate result has some reference value but some parts need adjustment, the initial intermediate result can be precisely modified through a result modification request to obtain the target intermediate result, which can then be used to generate the execution result of the target task. This improves both the fit between the execution result of the target task and the user's actual needs and the efficiency of obtaining the execution result of the target task.

[0060] In some optional implementations, the result determination request may also include a result selection request. Based on this, in this embodiment of the disclosure, step S104, namely, "in response to the result determination request, obtaining the target intermediate result corresponding to the initial intermediate result," may also include: (3) In the case of multiple initial intermediate results, in response to the result selection request, the target intermediate result is selected from the multiple initial intermediate results.

[0061] The result selection request can be generated by the terminal device in response to the result selection operation and sent to the service device. Here, the result selection operation can be triggered by the user.

[0062] In other words, in this embodiment of the disclosure, the user can select a target intermediate result from multiple initial intermediate results.

[0063] Furthermore, it should be noted that in this embodiment of the disclosure, when there are multiple initial intermediate results, the aforementioned "in response to a result modification request, modifying the initially selected intermediate results to obtain a pending result" can be: in response to a result modification request, modifying a specified intermediate result among the multiple initial intermediate results to obtain a pending result. The specified intermediate result can be determined in response to a result specification operation. Here, the result specification operation can be triggered by the user.

[0064] It should also be noted that, in this embodiment of the present disclosure, when there are multiple initial intermediate results, the multiple initial intermediate results can be obtained by the same current tool executing the current subtask. That is, when executing step S102, only one current tool can be determined and multiple initial intermediate results can be obtained using the current tool. Alternatively, multiple initial intermediate results can be obtained by multiple current tools executing the current task. That is, when executing step S102, multiple current tools can be determined and multiple initial intermediate results can be obtained one-to-one using the multiple current tools.

[0065] In this embodiment of the present disclosure, when there are multiple initial intermediate results, a target intermediate result can be selected from the multiple initial intermediate results in response to a result selection request. That is, in this embodiment of the present disclosure, users can select the most suitable target intermediate result from multiple initial intermediate results according to their actual needs, rather than being limited to a single initial intermediate result. This not only better meets the task execution needs under different task scenarios, but also further improves the adaptability between the execution result of the target task and the user's actual needs.

[0066] Furthermore, in this embodiment of the disclosure, when obtaining the target intermediate result based on the pending result, the pending result can be used as the target intermediate result, or: The results to be applied are verified to obtain the verification results; If the verification results indicate that the target result is available, the target result is used as the intermediate result.

[0067] In one example, if the result to be used is a verifiable result, the result to be used can be verified to obtain the verification result.

[0068] Verifiable results can include audiovisual results and executable programming results.

[0069] In this embodiment of the disclosure, audiovisual results can include at least one of text, images, and audio. For candidate results belonging to the audiovisual category, they can be verified from verification dimensions such as compliance, accuracy, and completeness. The final verification result (e.g., a verification score) is obtained by weighted fusion of the individual verification results under each verification dimension. If the verification result is greater than or equal to the score threshold, the candidate result is determined to be usable; otherwise, the candidate result is determined to be unusable. The score threshold can be set according to application requirements, and this embodiment of the disclosure does not limit it.

[0070] In this embodiment of the disclosure, for a pending result that is an executable programming result, a corresponding verification result can be obtained by running a test on it. For example, if running a test on an executable programming result can achieve the expected function, the pending result is determined to be usable; otherwise, the pending result is determined to be unusable.

[0071] Furthermore, in this embodiment of the disclosure, if the pending result is unavailable, an adjustment instruction can be generated for the pending result, carrying the reason for its unavailability, and sent to the terminal device. This allows the terminal device to display the adjustment instruction, enabling the user to promptly receive it and perform adjustment operations on the pending result based on the instruction. Subsequently, in response to the adjustment operation on the pending result, the terminal device can generate an adjustment request for the pending result and send it to the service device, allowing the service device to adjust the pending result in response to the adjustment request.

[0072] Through the above methods, in this embodiment of the disclosure, the result to be used can be verified to obtain a verification result. If the verification result indicates that the result to be used is usable, the result to be used can be used as the target intermediate result. In this way, on the one hand, the reliability and effectiveness of the target intermediate result can be improved, thereby improving the stability of subsequent task execution, reducing errors and failures caused by using unreliable target intermediate results, and improving the accuracy of the execution result of the target task.

[0073] Furthermore, the task execution method provided in this disclosure may also include: In response to a task intervention request, intervene in the current tool invocation.

[0074] The task intervention request can be generated by the terminal device in response to the task intervention operation and sent to the service device. Here, the intervention process can include at least one of pausing and resuming.

[0075] Through the above methods, in this embodiment of the disclosure, intervention processing can be performed on the current tool invocation in response to a task intervention request. On the one hand, this allows for timely responses to various unforeseen circumstances or changes in user needs during the execution of the target task, improving the flexibility and controllability of the target task execution. Specifically, in the event of unforeseen circumstances or changes in user needs, intervention processing can pause or resume the current tool invocation, avoiding more serious problems or resource waste caused by continuing to execute inappropriate tool invocation processes. This ensures that the target task can continue to progress in a more reasonable and realistic manner, improving the accuracy of the target task execution results. On the other hand, this intervention processing mechanism also provides users with more room for participation and control over the target task execution process, enabling users to adjust the target task execution process in real time according to actual needs, thereby enhancing the user experience.

[0076] This disclosure provides a task execution method that can be applied to a terminal device. The following will be combined with... Figure 3 The flowchart shown illustrates a task execution method provided by an embodiment of this disclosure. It should be noted that although a logical order is shown in the flowchart, in some cases, the steps shown or described in the flowchart may be executed in a different order.

[0077] Step S301: Obtain the initial intermediate results.

[0078] The initial intermediate result can be obtained by the service device based on the target task, obtaining a sub-task chain, and after determining the current sub-task from the sub-task chain, executing the current sub-task by calling the current tool corresponding to the current sub-task.

[0079] Step S302: Display the initial intermediate results.

[0080] Step S303: In response to the result confirmation operation, generate a result confirmation request.

[0081] The result determination operation can be triggered by the user, and it can be one of the result input operation and the result selection operation, or at least one of the result input operation, the result selection operation and the result modification operation. This disclosure does not limit this.

[0082] Step S304: Send the result confirmation request to the service device.

[0083] The service device can be used to respond to a result determination request, obtain a target intermediate result corresponding to the initial intermediate result, and obtain the execution result of the target task based on the target intermediate result.

[0084] The task execution method provided in this disclosure allows a service device to obtain a sub-task chain based on a target task. After determining the current sub-task from the sub-task chain, the service device executes the initial intermediate result obtained from the current sub-task by calling the current tool corresponding to the current sub-task, displays the initial intermediate result, generates a result determination request in response to a result determination operation, and sends the result determination request to the service device. This enables the service device to obtain a target intermediate result corresponding to the initial intermediate result in response to the result determination request, and obtains the execution result of the target task based on the target intermediate result. Compared to traditional solutions, the task execution method provided in this disclosure introduces an "initial intermediate result display process" and a "process for generating a result determination request in response to a result determination operation." This allows users to intervene in and / or adjust the initial intermediate result during the execution of the target task to obtain a target intermediate result corresponding to the initial intermediate result. This enhances the user's controllability over the execution process of the target task, improves the adaptability between the execution result of the target task and the user's actual needs, and reduces the occupancy of computing resources and the complexity of user operations.

[0085] In some optional implementations, the result determination request may include one of a result input request, a result modification request, and a result selection request. Based on this, in this embodiment of the disclosure, step S303, namely, "generating a result determination request in response to the result determination operation," may include: In response to a result input operation, a result input request is generated; In response to a result modification operation, a result modification request is generated; In response to a result selection operation, a result selection request is generated.

[0086] Among them, the result input operation, result modification operation, and result selection operation can be triggered by the user.

[0087] In this embodiment of the present disclosure, the server can obtain the target intermediate result by responding to at least one of a result input request, a result operation request, and a result selection request. Thus, on the one hand, users can flexibly and precisely intervene in the initial intermediate result during the execution of the target task according to their actual needs. Whether it's inputting entirely new result content, modifying existing result content, or selecting suitable result content from multiple candidate contents, this can be easily achieved, greatly improving user participation and autonomy. On the other hand, this flexible intervention mechanism ensures that the final target intermediate result better matches the user's actual needs, thereby effectively reducing repetitive operations and resource waste of service equipment caused by inaccurate initial intermediate results, improving the execution efficiency of the target task, and improving the accuracy of the target task's execution results.

[0088] In some alternative implementations, the task execution method may further include: Obtain the subtask chain; Display the sub-task chain; In response to the planning and determination operation for the sub-task chain, a planning and determination request is generated; The planning confirmation request is sent to the service device.

[0089] The service device can be used to respond to a planning determination request to determine a first target subtask from a subtask chain (that is, from at least one first subtask included in the subtask chain) and to determine a first target tool specified for the first target subtask from a plurality of candidate tools, so as to establish a first association between the first target subtask and the first target tool (in this process, the first association between the first target subtask and the first target tool is generated after the target task starts execution).

[0090] In one example, when generating a planning determination request in response to a planning determination operation for a sub-task chain, the process can involve determining a first target sub-task from the sub-task chain and a first target tool from multiple candidate tools, and then generating the planning determination request based on the task identifier of the first target sub-task and the tool identifier of the first target tool. Here, the task identifier can be a task number, task name, etc.; the tool identifier can be a tool number, tool name, etc.

[0091] In some alternative implementations, the task execution method may further include: In response to a specified input operation, the tool obtains the specified information. Send the tool-specified information to the service device.

[0092] The service device can be used to determine a second target subtask from the subtask chain (i.e., from at least one first subtask included in the subtask chain) based on tool-specified information, and to determine a second target tool from multiple candidate tools, establishing a first association between the second target subtask and the second target tool (in this process, the first association between the second target subtask and the second target tool is specified before the target task begins execution). Here, the specified information input operation can be triggered by the user, for example, when inputting the target task, and uploaded to the service device along with the target task to specify the invocation of the specified second target tool when facing a second target subtask of a specific task type. For example, the specified information input operation could specify the unified invocation of the PPT generation tool when facing a document output type subtask.

[0093] Through the above methods, in this embodiment of the disclosure, the association between target subtasks and target tools (e.g., a first target subtask and a first target tool; or, for example, a second target subtask and a second target tool) in a subtask chain can be established by user specification. That is, when faced with a target subtask, the user can specify that the target tool should be directly invoked for execution. Since the user can specify the association between target subtasks and target tools according to actual needs, the tool planning model no longer needs to spend more time and computing resources dynamically matching tools during the execution of the target task. This avoids potential problems with inaccurate tool matching, thereby ensuring that the target task can be completed smoothly in the way and with the desired effect. At the same time, this method of specifying tools also helps to optimize the overall execution flow of the target task, making the execution flow of the target task more stable.

[0094] The following, combined with Figure 4 and Figure 5 The present disclosure describes the system architecture and functional architecture of a task execution system (hereinafter referred to as the background system) deployed on a service device according to an embodiment of the present disclosure.

[0095] Intelligent Agent Service Layer Core functions: To manage the execution process of other layers (e.g., task planning layer, tool service layer, message communication layer, etc.) and process the execution results.

[0096] Task planning layer Core functionality: Efficiently and accurately identify key elements and potential needs in target tasks, thereby quickly generating sub-task chains that better meet the actual needs of users.

[0097] Implementation method: Train the target large model to perform task planning vertical category capabilities to obtain a task planning model, so that the task planning model has more proficient and in-depth analysis and planning capabilities for the target task, and use the task planning model to analyze the target task and obtain the sub-task chain.

[0098] The subtask chain may include only one first subtask, or it may include multiple first subtasks with an execution order. These multiple first subtasks may include at least some that need to be executed sequentially, or at least some that need to be executed in parallel. Based on this, it can be understood that in this embodiment, the aforementioned at least one first subtask is included in the output of the task planning model. Furthermore, the output of the task planning model may also include task parameters corresponding to at least some of the at least one first subtask.

[0099] In addition, in this embodiment of the disclosure, the function of the task planning layer also includes: classifying and judging each first subtask in the subtask chain to determine whether the first subtask belongs to the autonomous planning type of task.

[0100] Tool Service Layer Core functionality: Executes specific tool invocation operations, enabling serial invocation, parallel invocation, and flow control.

[0101] Implementation method: After determining the current subtask from the subtask chain: If the current subtask is a self-planning task, use the tool planning model to determine the current tool corresponding to the current subtask from multiple candidate tools; Alternatively, if the current subtask does not belong to the autonomous planning category, the candidate tool that has an association with the current subtask can be selected from multiple candidate tools based on the call rule engine and used as the current tool corresponding to the current subtask. Invoke the current tool to execute the current subtask in order to obtain initial intermediate results.

[0102] Message communication layer Core functionality: Enables message passing and history management between different layers, users, and tools.

[0103] Implementation: A stateless real-time communication protocol is adopted to ensure the real-time transmission of various messages, such as terminal requests, task parameters, execution status, and return results. A message middleware is built to classify, cache, and forward various messages to ensure the reliability and orderliness of message transmission. At the same time, a history record extraction module is designed to perform structured storage and key information extraction of user interaction records, tool call logs, and result data, forming user profiles and task knowledge bases to provide data support for subsequent task processing and meet the continuous needs of multiple user interactions.

[0104] File system layer Core functionality: To enable the storage, management, and access control of various file data generated during tool invocation.

[0105] Implementation: A distributed file storage system is adopted to classify and store various files such as tool input files, intermediate result files, execution result files, and system logs, supporting efficient reading, writing, and retrieval of massive files; a file metadata management mechanism is established to record information such as file creation time, associated tasks, owner, and storage path, enabling accurate location and traceability of various files; at the same time, in combination with the requirements of the risk control layer and the rights and interests layer, file access permissions are finely controlled to ensure data security.

[0106] Risk control layer Core functionality: Building a multi-level defense system to ensure the security of input and output, data compliance, and traceability of operations in the backend system.

[0107] Implementation method: (1) Multi-level interception mechanism: Input interception: Employing a dual mechanism of "regular expression matching + AI detection", the system identifies sensitive information (such as personal privacy data and illegal or non-compliant content) in the task description and reference data entered by the user, intercepts non-compliant input, and prompts the user to make corrections.

[0108] (2) Output interception: The intermediate results of each tool are subjected to security review through keyword matching and risk control classification models to ensure that they meet compliance requirements. The compliance requirements can be set according to application needs, and this embodiment does not limit them.

[0109] (3) Manual audit: Establish a manual audit channel for abnormal and sensitive information, manually review suspicious content detected by the back-end system, and implement a blocking mechanism for violating users and content.

[0110] (4) Compliance assurance mechanism: Privacy data compliance: Encrypt user privacy data (e.g., mobile phone numbers) and establish independent and secure channels for data transmission and storage to ensure compliance with relevant laws and regulations.

[0111] (5) Access control system: Based on user granularity, an access control model is built to assign differentiated tool usage permissions, file access permissions and operation permissions to users with different identities to prevent unauthorized operations.

[0112] (6) Audit trail system: Completely record user operation logs, tool call logs, data access logs and risk control interception logs. Log information is kept for at least 6 months and supports post-event traceability and audit analysis.

[0113] rights layer Core functionality: To enable accurate calculation of user rights and control over tool usage permissions.

[0114] Implementation: A user rights management system is built, linking user identity information (e.g., regular users, VIP users, enterprise users), paid packages, points, and usage history data to establish a rights calculation model. When a user initiates a request to use a specified target tool, the backend system can automatically determine whether the user has the necessary permissions to use that target tool (e.g., free users can only use basic tools, VIP users can use advanced tools) and usage limits based on the user's current rights status. Only after successful permission verification is the tool scheduling process triggered; otherwise, the user is prompted to upgrade their rights or purchase a service.

[0115] Please see Figure 6 This is a schematic diagram illustrating an application scenario of a task execution method provided in an embodiment of this disclosure.

[0116] The task execution method provided in this disclosure is applied to an electronic device. The electronic device is either a service device or a terminal device.

[0117] When an electronic device is a service device (e.g., a server, workbench, mainframe computer, or other similar computing device), it is used for: Based on the target task, a chain of subtasks is obtained; After determining the current subtask from the subtask chain, the current subtask is executed by calling the current tool corresponding to the current subtask in order to obtain the initial intermediate result; The initial intermediate results are sent to the terminal device; the terminal device is used to display the initial intermediate results. In response to the result determination request, a target intermediate result corresponding to the initial intermediate result is obtained; wherein, the result determination request is generated by the terminal device in response to the result determination operation and sent to the service device; Based on the intermediate results of the target, the execution result of the target task is obtained.

[0118] When an electronic device is a terminal device (e.g., a mainframe computer, a conventional computer, a portable terminal, a smart wearable device, or other similar computing device), it is used for: Obtain initial intermediate results; wherein, the initial intermediate results are obtained by the service device based on the target task, obtaining a sub-task chain, and after determining the current sub-task from the sub-task chain, executing the current sub-task by calling the current tool corresponding to the current sub-task; Display the initial intermediate results; In response to the result confirmation operation, a result confirmation request is generated; The result determination request is sent to the service device; wherein, in response to the result determination request, the service device obtains the target intermediate result corresponding to the initial intermediate result, and obtains the execution result of the target task based on the target intermediate result.

[0119] It should be noted that, in the embodiments disclosed herein, Figure 6 The application scenario diagrams shown are for illustrative purposes only and are not restrictive. Those skilled in the art can use them as a basis for their own interpretation. Figure 6 The examples may be modified in various obvious ways and / or substitutions, and the resulting technical solutions still fall within the scope of the disclosure of the embodiments of this disclosure.

[0120] To better implement the aforementioned task execution method applied to service devices, this disclosure also provides a task execution apparatus that can be integrated into the service device. The following will be combined with... Figure 7 The schematic block diagram shown illustrates a task execution device 700 provided in the disclosed embodiment.

[0121] Task execution device 700, including: Call chain determination unit 701 is used to obtain a subtask chain based on the target task; The first intermediate result acquisition unit 702 is used to obtain an initial intermediate result by calling the current tool corresponding to the current subtask after determining the current subtask from the subtask chain; The intermediate result sending unit 703 is used to send the initial intermediate result to the terminal device; wherein the terminal device is used to display the initial intermediate result; The second intermediate result acquisition unit 704 is used to obtain a target intermediate result corresponding to the initial intermediate result in response to a result determination request; wherein, the result determination request is generated by the terminal device in response to the result determination operation and sent to the service device. The execution result acquisition unit 705 is used to obtain the execution result of the target task based on the target intermediate result.

[0122] In some optional implementations, the first intermediate result acquisition unit 702 is used for: If the current subtask is a self-planning task, use the tool planning model to determine the current tool corresponding to the current subtask from multiple candidate tools; Alternatively, if the current subtask does not belong to the autonomous planning category, select the candidate tool that has an association with the current subtask from among multiple candidate tools and use it as the current tool corresponding to the current subtask. Invoke the current tool to execute the current subtask in order to obtain initial intermediate results.

[0123] In some alternative implementations, the task execution device 700 further includes a first tool binding unit for: Send the sub-task chain to the terminal device; the terminal device is used to display the sub-task chain. In response to the planning determination request, a first target subtask is determined from the subtask chain, and a first target tool is determined from multiple candidate tools to specify the first target subtask, so as to establish a first association between the first target subtask and the first target tool; wherein, the planning determination request is generated by the terminal device in response to the planning determination operation for the subtask chain and sent to the service device; the first association is one of the associations.

[0124] In some alternative implementations, the task execution device 700 further includes a second tool binding unit for: Get the specified information from the tool; Based on the tool specification information, determine the second target subtask from the subtask chain, and determine the second target tool from multiple candidate tools; Establish a first association between the second objective subtask and the second objective tool; wherein, the first association is one of the associations.

[0125] In some optional embodiments, the task execution device 700 further includes a task type determination unit, used for: If no candidate tool among multiple candidate tools is associated with the current subtask, the current subtask is determined to be a self-planning task. Alternatively, if among multiple candidate tools there is one that is related to the current subtask, it can be determined that the current subtask does not belong to the autonomous planning category.

[0126] In some optional implementations, the result determination request includes a result input request or a result modification request; the second intermediate result acquisition unit 704 is used for: In response to a result input request, a pending result is obtained, and based on the pending result, a target intermediate result is obtained; wherein, the result input request is generated by the terminal device in response to the result input operation and sent to the service device; Alternatively, in response to a result modification request, the initial intermediate result is modified to obtain a pending result, and the target intermediate result is obtained based on the pending result; wherein, the result modification request is generated by the terminal device in response to the result modification operation.

[0127] In some optional implementations, the second intermediate result acquisition unit 704 is used for: If it is determined that the result to be used is a verifiable result, then the result to be used is verified to obtain the verification result; If the verification results indicate that the target result is available, the target result is used as the intermediate result.

[0128] In some optional implementations, the result determination request includes a result selection request; the second intermediate result acquisition unit 704 is used for: When there are multiple initial intermediate results, in response to a result selection request, a target intermediate result is selected from the multiple initial intermediate results; wherein, the result selection request is generated by the terminal device in response to the result selection operation and sent to the service device.

[0129] In some optional implementations, the call chain determination unit 701 is used for: The task planning model, obtained by training the target large model, is used to analyze the target task and obtain the sub-task chain.

[0130] In some alternative implementations, the task execution device 700 further includes a task intervention unit for: In response to a task intervention request, the current tool invocation is intervened; wherein, the task intervention request is generated by the terminal device in response to the task intervention operation and sent to the service device; the intervention process includes at least one of pausing and resuming.

[0131] In this embodiment of the disclosure, the specific functions and examples of each unit in the task execution device 700 can be found in the relevant descriptions of the corresponding steps in the aforementioned embodiment of the task execution method applied to the service device, and will not be repeated here.

[0132] To better implement the aforementioned task execution method applied to terminal devices, this disclosure also provides a task execution apparatus that can be integrated into the terminal device. The following will be combined with... Figure 8 The schematic block diagram shown illustrates a task execution device 800 provided in the disclosed embodiment.

[0133] Task execution device 800, including: The third intermediate result acquisition unit 801 is used to acquire the initial intermediate result; wherein, the initial intermediate result is obtained by the service device based on the target task, obtaining a sub-task chain, and after determining the current sub-task from the sub-task chain, executing the current sub-task by calling the current tool corresponding to the current sub-task; Intermediate result display unit 802 is used to display the initial intermediate results; The request generation unit 803 is used to generate a result determination request in response to the result determination operation; The request sending unit 804 is used to send a result determination request to the service device; wherein, the service device is used to respond to the result determination request, obtain a target intermediate result corresponding to the initial intermediate result, and obtain the execution result of the target task based on the target intermediate result.

[0134] In some optional implementations, the result determination request includes one of a result input request, a result modification request, and a result selection request; the request generation unit 803 is used for one of the following: In response to a result input operation, a result input request is generated; In response to a result modification operation, a result modification request is generated; In response to a result selection operation, a result selection request is generated.

[0135] In some alternative implementations, the task execution device 800 further includes a first tool designation unit, used for: Obtain the subtask chain; Display the sub-task chain; In response to the planning and determination operation for the sub-task chain, a planning and determination request is generated; The planning determination request is sent to the service device; wherein, in response to the planning determination request, the service device determines a first target subtask from the subtask chain and determines a first target tool designated for the first target subtask from multiple candidate tools, so as to establish a first association between the first target subtask and the first target tool.

[0136] In some alternative implementations, the first tool designator unit is used for: In response to the planning and determination operation for the sub-task chain, the first target sub-task is determined from the sub-task chain, and the first target tool is determined from multiple candidate tools; Based on the task identifier of the first target subtask and the tool identifier of the first target tool, a planning determination request is generated.

[0137] In some alternative implementations, the task execution device 800 further includes a second tool designation unit for: In response to a specified input operation, the tool obtains the specified information. The tool specification information is sent to the service device; wherein, the service device is used to determine the second target subtask from the subtask chain and the second target tool from multiple candidate tools based on the tool specification information, and to establish a first association relationship between the second target subtask and the second target tool.

[0138] In this embodiment of the disclosure, the specific functions and examples of each unit in the task execution device 800 can be found in the relevant descriptions of the corresponding steps in the aforementioned embodiment of the task execution method applied to a terminal device, and will not be repeated here.

[0139] The collection, storage, use, processing, transmission, provision, and disclosure of any type of information, such as user personal information, in this technical solution comply with relevant laws and regulations and do not violate public order and good morals.

[0140] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0141] Figure 9A schematic structural block diagram of an example electronic device 900 that can be used to implement embodiments of the present disclosure is shown. Electronic device 900 is intended to represent various forms of digital computers, such as in-vehicle computing devices, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 900 can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0142] like Figure 9 As shown, the electronic device 900 includes a computing unit 901, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 902 or a computer program loaded from a storage unit 908 into a random access memory (RAM) 903. The RAM 903 may also store various programs and data required for the operation of the electronic device 900. The computing unit 901, ROM 902, and RAM 903 are interconnected via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0143] Multiple components in electronic device 900 are connected to I / O interface 905, including: input unit 906, such as keyboard, mouse, etc.; output unit 907, such as various types of renderers, speakers, etc.; storage unit 908, such as disk, optical disk, etc.; and communication unit 909, such as network card, modem, wireless transceiver, etc. Communication unit 909 allows electronic device 900 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0144] The computing unit 901 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose AI computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 901 performs the various methods and processes described above, such as task execution methods. For example, in some embodiments, the task execution method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 900 via ROM 902 and / or communication unit 909. When the computer program is loaded into RAM 903 and executed by the computing unit 901, one or more steps of the task execution methods described above may be performed. Alternatively, in other embodiments, the computing unit 901 may be configured as a task execution method by any other suitable means (e.g., by means of firmware).

[0145] Various embodiments of the systems and techniques described above 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), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transferring data and instructions to the storage system, the at least one input device, and the at least one output device.

[0146] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data optimization device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0147] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, RAM, ROM, erasable programmable read-only memory (EPROM) or flash memory, optical fibers, compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0148] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a rendering device (e.g., a cathode ray tube (CRT) renderer or a liquid crystal display (LCD)) for rendering information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices are also used to provide interaction with the user; for example, 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 sound input, voice input, or tactile input).

[0149] The systems and technologies described herein can be implemented in computing systems that include back-end components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0150] A computer system can include client and server components. Clients and servers are generally located far apart and typically interact via a communication network. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, a server in a distributed system, or a server incorporating blockchain technology.

[0151] This disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to perform a task execution method.

[0152] This disclosure also provides a computer program product, including a computer program that implements a task execution method when executed by a processor.

[0153] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure is achieved, and this is not limited herein. Furthermore, in this disclosure, relational terms such as "first," "second," and "third" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Additionally, "multiple" in this disclosure can be understood as at least two.

[0154] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A task execution method, comprising: Based on the target task, a chain of subtasks is obtained; After determining the current subtask from the subtask chain, the current subtask is executed by calling the current tool corresponding to the current subtask to obtain an initial intermediate result; The initial intermediate results are sent to a terminal device; wherein the terminal device is used to display the initial intermediate results; In response to a result determination request, a target intermediate result corresponding to the initial intermediate result is obtained; wherein, the result determination request is generated by the terminal device in response to the result determination operation and sent to the service device; Based on the intermediate results of the target, the execution result of the target task is obtained.

2. The method according to claim 1, wherein, The step of executing the current subtask by invoking the current tool corresponding to the current subtask to obtain initial intermediate results includes: If the current subtask belongs to the autonomous planning type, the tool planning model is used to determine the current tool corresponding to the current subtask from multiple candidate tools; Alternatively, if the current subtask does not belong to the autonomous planning type of task, the candidate tool that has an association relationship with the current subtask among multiple candidate tools shall be used as the current tool corresponding to the current subtask. The current tool is invoked to execute the current subtask in order to obtain the initial intermediate result.

3. The method according to claim 2, further comprising: The sub-task chain is sent to the terminal device; wherein the terminal device is used to display the sub-task chain. In response to a planning determination request, a first target subtask is determined from the subtask chain, and a first target tool is determined from the plurality of candidate tools to specify the first target subtask, thereby establishing a first association between the first target subtask and the first target tool; wherein, the planning determination request is generated by the terminal device in response to a planning determination operation for the subtask chain and sent to the service device; the first association is one of the associations.

4. The method according to claim 2, further comprising: Get the specified information from the tool; Based on the tool specification information, a second target subtask is determined from the subtask chain, and a second target tool is determined from the plurality of candidate tools; Establish a first association between the second target subtask and the second target tool; wherein the first association is one of the associations.

5. The method according to claim 2, further comprising: If none of the candidate tools among the multiple candidate tools are associated with the current subtask, the current subtask is determined to be a self-planning task. Alternatively, if among the multiple candidate tools there is a candidate tool that has an association with the current subtask, it is determined that the current subtask does not belong to the autonomous planning type of task.

6. The method according to claim 2, wherein, The result determination request includes a result input request or a result modification request; The step of responding to a result determination request and obtaining a target intermediate result corresponding to the initial intermediate result includes: In response to the result input request, a pending result is obtained, and based on the pending result, the target intermediate result is obtained; wherein, the result input request is generated by the terminal device in response to the result input operation and sent to the service device; Alternatively, in response to the result modification request, the initial intermediate result is modified to obtain a pending result, and the target intermediate result is obtained based on the pending result; wherein, the result modification request is generated by the terminal device in response to the result modification operation.

7. The method according to claim 6, wherein, The process of obtaining the target intermediate result based on the pending result includes: The proposed result is verified to obtain the verification result; If the verification result indicates that the pending result is available, the pending result is used as the target intermediate result.

8. The method according to claim 6, wherein, The result determination request includes a result selection request; the step of obtaining a target intermediate result corresponding to the initial intermediate result in response to the result determination request includes: In the case where there are multiple initial intermediate results, in response to the result selection request, the target intermediate result is selected from the multiple initial intermediate results; wherein, the result selection request is generated by the terminal device in response to the result selection operation and sent to the service device.

9. The method according to claim 1, wherein, The process of obtaining a sub-task chain based on the target task includes: The target task is analyzed using a task planning model obtained by training the target large model, and the sub-task chain is obtained.

10. The method according to claim 1, further comprising: In response to a task intervention request, the invocation of the current tool is intervened; wherein the task intervention request is generated by the terminal device in response to a task intervention operation and sent to the service device; the intervention process includes at least one of pausing and resuming.

11. A task execution method, comprising: Obtain initial intermediate results; wherein, the initial intermediate results are obtained by the service device based on the target task, obtaining a sub-task chain, and after determining the current sub-task from the sub-task chain, executing the current sub-task by calling the current tool corresponding to the current sub-task; Display the initial intermediate results; In response to the result confirmation operation, a result confirmation request is generated; The result determination request is sent to the service device; wherein, in response to the result determination request, the service device obtains a target intermediate result corresponding to the initial intermediate result, and obtains the execution result of the target task based on the target intermediate result.

12. The method according to claim 11, wherein, The result determination request includes one of the following: a result input request, a result modification request, and a result selection request; The response to the result determination operation, generating a result determination request, includes one of the following: In response to the result input operation, the result input request is generated; In response to the result modification operation, the result modification request is generated; In response to the result selection operation, the result selection request is generated.

13. The method of claim 11, further comprising: Obtain the sub-task chain; Display the sub-task chain; In response to the planning and determination operation for the sub-task chain, a planning and determination request is generated; The planning determination request is sent to the service device; wherein, in response to the planning determination request, the service device determines a first target subtask from the subtask chain and determines a first target tool designated for the first target subtask from a plurality of candidate tools, so as to establish a first association between the first target subtask and the first target tool.

14. The method according to claim 13, wherein, The step of generating a planning determination request in response to a planning determination operation for the sub-task chain includes: In response to the planning and determination operation for the sub-task chain, a first target sub-task is determined from the sub-task chain, and the first target tool is determined from the plurality of candidate tools; The planning determination request is generated based on the task identifier of the first target sub-task and the tool identifier of the first target tool.

15. The method of claim 11, further comprising: In response to a specified input operation, the tool obtains the specified information. The tool specification information is sent to the service device; wherein the service device is used to determine a second target subtask from the subtask chain based on the tool specification information, and to determine a second target tool from multiple candidate tools, and to establish a first association relationship between the second target subtask and the second target tool.

16. A task execution device, comprising: The call chain determination unit is used to obtain the subtask chain based on the target task; The first intermediate result acquisition unit is used to, after determining the current subtask from the subtask chain, execute the current subtask by calling the current tool corresponding to the current subtask to obtain an initial intermediate result; An intermediate result sending unit is used to send the initial intermediate result to a terminal device; wherein the terminal device is used to display the initial intermediate result; The second intermediate result acquisition unit is used to obtain a target intermediate result corresponding to the initial intermediate result in response to a result determination request; wherein the result determination request is generated by the terminal device in response to the result determination operation and sent to the service device. The execution result acquisition unit is used to obtain the execution result of the target task based on the target intermediate result.

17. A task execution device, comprising: The third intermediate result acquisition unit is used to acquire an initial intermediate result; wherein, the initial intermediate result is obtained by the service device based on the target task, obtaining a sub-task chain, and after determining the current sub-task from the sub-task chain, executing the current sub-task by calling the current tool corresponding to the current sub-task; An intermediate result display unit is used to display the initial intermediate results; The request generation unit is used to generate a result determination request in response to the result determination operation; A request sending unit is configured to send the result determination request to the service device; wherein, the service device is configured to, in response to the result determination request, obtain a target intermediate result corresponding to the initial intermediate result, and obtain the execution result of the target task based on the target intermediate result.

18. An electronic device comprising: At least one processor; A memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method according to any one of claims 1 to 15.

19. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1 to 15.

20. A computer program product comprising a computer program; wherein, When the computer program is executed by a processor, it can implement the method of any one of claims 1 to 15.

Images