Method and device for generating task planning information based on computing power of intelligent computing center

Abstract

The invention provides a method and device for generating task planning information based on computing power of an intelligent computing center, and relates to the technical field of intelligent computing centers, intelligent computing centers and computing power infrastructure, and the method comprises the steps: taking a to-be-processed first business demand as a task planning target, carrying out task planning based on a target model, and obtaining target planning data, wherein the target planning data is used for realizing a response to a first business demand, the first business demand corresponds to a computer vision business, the target model is obtained by performing fine tuning on the initial model based on a plurality of fine tuning data, and the fine tuning data comprises a second business demand, planning prompt data and business planning data; the planning prompt data is used for prompting the model reasoning logic from the second business demand to the business planning data. Through introduction of the planning prompt data, before the model is guided to generate the planning data, more thinking is carried out, and the execution reliability of the target planning data output by the model is improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of intelligent computing centers, intelligent computing centers and computing power infrastructure technologies, and particularly relates to a method and device for generating task planning information of computing power based on an intelligent computing center. Background Art

[0002] With the rapid development of artificial intelligence technology, "intelligent computing centers" and "intelligent computing centers" have emerged as the times require.

[0003] An "intelligent computing center" refers to a facility that uses large-scale heterogeneous computing power resources, including general computing power and intelligent computing power, and mainly provides the required computing power, data, and algorithms for artificial intelligence applications (such as scenarios of artificial intelligence deep learning model development, model training, and model inference, etc.). The intelligent computing center covers facilities, hardware, and software, and can provide full-stack capabilities from underlying computing power to top-level application enabling.

[0004] The "intelligent computing center" includes but is not limited to the "intelligent computing center".

[0005] An "intelligent computing center", that is, an artificial intelligence computing center, is a type of computing power infrastructure that provides computing power services, data services, and algorithm services required for artificial intelligence applications based on artificial intelligence theory and using an artificial intelligence computing architecture.

[0006] "Computing power" is the core of "intelligent computing centers" and "intelligent computing centers". It is the ability of computer devices or computing / data centers to process information, the ability of computer hardware and software to cooperate to jointly execute a certain computing requirement, the computing ability to achieve the output of target results by processing information data, and a new type of productive force integrating information computing power, network carrying capacity, and data storage capacity. It mainly provides services to society through computing power infrastructure.

[0007] Currently, how to quickly generate an intelligent agent (agent) that matches the requirements is an urgent problem to be solved. A possible solution is: by means of custom coding, a batch of general functions (functions) and the corresponding function patterns of the functions are pre-constructed in advance. This part of the pre-constructed functions can be called base functions (base function), and the corresponding function patterns of the functions can be called base function schemas (base function schema). By providing the base function schema to the large model so that the large model can master the uses and usages of the base function, the large model, as the business processing core of the pre-generated intelligent agent, can, after the intelligent agent receives the business requirements input by the user, dynamically form a static plan (plan) that meets the above business requirements based on the mastered uses and usages of the base function, and execute the static plan to achieve the response to the business requirements.

[0008] After the large model has mastered the uses and usages of the base function, the large model can also be fine-tuned. The model parameters of the large model are adjusted with the plan data that has been verified and can be successfully executed by the executor, the model output format during the subsequent application of the large model is standardized, and the probability that the static plan output by the large model can be successfully executed by the executor is increased. It is found in the application that after the large model is fine-tuned based on the related technology, the probability that the static plan output by the large model can be successfully executed by the executor is still not as expected.

[0009] It can be seen that based on the method of using the large model to generate task planning information in the prior art, the generated static plan has a very low execution reliability problem. Summary of the Invention

[0010] The purpose of the present invention is to provide a method and device for generating task planning information based on the computing power of an intelligent computing center, which is used to solve the problem that the execution reliability of the static planning information generated by the prior art is very low.

[0011] In order to solve the above technical problems, the present invention is implemented as follows:

[0012] In a first aspect, the present invention provides a method for generating task planning information based on the computing power of an intelligent computing center, including:

[0013] Step S1: Using the first service requirement to be processed as the task planning target, perform task planning based on the target model to obtain target planning data, where the target planning data is used to be executed by the task planning executor to achieve the response to the first service requirement. The first service requirement corresponds to the computer vision service, and the target model is a model obtained by fine-tuning the initial model based on multiple fine-tuning data. The fine-tuning data includes a second service requirement, planning prompt data, and service planning data. The planning prompt data is used to prompt the model inference logic from the second service requirement to the service planning data. The service planning data includes multiple step data bodies, and the multiple step data bodies correspond to multiple service steps one by one. The multiple service steps are used to achieve the response to the second service requirement.

[0014] In one embodiment, the planning prompt data is generated based on the service planning data.

[0015] In one embodiment, in step S1, the second service requirement is used as the input of the model, and the planning prompt data and the service planning data are used as the true value output corresponding to the input.

[0016] In one embodiment, the reading position of the planning prompt data in the fine-tuning data is the first position, and the reading position of the service planning data in the fine-tuning data is the second position, and the first position is before the second position.

[0017] In one embodiment, the number of the planning prompt data is less than or equal to a set number threshold.

[0018] In one embodiment, before the step S1, the method further includes:

[0019] Step S01': Fine-tune the initial model according to a plurality of fine-tuning data sets respectively to obtain a plurality of candidate models, where the plurality of candidate models and the plurality of fine-tuning data sets correspond one by one, the fine-tuning data set includes multiple groups of candidate fine-tuning data, different fine-tuning data sets correspond to different prompt rules, and different fine-tuning data sets include at least one of the following differences: the number of the planning prompt data is different, the data format of the planning prompt data is different, and the semantic depth of the planning prompt data is different;

[0020] Step S02': Use a preset test data set to perform task planning tests on the plurality of candidate models respectively to obtain a plurality of test results, and the test results are used to represent: the ratio of the successfully executed task planning data output by the corresponding candidate model;

[0021] Step S03': Among the plurality of candidate models, determine the candidate model corresponding to the highest ratio as the target model.

[0022] In a second aspect, the present invention further provides a device for generating task planning information based on the computing power of an intelligent computing center, including:

[0023] A task planning module, which uses a to-be-processed first service requirement as a task planning target, performs task planning based on a target model to obtain target planning data, where the target planning data is used to be executed by a task planning executor to realize the response to the first service requirement, the first service requirement corresponds to a computer vision service, the target model is a model obtained by fine-tuning an initial model based on a plurality of fine-tuning data, the fine-tuning data includes a second service requirement, planning prompt data, and service planning data, the planning prompt data is used to prompt: the model inference logic from the second service requirement to the service planning data; the service planning data includes a plurality of step data bodies, the plurality of step data bodies correspond to a plurality of service steps one by one, and the plurality of service steps are used to realize the response to the second service requirement.

[0024] In a third aspect, the present invention further provides a server, including: a processor, a memory, and a program stored on the memory and executable on the processor. When the program is executed by the processor, it implements the steps in the method for generating task planning information based on the computing power of an intelligent computing center as described in the first aspect above.

[0025] In a fourth aspect, the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps in the method for generating task planning information based on the computing power of an intelligent computing center as described in the first aspect above.

[0026] In a fifth aspect, the present invention provides a computer program product, including computer instructions. When the computer instructions are executed by a processor, they implement the steps in the method for generating task planning information based on the computing power of an intelligent computing center as described in the first aspect above.

[0027] In the present invention, through the introduction of planning prompt data, the model is guided to think more before generating planning data according to business requirements, improving the accuracy of the model's task planning, thereby increasing the probability that the planning data output by the target model is successfully executed, that is, improving the execution reliability of the output planning data. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0029] Figure 1 is an architecture diagram of a two-stage agent automation workflow provided by the present invention;

[0030] Figure 2 is a schematic diagram of the results of an experimental agent automation workflow using Qwen2.5 - 72B - Instruct provided by the present invention;

[0031] Figure 3 is a schematic flowchart of a method for generating task planning information based on the computing power of an intelligent computing center provided by the present invention;

[0032] Figure 4 is a schematic structural diagram of a device for generating task planning information based on the computing power of an intelligent computing center provided by the present invention;

[0033] Figure 5 is a schematic structural diagram of an electronic device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] The technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts fall within the protection scope of the present invention.

[0035] First, the technical terms related to the present invention will be briefly described below.

[0036] The "computing power" described in the present invention refers to the ability of a computer device or a computing / data center to process information. It is the ability of computer hardware and software to cooperate to execute a certain computing requirement together. It is the computing ability to process information data and output a target result. It is a new type of productive force integrating information computing power, network carrying capacity, and data storage capacity, and mainly provides services to society through computing power infrastructure.

[0037] The "computational power" (Computational Power, CP) described in the present invention is an ability of a data center server to process data and output a result. It is a comprehensive index to measure the computing ability of a data center, including general computing ability, supercomputing ability, and intelligent computing ability. The commonly used measurement unit is the number of floating-point operations per second (FLOPS: Floating Point Operations Per Second, 1EFLOPS = 10^18 FLOPS). The larger the value, the stronger the comprehensive computing ability. It is estimated that 1EFLOPS is approximately the computing power output of 5 Tianhe 2A or 500,000 mainstream server CPUs or 2 million mainstream laptops. The calculation formula is: CP = CP 通用 +CP 智能 +CP 超级 。

[0038] The "network power" (NP) described in the present invention is the performance of the data transmission ability of computing power facilities, including comprehensive capabilities such as network architecture, network bandwidth, transmission delay, intelligent management and scheduling, etc. The network power involves network transmission within and between data centers and is a comprehensive index to measure the network transmission scheduling ability. In the present invention, the network power uses the video memory bandwidth.

[0039] The "Storage Power" (SP) described in the present invention is the comprehensive ability of a data center in four aspects: data storage capacity, performance, security and reliability, and green and low-carbon. It is a comprehensive indicator to measure the data storage ability of a data center, including external storage devices such as storage arrays and built-in storage devices of servers. The commonly used measurement unit for storage capacity is exabyte (EB, 1EB = 2^60 bytes), the commonly used measurement unit for performance is the number of read / write operations per second per unit capacity (Input / Output Operations Per Second / TB, IOPS / TB), and the disaster recovery ratio is an important manifestation of security and reliability.

[0040] The "computing power infrastructure" described in the present invention is a new type of information infrastructure that integrates information computing power, network carrying capacity, and data storage power, and can realize the centralized computing, storage, transmission, and application of information.

[0041] The "new type of information infrastructure" described in the present invention refers to: mainly including network infrastructures such as 5G networks, fiber broadband networks, backbone networks, international communication networks, and satellite Internet, computing power infrastructures such as data centers, general computing power centers, intelligent computing centers, and supercomputing centers, and new technology facilities such as artificial intelligence, blockchain, and quantum computing.

[0042] The "computing power" described in the present invention includes general computing power, intelligent computing power, and super computing power.

[0043] The "general computing power" described in the present invention is the computing power provided by servers based on central processing unit (CPU) chips, and is used to support basic general computing such as cloud computing and edge computing.

[0044] The "intelligent computing power" described in the present invention is for various artificial intelligence innovation applications, and is a computing platform based on the large-scale deployment of dedicated chips such as graphics processing unit (GPU), field programmable gate array (FPGA), and application specific integrated circuit (ASIC), such as natural language processing and machine vision.

[0045] The "super computing power" described in the present invention is mainly the computing power provided by high-performance computing clusters such as supercomputers. It utilizes the centralized computing resources of multiple computer systems working in parallel and processes extremely complex or data-intensive problems through a dedicated operating system, and is mainly used for computing in cutting-edge scientific fields, such as planetary simulation, drug molecule design, and gene analysis.

[0046] The "Intelligent Computing Center" described in the present invention refers to a facility that uses large-scale heterogeneous computing power resources, including general computing power (CPU) and intelligent computing power (GPU, FPGA, etc.), and mainly provides the required computing power, data, and algorithms for artificial intelligence applications (such as scenarios like artificial intelligence deep learning model development, model training, and model inference). The intelligent computing center covers facilities, hardware, and software, and can provide full-stack capabilities from underlying computing power to top-level application enabling.

[0047] The "Intelligent Computing Center" described in the present invention includes, but is not limited to, the "Intelligent Computing Center".

[0048] The "Intelligent Computing Center" described in the present invention, namely the artificial intelligence computing center, is a type of computing power infrastructure based on artificial intelligence theory, adopting an artificial intelligence computing architecture, and providing computing power services, data services, and algorithm services required for artificial intelligence applications.

[0049] The "Computing Power Center" described in the present invention refers to a facility mainly composed of infrastructure such as wind, fire, water, and electricity and IT software and hardware devices, and having computing power, carrying capacity, and storage capacity, including general data centers, intelligent computing centers, supercomputing centers, etc.

[0050] The "Supercomputing Center" described in the present invention, that is, the supercomputing data center, is a data center based on supercomputers or large-scale computing clusters, capable of providing functions such as large-scale computing, storage, and network services, and is widely used in application scenarios such as aerospace, national defense, oil exploration, climate modeling, and genome sequencing.

[0051] The "Computing Power Resources" described in the present invention refer to technologies and facilities required for the development of the digital society and having information computing, transmission, storage, and application capabilities, including, but not limited to, computing resources such as CPU and GPU, network resources such as switches and routers, storage resources such as storage arrays and distributed storage, security resources such as firewalls and intrusion detection systems, and support and guarantee resources such as wind, fire, water, and electricity.

[0052] The "models" and "large models" described in the present invention include, but are not limited to, "large language models" and "multimodal large models".

[0053] The "Large Language Model" described in the present invention refers to a large language model (LLM), which is a language model with a relatively large number of parameters, aiming to understand and generate human language, trained through a large amount of text data, and can perform a wide range of tasks including text summarization, translation, sentiment analysis, etc.

[0054] The "multimodal large models" described in the present invention refer to models trained by jointly combining multimodal information such as text, images, videos, and audio, including but not limited to multimodal large language models.

[0055] The "planning prompt data" described in the present invention refers to the transitional data set between business requirements and business planning data, which is used to guide the model to perform more thinking before forming planning data according to business requirements.

[0056] The "execution reliability" described in the present invention refers to the probability that the planning data can be successfully executed by the task planning executor.

[0057] The "task planning information" described in the present invention can be understood as: taking the business input by the user as the input, obtaining through task planning by the task planning large model, and the planning data that can be executed by the task planning executor (in the present invention, the target planning data can be understood as an example of the task planning information).

[0058] The "agent" described in the present invention refers to an agent that can perceive the environment and take actions to achieve specific goals. It can be software, hardware, or a system, with autonomy, adaptability, and interaction capabilities. The agent perceives changes in the environment (such as through sensors or data input), makes judgments and decisions based on the knowledge and algorithms learned by itself, and then executes actions to affect the environment or achieve a predetermined goal. Agents are widely used in the field of artificial intelligence, commonly found in automation systems, robots, virtual assistants, and game characters, etc. The core lies in the ability to learn independently and evolve continuously to better complete tasks and adapt to complex environments.

[0059] The "base function" described in the present invention refers to the functions pre-constructed by developers according to the general / high-frequency business operations of the business scenarios faced by the agent. For example, when the agent is specifically a computer vision agent, the general / high-frequency business operations of the business scenarios faced by the agent may include: image preprocessing operations, image segmentation operations, image noise filtering operations, image target detection operations, etc.

[0060] The "base function schema" described in the present invention refers to the "base function user manual" formed by developers based on the functions and usage of base functions. The base function schema may include items such as name, description, type, etc. Among them, the name item may include the function name, the input parameter names of the function, the output parameter names of the function, etc. The description item can be understood as the function description, the semantic description of the input parameters of the function, and the semantic description of the output parameters of the function; the type item includes the data format description of the input parameters of the function and the data format description of the output parameters of the function. The large model can master the uses and usages of base functions through the "base function schema".

[0061] The "planner" described in the present invention refers to a large model for task planning. After the task rule large model masters the uses and usages of base functions through the base function schema, it can be used to generate a static plan in one go. The static plan contains all the business processing steps to complete the business requirements, and each business processing step includes its explanation (also known as the step description text), the method name called, and the input parameter information (each input parameter and its corresponding assignment).

[0062] The "autocoder" described in the present invention refers to a large model for code writing, which is used to write adapter functions.

[0063] The "static planning information" described in the present invention refers to the data representation of the static plan, which at least includes a plurality of step data bodies arranged in order. The plurality of step data bodies correspond one-to-one to a plurality of business processing steps, and the plurality of business processing steps are executed in sequence to complete the user's business requirements.

[0064] The "step data body" described in the present invention is the data representation of a corresponding business processing step, including but not limited to: the step description text of a corresponding business processing step (used to describe the business processing step to indicate the function, purpose, and / or execution method of the step, so as to facilitate users or developers to understand and implement), the step number of a corresponding business processing step among the plurality of business processing steps, the input data of a corresponding business processing step (such as the data name, data type requirements, data source, etc. of the input data), and the data processing method of a corresponding business processing step (calling a specific method through the method name).

[0065] The "pipeline" described in the present invention refers to the process of executing the static plan.

[0066] The "execution reliability" described in the present invention refers to the probability that the planning data can be successfully executed by the mission planning executor.

[0067] The "planning hint data" described in the present invention refers to the model inference logic used to hint from the second business requirement to the business planning data.

[0068] Regarding the problem of quickly generating an intelligent agent (agent) that matches user needs, the present invention provides an intelligent agent automation workflow solution, which specifically is: By means of custom coding, a batch of general functions (functions) and the function patterns corresponding to the functions are pre-constructed in advance. This part of the pre-constructed functions can be called base functions (basefunction), and the function patterns corresponding to the functions can be called base function schemas (base function schema). By providing the base function schema to the large model, so that the large model can master the uses and usages of the base function. The large model, as the business processing core of the pre-generated intelligent agent, can, after the intelligent agent receives the business requirements input by the user, based on the uses and usages of the mastered basefunction, dynamically form a static plan (plan) that meets the above-mentioned business requirements, and execute this static plan to achieve a dynamic response to the business requirements; In the actual scenario, the base function may not cover all business requirements. For this problem, the large model can be used to temporarily write code to generate an adapter function (adapter function) that can fill the gap of the base function, so as to cooperate with the base function to achieve full coverage of business requirements.

[0069] In an example, the powerful computing power and flexible resource configuration of the intelligent computing center are used to implement the foregoing intelligent agent automation workflow solution. The corresponding architecture of the two-stage auto-pipeline can be as Figure 1 shown, Figure 1 The relevant illustrations in

[0070] User requirements (one user query): Users input their requests or requirement content.

[0071] Task planning large model (planner): Used to parse user requirements and formulate an execution plan (plan). If the parsing fails, an error that cannot be converted into JSON format may be output.

[0072] Planning list (plan list): A list of specific steps generated according to user requirements.

[0073] Retrieve all-step function dependencies: Retrieve the function dependencies related to each step to ensure that all required functions are available.

[0074] Get executable input values:

[0075] Given values: The values directly provided to the function.

[0076] User inputs: The dynamic input values provided by the user in the query.

[0077] Previous function outputs: The function output values from the previous steps.

[0078] Media download: This link can also support the download of additional required media content.

[0079] Execute one step: It refers to the indication of the execution details of a specific step.

[0080] Execute function: Used to actually call the function to perform data processing work according to the obtained input values.

[0081] Base function: Execute specific basic functions to support the agent to have the set basic business processing capabilities.

[0082] Adapter function (autoencoder): As a supplement to the base function.

[0083] Retrieve depended function schemas: Obtain the structures or schemas of other dependent functions.

[0084] Autoencoder coding: Automatically generate the code involved in the above process.

[0085] Exception code: If an error or exception occurs, the system will return a specific exception code for subsequent analysis and debugging. For example:

[0086] The exceptions can be classified into the following 6 categories:

[0087] 0: No exception, the pipeline execution is successful

[0088] 1: Other pipeline exceptions

[0089] 2: Automatic coding execution failed

[0090] 3: Basic function execution failed

[0091] 4: Planning (plan) acquisition failed (usually caused by the failure of converting plan to json)

[0092] 5: The variable placeholder in the plan is expressed incorrectly and the parsing fails.

[0093] It should be noted that the two-stage auto-pipeline has at least the following advantages:

[0094] 1. Higher accuracy: There are more cases where the given plan can be successfully executed;

[0095] 2. By decomposing tasks, the total inference time of the LLM can be shortened;

[0096] 3. By manually handling the core information transfer between steps, the scope of effective information can be clarified and the transparency of the system can be increased.

[0097] The results of experimenting with the two-stage auto-pipeline using the Qwen2.5-72B-Instruct tool for different data sources are as Figure 2 shown.

[0098] Experiments found that errors mainly occurred in the basic function execution stage and the variable placeholder expression stage. Among them, the errors in the basic function execution stage included: incorrect parameter passing, confusing use of picture lists and video paths, and some parameters unrelated to the input (such as thresholds), etc.; the errors in the variable placeholder expression stage included: importing external packages (lacking package dependencies), incorrect understanding and handling of complex parameters, etc. In addition, some errors were caused by poor compliance with the json format output by the plan.

[0099] In the above solution, the task planning large model is obtained by pre-training and fine-tuning a large model. Among them, the large model after pre-training is the initial model described in the present invention. The initial model has a certain task planning ability. The initial model can output corresponding planning data (plan) for the input business requirements (also known as user requirements). In the application, the available planning data and the corresponding business requirements are used as fine-tuning data to further fine-tune the model parameters and the model output format of the initial model, which can standardize the data format of the planning data output by the model and increase the probability that the planning data output by the model can be successfully executed by the task planning executor. Among them, the available planning data refers to the planning data that can be successfully executed by the task planning executor. The task planning executor is used to execute the planning data to generate an execution result for responding to the corresponding business requirement.

[0100] In the prior art, using the available planning data and the corresponding business requirements as fine-tuning data to further fine-tune the model parameters and the model output format of the initial model, the obtained model fine-tuning effect is lower than expected. More specifically, for the task planning large model obtained based on the fine-tuning solution provided by the prior art, the execution reliability of the output planning data is very low.

[0101] To address the above problems, in view of the problem that the model fine-tuning effect of the model fine-tuning solution provided by the prior art is poor, resulting in very low execution reliability of the static planning information output by the fine-tuned model, the present invention provides a method for generating task planning information based on the computing power of an intelligent computing center. On the premise of utilizing the powerful computing power of the intelligent computing center, in the model fine-tuning stage, by introducing planning hint data to prompt the model with the model inference logic from business requirements to planning data, thereby guiding the model to think more before generating planning data according to business requirements, improving the accuracy of the model for task planning, and thus increasing the probability that the planning data output by the target model can be successfully executed, that is, improving the execution reliability of the output planning data.

[0102] Please refer to Figure 3 , Figure 3 which is a method for generating task planning information based on the computing power of an intelligent computing center provided by the present invention. As Figure 3 shown, it includes the following steps:

[0103] Step S1: Using the first business requirement to be processed as the task planning target, perform task planning based on the target model to obtain target planning data.

[0104] Among them, the target planning data is used to be executed by the task planning executor to achieve the response to the first business requirement, and the first business requirement corresponds to a computer vision business.

[0105] The target model is a model obtained by fine-tuning an initial model based on multiple pieces of fine-tuning data, where the fine-tuning data includes a second business requirement, planning prompt data, and business planning data. The planning prompt data is used to prompt the model inference logic from the second business requirement to the business planning data. The business planning data includes a plurality of step data bodies, and the plurality of step data bodies correspond one-to-one to a plurality of business steps, and the plurality of business steps are used to implement a response to the second business requirement.

[0106] Exemplarily, the computer vision service may be an image recognition service, an object detection service, an image segmentation service, a pose estimation service, a face recognition service, an image generation service, a video analysis service, an augmented reality service, an autonomous driving service, etc.

[0107] The above-mentioned first business requirement is used to represent: in the actual application process, the real business requirement input by the user to the computer vision agent. It should be noted that the computer vision agent uses the target model as the core of business processing.

[0108] In the application, any business requirement (such as the first business requirement or the second business requirement) at least includes the image / image set to be processed (or the acquisition address of the image / image set to be processed), and the information expected to be obtained from the image / image set to be processed. For example, the first business requirement may be: performing oil stain detection on the image / image set input by the user to determine whether the input image / image set includes oil stains; or performing face recognition and verification on the image / image set provided by the user, etc., to determine whether the provided image / image set includes a set face.

[0109] The task planning referred to in the present invention should be understood as: a process of performing requirement analysis on a business requirement to determine a requirement goal, and several steps to achieve the requirement goal, and forming planning data accordingly.

[0110] The above-mentioned second business requirement can be understood as a business requirement used to simulate the first business requirement. The business planning data is the aforementioned available planning data, and after being executed by the task planning executor, the business planning data can implement a response to the second business requirement.

[0111] The step data body is a data representation of a corresponding business step, and it at least includes: first data and second data, where the first data is used to represent: the function name of the execution function for implementing the corresponding business step, and the function function to be completed by the execution function; the second data is used to represent: the parameter name of the input parameter of the aforementioned execution function, the parameter name of the output parameter, the parameter format of the input parameter, and the parameter format of the output parameter.

[0112] In some embodiments, the planning prompt data can be obtained by analyzing the second business requirement (which can be approximately understood as the requirement splitting and refinement processing of the second business requirement); the planning prompt data can also be obtained by summarizing the business planning data; or the planning prompt data can be obtained by performing an association analysis on the second business requirement and the business planning data (which can be approximately understood as the derivation of the processing logic from the second business requirement to the business planning data).

[0113] For example, if the second business requirement is set as: "Perform oil stain detection on the image to be processed"; the business planning data is as follows:

[0114]

[0115]

[0116] Then, by only analyzing the second business requirement, the obtained planning prompt data can be: "1. Requirement analysis and data collection; 2. Image preprocessing and feature extraction; 3. Select and train a detection model; 4. System integration and testing; 5. Deployment implementation and maintenance to ensure efficient and accurate oil stain detection";

[0117] By only summarizing the business planning data, the obtained planning prompt data can be: "The oil stain detection agent made for the power company needs to process the input pictures and detect whether there is oil stain on 4 specific components (riser, pressure relief valve, cooler, bushing) in the transformer and reactor. The task can be divided into the following steps: image preprocessing, object detection, segmentation to detect oil stain, output the detection result and generate an alarm";

[0118] By deriving the processing logic from the second business requirement to the business planning data, the obtained planning prompt data can be: "For the task requirement of performing oil stain detection on the image to be processed, the execution plan first ensures that the data is suitable for analysis through image preprocessing. Then, use the detection algorithm to identify the key components of the transformer and reactor, and then perform region segmentation to identify the oil stain area. Finally, perform post-processing and classification to ensure the detection accuracy and send an alarm to notify relevant personnel."

[0119] It should be understood that since the fine-tuning data includes planning prompt data, the target planning data also includes target prompt data corresponding to the aforementioned planning prompt data. Therefore, before sending the target planning data to the task planning executor for task execution, it is necessary to clear the target prompt data included in the target planning data. In the application, the planning prompt data will be indicated in the fine-tuning data through specific prompt characters. Therefore, after obtaining the target planning data, the target prompt data in the target planning data can also be determined through the aforementioned set prompt characters, and the target prompt data can be cleared.

[0120] In the present invention, through the introduction of the planning prompt data, the model is guided to think more before generating the planning data according to the business requirements, improving the accuracy of the model for task planning, and further increasing the probability that the planning data output by the target model can be successfully executed, that is, improving the execution reliability of the output planning data.

[0121] In one embodiment, the planning prompt data is generated based on the business planning data.

[0122] In this embodiment, specifically, the planning prompt data is generated according to the business planning data, so as to use the planning prompt data to guide the model to think more effectively during the task planning stage while avoiding introducing interference to promote the generation quality of subsequent plans.

[0123] Furthermore, the planning prompt data is used to indicate the business execution order corresponding to the business planning data. Among them, the business execution order can be understood as a summary description of the multiple business steps indicated by the business planning data and the execution order of the multiple business steps.

[0124] More specifically, the planning prompt data can be understood as the summary of the business planning data.

[0125] In one embodiment, in the step S1, the second business requirement is used as the input of the model, and the planning prompt data and the business planning data are used as the true value outputs corresponding to the input.

[0126] In this embodiment, the second business requirement in the fine-tuning data is used as the input of the initial model in the fine-tuning stage, and the planning prompt data and the business planning data in the fine-tuning data are used as the true value outputs of the initial model in the fine-tuning stage. Before guiding the model to output the corresponding planning data for the business requirement, the prompt data is output first as an outline / summary of the subsequent output planning data, so as to promote the model to think more during the task planning process, thereby achieving the purpose of promoting the generation quality of subsequent plans.

[0127] Exemplarily, before the step S1, the method further includes:

[0128] Step S0: Based on the m fine-tuning data, perform m iterations of fine-tuning training on the initial model to obtain the target model;

[0129] Among them, the i-th iteration in the iterative fine-tuning includes:

[0130] Input the second business requirement in the i-th fine-tuning data into the i-th model for task planning to obtain the i-th predicted planning information, where i is an integer greater than 0 and less than or equal to m. When i is equal to 1, the i-th model is the initial model, and m is used to represent the number of the foregoing multiple fine-tuning data, and m is an integer greater than 1;

[0131] Generate the i-th loss data according to the i-th predicted planning information, the planning prompt data in the i-th fine-tuning data, and the business planning data in the i-th fine-tuning data;

[0132] Adjust the model parameters of the i-th model according to the i-th loss data to obtain the (i + 1)-th model. The above loss data is used to represent the data difference between the predicted planning information and the business planning data.

[0133] Among them, adjusting the model parameters of the i-th model according to the i-th loss data to obtain the (i + 1)-th model includes: using the i-th loss data as the input of backpropagation to adjust the model parameters of the i-th model to obtain the (i + 1)-th model.

[0134] In one embodiment, the reading position of the planning prompt data in the fine-tuning data is the first position, and the reading position of the business planning data in the fine-tuning data is the second position, and the first position is before the second position.

[0135] Among them, the reading position of the planning prompt data in the fine-tuning data is: the setting position of the planning prompt data in the fine-tuning data; the reading position of the business planning data in the fine-tuning data is: the setting position of the business planning data in the fine-tuning data.

[0136] It should be understood that the initial model will sequentially read the content in the fine-tuning data. Therefore, the earlier the setting position is, the earlier the corresponding content will be read by the initial model.

[0137] In this embodiment, make the planning prompt data be located before the corresponding business planning data to guide the initial model to think more before generating the planning data according to the business requirements, improve the accuracy of the model for task planning, and thus increase the probability that the planning data output by the target model is successfully executed.

[0138] In one embodiment, the number of pieces of the planning prompt data is less than or equal to a set number threshold.

[0139] The number of pieces of the planning prompt data can also be understood as the number of characters of the planning prompt data.

[0140] In this embodiment, by restricting the number of pieces of the planning prompt data, it is avoided that too long planning prompt data contains too much semantics, thereby imposing unnecessary constraints or interferences on subsequent planning generation. While guiding the model to do some thinking before task planning, it is also avoided that the model overthinks before task planning, thereby improving the execution reliability of the finally output task planning data.

[0141] In some embodiments, on the premise of implementing the above-mentioned restriction on the number of pieces of data, the planning prompt data can be further set as the summary of the business planning data, and this summary is only used to represent the business processing ideas corresponding to multiple business steps of the business planning data, but does not include the specific business to be processed in each business step, the specific functions to be called, the specific parameters involved, etc., so as to further increase the probability that the planning data output by the model is successfully executed.

[0142] Specifically, through experiments, it is found that when the planning prompt data includes the specific business to be processed in each business step, the specific functions to be called, the specific parameters involved, etc., it will cause the model to overthink before task planning and reduce the execution reliability of the finally output task planning data (compared with the case of only using the second business requirement and business planning data as the model fine-tuning data).

[0143] Therefore, setting the planning prompt data as the summary of the business planning data, and this summary is only used to represent the business processing ideas (or a simple summary of the business processing ideas) corresponding to multiple business steps of the business planning data, can guide the model to do more thinking while avoiding the problem of the model overthinking, so as to achieve the purpose of increasing the probability that the planning data output by the model is successfully executed.

[0144] In one embodiment, before the step S1, the method further includes:

[0145] Step S01’: Fine-tuning the initial model respectively according to a plurality of fine-tuning data sets to obtain a plurality of candidate models.

[0146] Among them, the plurality of candidate models and the plurality of fine-tuning data sets correspond one by one. The fine-tuning data set includes multiple groups of candidate fine-tuning data. Different fine-tuning data sets correspond to different prompting rules. Different fine-tuning data sets include at least one of the following differences: the number of pieces of the planning prompt data is different, the data format of the planning prompt data is different, and the semantic depth of the planning prompt data is different.

[0147] The above-mentioned prompting rules may include: a data quantity constraint rule, a data format constraint rule, and a semantic depth constraint rule for data.

[0148] Exemplarily, the data quantity constraint rule is used to represent: the quantity interval corresponding to the data quantity of the planned prompting data. Different quantity intervals may include: [0, 30], [31, 50], [51, 100]. It should be noted that any two different quantity intervals do not overlap.

[0149] The data format constraint rule is used to represent: the data format adopted by the planned prompting data. Different data formats may include: plain text form, process form, table form, array form, etc.

[0150] The semantic depth constraint rule for data is used to represent: the semantic depth interval corresponding to the semantics of the planned prompting data. The different semantic depth intervals include: brief semantics, deep semantics. Among them, the brief semantics can be understood as a brief description of the business execution process of the business planning data, and the deep semantics can be understood as a step description of each business execution step of the business planning data (including parameter description, function call description, etc.).

[0151] The above different constraint rules can be combined arbitrarily to form multiple different prompting rules. For any two different prompting rules, the number of constraint rules included in the two is different, and / or, the different constraint rules included in the two corresponding to the same dimension are different.

[0152] Step S02': Use a pre-set test data set to perform task planning tests on the multiple candidate models respectively, and obtain multiple test results.

[0153] The test results are used to represent: the ratio of the successfully executed task planning data output by the corresponding candidate model.

[0154] Specifically, the ratio of the successfully executed task planning data output by the foregoing corresponding candidate model is: among the multiple task planning data output by the corresponding candidate model, the ratio of the task planning data successfully executed by the task planning executor to the quantity of the multiple task planning data.

[0155] Step S03': Among the multiple candidate models, determine the candidate model corresponding to the highest ratio as the target model.

[0156] It should be noted that each test data in the test dataset includes at least one test business requirement for testing the model. In this case, a candidate model takes the test business requirement as the model input and can output test plan data that can be successfully executed by the task planning executor, that is, the candidate model is considered to pass a test, and the test plan data is also used as the task plan data successfully executed by the task planning executor.

[0157] In this embodiment, different fine-tuning datasets corresponding to different prompting rules are pre-configured, and the same initial model is fine-tuned respectively to obtain multiple candidate models. Then, the same test dataset is used to test the multiple candidate models respectively, and the candidate model with the best test effect is determined as the target model to ensure the model fine-tuning effect obtained by the initial model.

[0158] Please refer to Figure 4 , Figure 4 which is a schematic structural diagram of a device for generating task planning information based on the computing power of an intelligent computing center provided by the present invention. As Figure 4 shown, the device 400 for generating task planning information based on the computing power of an intelligent computing center includes:

[0159] A task planning module 401, which takes the first business requirement to be processed as the task planning target, performs task planning based on the target model to obtain target planning data. Among them, the target planning data is used to be executed by the task planning executor to realize the response to the first business requirement. The first business requirement corresponds to a computer vision business, and the target model is a model obtained by fine-tuning the initial model based on multiple fine-tuning data. The fine-tuning data includes a second business requirement, planning prompt data, and business planning data. The planning prompt data is used to prompt the model inference logic from the second business requirement to the business planning data. The business planning data includes multiple step data bodies, and the multiple step data bodies correspond to multiple business steps one by one. The multiple business steps are used to realize the response to the second business requirement.

[0160] In one embodiment, the planning prompt data is generated based on the business planning data.

[0161] In one embodiment, in step S1, the second business requirement is used as the input of the model, and the planning prompt data and the business planning data are used as the true value outputs corresponding to the input.

[0162] In one embodiment, the reading position of the planning prompt data in the fine-tuning data is the first position, and the reading position of the business planning data in the fine-tuning data is the second position, and the first position is before the second position.

[0163] In one embodiment, the number of pieces of the planning prompt data is less than or equal to a set number threshold.

[0164] In one embodiment, the apparatus 400 for generating task planning information based on the computing power of an intelligent computing center further includes:

[0165] A model acquisition unit, configured to respectively fine-tune the initial model according to a plurality of fine-tuning data sets to obtain a plurality of candidate models, where the plurality of candidate models and the plurality of fine-tuning data sets correspond one by one, the fine-tuning data set includes multiple groups of candidate fine-tuning data, different fine-tuning data sets correspond to different prompting rules, and different fine-tuning data sets include at least one of the following differences: the number of pieces of the planning prompt data is different, the data format of the planning prompt data is different, and the semantic depth of the planning prompt data is different;

[0166] A model testing unit, configured to respectively perform task planning tests on the plurality of candidate models using a preset test data set to obtain a plurality of test results, where the test results are used to represent: the ratio of the successfully executed task planning data output by the corresponding candidate model;

[0167] A model determination unit, configured to determine, among the plurality of candidate models, the candidate model corresponding to the highest ratio as the target model.

[0168] The apparatus for generating task planning information based on the computing power of an intelligent computing center provided by the present invention can implement each process of the above embodiments of the method for generating task planning information based on the computing power of an intelligent computing center, and the technical features correspond one by one and can achieve the same technical effects. To avoid repetition, details are not described herein again.

[0169] It should be noted that the apparatus for generating task planning information based on the computing power of an intelligent computing center in the present invention can be a device, or a component, an integrated circuit, or a chip in an electronic device.

[0170] The present invention also provides an electronic device. Refer to Figure 5 , Figure 5 is a schematic structural diagram of an electronic device provided in an embodiment of the present invention. The electronic device includes a memory 501, a processor 502, and a program or instruction stored on the memory 501 and running. When the program or instruction is executed by the processor 502, it can implement Figure 1 any step in the corresponding method embodiment of generating task planning information based on the computing power of an intelligent computing center and achieve the same beneficial effects. Details are not described herein again.

[0171] Among them, the processor 502 can be a CPU, an ASIC, an FPGA, or a GPU.

[0172] A person skilled in the art can understand that all or part of the steps of the above-mentioned method embodiment for generating task planning information based on the computing power of an intelligent computing center can be completed by hardware related to program instructions, and the program can be stored in a readable medium.

[0173] The present invention also provides a readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the above Figure 4 Any steps in the corresponding method embodiment for generating task planning information based on the computing power of an intelligent computing center can achieve the same technical effect, and will not be repeated here to avoid repetition. The storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0174] The terms "first", "second", etc. in the present invention are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. In addition, the terms "include" and "have" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices. In addition, "and / or" is used in the present invention to represent at least one of the connected objects, such as A and / or B and / or C, which means including 7 situations including single A, single B, single C, and both A and B exist, both B and C exist, both A and C exist, and both A, B and C exist.

[0175] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0176] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described embodiment methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases, the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal (which can be a mobile phone, computer, server, air conditioner, or a second terminal device, etc.) to execute the methods of the various embodiments of the present invention.

[0177] The embodiments of the present invention are described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the purpose of the present invention and the scope protected by the claims, and all of them belong to the protection scope of the present invention.

Claims

1. A method for generating task planning information of computing power based on an intelligent computing center, characterized in that Including: Step S1: Using the first business requirement to be processed as the task planning goal, perform task planning based on the target model to obtain target planning data; Among them, the target planning data is used to be executed by the task planning executor to achieve the response to the first business requirement, and the first business requirement corresponds to the computer vision business; The target model is a model obtained by fine-tuning the initial model based on multiple fine-tuning data. The fine-tuning data includes the second business requirement, planning prompt data, and business planning data. The planning prompt data is used to prompt the model inference logic from the second business requirement to the business planning data. The business planning data includes multiple step data bodies, and the multiple step data bodies correspond to multiple business steps one by one. The multiple business steps are used to achieve the response to the second business requirement.

2. The method according to claim 1, characterized in that, The planning prompt data is generated based on the business planning data.

3. The method according to claim 2, wherein In the step S1, the second business requirement is used as the input of the model, and the planning prompt data and the business planning data are used as the true value output corresponding to the input.

4. The method according to claim 3, characterized in that, The reading position of the planning prompt data in the fine-tuning data is the first position, and the reading position of the business planning data in the fine-tuning data is the second position. The first position is before the second position.

5. The method according to any one of claims 1 to 4, characterized in that, The number of data of the planning prompt data is less than or equal to the set number threshold.

6. The method according to claim 1, characterized in that, Before the step S1, the method further includes: Step S01’: According to multiple fine-tuning data sets, respectively fine-tune the initial model to obtain multiple candidate models. Among them, the multiple candidate models and the multiple fine-tuning data sets correspond one by one. The fine-tuning data set includes multiple groups of candidate fine-tuning data. Different fine-tuning data sets correspond to different prompting rules. Different fine-tuning data sets include at least one of the following differences: different numbers of data of the planning prompt data, different data formats of the planning prompt data, different semantic depths of the planning prompt data; Step S02’: Use a pre-set test data set to respectively perform task planning tests on the multiple candidate models to obtain multiple test results. The test results are used to represent the ratio of the successfully executed task planning data output by the corresponding candidate model; Step S03’: Among the multiple candidate models, determine the candidate model corresponding to the highest ratio as the target model.

7. An apparatus for generating task planning information of computing power based on an intelligent computing center, characterized in that, Including: A task planning module, which uses the first business requirement to be processed as the task planning target, performs task planning based on the target model to obtain target planning data. Among them, the target planning data is used to be executed by a task planning executor to realize the response to the first business requirement. The first business requirement corresponds to the computer vision business. The target model is a model obtained by fine-tuning an initial model based on multiple fine-tuning data. The fine-tuning data includes a second business requirement, planning prompt data, and business planning data. The planning prompt data is used to prompt the model inference logic from the second business requirement to the business planning data. The business planning data includes multiple step data bodies, and the multiple step data bodies correspond to multiple business steps one by one. The multiple business steps are used to realize the response to the second business requirement.

8. A server, characterized in that, It includes: A processor, a memory, and a program stored on the memory and executable on the processor. When the program is executed by the processor, it realizes the steps of the method for generating task planning information based on the computing power of the intelligent computing center as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium. When the computer program is executed by a processor, it realizes the steps of the method for generating task planning information based on the computing power of the intelligent computing center as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes computer instructions. When the computer instructions are executed by a processor, it realizes the steps of the method for generating task planning information based on the computing power of the intelligent computing center as described in any one of claims 1 to 6.

Images