Model recommendation method and device based on multi-model cooperation, equipment and storage medium

Abstract

The invention relates to the technical field of large language models, and discloses a model recommendation method, device and equipment based on multi-model collaboration and a storage medium, the method comprises the following steps: in response to an input user instruction, determining a model collaboration scheme according to an instruction type of the user instruction, the model cooperation scheme comprises the model number and model roles of a plurality of cooperation models; according to the model role, selecting a corresponding to-be-recommended model for each cooperation model from a preset model library; and displaying the model role of each cooperation model and the corresponding to-be-recommended model. According to the method and the device, the accuracy and the matching degree of model selection can be effectively improved, so that more accurate answers with stronger correlation can be generated, and the answer quality is remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of large language model, and particularly relates to a model recommendation method and device based on multi-model cooperation, equipment and a storage medium. BACKGROUND

[0002] In recent years, natural language processing technology based on large language models (LLMs) has been widely applied in various artificial intelligence application scenarios, such as intelligent assistants, search engines, automatic customer service, content generation and intelligent recommendation, etc. These models, through training on large-scale data sets, can generate high-quality natural language responses, showing strong language understanding and generation capabilities. However, as the scale and application fields of large language models continue to expand, their performance in diversified applications also exposes some challenges, especially when users choose large language models themselves, the mismatch between the selected model and the actual problem may result in low accuracy of the generated answers, which cannot effectively meet the needs of users.

[0003] Therefore, how to effectively improve the accuracy and matching degree of model selection is a problem that needs to be solved at present.

[0004] The above content is only used to assist in understanding the technical solutions of the present application, and does not represent the acknowledgement of the above content as prior art. SUMMARY

[0005] The main purpose of the present application is to provide a model recommendation method, device, equipment and storage medium based on multi-model cooperation, aiming at solving the technical problem that when users choose large language models themselves, the mismatch between the selected model and the actual problem leads to low accuracy of the generated answers.

[0006] To achieve the above purpose, the present application provides a model recommendation method based on multi-model cooperation, which comprises:

[0007] In response to the input user instruction, determine the model cooperation scheme according to the instruction type of the user instruction, the model cooperation scheme comprising the model quantity and model role of multiple cooperation models;

[0008] According to the model role, select corresponding recommended models for each cooperation model from the preset model library;

[0009] The model role of each cooperation model and the corresponding recommended model are displayed.

[0010] In an embodiment, the method further comprises:

[0011] Obtain the attribute information and label information of each model in the preset model library;

[0012] determine a candidate model corresponding to each collaborative model in the preset model library based on the label information and the model role;

[0013] determine a matching degree between the model role and the candidate model based on the attribute information and the candidate model;

[0014] select a corresponding to-be-recommended model for each collaborative model from the preset model library according to the matching degree.

[0015] In an embodiment, the determining a candidate model corresponding to each collaborative model in the preset model library based on the label information comprises:

[0016] determining a good-at problem type of each model in the preset model library according to the label information;

[0017] determining a candidate model corresponding to each collaborative model in the preset model library based on the good-at problem type of each model and the model role.

[0018] In an embodiment, the determining a matching degree between the model role and the candidate model based on the attribute information and the candidate model comprises:

[0019] determining an accuracy, a response time length, and a device performance consumption parameter of the candidate model according to the attribute information;

[0020] determining a matching degree between the model role and the candidate model according to the accuracy, the response time length, and the device performance consumption parameter of the candidate model.

[0021] In an embodiment, the determining a model collaboration scheme according to an instruction type of the input user instruction comprises:

[0022] determining a corresponding task type according to an instruction type of the input user instruction;

[0023] determining a task complexity based on the task type;

[0024] determining whether multiple models are needed to collaborate to complete a task according to the task complexity;

[0025] in a case where multiple models are needed to collaborate to complete the task, determining a model collaboration scheme according to the task complexity and the task type.

[0026] In an embodiment, the determining a model collaboration scheme according to the task complexity and the task type comprises:

[0027] determining a model quantity of multiple collaborative models according to the task complexity;

[0028] determine model roles of a plurality of collaboration models according to the task type;

[0029] determine a model collaboration scheme according to the model quantity and the model roles.

[0030] In an embodiment, the determining the model roles of the plurality of collaboration models according to the task type comprises:

[0031] determining a task demand based on the task type;

[0032] decomposing a corresponding task into a plurality of sub-tasks according to the task demand;

[0033] determining model roles of corresponding collaboration models according to each of the sub-tasks.

[0034] In an embodiment, the displaying the model roles of each of the collaboration models and the corresponding to-be-recommended models comprises:

[0035] obtaining a working order of each of the collaboration models and a user equipment performance parameter;

[0036] displaying the model roles of each of the collaboration models according to the working order of each of the collaboration models;

[0037] determining a display priority of a to-be-recommended model corresponding to each of the collaboration models according to the user equipment performance parameter;

[0038] arranging and displaying the to-be-recommended model corresponding to each of the collaboration models based on the display priority.

[0039] In an embodiment, after the displaying the model roles of each of the collaboration models and the corresponding to-be-recommended models, the method further comprises:

[0040] determining a target collaboration model combination based on the to-be-recommended model corresponding to each of the collaboration models;

[0041] collaborating through a target collaboration model in the target collaboration model combination to generate and display reply information of the user instruction.

[0042] In an embodiment, the determining the target collaboration model combination based on the to-be-recommended model corresponding to each of the collaboration models comprises:

[0043] in response to an input selection instruction, determining a target recommended model corresponding to each of the collaboration models according to the selection instruction and the to-be-recommended model corresponding to each of the collaboration models;

[0044] generating the target collaboration model combination based on the target recommended model corresponding to each of the collaboration models.

[0045] In an embodiment, the determining the target collaboration model combination based on the to-be-recommended model corresponding to each of the collaboration models further comprises:

[0046] determining a plurality of collaboration model combinations based on permutation and combination of the to-be-recommended models corresponding to the collaboration models;

[0047] performing comprehensive performance evaluation on the collaboration model combinations to obtain comprehensive performance scores of the collaboration model combinations;

[0048] determining a target collaboration model combination from the plurality of collaboration model combinations according to the comprehensive performance scores.

[0049] In addition, to achieve the above object, the application further provides a model recommendation device based on multi-model collaboration, which comprises:

[0050] a determination module configured to determine a model collaboration scheme according to an instruction type of an input user instruction in response to the user instruction, the model collaboration scheme comprising a model number and a model role of a plurality of collaboration models;

[0051] a selection module configured to select a corresponding to-be-recommended model for each collaboration model from a preset model library according to the model role;

[0052] a display module configured to display the model role of each collaboration model and the corresponding to-be-recommended model.

[0053] In an embodiment, the selection module is further configured to acquire attribute information and label information of each model in the preset model library;

[0054] determine a candidate model corresponding to each collaboration model in the preset model library based on the label information and the model role;

[0055] determine a matching degree of the model role and the candidate model based on the attribute information and the candidate model;

[0056] select a corresponding to-be-recommended model for each collaboration model from the preset model library according to the matching degree.

[0057] In an embodiment, the selection module is further configured to determine a good problem type of each model in the preset model library according to the label information;

[0058] determine a candidate model corresponding to each collaboration model in the preset model library based on the good problem type of each model and the model role.

[0059] In an embodiment, the selection module is further configured to determine an accuracy rate, a response time length and a device performance consumption parameter of a candidate model according to the attribute information;

[0060] The matching degree of the model role and the candidate model is determined according to the accuracy, response time length and device performance consumption parameter of the candidate model.

[0061] In an embodiment, the determining module is further configured to determine a corresponding task type according to an instruction type of an input user instruction in response to the input user instruction.

[0062] A task complexity is determined based on the task type.

[0063] It is determined whether multiple models need to cooperate to complete a task according to the task complexity.

[0064] In a case where multiple models need to cooperate to complete the task, a model cooperation scheme is determined according to the task complexity and the task type.

[0065] In an embodiment, the determining module is further configured to determine a model number of multiple cooperation models according to the task complexity.

[0066] A model role of the multiple cooperation models is determined according to the task type.

[0067] The model cooperation scheme is determined according to the model number and the model role.

[0068] In an embodiment, the determining module is further configured to determine a task demand based on the task type.

[0069] The corresponding task is decomposed into multiple sub-tasks according to the task demand.

[0070] The model role of a corresponding cooperation model is determined according to each of the sub-tasks.

[0071] In addition, to achieve the above object, the present application further provides a model recommendation device based on multiple model cooperation, which comprises a memory, a processor and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the model recommendation method based on multiple model cooperation.

[0072] In addition, to achieve the above object, the present application further provides a storage medium, which is a computer readable storage medium, and the storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the model recommendation method based on multiple model cooperation.

[0073] In addition, to achieve the above object, the present application further provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the steps of the model recommendation method based on multiple model cooperation.

[0074] The application provides a model recommendation method based on multi-model cooperation. The application determines a model cooperation scheme according to the instruction type of an input user instruction, selects corresponding to-be-recommended models for each cooperation model from a preset model library according to the model roles, and displays the model roles of each cooperation model and the corresponding to-be-recommended models, thereby effectively improving the accuracy and matching degree of model selection, generating more accurate and more relevant answers, and significantly improving the answer quality.

[0075] In conclusion, the application determines a model cooperation scheme according to the instruction type of a user instruction, accurately allocates corresponding model roles to each cooperation model, selects corresponding to-be-recommended models for each cooperation model from a preset model library according to the model roles, effectively improves model adaptability and reduces model selection errors, displays the model roles of each cooperation model and the corresponding to-be-recommended models, makes the selected model accurate and highly matched with the problem, overcomes the technical defect that the selected model is not matched with the actual problem when a user selects a large language model, effectively improves the accuracy and matching degree of model selection, generates more accurate and more relevant answers, and significantly improves the answer quality. BRIEF DESCRIPTION OF DRAWINGS

[0076] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and serve to explain the principles of the application together with the specification.

[0077] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the accompanying drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, for those of ordinary skill in the art, the other drawings can be obtained based on these drawings without any creative work.

[0078] Figure 1 A flowchart is provided for the model recommendation method based on multi-model cooperation according to the first embodiment of the application.

[0079] Figure 2 A graphical to-be-recommended model display page is provided for the model recommendation method based on multi-model cooperation according to the first embodiment of the application.

[0080] Figure 3 A text description to-be-recommended model display page is provided for the model recommendation method based on multi-model cooperation according to the first embodiment of the application.

[0081] Figure 4A flowchart of the second embodiment of the model recommendation method based on multi-model collaboration provided in this application;

[0082] Figure 5 A flowchart of the third embodiment of the model recommendation method based on multi-model collaboration provided in this application;

[0083] Figure 6 This is a schematic diagram of the module structure of a model recommendation device based on multi-model collaboration according to an embodiment of the present application;

[0084] Figure 7 This is a schematic diagram of the device structure of the hardware operating environment involved in the model recommendation method based on multi-model collaboration in an embodiment of the present application.

[0085] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0086] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0087] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0088] The main solution of the embodiment of the present application is: in response to an input user instruction, determining a model collaboration scheme according to the instruction type of the user instruction, the model collaboration scheme including the number of models and model roles of multiple collaborative models; selecting corresponding models to be recommended for each collaborative model from a preset model library according to the model roles; and displaying the model roles of each collaborative model and the corresponding models to be recommended.

[0089] In recent years, natural language processing technologies based on large language models (LLMs) have been widely used in various AI application scenarios, such as intelligent assistants, search engines, automated customer service, content generation, and intelligent recommendations. These models, trained on large datasets, are able to generate high-quality natural language responses, demonstrating strong language understanding and generation capabilities. However, as the scale and application areas of large language models continue to expand, their performance in diverse applications has also exposed some challenges. In particular, when users select large language models on their own, a mismatch between the selected model and the actual question may result in inaccurate answers that fail to effectively meet user needs. Therefore, how to effectively improve the accuracy and matching of model selection is a problem that needs to be urgently addressed.

[0090] The application determines a model cooperation scheme according to the instruction type of the user instruction, can accurately assign corresponding model roles to each cooperation model, and accordingly selects corresponding to-be-recommended models for each cooperation model from a preset model library according to the model roles, can effectively improve model adaptability and reduce model selection errors, and accordingly displays the model roles of each cooperation model and the corresponding to-be-recommended models, so that the model selected by the user is accurate and has high matching degree with the problem, overcomes the technical defects that when the user selects a large language model by himself / herself, the selected model is not matched with the actual problem, resulting in low accuracy of the generated answer, can effectively improve the accuracy and matching degree of model selection, and can further generate more accurate and more relevant answers, and significantly improves the answer quality.

[0091] It should be noted that the execution subject of the embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a model recommendation device based on multi-model cooperation, etc. The embodiments will be described below with the model recommendation device based on multi-model cooperation as an example.

[0092] Based on this, the embodiment of the application provides a model recommendation method based on multi-model cooperation, referring to Figure 1 , Figure 1 The flowchart of the first embodiment of the model recommendation method based on multi-model cooperation of the application is shown in FIG. 1.

[0093] In the embodiment, the model recommendation method based on multi-model cooperation includes steps S10-S30:

[0094] Step S10, in response to an input user instruction, a model cooperation scheme is determined according to the instruction type of the user instruction, and the model cooperation scheme includes the model quantity and model role of a plurality of cooperation models.

[0095] It should be noted that the user instruction can be a text instruction or an instruction input through a user interface. The instruction type of the user instruction can include but is not limited to natural language query, request for specific task, data processing requirement, etc., which is not limited in the embodiment.

[0096] It can be understood that multiple collaborative models refer to large language models composed of different functions and characteristics, such as ChatGPT, TongYiQian, WenXinYiYan, etc. These models are good at handling different types of tasks according to their design and training data. For example, ChatGPT may perform well in dialogue generation, TongYiQian may be more skilled in knowledge question and answer, and WenXinYiYan may be more skilled in text understanding and generation. Multiple collaborative models can be large language models of the same type but different parameters, or combinations of large language models of different types, for example, ChatGPT for dialogue generation and WenXinYiYan for text understanding and generation. These models can complement each other when collaborating to improve overall processing capacity. The number of collaborative models can be determined according to the complexity of the task and the type of data to be processed, and the model role defines the specific responsibilities and functions of each collaborative model in the task completion process. For example, in a content generation task, one model may be responsible for understanding user instructions and generating preliminary content, while another model may focus on optimizing and polishing the content. By defining the role of each model, the efficiency and goal orientation of the entire collaboration process can be ensured.

[0097] Specifically, according to the different types of instructions, the number of models and role allocation will be adjusted accordingly to generate a model collaboration scheme to meet the needs of different application scenarios. For example, for natural language queries, the model collaboration scheme may include a language understanding model and a language generation model; while for data processing requirements, a data preprocessing model and a data analysis model may be needed. The determination of the model role is based on the specific content and goal of the user instruction to ensure that each collaborative model can maximize its effectiveness in its area of expertise.

[0098] Step S20, according to the model role, selecting a corresponding to-be-recommended model for each collaborative model from a preset model library.

[0099] It should be noted that the preset model library stores a variety of types of large language models, which have undergone strict testing and evaluation and can cover different application scenarios and functional requirements. Each model is classified and labeled according to its function and attribute. The models in the preset model library can include but are not limited to text generation models, text classification models, sentiment analysis models, machine translation models, speech recognition models, etc., which are not specifically limited in this embodiment.

[0100] It is understood that the recommended model refers to the model selected from the preset model library to meet the requirements of a specific collaborative model role. When selecting the recommended model, the most suitable model will be screened from the preset model library based on the specific requirements of the model role. For example, if the model role is language understanding, then a model with excellent performance in language understanding will be selected from the preset model library; if the model role is language generation, a model with strong language generation capabilities will be selected. In this way, it can be ensured that each collaborative model can maximize its effectiveness in the area where it is best excelling, thereby improving the processing efficiency and accuracy of the overall task.

[0101] Specifically, the model selection process considers the model's applicability, efficiency, and accuracy to ensure that each collaborative model can achieve its optimal performance. For example, for tasks requiring advanced language understanding, a deep learning language model might be selected; for tasks requiring creative content generation, a model based on generative adversarial networks might be selected. In this way, each collaborative model can be guaranteed to provide the most appropriate solution for its specific role and task requirements.

[0102] Step S30: Display the model role of each collaborative model and the corresponding model to be recommended.

[0103] It should be noted that the display method can be a graphical interface or a text description. The purpose is to enable users to intuitively understand the role of each collaborative model and the corresponding recommended models. The graphical interface can take the form of a flowchart or tree diagram to clearly display the relationship and collaboration between models. The text description can list the model role name and function of each collaborative model in detail, as well as the detailed information of the corresponding recommended model, including the model name, type, characteristics, etc. This display not only helps users better understand the model collaboration mechanism, but also facilitates users to select and adjust according to their needs.

[0104] like Figure 2 As shown, Figure 2 This is a graphical diagram of the page displaying models to be recommended. In the figure, model roles A, B, and C are arranged in sequence according to their workflows. Each role includes corresponding models to be recommended. The models to be recommended for model role A include model 1 to be recommended, model 2 to be recommended, and model 3 to be recommended. The models to be recommended for model role B include model 4 to be recommended and model 5 to be recommended. The models to be recommended for model role C include model 6 to be recommended, model 7 to be recommended, and model 8 to be recommended.

[0105] like Figure 3 As shown, Figure 3For the text description of the recommended model display page, the name of each model role, the model role function, the corresponding recommended model name and the recommended model type are described in turn. The model role function of model role A is to generate the answer to the question, the corresponding recommended model name is recommended model 1 and recommended model 2. The model role function of model role B is to generate the correction information of the answer, the corresponding recommended model name is recommended model 3, recommended model 4 and recommended model 5. The model role function of model role C is to correct the original answer according to the correction information, and the corresponding recommended model name is recommended model 6 and recommended model 7.

[0106] It can be understood that in the display process, interactive elements can also be provided, such as clicking on a certain model to view more detailed introductions or to make model replacement and optimization suggestions, thereby improving the user experience.

[0107] In a feasible implementation, the step S30 can include: obtaining the working order of each collaborative model and the user device performance parameter; displaying the model roles of each collaborative model according to the working order of each collaborative model; determining the display priority of the recommended model corresponding to each collaborative model according to the user device performance parameter; and arranging and displaying the recommended model corresponding to each collaborative model based on the display priority.

[0108] It should be noted that the positions and roles of each collaborative model in the overall workflow are identified, and then the display content is organized according to their order in actual operation. For example, if one model is responsible for preliminary screening and another model is responsible for final decision-making, the preliminary screening model will be displayed before the decision-making model in the graphical interface or text description to reflect their sequence in actual work. Such sequential display helps users understand the logical relationship between models and the data flow. At the same time, by considering the performance parameters of the user device, the display order of the recommended models can be intelligently adjusted to ensure that the recommended models not only meet the user's functional needs but also have good running effects on the user's device in terms of performance. Ultimately, the user can select the most suitable model for use according to the display priority and detailed information.

[0109] It can be understood that the user device performance parameters refer to the computing power, storage capacity, network connection speed and other key performance indicators of the user device. These parameters can affect the running efficiency and response speed of the model on a specific device, so the performance limitations of the user device need to be considered when determining the display priority of the recommended model. For example, for devices with weak computing power, models with low resource requirements are preferred; while for high-performance devices, models that are computationally intensive but have better results can be displayed. In addition, the user's historical choices and preferences can also be used as a basis for adjusting the display priority to provide more personalized recommendations. In this way, users can quickly find the most suitable model for their needs, thereby improving work efficiency and satisfaction.

[0110] In a feasible implementation, after the step S30, the method further comprises: determining a target collaboration model combination based on the to-be-recommended models corresponding to each collaboration model; and generating and displaying the reply information of the user instruction by collaboration of the target collaboration models in the target collaboration model combination.

[0111] It should be noted that each collaboration model can correspond to multiple to-be-recommended models, and the determination of the target collaboration model combination can be user self-selection or automatic selection according to the performance evaluation results of different model combinations. Users can select the most suitable model combination according to their specific needs and preferences to achieve the best work effect, for example, if the user needs a model with fast response, they can select a model combination that optimizes the response time; if the user pays more attention to the accuracy of the results, they can select a model combination that has an advantage in accuracy. In this way, users not only can customize the model combination according to their needs, but also can ensure the efficiency and accuracy of the selected model combination in actual application.

[0112] It can be understood that after determining the target collaboration model combination, the models are used for collaboration processing to generate the reply information of the user instruction, for example, multiple models are used for collaboration to process complex problems, and finally generate answers with higher accuracy. The display of the reply information can be in the form of text, chart or multimedia to adapt to the information receiving preferences of different users.

[0113] In a feasible implementation, the method further comprises: in response to an input selection instruction, determining a target recommended model corresponding to each collaboration model according to the selection instruction and the to-be-recommended model corresponding to each collaboration model; and generating a target collaboration model combination based on the target recommended model corresponding to each collaboration model.

[0114] It should be noted that in the present embodiment, a target recommendation model is selected from each of the to-be-recommended models corresponding to the collaboration models through a selection instruction input by the user, and then the target recommendation models corresponding to each of the collaboration models are combined to form a comprehensive model combination scheme, i.e., a target collaboration model combination.

[0115] It can be understood that the selection process of the user can be interactive, allowing the user to see the expected effect and performance indicators of each model combination in real time during the selection process, so as to make a more intelligent decision. For example, the user can intuitively see the performance comparison of different model combinations through a graphical interface, including key indicators such as processing speed, accuracy, and resource consumption. In this way, the user can more intuitively understand the advantages and disadvantages of each model combination and select the most suitable model combination according to his actual needs and preferences to achieve the best work efficiency and result quality.

[0116] In a feasible embodiment, the determining of the target collaboration model combination based on the to-be-recommended models corresponding to the collaboration models further includes: performing permutation and combination on the to-be-recommended models corresponding to the collaboration models to determine a plurality of collaboration model combinations; performing comprehensive performance evaluation on each of the collaboration model combinations to obtain a comprehensive performance score of each of the collaboration model combinations; and determining the target collaboration model combination from the plurality of collaboration model combinations according to the comprehensive performance score.

[0117] It should be noted that in the present embodiment, the optimal collaboration model combination is automatically selected according to the performance evaluation results of different model combinations. Each of the to-be-recommended models of the collaboration models is permuted and combined to generate a plurality of different collaboration model combinations.

[0118] It can be understood that by performing performance testing and evaluation on each of the collaboration model combinations, these combinations can be scored to determine their performance in terms of processing speed, accuracy, and resource consumption. For example, one combination may perform excellently in accuracy but slightly less in response time; and another combination may be faster in response time but slightly less accurate.

[0119] It should be noted that one or more optimal collaboration model combinations can be recommended to the user according to the comprehensive performance score, ensuring that the user can obtain the best performance and work efficiency when selecting the model combination.

[0120] The embodiment provides a model recommendation method based on multi-model cooperation. The embodiment can determine a model cooperation scheme according to the instruction type of an input user instruction, in response to the input user instruction, the model cooperation scheme including the model quantity and model role of a plurality of cooperation models; and select a corresponding to-be-recommended model for each cooperation model from a preset model library according to the model role. The model role and the corresponding to-be-recommended model of each cooperation model are displayed, which can effectively improve the accuracy and matching degree of model selection, and further generate a more accurate and more relevant answer, thereby significantly improving the answer quality.

[0121] In conclusion, the embodiment can accurately allocate a corresponding model role to each cooperation model according to the instruction type of a user instruction, thereby selecting a corresponding to-be-recommended model for each cooperation model from a preset model library according to the model role, which can effectively improve model adaptability and reduce model selection errors, and further display the model role and the corresponding to-be-recommended model of each cooperation model, so that the model selected by the user is accurate and has a high matching degree with the problem. The technical defect that the accuracy of the generated answer is low due to the mismatch between the selected model and the actual problem when the user selects a large language model is overcome, the accuracy and matching degree of model selection are effectively improved, and a more accurate and more relevant answer can be generated, thereby significantly improving the answer quality.

[0122] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as the above-mentioned first embodiment can be referred to the above introduction, and will not be described in detail. On this basis, please refer to Figure 4 , the step S20 further includes steps S201-S204:

[0123] In step S201, attribute information and label information of each model in the preset model library are acquired.

[0124] It should be noted that the attribute information of the model includes accuracy, response time, and consumption of device performance, and the label information includes the problem type that the model is good at, the applicable scene, and the specific function of the model, and the embodiment does not make a specific limitation. By acquiring these information, the characteristics and application range of each model can be more accurately understood.

[0125] In step S202, the candidate model corresponding to each cooperation model in the preset model library is determined based on the label information and the model role.

[0126] It should be noted that the candidate model matched with the model role can be preliminarily screened out by matching the model role of each cooperation model with the label information of each model in the preset model library.

[0127] In an implementation, the step S202 specifically comprises: determining the problem types that each model in the preset model library is good at according to the label information; and determining the candidate model corresponding to each collaborative model in the preset model library based on the problem types that each model is good at and the model role.

[0128] It should be noted that the problem types that each model is good at refer to the excellent performance and high accuracy of the model in a specific field or problem type. For example, some models may perform well in processing natural language understanding tasks, while others may be more skilled in image recognition or speech processing.

[0129] It can be understood that the applicable scenario and functional requirement of the corresponding collaborative model can be determined according to the model role, thereby effectively narrowing the range of candidate models. For example, if the model role is for real-time speech recognition, the candidate model library should contain models that are good at processing speech data. In this way, it can be ensured that the selected model not only meets the technical requirements, but also achieves the expected performance standard in actual application.

[0130] The step S203 comprises: determining the matching degree between the model role and the candidate model based on the attribute information and the candidate model.

[0131] It should be noted that after determining the candidate model, the attribute information of each candidate model is analyzed to evaluate the matching degree between the model role and the candidate model. The attribute information can include the processing speed, accuracy, resource consumption, scalability and other key performance indicators of the model, which are not specifically limited in this embodiment. By comparing these attributes with the specific requirements of the model role, the applicability of each candidate model can be quantitatively evaluated. For example, if the model role requires high accuracy and low latency, those models that perform better in these attributes will be given a higher matching score among the candidate models. Finally, according to the matching degree, the collaborative model that best meets the requirements of the model role can be selected to ensure the efficient and stable operation of the entire system.

[0132] In an implementation, the step S203 specifically comprises: determining the accuracy, response time and device performance consumption parameters of the candidate model according to the attribute information; and determining the matching degree between the model role and the candidate model according to the accuracy, response time and device performance consumption parameters of the candidate model.

[0133] It should be noted that the accuracy of the model refers to the proportion of the correctness of the output result of the model in processing a specific task, which is directly related to the reliability of the model in actual application. The response time reflects the speed of the model in processing the task, which is crucial for application scenarios that require fast response. The device performance consumption parameter is the performance consumption of the model running on the device, including but not limited to the use of CPU, memory and storage space.

[0134] It can be understood that each model role may have different requirements for accuracy, response time and device performance consumption. According to the model role, weights can be assigned to attribute information such as accuracy, response time and device performance consumption parameters, and then according to the accuracy, response time and device performance consumption parameters and the corresponding weights, the matching score of the candidate model and the model role, i.e. the matching degree, is calculated, so as to realize the comprehensive evaluation of the candidate model.

[0135] In step S204, a corresponding to-be-recommended model is selected for each collaborative model from the preset model library according to the matching degree.

[0136] It should be noted that by comparing the matching degrees of each candidate model, one or more to-be-recommended models with the highest matching degree can be selected for each collaborative model.

[0137] It can be understood that the candidate model with a matching degree reaching a preset matching degree threshold can be taken as a to-be-recommended model. The preset matching degree threshold can be a standard set by the user in advance, and only when the matching degree of the candidate model exceeds this standard, the model is considered to be a suitable to-be-recommended model. The matching degrees of each candidate model can also be sorted, and the top N candidate models in the sorting result can be taken as to-be-recommended models, and the embodiment does not make specific limitations.

[0138] In the embodiment, by determining the candidate model corresponding to each collaborative model according to the tag information and model role of each model in the preset model library, the range of to-be-recommended models can be effectively reduced, thereby improving the recommendation efficiency, and then determining the matching degree of the model role and the candidate model according to the attribute information of each model and the candidate model, and then selecting a corresponding to-be-recommended model for each collaborative model according to the matching degree, the model recommendation efficiency and accuracy are effectively improved, thereby improving the model adaptability and reducing the model selection error.

[0139] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the model recommendation method based on multi-model collaboration of the present application. More forms of simple transformation based on this technical concept are within the protection scope of the present application.

[0140] Based on the first embodiment of the present application, in the third embodiment of the present application, the same or similar contents as the above embodiment one can be referred to the above introduction, and the subsequent will not be described. On this basis, please refer to Figure 5 , the step S10 further includes steps S101-S104:

[0141] Step S101, in response to the input user instruction, determining the corresponding task type according to the instruction type of the user instruction.

[0142] It should be noted that the task type refers to the specific operation or demand category involved in the user instruction, including but not limited to information retrieval task, generation task, language understanding task, dialogue system task, reasoning task, multi-modal task, etc., which is not limited in this embodiment.

[0143] It can be understood that the instruction type of the user instruction can be identified by the keywords, phrases or sentence structure in the instruction. For example, if the instruction contains keywords such as "query" and "search", the instruction type is query instruction, and the corresponding task type may be information retrieval task.

[0144] Step S102, determining the task complexity based on the task type.

[0145] It should be noted that the task complexity can be evaluated according to the data volume and processing difficulty corresponding to different task types, or according to the number of steps required to be executed corresponding to different task types, or according to the time required to complete the task, which is not limited in this embodiment.

[0146] Step S103, determining whether multiple models are needed to cooperate to complete the task according to the task complexity.

[0147] It should be noted that complex tasks may require the cooperation of multiple models to achieve more efficient and accurate processing, while simple tasks may only require one model to complete.

[0148] It can be understood that when the task complexity reaches a certain threshold, a single model may not be able to provide sufficient processing power, at which point introducing multiple models to cooperate can significantly improve the processing efficiency and accuracy of the results. For example, in processing large-scale data sets or complex pattern recognition, multiple models can work in parallel, sharing the computing load, and through the synergistic effect between models, they can complement each other's shortcomings, thus achieving better processing results. After determining that multiple models are needed to cooperate, the appropriate number and type of models need to be selected to ensure that the task can be completed efficiently and accurately.

[0149] Step S104, in the case of multiple models cooperating to complete the task, determining the model cooperation scheme according to the task complexity and task type.

[0150] It should be noted that the model cooperation scheme includes the number of models of the plurality of cooperation models and the model roles, and the allocation of the model roles should be based on the task requirements of the cooperation models. According to the task complexity and the task type, the number of models of the plurality of cooperation models and the model roles are determined, so as to obtain the model cooperation scheme.

[0151] In a feasible implementation, the step S104 can include: determining the number of models of the plurality of cooperation models according to the task complexity; determining the model roles of the plurality of cooperation models according to the task type; and determining the model cooperation scheme according to the number of models and the model roles.

[0152] It should be noted that the number of models of the plurality of cooperation models is determined by the task complexity, and the higher the task complexity is, the more cooperation models are needed to jointly process the task, so as to ensure that the processing capability matches the task requirement. The model roles are determined by the task type, for example, when processing an image recognition task, a model responsible for feature extraction and a model responsible for classification decision are needed. In a natural language processing task, a language model is needed to understand the context, and another model is needed to generate or translate text, and the allocation of the model roles should ensure that each model can play the maximum in its own field.

[0153] In a feasible implementation, the determination of the model roles of the plurality of cooperation models according to the task type includes: determining the task requirement based on the task type; decomposing the corresponding task into a plurality of subtasks according to the task requirement; and determining the model roles of the corresponding cooperation models according to each of the subtasks.

[0154] It should be noted that the task requirement can be determined according to the task type, for example, the task requirement corresponding to an information retrieval task is to quickly and accurately retrieve relevant information from a large amount of data, the task requirement corresponding to a language understanding task is to understand the meaning and context of language, and the task requirement corresponding to a dialogue system task is to effectively interact with the user and provide accurate answers.

[0155] It can be understood that after the task is decomposed into subtasks, the model roles can be more accurately allocated, for example, for a question and answer task, it can be split into an information retrieval subtask, an information correction subtask and an answer generation subtask, and then different models are allocated to them, one model can be used to generate an initial answer, one model can be used to generate correction information of the initial answer, and another model can be used to correct the initial answer according to the correction information, so as to generate the final answer of the question. In this way, each model can work in its most proficient field, so as to improve the efficiency and accuracy of the whole task processing.

[0156] In this embodiment, the corresponding task type is determined according to the instruction type of the user instruction, the corresponding task complexity is determined, and then it is determined whether multiple models need to cooperate to complete the task. In the case where multiple models need to cooperate to complete the task, the number of cooperative models is accurately determined according to the task complexity and the task type, and each cooperative model is allocated a corresponding model role, and an accurate model cooperation scheme is generated, thereby effectively improving the model recommendation efficiency and accuracy.

[0157] It should be noted that the above examples are only used for understanding the present application and do not constitute a limitation on the model recommendation method based on multi-model cooperation of the present application. More forms of simple transformation based on this technical concept are within the protection scope of the present application.

[0158] The present application also provides a model recommendation device based on multi-model cooperation, please refer to Figure 6 The model recommendation device based on multi-model cooperation comprises:

[0159] A determination module 10 is configured to determine a model cooperation scheme according to the instruction type of an input user instruction, wherein the model cooperation scheme comprises the number of models and the model role of multiple cooperative models.

[0160] A selection module 20 is configured to select a corresponding to-be-recommended model for each cooperative model from a preset model library according to the model role.

[0161] A display module 30 is configured to display the model role of each cooperative model and the corresponding to-be-recommended model.

[0162] The present embodiment provides a model recommendation device based on multi-model cooperation. In response to an input user instruction, the present embodiment determines a model cooperation scheme according to the instruction type of the user instruction, wherein the model cooperation scheme comprises the number of models and the model role of multiple cooperative models. The present embodiment selects a corresponding to-be-recommended model for each cooperative model from a preset model library according to the model role. The present embodiment displays the model role of each cooperative model and the corresponding to-be-recommended model, which can effectively improve the accuracy and matching degree of model selection, thereby generating a more accurate and more relevant answer, and significantly improving the answer quality.

[0163] In conclusion, the embodiment can accurately allocate corresponding model roles to each collaborative model by determining the model collaboration scheme according to the instruction type of the user instruction, thereby selecting corresponding to-be-recommended models for each collaborative model from the preset model library according to the model roles, effectively improving model adaptability and reducing model selection errors, and further displaying the model roles of each collaborative model and the corresponding to-be-recommended models, so that the selected model is accurate and has high matching degree with the problem, overcoming the technical defects that when the user selects a large language model by himself / herself, the selected model is not matched with the actual problem, resulting in low accuracy of the generated answer, and effectively improving the accuracy and matching degree of model selection, thereby generating more accurate and more relevant answers, and significantly improving the answer quality.

[0164] Optionally, the selecting module 20 is further configured to acquire attribute information and label information of each model in the preset model library; determine candidate models corresponding to each collaborative model in the preset model library based on the label information and the model roles; determine a matching degree between the model roles and the candidate models based on the attribute information and the candidate models; and select corresponding to-be-recommended models for each collaborative model from the preset model library according to the matching degree.

[0165] Optionally, the selecting module 20 is further configured to determine problem types that each model in the preset model library is good at according to the label information; and determine candidate models corresponding to each collaborative model in the preset model library based on the problem types that each model is good at and the model roles.

[0166] Optionally, the selecting module 20 is further configured to determine an accuracy, a response time length, and a device performance consumption parameter of a candidate model according to the attribute information; and determine a matching degree between the model roles and the candidate models according to the accuracy, the response time length, and the device performance consumption parameter of the candidate model.

[0167] Optionally, the determining module 10 is further configured to determine a corresponding task type according to an instruction type of an input user instruction in response to the input user instruction; determine a task complexity based on the task type; determine whether multiple models are needed to collaborate to complete a task according to the task complexity; and determine a model collaboration scheme according to the task complexity and the task type in a case where the multiple models are needed to collaborate to complete the task.

[0168] Optionally, the determining module 10 is further configured to determine a model quantity of multiple collaborative models according to the task complexity; determine model roles of the multiple collaborative models according to the task type; and determine a model collaboration scheme according to the model quantity and the model roles.

[0169] Optionally, the determining module 10 is further configured to determine a task demand based on the task type, decompose the corresponding task into a plurality of sub-tasks according to the task demand, and determine a model role of a corresponding collaboration model according to each sub-task.

[0170] Optionally, the displaying module 30 is further configured to obtain a working order of each collaboration model and a user equipment performance parameter, display the model role of each collaboration model according to the working order of each collaboration model, determine a display priority of a to-be-recommended model corresponding to each collaboration model according to the user equipment performance parameter, and arrange and display the to-be-recommended model corresponding to each collaboration model based on the display priority.

[0171] Optionally, the model recommendation device based on multi-model collaboration further comprises a collaboration module, which is configured to determine a target collaboration model combination based on the to-be-recommended model corresponding to each collaboration model, and generate and display reply information of the user instruction by collaboration of the target collaboration model in the target collaboration model combination.

[0172] Optionally, the collaboration module is further configured to determine a target recommended model corresponding to each collaboration model according to the selection instruction and the to-be-recommended model corresponding to each collaboration model in response to an input selection instruction, and generate a target collaboration model combination based on the target recommended model corresponding to each collaboration model.

[0173] Optionally, the collaboration module is further configured to arrange and combine the to-be-recommended model corresponding to each collaboration model to determine a plurality of collaboration model combinations, perform comprehensive performance evaluation on each collaboration model combination to obtain a comprehensive performance score of each collaboration model combination, and determine a target collaboration model combination in the plurality of collaboration model combinations according to the comprehensive performance score.

[0174] The model recommendation device based on multi-model collaboration provided in the application adopts the model recommendation method based on multi-model collaboration in the above embodiments, and can solve the technical problem that when a user selects a large language model by himself, the selected model does not match the actual problem, resulting in low accuracy of the generated answer. Compared with the prior art, the model recommendation device based on multi-model collaboration provided in the application has the same beneficial effects as the model recommendation method based on multi-model collaboration provided in the above embodiments, and other technical features in the model recommendation device based on multi-model collaboration are the same as the features disclosed in the above method embodiments, which will not be repeated here.

[0175] The application provides a model recommendation device based on multi-model cooperation, which comprises at least one processor and a memory connected with the at least one processor; the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the model recommendation method based on multi-model cooperation in the above embodiment one.

[0176] Reference will be made to the following Figure 7 which shows a structural diagram of the model recommendation device based on multi-model cooperation suitable for being used to implement the embodiments of the application. The model recommendation device based on multi-model cooperation in the embodiments of the application can include but is not limited to mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (Personal Digital Assistant), PADs (Portable Application Description), PMPs (Portable MediaPlayer), vehicle-mounted terminals (such as vehicle-mounted navigation terminals) and the like, and fixed terminals such as digital TVs, desktop computers and the like. Figure 7 The model recommendation device based on multi-model cooperation shown is only an example, and should not bring any limitation to the functions and use range of the embodiments of the application.

[0177] As Figure 7As shown, the model recommendation device based on multi-model collaboration can include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 1002 or loaded from a storage device 1003 into a random access memory (RAM) 1004. In the RAM 1004, various programs and data required for the model recommendation device based on multi-model collaboration to operate are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; the storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the model recommendation device based on multi-model collaboration to communicate with other devices wirelessly or by wire to exchange data. Although the model recommendation device based on multi-model collaboration with various systems is shown in the figure, it should be understood that all the systems shown are not required to be implemented or possessed. More or fewer systems can be alternatively implemented or possessed.

[0178] In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network through the communication device, or installed from the storage device 1003, or installed from the ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the methods of the embodiments of the present disclosure are performed.

[0179] The model recommendation device based on multi-model cooperation provided in the application adopts the model recommendation method based on multi-model cooperation in the above embodiment, and can solve the technical problem that when a user selects a large language model by himself / herself, the selected model does not match the actual problem, resulting in low accuracy of the generated answer. Compared with the prior art, the beneficial effects of the model recommendation device based on multi-model cooperation provided in the application are the same as those of the model recommendation method based on multi-model cooperation provided in the above embodiment, and other technical features in the model recommendation device based on multi-model cooperation are the same as those disclosed in the above embodiment method, and will not be repeated here.

[0180] It should be understood that parts of the present application can be realized by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0181] The above is merely specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0182] The present application provides a computer readable storage medium having computer readable program instructions (i.e. computer programs) stored thereon, the computer readable program instructions being used to execute the model recommendation method based on multi-model cooperation in the above embodiment.

[0183] The computer readable storage medium provided in the application may, for example, be a U disk, but is not limited to an electric, magnetic, optical, electromagnetic, infrared, or semiconductor system, system, or device, or any combination of the above. More specific examples of the computer readable storage medium may include, but are not limited to, an electric connection with one or more conductive wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the embodiment, the computer readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system, or device. The program code contained on the computer readable storage medium can be transmitted by any suitable medium, including but not limited to an electric wire, an optical cable, an RF (Radio Frequency), and the like, or any suitable combination of the above.

[0184] The above computer readable storage medium may be contained in the model recommendation device based on multi-model cooperation, or may exist independently without being assembled into the model recommendation device based on multi-model cooperation.

[0185] The above computer readable storage medium carries one or more programs, when the one or more programs are executed by the model recommendation device based on multi-model cooperation, the model recommendation device based on multi-model cooperation is caused to: in response to an input user instruction, determine a model cooperation scheme according to an instruction type of the user instruction, the model cooperation scheme including a model number and a model role of a plurality of cooperation models; select a corresponding to-be-recommended model for each cooperation model from a preset model library according to the model role; and display the model role and the corresponding to-be-recommended model of each cooperation model.

[0186] Computer program code for carrying out operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0187] The flow and block diagrams in the drawings show architectural, functional and operational architectures of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flow and block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may

[0188] The modules involved in the embodiments of the present application can be implemented in the form of software or in the form of hardware. In some cases, the name of the module does not constitute a limitation on the module itself.

[0189] The readable storage medium provided by the present application is a computer readable storage medium, which stores computer readable program instructions (i.e., computer programs) for executing the above-mentioned model recommendation method based on multi-model cooperation, and can solve the technical problem of low accuracy of generated answers caused by the mismatch between the selected model and the actual problem when the user selects a large language model by himself. Compared with the prior art, the computer readable storage medium provided by the present application has the same beneficial effects as the model recommendation method based on multi-model cooperation provided by the above-mentioned embodiments, and will not be described here.

[0190] The application also provides a computer program product comprising a computer program which, when executed by a processor, implements the steps of the model recommendation method based on multi-model cooperation as described above.

[0191] The computer program product provided by the application can solve the technical problem of low accuracy of the generated answer caused by the mismatch between the selected model and the actual problem when the user selects a large language model by himself. Compared with the prior art, the beneficial effects of the computer program product provided by the application are the same as those of the model recommendation method based on multi-model cooperation provided by the above-mentioned embodiments, and are not described here.

[0192] The above only describes some embodiments of the application, and does not limit the patent scope of the application. Any equivalent structural transformation made by using the content of the application specification and drawings, or direct / indirect application in other related technical fields is included in the patent protection scope of the application.

[0193] A1. A model recommendation method based on multi-model cooperation, the method comprising:

[0194] In response to an input user instruction, determining a model cooperation scheme according to the instruction type of the user instruction, the model cooperation scheme comprising a model quantity and a model role of a plurality of cooperation models;

[0195] Selecting a corresponding to-be-recommended model for each cooperation model from a preset model library according to the model role;

[0196] Displaying the model role of each cooperation model and the corresponding to-be-recommended model.

[0197] A2. The method of A1, wherein the selecting a corresponding to-be-recommended model for each cooperation model from a preset model library according to the model role comprises:

[0198] Obtaining attribute information and label information of each model in the preset model library;

[0199] Determining a candidate model corresponding to each cooperation model in the preset model library based on the label information and the model role;

[0200] Determining a matching degree of the model role and the candidate model based on the attribute information and the candidate model;

[0201] Selecting a corresponding to-be-recommended model for each cooperation model from the preset model library according to the matching degree.

[0202] A3. The method of A2, wherein the determining a candidate model corresponding to each cooperation model in the preset model library based on the label information comprises:

[0203] determine, according to the label information, a problem type that each model in the preset model library is good at;

[0204] determine, based on the problem type that each model is good at and the model role, a candidate model corresponding to each collaborative model in the preset model library.

[0205] A4. The method of A2, wherein the determining, based on the attribute information and the candidate model, a matching degree of the model role and the candidate model comprises:

[0206] determining, according to the attribute information, an accuracy, a response time length, and a device performance consumption parameter of the candidate model;

[0207] determining, according to the accuracy, the response time length, and the device performance consumption parameter of the candidate model, the matching degree of the model role and the candidate model.

[0208] A5. The method of A1, wherein the determining, according to an instruction type of the input user instruction, a model collaboration scheme in response to the input user instruction comprises:

[0209] determining, according to an instruction type of the input user instruction, a corresponding task type in response to the input user instruction;

[0210] determining a task complexity based on the task type;

[0211] determining, according to the task complexity, whether multiple models are needed to collaborate to complete a task;

[0212] determining, according to the task complexity and the task type, a model collaboration scheme in a case where multiple models are needed to collaborate to complete the task.

[0213] A6. The method of A5, wherein the determining, according to the task complexity and the task type, a model collaboration scheme comprises:

[0214] determining, according to the task complexity, a model quantity of multiple collaborative models;

[0215] determining, according to the task type, a model role of the multiple collaborative models;

[0216] determining, according to the model quantity and the model role, a model collaboration scheme.

[0217] A7. The method of A6, wherein the determining, according to the task type, a model role of the multiple collaborative models comprises:

[0218] determining, based on the task type, a task requirement;

[0219] determining, according to the task requirement, a plurality of subtasks by decomposing a corresponding task;

[0220] determining the model roles of the corresponding collaboration models according to the subtasks.

[0221] A8. The method of A1, wherein the displaying the model roles of the collaboration models and the corresponding recommended models comprises:

[0222] obtaining the working sequences of the collaboration models and the user device performance parameters;

[0223] displaying the model roles of the collaboration models according to the working sequences of the collaboration models;

[0224] determining display priorities of the recommended models corresponding to the collaboration models according to the user device performance parameters;

[0225] arranging and displaying the recommended models corresponding to the collaboration models based on the display priorities.

[0226] A9. The method of A1, wherein after the displaying the model roles of the collaboration models and the corresponding recommended models, the method further comprises:

[0227] determining a target collaboration model combination based on the recommended models corresponding to the collaboration models;

[0228] collaborating through the target collaboration models in the target collaboration model combination to generate and display reply information of the user instruction.

[0229] A10. The method of A9, wherein the determining the target collaboration model combination based on the recommended models corresponding to the collaboration models comprises:

[0230] in response to an input selection instruction, determining target recommended models corresponding to the collaboration models according to the selection instruction and the recommended models corresponding to the collaboration models;

[0231] generating the target collaboration model combination based on the target recommended models corresponding to the collaboration models.

[0232] A11. The method of A9, wherein the determining the target collaboration model combination based on the recommended models corresponding to the collaboration models further comprises:

[0233] determining a plurality of collaboration model combinations based on the recommended models corresponding to the collaboration models;

[0234] performing comprehensive performance evaluation on the collaboration model combinations to obtain comprehensive performance scores of the collaboration model combinations;

[0235] determining the target collaboration model combination from the plurality of collaboration model combinations according to the comprehensive performance scores.

[0236] The application further discloses B12. A model recommendation device based on multi-model cooperation, comprising:

[0237] A determination module is configured to determine a model cooperation scheme according to an instruction type of an input user instruction in response to the user instruction, wherein the model cooperation scheme comprises a model number and a model role of a plurality of cooperation models;

[0238] A selection module is configured to select a corresponding to-be-recommended model for each cooperation model from a preset model library according to the model role;

[0239] A display module is configured to display the model role of each cooperation model and the corresponding to-be-recommended model.

[0240] B13. The device of B12, wherein the selection module is further configured to acquire attribute information and label information of each model in the preset model library;

[0241] determine a candidate model corresponding to each cooperation model in the preset model library based on the label information and the model role;

[0242] determine a matching degree of the model role and the candidate model based on the attribute information and the candidate model;

[0243] select a corresponding to-be-recommended model for each cooperation model from the preset model library according to the matching degree.

[0244] B14. The device of B13, wherein the selection module is further configured to determine a good-at problem type of each model in the preset model library according to the label information;

[0245] determine a candidate model corresponding to each cooperation model in the preset model library based on the good-at problem type of each model and the model role.

[0246] B15. The device of B13, wherein the selection module is further configured to determine an accuracy rate, a response time length and a device performance consumption parameter of the candidate model according to the attribute information;

[0247] determine a matching degree of the model role and the candidate model according to the accuracy rate, the response time length and the device performance consumption parameter of the candidate model.

[0248] B16. The device of B12, wherein the determination module is further configured to determine a corresponding task type according to an instruction type of an input user instruction in response to the user instruction;

[0249] determine a task complexity based on the task type; and determine whether a plurality of models are needed to cooperate to complete a task according to the task complexity;

[0250] In a case where multiple models are required to cooperate to complete a task, a model cooperation scheme is determined according to the task complexity and the task type.

[0251] B17. The apparatus of B16, and the determining module is further configured to determine a model number of the multiple cooperation models according to the task complexity.

[0252] determine a model role of the multiple cooperation models according to the task type; and determine the model cooperation scheme according to the model number and the model role.

[0253] B18. The apparatus of B17, and the determining module is further configured to determine a task requirement based on the task type.

[0254] determine a task requirement according to the task type; and decompose a corresponding task into multiple sub-tasks according to the task requirement.

[0255] determine a model role of a corresponding cooperation model according to each of the sub-tasks.

[0256] The application further discloses C19. A model recommendation device based on multiple model cooperation, comprising a memory, a processor, and a model recommendation program based on multiple model cooperation stored on the memory and executable on the processor, the model recommendation program based on multiple model cooperation being configured to implement the model recommendation method based on multiple model cooperation.

[0257] The application further discloses D20. A storage medium, the storage medium storing a model recommendation program based on multiple model cooperation, the model recommendation program based on multiple model cooperation being executed by a processor to implement the model recommendation method based on multiple model cooperation.

Claims

1. A model recommendation method based on multi-model collaboration, characterized in that: The method comprises: In response to an input user instruction, determining a model collaboration scheme according to an instruction type of the user instruction, the model collaboration scheme including the number of models and model roles of the plurality of collaboration models; Selecting a corresponding model to be recommended for each collaborative model from a preset model library according to the model role; The model role of each collaborative model and the corresponding model to be recommended are displayed.

2. The method according to claim 1, wherein The selecting a corresponding model to be recommended for each collaborative model from a preset model library according to the model role includes: Get the attribute information and label information of each model in the preset model library; Determine, based on the label information and the model role, a candidate model corresponding to each collaborative model in a preset model library; determining a degree of matching between the model role and the candidate model based on the attribute information and the candidate model; According to the matching degree, a corresponding model to be recommended is selected from the preset model library for each collaborative model.

3. The method according to claim 2, wherein The determining, based on the tag information, candidate models corresponding to each collaborative model in the preset model library includes: Determine the problem type that each model in the preset model library is good at based on the label information; The candidate models corresponding to the collaborative models in the preset model library are determined based on the problem types that the models are good at and the model roles.

4. The method according to claim 2, wherein The determining, based on the attribute information and the candidate model, a degree of matching between the model role and the candidate model, includes: Determining the accuracy, response time, and device performance consumption parameters of the candidate model based on the attribute information; The degree of matching between the model role and the candidate model is determined according to the accuracy, response time and device performance consumption parameters of the candidate model.

5. The method according to claim 1, wherein The step of determining, in response to an input user instruction and according to an instruction type of the user instruction, a model collaboration scheme includes: In response to an input user instruction, determining a corresponding task type according to an instruction type of the user instruction; determining a task complexity based on the task type; Determine whether multiple models need to collaborate to complete the task based on the complexity of the task; When multiple models need to collaborate to complete a task, a model collaboration scheme is determined based on the complexity and type of the task.

6. The method according to claim 5, wherein The determining of the model collaboration scheme according to the task complexity and task type includes: Determining the number of models of the multiple collaborative models according to the complexity of the task; determining model roles of a plurality of collaborative models according to the task type; A model collaboration scheme is determined according to the number of models and the model roles.

7. The method according to claim 6, wherein The determining of the model roles of the plurality of collaborative models according to the task type includes: determining task requirements based on the task type; Decompose the corresponding task into multiple subtasks according to the task requirements; The model role of the corresponding collaboration model is determined according to each of the subtasks.

8. A model recommendation device based on multi-model collaboration, characterized in that: The model recommendation device based on multi-model collaboration includes: a determination module, configured to determine, in response to an input user instruction, a model collaboration scheme according to an instruction type of the user instruction, the model collaboration scheme including the number of models and model roles of the plurality of collaboration models; A selection module, configured to select a corresponding model to be recommended for each collaborative model from a preset model library according to the model role; The display module is used to display the model roles of each collaborative model and the corresponding models to be recommended.

9. A model recommendation device based on multi-model collaboration, characterized in that: The model recommendation device based on multi-model collaboration includes: a memory, a processor, and a model recommendation program based on multi-model collaboration stored on the memory and executable on the processor, wherein the model recommendation program based on multi-model collaboration is configured to implement the model recommendation method based on multi-model collaboration as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium stores a model recommendation program based on multi-model collaboration, and when the model recommendation program based on multi-model collaboration is executed by the processor, the model recommendation method based on multi-model collaboration as described in any one of claims 1 to 7 is implemented.

Images