Construction method and device of industry large model, tool chain system, storage medium, system and program product

Abstract

The invention provides a construction method and device of an industry large model, a tool chain system, a storage medium, a system and a program product, and the construction method of the industry large model comprises the steps: responding to a generation object configuration operation executed by a user on a first configuration interface, and determining object information of at least one generation object; in response to an input configuration operation executed by the user on the second configuration interface, determining an input data format; tool configuration operation executed by the user on the third configuration interface is responded, an execution tool is configured for each generated object, and the types of the execution tools comprise at least one of a large language model, a machine learning model, a deep learning model, a function and a mechanism model; in response to an output configuration operation executed by the user on the fourth configuration interface, determining a data output style; and constructing an industry large model based on the determined information. The method can support integration of different execution tools and is jointly used for solving actual industry problems, and landing application of an industry large model is effectively achieved.

Description

Technical Field

[0001] The present disclosure generally relates to the field of big model technology, and more specifically, to a method, device, tool chain system, storage medium, system and program product for constructing an industry big model. Background Art

[0002] With the rise of large language models represented by GPT, the industry has begun to explore the construction of industry large model applications. The so-called industry large model can be understood as an industry-specific large model and application with dedicated knowledge and capabilities, such as Alibaba's geographic big model DDE (Deep-time Digital Earth).

[0003] However, the common solutions for existing industry large-scale model applications are all knowledge-layer application explorations based on RAG (Retrieval Augmented Generation, which uses industry-related textbooks and knowledge information as external knowledge bases) technology or SFT (Supervised Fine-Tuning, which fine-tunes open source or closed source base models based on data marked manually or by other means) technology. These technologies do not provide good coverage of the application and tool layers within the industry, and are unable to integrate various tools such as large language models, traditional machine learning models, deep learning models, function calculations, and mechanism models. They are unable to solve further business problems and implement them, such as earthquake prediction, geomorphic analysis, geological risk assessment, oil and gas resource exploration, and other business problems actually encountered in the industry. Summary of the invention

[0004] The present disclosure provides a method, device, tool chain system, storage medium, system and program product for constructing an industry big model, which are used to solve at least one of the above problems.

[0005] According to one aspect of the present disclosure, a method for constructing an industry big model is provided, the method comprising: displaying a first configuration interface, and determining object information of at least one generated object of the industry big model in response to a generated object configuration operation performed by a user on the first configuration interface; displaying a second configuration interface, and determining an input data format of the industry big model in response to an input configuration operation performed by a user on the second configuration interface; displaying a third configuration interface, and configuring an execution tool for each of the at least one generated object in response to a tool configuration operation performed by a user on the third configuration interface, wherein the execution tool is used to perform data processing to obtain the generated object, and the type of the execution tool includes at least one of a large language model, a machine learning model, a deep learning model, a function, and a mechanism model; displaying a fourth configuration interface, and determining a data output style of the industry big model in response to an output configuration operation performed by a user on the fourth configuration interface, wherein the data output style is a display style of the output data of the industry big model on an output interface; constructing the industry big model based on the object information of the at least one generated object of the industry big model, the input data format, the execution tool of each of the at least one generated object, and the data output style.

[0006] Optionally, the object information includes at least one of a name and a data type, and the data type includes at least one of a value and a text.

[0007] Optionally, the first configuration interface includes at least two object type options; wherein, one of the at least two object type options is a complex object type, and the complex object type indicates that the generation data of the industry big model includes at least two generation objects; and at least one of the at least two object type options is a non-complex object type, and the non-complex object type indicates that the generation data of the industry big model is one generation object.

[0008] Optionally, the at least one non-complex object type includes at least one of the following: a text object type, an audio object type, and a video object type.

[0009] Optionally, in response to the generation object configuration operation performed by the user in the first configuration interface, the object information of at least one generation object of the industry large model is determined, including: in response to the user selecting the complex object type in the first configuration interface, displaying the generation object input interface in the first configuration interface; in response to the user input operation on the generation object input interface, determining the object information of at least two generation objects of the industry large model.

[0010] Optionally, the generated object input interface includes a second input interface for inputting task information of a target task to be performed by the industry big model; determining object information of at least two generated objects of the industry big model in response to an input operation of the user on the generated object input interface includes: receiving the task information in response to an input operation of the user on the second input interface; processing the task information using a task decomposition model to decompose the target task into at least two subtasks, and obtaining object information of the at least two generated objects that can be generated by the at least two subtasks, wherein the subtasks are used to generate the generated objects.

[0011] Optionally, displaying the second configuration interface includes: in response to the user selecting the complex object type in the first configuration interface, displaying the second configuration interface corresponding to the complex object type, and displaying specified format options and other format options in the second configuration interface corresponding to the complex object type, wherein the specified format options include specified format information and a format preview interface, the format preview interface is used to provide a file preview of the specified format, and the other format options include other format information and a parsing script upload interface, the parsing script upload interface is used to receive parsing scripts for files in other formats uploaded by the user; in response to the user selecting the non-complex object type in the first configuration interface, displaying the second configuration interface corresponding to the non-complex object type, and displaying at least one data format option of the non-complex object type selected by the user in the second configuration interface corresponding to the non-complex object type, and the at least one data format option is used for the user to select a data format for the non-complex object type.

[0012] Optionally, the third configuration interface includes a tool type selection interface and a configuration editing interface for each generation object of the at least one generation object; wherein the tool type selection interface is used to select the type of execution tool of the corresponding generation object; the configuration editing interface is used to edit the configuration information of the execution tool of the corresponding generation object, the configuration information of the execution tool of the model type includes the model URL and the model input, and the editing process includes at least one of the following: addition, deletion, and modification.

[0013] Optionally, the fourth configuration interface includes a component navigation area, a component configuration interface, and an output board preview area; wherein, the component navigation area displays a plurality of selectable display components, the display components are used to display the generated objects of the industry large model, and the types of the display components include at least one of the following: text, picture, data chart; the component configuration interface is used to configure the selected display component; the output board preview area is used to display the preview effect of the selected display component, and the position of the display component in the output board preview area is movable so that the user can edit the display position of the display component.

[0014] According to another aspect of the present disclosure, a construction device for an industry big model is provided, the construction device comprising: a first configuration unit, configured to display a first configuration interface, and determine the object information of at least one generation object of the industry big model in response to a generation object configuration operation performed by a user on the first configuration interface; a second configuration unit, configured to display a second configuration interface, and determine the input data format of the industry big model in response to an input configuration operation performed by a user on the second configuration interface; a third configuration unit, configured to display a third configuration interface, and configure an execution tool for each generation object of the at least one generation object in response to a tool configuration operation performed by a user on the third configuration interface, wherein the execution tool is used The processing unit executes data to obtain the generated object, and the type of the execution tool includes at least one of a large language model, a machine learning model, a deep learning model, a function, and a mechanism model; the fourth configuration unit is configured to display a fourth configuration interface, and determine the data output style of the industry big model in response to the output configuration operation performed by the user on the fourth configuration interface, wherein the data output style is the display style of the output data of the industry big model on the output interface; the construction unit is configured to construct the industry big model based on the object information of the at least one generated object of the industry big model, the input data format, the execution tool of each generated object in the at least one generated object, and the data output style.

[0015] Optionally, the object information includes at least one of a name and a data type, and the data type includes at least one of a value and a text.

[0016] Optionally, the first configuration interface includes at least two object type options; wherein, one of the at least two object type options is a complex object type, and the complex object type indicates that the generation data of the industry big model includes at least two generation objects; and at least one of the at least two object type options is a non-complex object type, and the non-complex object type indicates that the generation data of the industry big model is one generation object.

[0017] Optionally, the at least one non-complex object type includes at least one of the following: a text object type, an audio object type, and a video object type.

[0018] Optionally, the first configuration unit is further configured to: display a generation object input interface on the first configuration interface in response to a user selecting the complex object type on the first configuration interface; and determine object information of at least two generation objects of the industry large model in response to a user input operation on the generation object input interface.

[0019] Optionally, the generated object input interface includes a second input interface for inputting task information of a target task to be performed by the industry large model; the first configuration unit is also configured to: receive the task information in response to a user's input operation on the second input interface; process the task information using a task decomposition model to decompose the target task into at least two subtasks, and obtain object information of the at least two generated objects that can be generated by the at least two subtasks, wherein the subtasks are used to generate the generated objects.

[0020] Optionally, the second configuration unit is further configured to: in response to the user selecting the complex object type in the first configuration interface, display the second configuration interface corresponding to the complex object type, and display specified format options and other format options in the second configuration interface corresponding to the complex object type, wherein the specified format options include specified format information and a format preview interface, the format preview interface is used to provide a file preview of the specified format, and the other format options include other format information and a parsing script upload interface, the parsing script upload interface is used to receive parsing scripts for other format files uploaded by the user; in response to the user selecting the non-complex object type in the first configuration interface, display the second configuration interface corresponding to the non-complex object type, and display at least one data format option of the non-complex object type selected by the user in the second configuration interface corresponding to the non-complex object type, and the at least one data format option is used for the user to select a data format for the non-complex object type.

[0021] Optionally, the third configuration interface includes a tool type selection interface and a configuration editing interface for each generation object of the at least one generation object; wherein the tool type selection interface is used to select the type of execution tool of the corresponding generation object; the configuration editing interface is used to edit the configuration information of the execution tool of the corresponding generation object, the configuration information of the execution tool of the model type includes the model URL and the model input, and the editing process includes at least one of the following: addition, deletion, and modification.

[0022] Optionally, the fourth configuration interface includes a component navigation area, a component configuration interface, and an output board preview area; wherein, the component navigation area displays a plurality of selectable display components, the display components are used to display the generated objects of the industry large model, and the types of the display components include at least one of the following: text, picture, data chart; the component configuration interface is used to configure the selected display component; the output board preview area is used to display the preview effect of the selected display component, and the position of the display component in the output board preview area is movable so that the user can edit the display position of the display component.

[0023] According to another aspect of the present disclosure, a tool chain system for an industry big model is provided, the tool chain system comprising: a construction module, used to execute the construction method of the industry big model as described above, so as to construct the industry big model; and a calling module, used to receive input data and call the industry big model to perform data processing on the input data.

[0024] Optionally, the tool chain system further includes: a tool module for developing and optimizing the execution tool.

[0025] According to another aspect of the present disclosure, a computer-readable storage medium storing instructions is provided, wherein when the instructions are executed by at least one computing device, the at least one computing device is prompted to execute the method for constructing an industry big model as described above.

[0026] According to another aspect of the present disclosure, a system is provided, comprising at least one computing device and at least one storage device storing instructions, wherein the instructions, when executed by the at least one computing device, prompt the at least one computing device to execute the method for constructing an industry big model as described above.

[0027] According to another aspect of the present disclosure, a computer program product is provided, comprising instructions, wherein when the instructions are executed by at least one computing device, the at least one computing device is prompted to execute the method for constructing the industry big model as described above.

[0028] According to the method, device, tool chain system, storage medium, system and program product for constructing the industry big model of the exemplary embodiment of the present disclosure, a systematic model construction method is proposed. On the one hand, it provides standardized configuration functions for the input and output of the model. On the other hand, it is based on decomposing the generation result of the industry big model into at least one generation object, and configuring a corresponding execution tool for each generation object, so as to use the specified execution tool to obtain the corresponding generation object. It can support the integration of different execution tools such as large language models, machine learning models, deep learning models, functions, mechanism models, etc., which are used together to solve actual industry problems, effectively realizing the practical application of industry big models.

[0029] Additional aspects and / or advantages of the present general inventive concept will be set forth in part in the following description and in part will be apparent from the description or may be learned through practice of the present general inventive concept. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] These and / or other aspects and advantages of the present disclosure will become clear and more easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0031] Figure 1 is a flowchart illustrating a method for constructing an industry big model according to an exemplary embodiment of the present disclosure;

[0032] Figure 2 is a schematic diagram showing a homepage of a construction platform according to a specific embodiment of the present disclosure;

[0033] Figure 3 is a schematic diagram showing a first configuration interface according to a specific embodiment of the present disclosure;

[0034] Figure 4 is a schematic diagram showing a second configuration interface corresponding to an audio object type according to a specific embodiment of the present disclosure;

[0035] Figure 5 is a schematic diagram showing a third configuration interface corresponding to a non-complex object type according to a specific embodiment of the present disclosure;

[0036] Figure 6 is a schematic diagram showing a fourth configuration interface corresponding to an audio object type according to a specific embodiment of the present disclosure;

[0037] Figure 7 is a schematic diagram showing a construction success prompt window according to a specific embodiment of the present disclosure;

[0038] Figure 8 is a schematic diagram showing a generated object input interface according to a specific embodiment of the present disclosure;

[0039] Fig. 9 is a schematic diagram showing a second configuration interface corresponding to a complex object type according to a specific embodiment of the present disclosure;

[0040] Fig.10 is a schematic diagram showing a third configuration interface corresponding to a complex object type according to a specific embodiment of the present disclosure;

[0041] Fig.11 is a schematic diagram showing a tool editing window displaying a fixed value tool on a third configuration interface according to a specific embodiment of the present disclosure;

[0042] Fig.12is a schematic diagram showing a tool editing window of a professional model tool displayed on a third configuration interface according to a specific embodiment of the present disclosure;

[0043] Fig.13 is a schematic diagram showing a tool editing window of a large language model tool displayed on a third configuration interface according to a specific embodiment of the present disclosure;

[0044] Fig.14 is a schematic diagram showing a fourth configuration interface according to a specific embodiment of the present disclosure;

[0045] Fig.15 is a schematic diagram showing an input interface of a large model of a chronic disease management industry according to a specific embodiment of the present disclosure;

[0046] Fig.16 is a schematic diagram showing an output dashboard of a large model for the chronic disease management industry according to a specific embodiment of the present disclosure;

[0047] Fig.17 is a block diagram illustrating an apparatus for constructing an industry big model according to an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0048] The following description with reference to the accompanying drawings is provided to assist in a comprehensive understanding of the embodiments of the present invention as defined by the claims and their equivalents. Various specific details are included to assist in understanding, but these details are considered to be exemplary only. Therefore, one of ordinary skill in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present invention. In addition, descriptions of well-known functions and structures are omitted for clarity and brevity.

[0049] It should be noted that the phrase "at least one of the items" in the present disclosure includes three types of parallel situations: "any one of the items", "a combination of any number of the items", and "all of the items". For example, "including at least one of A and B" includes the following three parallel situations: (1) including A; (2) including B; (3) including A and B. Another example is "executing at least one of step 1 and step 2" which means the following three parallel situations: (1) executing step 1; (2) executing step 2; (3) executing step 1 and step 2.

[0050] Refer to the following Figures 1 to 17 The method, apparatus, tool chain system, storage medium, system and program product for constructing an industry big model according to exemplary embodiments of the present disclosure are described in detail.

[0051] Figure 11 is a flowchart showing a method for constructing a large industry model according to an exemplary embodiment of the present disclosure. The method for constructing a large industry model according to an exemplary embodiment of the present disclosure can be implemented in a computing device with sufficient computing power.

[0052] Reference Figure 1 In step S101, a first configuration interface is displayed, and in response to a generated object configuration operation performed by a user in the first configuration interface, object information of at least one generated object of the industry macro model is determined.

[0053] The first configuration interface provides a standardized configuration function for the model generation object, which is used to configure the information of at least one generated object to be generated for the constructed industry big model. It should be understood that at least one generated object means that the industry big model can have the ability to generate one or more objects.

[0054] Optionally, the object information includes at least one of a name and a data type, the name is used to name the generated object of the industry big model, and the data type is used to indicate that the industry big model is expected to be used to generate objects of the corresponding data type. Among them, since the data type will affect the execution process of the constructed industry big model, it is often indispensable information. The data type includes at least one of a value and a text. For the value, it can be further refined to a numerical type such as an integer, a floating point type, and for the text, it can correspond to a string. By not limiting the data type to the text predicted by the conventional large language model, but also extending it to a numerical value, the constructed industry big model can have the ability to predict different types of data, which is more helpful to solve practical industry problems. It should be understood that the data types introduced here are intended to emphasize that the industry big model transcends the limitations of text prediction, and the numerical value is the data type usually involved in solving practical industry problems, but according to the needs of the actual problem to be solved, the data type can also include other types, such as array data (Array), categorical data (Categorical), Boolean data (Boolean), etc., and the present disclosure does not limit this.

[0055] In step S102, a second configuration interface is displayed, and in response to an input configuration operation performed by a user on the second configuration interface, an input data format of the industry macro model is determined.

[0056] The second configuration interface provides a standardized configuration function for the input data format. It should be understood that since the basic purpose of the industry model is to make homogeneous output based on the input, that is, the input data is consistent with the output data format, the output data format is the same as the input data format configured here, and no additional configuration is required. In the case of changes in the output data format, the corresponding configuration interface and configuration function can be provided additionally.

[0057] In step S103, a third configuration interface is displayed, and in response to a tool configuration operation performed by a user on the third configuration interface, an execution tool is configured for each generation object of at least one generation object.

[0058] The third configuration interface provides a standardized configuration function for the execution tool of each generated object. The execution tool is used to perform data processing to obtain a generated object. A generated object can be generated using at least one execution tool. Different generated objects can use the same execution tool or different execution tools. The configuration can be based on the needs of the actual problem to be solved, and the present disclosure does not limit this.

[0059] The types of execution tools include at least one of large language models, machine learning models, deep learning models, functions, and mechanism models, so that it is possible to integrate multiple different types of execution tools, and use these different types of execution tools together to solve actual industry problems, effectively realizing the practical application of large industry models.

[0060] Optionally, the third configuration interface includes a tool type selection interface and a configuration editing interface for each of the at least one generation object; wherein the tool type selection interface is used to select the type of execution tool of the corresponding generation object; the configuration editing interface is used to edit the configuration information of the execution tool of the corresponding generation object, and the configuration information of the execution tool of the model type (such as a large language model, a machine learning model, a deep learning model, etc.) includes a model URL and a model input, and the editing process includes at least one of the following: adding, deleting, and modifying. By setting the tool type selection interface and the configuration editing interface in the third configuration interface, it is possible to link the corresponding configuration information for different types of execution tools at the corresponding configuration editing interface, so that the user can perform editing processes such as adding, deleting, and modifying the configuration information of each execution tool, wherein the adding process and the deleting process can respectively realize the addition and removal of the execution tool, and the modifying process can realize the correction and optimization of the execution tool. Regarding configuration information, for example, for model type execution tools, model URL and model input can be provided as configuration information. The model URL enables the industry big model to call the corresponding model execution tool during operation, and the model input enables the industry big model to pass the specified input to the called model execution tool, thereby ensuring the reliable operation of the constructed industry big model.

[0061] Accordingly, the tool configuration operation performed by the user in the third configuration interface may include an operation of triggering a tool type selection interface, which can trigger the third configuration interface to display multiple available tool types, such as displaying through a drop-down menu, or displaying through a table with a selection function. On this basis, the tool configuration operation also includes an operation of selecting a tool type, which can trigger the third configuration interface to display a configuration editing interface of the corresponding type of execution tool. When the user selects the interface, the third configuration interface can be triggered to display a tool editing window. For example, for a model type execution tool, the tool editing window can display an input interface for the model URL and model input, such as a text input box, for the user to fill the model URL and model input into the corresponding text input box. Of course, according to actual needs, the tool editing window may include different types of input interfaces, as long as it can allow the user to input the corresponding configuration information, and the present disclosure does not limit this.

[0062] In addition, it will be recorded later that the present disclosure may also provide a tool chain system for an industry big model, which includes a construction module for executing the construction method of the industry big model according to the exemplary embodiment of the present disclosure, and may also include a tool module for developing and optimizing an execution tool, which may use relevant technologies to develop and optimize the execution tool, and the configuration information of the optimized execution tool (such as the model URL) may be refilled into the configuration editing interface to modify the configuration information, thereby optimizing the execution tool. Of course, in the absence of a configuration tool module, or the user does not use the tool module to develop the execution tool, the user may develop and optimize the execution tool by himself, and use the function of modifying the configuration information to modify the originally input configuration information to the optimized configuration information of the execution tool, thereby optimizing the execution tool.

[0063] In step S104, a fourth configuration interface is displayed, and in response to an output configuration operation performed by a user in the fourth configuration interface, a data output style of the industry macro model is determined.

[0064] The data output style is the display style of the output data of the industry big model on the output interface. The fourth configuration interface provides a standardized configuration function for the data output style. As mentioned above, the input and output of the industry big model are usually homogeneous objects, but how to display the generated object cannot be determined based on the input data. By providing the fourth configuration interface, users can perform output configuration operations, so that the created industry big model can display the generated object in the style expected by the user, meet practical needs, and promote better implementation of the industry big model.

[0065] Optionally, the fourth configuration interface includes a component navigation area, a component configuration interface, and an output kanban preview area; wherein the component navigation area displays a plurality of display components for selection, and the display components are used to display the generated objects of the industry large model, and the types of display components include at least one of the following: text, picture, data chart, and the icons of multiple types of display components displayed in the component navigation area; the component configuration interface is used to configure the selected display component, such as naming the display component, setting the data associated with the corresponding display component in the generated object (i.e., the data source of the display component, or which or which data in the generated object the display component is used to display), and setting the parameters of the display component (such as the font of the text, the resolution of the picture, the header, number of rows and columns of the data icon, etc.); the output kanban preview area is used to display the preview effect of the selected display component, and the position of the display component in the output kanban preview area is movable, so that the user can edit the display position of the display component. By specifically designing the fourth configuration interface to include the above-mentioned component navigation area, component configuration interface and output kanban preview area, users can set and preview the output kanban as needed, and realize the standardized configuration of data output style.

[0066] Accordingly, the output configuration operation performed by the user in the fourth configuration interface may include dragging the required type of display component icon from the component navigation area and placing it in a position in the output kanban preview area. At this time, the display component may be activated by default, and the component configuration interface for the display component may be displayed in the designated area on the fourth configuration interface for the user to perform configuration operations on the display component. In addition, the display position of the display component in the output kanban preview area can continue to be moved and adjusted. Specifically, the display component can be activated by clicking on a display component in the output kanban preview area, and then the display component can be dragged. Of course, if a display component itself is in an activated state, for example, the user has just dragged it into the output kanban preview area, the display component can be directly dragged.

[0067] In step S105, the industry big model is constructed based on the object information of at least one generation object of the industry big model, the input data format, the execution tool of each generation object in the at least one generation object, and the data output style.

[0068] Based on the information configured in the above steps, this step finally realizes the construction of the industry big model.

[0069] It should also be noted that in order to distinguish different steps, this article adds serial numbers to each step, but the serial numbers are not a restriction on the execution order of each step. As long as it is logically allowed, the execution order of the above steps can be adjusted. For example, step S101 implements the configuration of the generated object, which is the basis for building the industry model. The subsequent steps S102, S103, and S104 configure the input data format, the execution tool of the generated object, and the data output style respectively. There is no logical constraint between the three, so the execution order of the three can be swapped, and the last step S105 is to build the industry model based on the information determined by each configuration operation, so it can only be executed at the end.

[0070] Next, step S101 and step S102, that is, generating an object configuration operation and a corresponding input configuration operation, will be further introduced.

[0071] Optionally, the first configuration interface includes at least two object type options; wherein, one of the at least two object type options is a complex object type, and the complex object type indicates that the generation data of the industry big model includes at least two generation objects; and at least one of the at least two object type options is a non-complex object type, and the non-complex object type indicates that the generation data of the industry big model is one generation object. In other words, the constructed industry big model can be used to generate a variety of different types of single object types (i.e., non-complex object types), and can also be used to generate complex object types including multiple generation objects (the generation objects included in the complex object type can also be understood as object elements), thereby meeting different data generation requirements and helping to promote the implementation of the industry big model.

[0072] Regarding non-complex object types, optionally, at least one non-complex object type includes at least one of the following: a text object type, an audio object type, and a video object type. By configuring a variety of different media object types for non-complex object types, a single generated object can have a richer media form to meet different data generation needs. It should be understood that, similarly, although the complex object type includes multiple generated objects, it can also be displayed in the form of text, audio, or video for multimedia. It should also be understood that, as mentioned above, the data type of the generated object includes at least one of a numerical value and a text, and the data type focuses on the attributes of the data content, while the object type introduced here focuses on media attributes, that is, the attributes of the output form, whether it is a numerical value or text, it can be output in the form of text, audio, or video.

[0073] Regarding the object generation configuration operation, optionally, in step S101, in response to the object generation configuration operation performed by the user on the first configuration interface, the operation of determining the object information of at least one generated object of the industry large model includes: in response to the user selecting the complex object type on the first configuration interface, displaying the generated object input interface on the first configuration interface; in response to the user's input operation on the generated object input interface, determining the object information of at least two generated objects of the industry large model. For a complex object type including at least two generated objects, it is usually necessary to split the actual industry problem to be solved into multiple simple problems, and generate corresponding generated objects respectively. However, the problem is solved by predicting the generated objects and combining the generated objects. At this time, the generated objects split out may have different data types, and each generated object needs to be named to achieve its distinction and use. To this end, by configuring the generated object input interface, the user can configure each generated object separately, thereby realizing the configuration of the generated objects of the entire complex object type and determining its object information.

[0074] Further optionally, the generated object input interface includes at least one of the following: a first input interface for inputting object information of the generated object, and a second input interface for inputting task information of the target task performed by the industry large model, that is, different types of generated object input interfaces can be configured to achieve different forms of generated object configuration. As an example, the first configuration interface may only display one of the first input interface and the second input interface to clarify the configuration method of the generated object, or it may display the first input interface and the second input interface at the same time for the user to choose as needed, and it may also configure a selection button first for the user to select different configuration methods, and then display the first input interface or the second input interface according to the user's choice, and the present disclosure does not limit this.

[0075] Specifically, in the case where the generated object input interface includes a first input interface, as an example, a first input interface can be configured for each generated object to constitute a generated object entry, and each first input interface can include a name input interface and a data type input interface, and of course, can also include input interfaces for other information, and can also set a "add" button to add a generated object entry, thereby adding a new first input interface, and can also set a "delete" button for each existing first input interface to delete the corresponding generated object entry, and the present disclosure does not limit this. The first input interface allows users to split the actual industry problems to be solved by themselves, ensuring the reliability of the split.

[0076] In the case where the generated object input interface includes a second input interface, the operation of determining the object information of at least two generated objects of the industry large model in response to the user's input operation on the generated object input interface in step S101 includes: receiving the task information of the target task in response to the user's input operation on the second input interface; using the task disassembly model to process the task information to disassemble the target task into at least two subtasks, and obtain the object information of at least two generated objects that can be generated by at least two subtasks, wherein the subtask is used to generate the generated object. By configuring the second input interface for receiving the task information, the task disassembly model is called to realize the task disassembly, which can reduce manual operations, improve the automation level of the industry large model construction, and improve the construction efficiency. As an example, when the task disassembly model is used to realize the task disassembly, it can be completely realized automatically by the task disassembly model, or it can be realized manually with the assistance of the task disassembly model, for example, the physical mechanism of task disassembly is provided manually in the form of rules or functions, and the task disassembly model realizes the disassembly accordingly, or the task disassembly model first completes the preliminary disassembly, and then the manual review and correction is performed, and the present disclosure does not limit this.

[0077] In addition, for a non-complex object type, as an example, default object information may be configured for it, such as not configuring a name by default, and setting its data type by default (e.g., setting it to text type by default), and then not configuring a generated object input interface. Of course, the generated object input interface may also be configured as needed, and the generated object input interface of the complex object type may be referred to. The difference is that the non-complex object type only involves one generated object, and thus only the generated object input interface for one generated object needs to be configured. The present disclosure does not limit this.

[0078] Regarding the input configuration operations of different object types, optionally, the operation of displaying the second configuration interface in step S102 may include: in response to the user selecting a complex object type in the first configuration interface, displaying the second configuration interface corresponding to the complex object type (it should be understood that what is defined here is the display content rather than the display timing. In other words, the display timing is still the display timing of the second configuration interface, rather than the timing when the user selects the complex object type in the first configuration interface, and the display content is specifically the second configuration interface corresponding to the complex object type), and displaying the specified format options and other format options in the second configuration interface corresponding to the complex object type.

[0079] Among them, the specified format option includes specified format information and a format preview interface. The format preview interface is used to provide a preview of the file in the specified format, which can provide the input data format that the industry large model can process by default, and allow users to confirm their own file formats through the preview function, thereby ensuring that the constructed industry large model can realize the normal reading of input files. Correspondingly, the specified format information displayed in the second configuration interface can be, for example, the name of the specified format, such as the JSON format, so that the user can understand the specific format of the specified format. When the user clicks on the format preview interface displayed in the second configuration interface, a file preview image of the specified format can be displayed. The specified format option can be a selection button. If the user performs the input configuration operation of clicking the button, the input data format of the industry large model will be determined as the specified format.

[0080] Other format options include other format information and a parsing script upload interface. The parsing script upload interface is used to receive parsing scripts of other format files uploaded by users, so that the constructed industry model can process input data in formats other than the default specified data format, so that users can flexibly configure different input data formats to meet different format requirements. Correspondingly, the other format option can also be a selection button. If the user performs the input configuration operation of clicking the button, the input data format of the industry model will be determined as other formats. Other format input interfaces can be further configured here for users to enter the names of other formats, or they can be unconfigured. The other format information displayed in the second configuration interface can be, for example, examples of several other formats, such as PDF / Word / txt / excel, etc., for users to refer to other formats that can be used. When the user successfully uploads the parsing script of other format files through the parsing script upload interface, the constructed industry model can subsequently parse the input data according to the parsing script.

[0081] In addition, the operation of displaying the second configuration interface in step S102 may also include: in response to the user selecting a non-complex object type in the first configuration interface, displaying a second configuration interface corresponding to the non-complex object type (similar to the above, the display content rather than the display timing is defined here), and displaying at least one data format option of the non-complex object type selected by the user in the second configuration interface corresponding to the non-complex object type, and the at least one data format option is used for the user to select a data format for the non-complex object type. By configuring the second configuration interface corresponding to the non-complex object type and displaying at least one data format option of the non-complex object type therein, the user can complete the input data format setting for the non-complex object type in a targeted manner. Accordingly, if the user performs an input configuration operation by clicking on any data format option, the input data format of the industry model is determined to be the data format corresponding to the option. As an example, when the user selects a text object type, data format options such as PDF, Word, txt, xml, html, etc. can be displayed; when the user selects an audio object type, data format options such as PCM, MP3, WAV, AAC, etc. can be displayed; when the user selects a video object type, data format options such as MP4, AVI, MKV, MOV, WMV, etc. can be displayed.

[0082] Next, a method for constructing an industry big model according to the present disclosure is introduced in conjunction with a specific embodiment.

[0083] In this specific embodiment, in order to solve the problem of implementing the big model in actual industry applications, the present disclosure returns to problem definition, tool construction, tool calling, and tool optimization, and proposes a system implementation method based on the user's disassembly definition of historically complex objects (such as physical state, hydrological state, environmental state, etc.), combined with the business meaning and technology selection of each element disassembled, to support the model development, launch, and optimization of different algorithms such as large language models, traditional machine learning models, deep learning models, function calculations, mechanism models, etc., and integrate various algorithms to generate subsequent time-series complex objects to create a feasible human-computer interaction industry big model, and provide a series of mature components and configuration items, so that users can customize the kanban style of generated objects based on business requirements. The construction method of the industry big model of this specific embodiment is implemented based on the construction platform of the industry big model, and the specific construction process is as follows.

[0084] Figure 2 is a schematic diagram showing a homepage of a construction platform according to a specific embodiment of the present disclosure. Figure 2, multiple industry models that have been built are displayed in the form of cards on the homepage of the construction platform, and an interface for creating a new industry model can be configured. In response to the operation used to trigger the new interface, a new window for inputting new model information can be highlighted on the homepage of the construction platform in the form of a mask, so that users can define the name and description of the industry model based on actual needs. After the industry model is built, the name and description of the industry model can be displayed in a card on the homepage of the construction platform, which is convenient for users to manage and call it.

[0085] like Figure 2 The new window shown is also provided with a "Cancel" button and an "OK" button, for the user to choose whether to abandon the new model or continue with the subsequent building operations.

[0086] In response to the user clicking the "OK" button, the following is displayed: Figure 3 The first configuration interface shown allows the user to first define the object type of the generated object. There are four object type options displayed in the first configuration interface, from left to right, they are text object type option, audio object type option, video object type option, and complex object type option. The first three belong to non-complex object types, and their configurations are similar, while the configuration of the last complex object type is different. Next, we will take the audio object type as an example to introduce the model configuration based on the non-complex object type, and introduce the model configuration based on the complex object type.

[0087] It should be noted that this specific embodiment does not support COT (Chain of Thought, that is, breaking down a command into multiple steps and implementing them one by one). What it implements is to generate subsequent time series object sequences based on historical sequence objects, and the input and output are homogeneous objects. For example, health reports for the next few years are generated based on health reports for the past few years.

[0088] For model configuration based on non-complex object types such as audio object types, users can Figure 3 The first configuration interface shown selects the audio object type option. Based on this operation, the second configuration interface will be displayed, for example Figure 4 The following table shows four data format options: PCM, MP3, WAV, and AAC, for users to confirm the input data format of the industry model. Since the input and output are homogeneous objects, this configuration also determines the encoding method of the generated audio object.

[0089] After completing the input data format setting, you will be redirected to the third configuration interface, for example Figure 5As shown, a text input box for the model URL and a text input box for the AIP key (a string used to implement identity authentication when accessing the application programming interface (API)) can be displayed to implement the configuration of the execution tool. The text input box for the model URL can specifically include two text input boxes, one for inputting the address port number and the other for inputting the model path.

[0090] After completing the configuration of the execution tool, you will be redirected to the fourth configuration interface, for example Figure 6 As shown, from left to right are the component navigation area, the output dashboard preview area, and the component configuration interface. The component navigation area displays multiple audio styles. The component configuration interface can configure the output audio language, speaking style, timbre, whether to automatically play information and generate text, speaking speed, volume and other parameters, thereby supporting users to customize data output styles.

[0091] After completing the configuration of the data output style, you can Figure 7 As shown, the successful construction prompt window is highlighted in the fourth configuration interface in the form of a mask, and the construction of the industry model is completed.

[0092] For model configuration based on complex object types, users need to first define what elements the objects generated by the industry large model contain, that is, what generated objects the complex object type specifically includes, as well as the names (such as English names), data types, and descriptions of the included elements, and provide examples for reference. Specifically, users can Figure 3 The first configuration interface shown in the figure selects the complex object type option. Based on this operation, Figure 8 The jump display generates an object input interface, specifically, multiple generated object entries are displayed, each generated object entry includes a name input interface, a data type input interface and a description information input interface, and the data type input interface is specifically a drop-down menu from which the user can select a corresponding data type. The data types provided in the drop-down menu may include, for example Figure 8 The examples in the example include int (integer), double (double-precision floating point number), string (string), and array (array), and of course other data types can also be included, such as float (single-precision floating point number). A "Delete" button is set at the right end of each generated object entry to delete the corresponding generated object entry, and a "Add" button is also set at the bottom of these generated object entries to add new generated object entries. It should be noted that Figure 8 Taking the health report as an example, several elements are listed. Here, if we want to generate future reports based on historical health reports, which contain multiple elements, they need to be defined by the user.

[0093] In such Figure 8 After defining the generated object in the generated object input interface of the first configuration interface shown in FIG. Fig. 9 The second configuration interface shown is for users to confirm the input data format of the industry big model. For the generation of complex objects, the big model only supports interaction with json files, that is, in order to realize the user's business questions, the user needs to upload (or the big model calls it by itself) the corresponding json format historical object file. If the user uploads the json file directly, he needs to ensure that the format in the json is consistent with the "preview". If the user uploads a file type in other formats (such as a PDF health report), the corresponding parsing script must be uploaded through the parsing script upload interface so that other file types can be converted to the json format defined in the previous step during execution before interacting with the industry big model.

[0094] Next jump to Fig.10 The third configuration interface shown in FIG. 1 is used by the user to configure the execution tool in the industry model. Figure 8 The first configuration interface shown has defined the first three columns of each element, so this part of the content is grayed out, making it uneditable, and provides a tool type selection interface and a configuration editing interface for users to define the generation method and configuration information of each element one by one.

[0095] Specifically, if the element contains basic information such as name and ID, which will not change with the sequence, you can choose the fixed value tool, that is, choose the fixed value generation method. For this, you can provide Fig.11 The tool editing window of the fixed value tool is shown, allowing the user to input the fixed value.

[0096] If an element needs to be generated by a traditional professional model, it is necessary to configure basic information such as service URL, model path (the corresponding path can be configured), model input (the parameter name and parameter value of the model input can be configured), etc. Fig.12 The tool editing window of the professional model tool shown is for the user to input these basic information. Specifically, the professional model can be implemented and optimized using the modeling tool of the construction platform, or can be implemented and optimized by the user using his own modeling platform.

[0097] For situations where a large language model is needed, in addition to the basic information such as the service URL, model path, and API key (the corresponding string itself can be configured, and the path where the string is stored can also be configured), such as Fig.13As shown, the prompt word prompt needs to be configured, which may involve some variable values ​​in the historical sequence object, and can support information calling through escape characters, etc. Specifically, similar to the aforementioned professional model, the language large model can also be implemented and optimized using the modeling tools of the construction platform, or it can be implemented and optimized by the user using his own modeling platform.

[0098] After completing the execution tool configuration, you will be redirected to Fig.14 The fourth configuration interface shown, and Figure 6 Similar, such as Fig.14 The fourth configuration interface shown also displays the component navigation area, output board preview area, and component configuration interface from left to right, and provides different component configuration interfaces according to the selected display component type. Users can select different display components from the left and drag them into the output board preview area, and configure the configuration items of the corresponding components on the right to complete the customized configuration of the board. It should be noted that Fig.14 Taking the health report as an example, several display components are listed, which can be defined by the user.

[0099] Fig.15 and Fig.16 The schematic diagram of the input interface and output dashboard of the large model of the chronic disease management industry is shown. After the model is built, the user can call the model and input multiple historical health reports. The model can output predicted future health reports and can be used in Fig.16 The output dashboard shown displays core indicators, future health risks, doctor's advice, personalized diet plans, personalized exercise plans, etc., so that users can intuitively view predictions and suggestions.

[0100] Fig.17 is a block diagram illustrating an apparatus for constructing an industry big model according to an exemplary embodiment of the present disclosure.

[0101] Reference Fig.17 The construction device 1700 of the industry model includes a first configuration unit 1701, a second configuration unit 1702, a third configuration unit 1703, a fourth configuration unit 1704, and a construction unit 1705.

[0102] The first configuration unit 1701 can display a first configuration interface, and determine object information of at least one generated object of the industry macro model in response to a generated object configuration operation performed by a user on the first configuration interface.

[0103] The second configuration unit 1702 can display a second configuration interface, and determine the input data format of the industry big model in response to an input configuration operation performed by the user on the second configuration interface.

[0104] The third configuration unit 1703 can display a third configuration interface, and in response to a tool configuration operation performed by the user in the third configuration interface, configure an execution tool for each generation object of at least one generation object, wherein the execution tool is used to perform data processing to obtain the generation object, and the type of the execution tool includes at least one of a large language model, a machine learning model, a deep learning model, a function, and a mechanism model.

[0105] The fourth configuration unit 1704 can display the fourth configuration interface, and determine the data output style of the industry big model in response to the output configuration operation performed by the user in the fourth configuration interface, wherein the data output style is the display style of the output data of the industry big model on the output interface.

[0106] The construction unit 1705 can construct the industry big model based on the object information of at least one generation object of the industry big model, the input data format, the execution tool of each generation object in the at least one generation object, and the data output style.

[0107] Optionally, the object information includes at least one of a name and a data type, and the data type includes at least one of a value and a text.

[0108] Optionally, the first configuration interface includes at least two object type options; wherein, one of the at least two object type options is a complex object type, and the complex object type indicates that the generation data of the industry large model includes at least two generation objects; and at least one of the at least two object type options is a non-complex object type, and the non-complex object type indicates that the generation data of the industry large model is one generation object.

[0109] Optionally, the at least one non-complex object type includes at least one of the following: a text object type, an audio object type, and a video object type.

[0110] Optionally, the first configuration unit 1701 may also: in response to the user selecting a complex object type in the first configuration interface, display a generation object input interface in the first configuration interface; in response to the user input operation on the generation object input interface, determine the object information of at least two generation objects of the industry large model.

[0111] Optionally, the generated object input interface includes at least one of the following: a first input interface for inputting object information of the generated object, and a second input interface for inputting task information of the target task performed by the industry large model; when the generated object input interface includes the second input interface, the first configuration unit 1701 may also: receive task information in response to the user's input operation on the second input interface; use the task decomposition model to process the task information to decompose the target task into at least two subtasks, and obtain object information of at least two generated objects that can be generated by the at least two subtasks, wherein the subtasks are used to generate the generated object.

[0112] Optionally, the second configuration unit 1702 may also: in response to the user selecting a complex object type in the first configuration interface, display the second configuration interface corresponding to the complex object type, and display specified format options and other format options in the second configuration interface corresponding to the complex object type, wherein the specified format options include specified format information and a format preview interface, the format preview interface is used to provide a file preview of the specified format, and other format options include other format information and a parsing script upload interface, the parsing script upload interface is used to receive parsing scripts for other format files uploaded by the user; in response to the user selecting a non-complex object type in the first configuration interface, display the second configuration interface corresponding to the non-complex object type, and display at least one data format option for the non-complex object type selected by the user in the second configuration interface corresponding to the non-complex object type, and the at least one data format option is used for the user to select a data format for the non-complex object type.

[0113] Optionally, the third configuration interface includes a tool type selection interface and a configuration editing interface for each generation object in at least one generation object; wherein the tool type selection interface is used to select the type of execution tool of the corresponding generation object; the configuration editing interface is used to edit the configuration information of the execution tool of the corresponding generation object, the configuration information of the execution tool of the model type includes the model URL and the model input, and the editing process includes at least one of the following: addition, deletion, and modification.

[0114] Optionally, the fourth configuration interface includes a component navigation area, a component configuration interface, and an output board preview area; wherein, a plurality of selectable display components are displayed in the component navigation area, and the display components are used to display the generated objects of the industry large model, and the types of display components include at least one of the following: text, picture, data chart; the component configuration interface is used to configure the selected display component; the output board preview area is used to display the preview effect of the selected display component, and the position of the display component in the output board preview area is movable so that the user can edit the display position of the display component.

[0115] According to an exemplary embodiment of the present disclosure, a tool chain system for an industry big model may also be provided, the tool chain system comprising: a construction module for executing the construction method of the industry big model as described above to construct the industry big model; a calling module for receiving input data and calling the industry big model to perform data processing on the input data. The tool chain system provides an integrated industry big model construction and calling function, and because it can execute the construction method of the industry big model as described above, it has all the beneficial technical effects of the method, which will not be elaborated here.

[0116] Optionally, the tool chain system also includes a tool module for developing and optimizing execution tools, which can further provide integrated tool development and optimization functions, supporting users to achieve one-stop industry large model construction and calling from scratch.

[0117] The above has been referred to Figures 1 to 17 A method and apparatus for constructing an industry big model according to an exemplary embodiment of the present disclosure are described.

[0118] Fig.17 Each unit in the construction device of the industry big model shown can be configured as software, hardware, firmware or any combination of the above items to perform specific functions. For example, each unit can correspond to a dedicated integrated circuit, or a pure software code, or a module that combines software and hardware. In addition, one or more functions implemented by each unit can also be uniformly executed by components in a physical entity device (for example, a processor, a client or a server, etc.).

[0119] In addition, refer to Figure 1 The described method for constructing a large industry model can be implemented by a program (or instruction) recorded on a computer-readable storage medium. For example, according to an exemplary embodiment of the present disclosure, a computer-readable storage medium storing instructions may be provided, wherein when the instructions are executed by at least one computing device, the at least one computing device is prompted to execute the method for constructing a large industry model according to the present disclosure.

[0120] The computer program in the computer-readable storage medium can be run in an environment deployed in a computer device such as a client, a host, an agent device, a server, etc. It should be noted that the computer program can also be used to perform additional steps in addition to the above steps or perform more specific processing when performing the above steps. The contents of these additional steps and further processing have been described in reference to Figure 1 It is mentioned in the description of the related method, so it will not be repeated here to avoid repetition.

[0121] It should be noted that each unit in the construction device of the industry big model according to the exemplary embodiment of the present disclosure can completely rely on the operation of the computer program to realize the corresponding function, that is, each unit corresponds to each step in the functional architecture of the computer program, so that the entire system is called through a special software package (for example, lib library) to realize the corresponding function.

[0122] on the other hand, Fig.17 The various units shown may also be implemented by hardware, software, firmware, middleware, microcode, or any combination thereof. When implemented by software, firmware, middleware, or microcode, the program code or code segment for performing the corresponding operation may be stored in a computer-readable medium such as a storage medium, so that the processor may perform the corresponding operation by reading and running the corresponding program code or code segment.

[0123] For example, an exemplary embodiment of the present disclosure may also be implemented as a computing device, which includes a storage component and a processor, wherein a set of computer executable instructions is stored in the storage component, and when the set of computer executable instructions is executed by the processor, a method for constructing an industry big model according to the exemplary embodiment of the present disclosure is executed.

[0124] Specifically, the computing device can be deployed in a server or client, or can be deployed on a node device in a distributed network environment. In addition, the computing device can be a PC, a tablet device, a personal digital assistant, a smart phone, a web application, or other device capable of executing the above instruction set.

[0125] Here, the computing device is not necessarily a single computing device, but may also be any device or circuit collection that can execute the above instructions (or instruction sets) individually or jointly. The computing device may also be part of an integrated control system or system manager, or may be configured as a portable electronic device that is interconnected with a local or remote (e.g., via wireless transmission) interface.

[0126] In a computing device, a processor may include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller, or a microprocessor. By way of example and not limitation, a processor may also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, etc.

[0127] According to the exemplary embodiment of the present disclosure, some operations described in the method for constructing an industry big model can be implemented by software, some operations can be implemented by hardware, and in addition, these operations can also be implemented by a combination of software and hardware.

[0128] The processor may execute instructions or codes stored in one of the storage components, which may also store data. Instructions and data may also be sent and received over a network via a network interface device, which may employ any known transmission protocol.

[0129] The storage component may be integrated with the processor, for example, RAM or flash memory is arranged within an integrated circuit microprocessor, etc. In addition, the storage component may include a separate device, such as an external disk drive, a storage array, or any other storage device that can be used by a database system. The storage component and the processor may be operatively coupled, or may communicate with each other, such as through an I / O port, a network connection, etc., so that the processor can read files stored in the storage component.

[0130] In addition, the computing device may also include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, a mouse, a touch input device, etc.) All components of the computing device may be connected to each other via a bus and / or a network.

[0131] The method for constructing an industry big model according to an exemplary embodiment of the present disclosure can be described as various interconnected or coupled functional blocks or functional diagrams. However, these functional blocks or functional diagrams can be equally integrated into a single logical device or operate according to non-exact boundaries.

[0132] Therefore, refer to Figure 1 The described method for constructing a large industry model may be implemented by a system including at least one computing device and at least one storage device storing instructions.

[0133] According to an exemplary embodiment of the present disclosure, at least one computing device is a computing device for executing the method for constructing an industry large model according to an exemplary embodiment of the present disclosure, and a computer executable instruction set is stored in the storage device. When the computer executable instruction set is executed by the at least one computing device, the execution reference Figure 1 Described method for building a large industry model.

[0134] According to an exemplary embodiment of the present disclosure, a computer program product may also be provided, including instructions, which, when executed by at least one computing device, prompt the at least one computing device to execute the method for constructing an industry big model according to the present disclosure.

[0135] The above describes various exemplary embodiments of the present disclosure, and it should be understood that the above description is only exemplary and not exhaustive, and the present disclosure is not limited to the disclosed exemplary embodiments. Without departing from the scope and spirit of the present disclosure, many modifications and changes are obvious to those of ordinary skill in the art. Therefore, the scope of protection of the present disclosure should be based on the scope of the claims.

Claims

1. A method for constructing a large industry model, characterized in that: The construction method comprises: Displaying a first configuration interface, and determining object information of at least one generated object of the industry macro model in response to a generated object configuration operation performed by a user in the first configuration interface; Displaying a second configuration interface, and determining an input data format of the industry macro model in response to an input configuration operation performed by a user on the second configuration interface; Displaying a third configuration interface, and in response to a tool configuration operation performed by a user on the third configuration interface, configuring an execution tool for each of the at least one generated object, wherein the execution tool is used to perform data processing to obtain the generated object, and the type of the execution tool includes at least one of a large language model, a machine learning model, a deep learning model, a function, and a mechanism model; Displaying a fourth configuration interface, and determining a data output style of the industry model in response to an output configuration operation performed by a user on the fourth configuration interface, wherein the data output style is a display style of output data of the industry model on the output interface; The industry big model is constructed based on the object information of the at least one generation object of the industry big model, the input data format, the execution tool of each generation object in the at least one generation object, and the data output style.

2. The construction method according to claim 1, characterized in that: The object information includes at least one of a name and a data type, and the data type includes at least one of a value and a text.

3. The construction method according to claim 1, characterized in that: The first configuration interface includes at least two object type options; Among them, one of the at least two object type options is a complex object type, and the complex object type indicates that the generation data of the industry large model includes at least two generation objects; At least one of the at least two object type options is a non-complex object type, and the non-complex object type indicates that the generation data of the industry big model is a generation object.

4. The construction method according to claim 3, characterized in that: The at least one non-complex object type includes at least one of the following: a text object type, an audio object type, and a video object type.

5. The construction method according to claim 3, characterized in that: The step of determining object information of at least one generated object of the industry macro model in response to the generated object configuration operation performed by the user on the first configuration interface includes: In response to a user selecting the complex object type in the first configuration interface, displaying a generate object input interface in the first configuration interface; In response to a user input operation on the generated object input interface, object information of at least two generated objects of the industry master model is determined.

6. A device for constructing a large industry model, characterized in that: The construction device comprises: A first configuration unit is configured to display a first configuration interface, and determine object information of at least one generated object of the industry macro model in response to a generated object configuration operation performed by a user in the first configuration interface; A second configuration unit is configured to display a second configuration interface, and determine the input data format of the industry macro model in response to an input configuration operation performed by a user on the second configuration interface; A third configuration unit is configured to display a third configuration interface, and in response to a tool configuration operation performed by a user on the third configuration interface, configure an execution tool for each of the at least one generated object, wherein the execution tool is used to perform data processing to obtain the generated object, and the type of the execution tool includes at least one of a large language model, a machine learning model, a deep learning model, a function, and a mechanism model; A fourth configuration unit is configured to display a fourth configuration interface, and determine a data output style of the industry macro model in response to an output configuration operation performed by a user on the fourth configuration interface, wherein the data output style is a display style of output data of the industry macro model on the output interface; A construction unit is configured to construct the industry big model based on the object information of the at least one generation object of the industry big model, the input data format, the execution tool of each generation object in the at least one generation object, and the data output style.

7. A tool chain system for industry large models, characterized in that: The tool chain system includes: A construction module, used to execute the method for constructing an industry big model as claimed in any one of claims 1 to 5, so as to construct the industry big model; The calling module is used to receive input data and call the industry big model to perform data processing on the input data.

8. A computer-readable storage medium storing instructions, characterized in that: When the instructions are executed by at least one computing device, the at least one computing device is prompted to execute the method for constructing an industry big model as claimed in any one of claims 1 to 5.

9. A system comprising at least one computing device and at least one storage device storing instructions, characterized in that: When the instructions are executed by the at least one computing device, the at least one computing device is prompted to execute the method for constructing an industry big model as claimed in any one of claims 1 to 5.

10. A computer program product comprising instructions, characterized in that When the instructions are executed by at least one computing device, the at least one computing device is prompted to execute the method for constructing an industry big model as claimed in any one of claims 1 to 5.

Images