Method and system for realizing self-evolution of intelligent agents through computing power on the intelligent computing center cloud platform

Abstract

The present invention provides a method and system for realizing self-evolution of intelligent bodies through computing power on an intelligent computing center cloud platform, which relates to the technical fields of intelligent computing centers, intelligent computing centers and computing power infrastructure. The method is executed by an embedded intelligent body in a self-evolving intelligent body, and includes: step S1: determining that a functional module needs to be developed and determining the development requirements, the self-evolving intelligent body is a target intelligent body embedded with a development intelligent body; step S2: determining a design scheme for the functional module according to the development requirements; step S3: generating code for the functional module according to the development requirements and the design scheme; step S4: debugging the code of the functional module; when the development requirements are not met, checking the design scheme and code of the functional module, and returning to step S2 if the design scheme needs to be modified, and returning to step S3 if the code needs to be modified, until the development requirements are met or the number of debugging times exceeds a preset threshold; step S5: feeding back development result information to the user; step S6: embedding the intelligent body to launch the functional module.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent computing centers, smart computing centers and computing power infrastructure, and specifically to a method and system for an intelligent computing center cloud platform to achieve self-evolution of intelligent entities through computing power. Background Art

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

[0003] An "Intelligent Computing Center" is a facility that utilizes large-scale heterogeneous computing resources, including general-purpose and intelligent computing power, to provide the computing power, data, and algorithms required for AI applications (such as AI deep learning model development, model training, and model inference). The Intelligent Computing Center encompasses facilities, hardware, and software, providing a full stack of capabilities, from bottom-level computing power to top-level application enablement.

[0004] “Intelligent Computing Center” includes but is not limited to “Smart Computing Center”.

[0005] "Intelligent Computing Center" refers to an artificial intelligence computing center. It is a type of computing power infrastructure that is based on artificial intelligence theory, adopts artificial intelligence computing architecture, and provides computing power services, data services, and algorithm services required for artificial intelligence applications.

[0006] "Computing power" is the core of "intelligent computing center" and "intelligent computing center". It is the ability of computer equipment or computing / data center to process information. It is the ability of computer hardware and software to work together to perform certain computing needs. It is the computing power to achieve target result output by processing information data. It is a new type of productivity that integrates information computing power, network carrying capacity, and data storage capacity. It mainly provides services to society through computing power infrastructure.

[0007] An "agent" is an agent capable of perceiving its environment and taking actions to achieve specific goals. It can be software, hardware, or a system, possessing autonomy, adaptability, and interactivity. An agent perceives changes in its environment (e.g., through sensors or data input), makes judgments and decisions based on learned knowledge and algorithms, and then executes actions to influence the environment or achieve predetermined goals. Agents are widely used in the field of artificial intelligence, often found in automated systems, robots, virtual assistants, and game characters. Their core capability lies in their ability to autonomously learn and continuously evolve to better complete tasks and adapt to complex environments.

[0008] In existing technologies, intelligent agents can often only perform specific tasks. If their own functions (or capabilities) need to be upgraded, expanded, or optimized, they usually need to develop the required functional modules through manual development. However, the development cost of manually developing the functional modules of intelligent agents is high and the efficiency is low. Summary of the Invention

[0009] The present invention provides a method and system for realizing self-evolution of intelligent bodies through computing power in an intelligent computing center cloud platform, which is used to solve the problems of high cost and low efficiency of the existing artificial development of functional modules of intelligent bodies.

[0010] In order to solve the above-mentioned technical problems, the present invention is achieved as follows:

[0011] In a first aspect, the present invention provides a method for realizing self-evolution of intelligent agents through computing power on an intelligent computing center cloud platform, comprising:

[0012] Step S1: An embedded agent embedded in a self-evolving agent determines that a target functional module needs to be developed for the self-evolving agent, and determines the development requirements of the target functional module. The self-evolving agent is the target agent in which a development agent is embedded, and the embedded development agent is the embedded agent.

[0013] Step S2: The embedded agent determines the design scheme of the target functional module according to the development requirements and in combination with the large language model method;

[0014] Step S3: The embedded agent generates the code of the target functional module according to the development requirements and design plan in combination with the large language model method;

[0015] Step S4: The embedded agent deploys the operating environment of the target functional module in combination with the large language model method, and debugs the code of the target functional module based on the operating environment; when the debugging result indicates that the target functional module does not meet the development requirements, the design scheme and code of the target functional module are checked in combination with the large language model method, and when the design scheme needs to be modified, the process returns to step S2; when the code needs to be modified, the process returns to step S3, until the debugging result indicates that the target functional module meets the development requirements or the number of debugging times exceeds a preset threshold;

[0016] Step S5: the embedded agent feeds back the development result information of the target functional module to the user through the interactive interface of the self-evolving agent;

[0017] Step S6: The embedded agent brings the target functional module online.

[0018] Optionally, before step S1, the following steps may be further performed:

[0019] Step S01: The development agent receives the task information embedded in the target agent given by the user through the interaction interface of the development agent;

[0020] Step S02: The development agent automatically identifies the relevant information of the target agent, and embeds itself into the target agent according to the relevant information of the target agent to form the self-evolving agent.

[0021] Optionally, step S02 includes:

[0022] Step S021: The development agent combines the large language model method to read the documents and codes of the target agent, and determines the relevant information of the target agent based on the reading results. The relevant information includes at least one of the following: architecture, interface, dependent model, extensible functional module, and functional module that the target agent already has.

[0023] Optionally, step S1 includes:

[0024] Step S11: The embedded agent determines the need to develop a target functional module for the self-evolving agent and determines the development requirements of the target functional module based on the user's requirements for functional upgrade, expansion or optimization of the self-evolving agent; or, the embedded agent reflects on the current task or historical task of the self-evolving agent, determines the need to develop a target functional module for the self-evolving agent based on the reflection results, and determines the development requirements of the target functional module.

[0025] Optionally, step S1 includes:

[0026] Step S12: The embedded intelligent agent compares the development requirements with the historical development requirements in the database, and determines whether it is necessary to continue developing the target functional module based on the comparison results, wherein the database records the historical development requirements and the status of the historical development requirements, and the status includes at least one of the following: collected, confirmed, developed, online, and abandoned.

[0027] Optionally, the target function module includes at least one of the following: a reflection system, a memory system, a tool function, a thinking framework, a demand module and a learning system.

[0028] In a second aspect, the present invention provides a system for realizing self-evolution of intelligent agents through computing power on an intelligent computing center cloud platform, including a self-evolving intelligent agent, wherein the self-evolving intelligent agent is a target intelligent agent embedded with a development intelligent agent, and the embedded development intelligent agent is an embedded intelligent agent;

[0029] The embedded agent includes:

[0030] A requirement confirmation module is used to determine the need to develop a target functional module for the self-evolving intelligent agent and determine the development requirements of the target functional module;

[0031] A design module is used to determine the design scheme of the target functional module based on the development requirements and in combination with the large language model method;

[0032] An encoding module, configured to generate the code of the target functional module according to the development requirements and design plan in combination with a large language model method;

[0033] a debugging module for deploying the operating environment of the target functional module in combination with the large language model method, and debugging the code of the target functional module based on the operating environment; when the debugging result indicates that the target functional module does not meet the development requirements, checking the design scheme and code of the target functional module in combination with the large language model method; when the design scheme needs to be modified, triggering the design module to continue working; when the code needs to be modified, triggering the encoding module to continue working, until the debugging result indicates that the target functional module meets the development requirements or the number of debugging times exceeds a preset threshold;

[0034] A feedback module, configured to provide feedback to a user on the development result of the target functional module via an interactive interface of the self-evolving intelligent agent;

[0035] The online module is used to bring the target functional module online.

[0036] Optionally, the system further includes:

[0037] The development agent is used to receive task information embedded in the target agent given by the user through the interactive interface of the development agent; automatically identify the relevant information of the target agent, and embed itself into the target agent according to the relevant information of the target agent to form the self-evolving agent.

[0038] In a third aspect, the present invention provides an electronic device comprising: a processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the steps of the method for realizing self-evolution of intelligent bodies through computing power of the intelligent computing center cloud platform as described in the first aspect above are implemented.

[0039] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for realizing self-evolution of intelligent agents through computing power of an intelligent computing center cloud platform as described in the first aspect above.

[0040] In a fifth aspect, the present invention provides a computer program product comprising computer instructions, which, when executed by a processor, implement the steps of the method for realizing self-evolution of intelligent agents through computing power of an intelligent computing center cloud platform as described in the first aspect above.

[0041] In the present invention, a self-evolving intelligent agent with self-evolving ability is formed by embedding a development intelligent agent in the target intelligent agent. The development intelligent agent embedded in the self-evolving intelligent agent can confirm whether the target functional module is developed for the self-evolving intelligent agent, and determine the development requirements of the target functional module, and automatically complete the development, debugging and online launch of the target functional module, replacing the process of manually developing the target functional module, realizing the self-evolution of the intelligent agent, thereby effectively saving development costs, and the intelligent computing center cloud platform can provide sufficient computing power resources for the self-evolution of the intelligent agent, greatly improving the efficiency of the self-evolution of the development intelligent agent. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0043] Figure 1 Schematic diagram of the process of the method for realizing self-evolution of intelligent agents through computing power on the intelligent computing center cloud platform of the present invention;

[0044] Figure 2 This is a schematic diagram of the structure of the system of the intelligent computing center cloud platform of the present invention that realizes the self-evolution of intelligent agents through computing power;

[0045] Figure 3 Schematic diagram of the structure of the electronic device of the present invention. DETAILED DESCRIPTION

[0046] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0047] First, the technical terms involved in the present invention are briefly explained below.

[0048] The "computing power" mentioned in the present invention refers to: the ability of computer equipment or computing / data centers to process information, the ability of computer hardware and software to work together to execute certain computing requirements, and the computing power to achieve target result output by processing information data. It is a new type of productivity that integrates information computing power, network carrying capacity, and data storage capacity, and mainly provides services to society through computing power infrastructure.

[0049] The "computing power" (CP) mentioned in this invention refers to the ability of a data center server to process data and output results. It is a comprehensive indicator for measuring the computing power of a data center, including general computing power, supercomputing power, and intelligent computing power. The commonly used unit of measurement is the number of floating-point operations per second (FLOPS, 1EFLOPS=10^18 FLOPS). The larger the value, the stronger the comprehensive computing power. According to calculations, 1 EFLOPS is approximately the computing power output of 5 Tianhe-2A units, 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 .

[0050] The "Network Power" (NP) mentioned in this invention refers to: it is a manifestation of the data transmission capability of computing power facilities, including comprehensive capabilities such as network architecture, network bandwidth, transmission latency, intelligent management and scheduling, etc. It involves network transmission within and between data centers, and is a comprehensive indicator for measuring network transmission scheduling capabilities.

[0051] "Storage Power" (SP) as used in this document refers to the comprehensive capabilities of a data center in four areas: data storage capacity, performance, security and reliability, and environmental friendliness. It serves as a comprehensive indicator for measuring a data center's data storage capabilities, encompassing both external storage devices like storage arrays and internal server storage. Storage capacity is commonly measured in exabytes (EB, 1EB = 2^60 bytes), while performance is commonly measured in IOPS / TB (Input / Output Operations Per Second). Disaster recovery ratio is a key indicator of security and reliability.

[0052] The "computing power infrastructure" mentioned in the present invention refers to a new type of information infrastructure that integrates information computing power, network carrying capacity, and data storage capacity, and can realize the centralized calculation, storage, transmission and application of information.

[0053] The "new information infrastructure" mentioned in the present invention refers to: mainly including network infrastructure such as 5G networks, fiber-optic broadband networks, backbone networks, international communication networks, satellite Internet, computing power infrastructure such as data centers, general computing power centers, intelligent computing centers, supercomputing centers, and new technology facilities such as artificial intelligence, blockchain, and quantum computing.

[0054] The "computing power" mentioned in the present invention includes: general computing power, intelligent computing power and super computing power.

[0055] The "general computing power" mentioned in the present invention refers to the computing power provided by servers based on CPU (Central Processing Unit) chips, which is used to support basic general computing such as cloud computing and edge computing.

[0056] The "intelligent computing power" mentioned in this invention refers to: a computing platform based on specialized chips such as GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), and ASIC (Application Specific Integrated Circuit) for various innovative artificial intelligence applications, such as natural language processing and machine vision.

[0057] The "supercomputing power" mentioned in the present invention refers to 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 uses a dedicated operating system to handle extremely complex or data-intensive problems. It is mainly used for calculations in cutting-edge scientific fields, such as planetary simulation, drug molecule design, genetic analysis, etc.

[0058] The term "intelligent computing center" as used in this document refers to a facility that utilizes large-scale heterogeneous computing resources, including general-purpose computing power (CPUs) and intelligent computing power (GPUs, FPGAs, ASICs, etc.), primarily to provide the computing power, data, and algorithms required for AI applications (such as AI deep learning model development, model training, and model inference). An intelligent computing center encompasses facilities, hardware, and software, providing a full stack of capabilities, from bottom-level computing power to top-level application enablement.

[0059] The "intelligent computing center cloud platform" mentioned in the present invention refers to: a cloud computing platform that provides comprehensive services based on the hardware resources and software resources of the intelligent computing center.

[0060] The "intelligent computing center" mentioned in the present invention includes but is not limited to the "intelligent computing center".

[0061] The "intelligent computing center" mentioned in the present invention is an artificial intelligence computing center, which is a type of computing power infrastructure based on artificial intelligence theory, adopts artificial intelligence computing architecture, and provides computing power services, data services and algorithm services required for artificial intelligence applications.

[0062] The "computing power center" mentioned in the present invention refers to: a facility that is mainly composed of infrastructure such as wind, fire, water, electricity, and IT hardware and software equipment, and has computing power, transportation capacity, and storage capacity, including general data centers, intelligent computing centers, supercomputing centers, etc.

[0063] The "supercomputing center" mentioned in the present invention refers to: a supercomputing data center, which is a data center based on a supercomputer or a large-scale computing cluster, which can provide large-scale computing, storage and network services and other functions, and is widely used in application scenarios such as aerospace, national defense, oil exploration, climate modeling and genome sequencing.

[0064] The "computing resources" mentioned in the present invention refer to: technologies and facilities with information calculation, transmission, storage and application capabilities required for the development of a digital society, including but not limited to computing resources such as CPUs and GPUs, 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 supporting and guarantee resources such as wind, fire, water, and electricity.

[0065] The "large language model" mentioned in this invention refers to a large language model (LLM), which is a language model with a large parameter scale. It is designed to understand and generate human language. It is trained with large amounts of text data and can perform a wide range of tasks including text summarization, translation, sentiment analysis, etc.

[0066] The "agent" described in this article refers to an agent capable of perceiving its environment and taking actions to achieve specific goals. It is a crucial component of the intelligent computing center cloud platform. It can be software, hardware, or a system, possessing autonomy, adaptability, and interactivity. An agent perceives changes in its environment (e.g., through sensors or data input), makes judgments and decisions based on learned knowledge and algorithms, and then executes actions to influence the environment or achieve predetermined goals. Agents are widely used in the field of artificial intelligence, often found in automated systems, robots, virtual assistants, and game characters. Their core capability lies in their ability to autonomously learn and continuously evolve to better complete tasks and adapt to complex environments.

[0067] The "self-evolution of the intelligent agent" mentioned in the present invention means that the intelligent agent can develop the required functional modules for itself. Self-evolution can also be called self-upgrade, self-expansion or self-optimization.

[0068] The "reflection system" mentioned in the present invention refers to the ability of the intelligent agent to review, evaluate, correct and summarize its own behavioral feedback, thinking or decision-making, aiming to discover and promptly correct errors, summarize experience for future application, and thus continuously improve the ability to handle tasks.

[0069] The "self-evolving intelligent agent" mentioned in the present invention refers to an intelligent agent embedded with a development intelligent agent, which can use the embedded development intelligent agent to develop the functional modules required for itself and has the ability of self-evolution (or self-upgrade or self-expansion or self-optimization).

[0070] The "development agent" mentioned in the present invention refers to the integration of multiple manufacturing agents with different capabilities, modularization and standardization of these capabilities, to form a reusable and embeddable agent. The development agent can be embedded in any target agent, so that the target agent has the ability to self-evolve.

[0071] The "target agent" mentioned in the present invention refers to an agent that is not embedded in a development agent and is used to complete a specified task.

[0072] The "embedded agent" mentioned in the present invention refers to a development agent embedded in a target agent, and the embedded agent and the target agent constitute a self-evolving agent.

[0073] The "memory system" mentioned in the present invention refers to a functional module in the intelligent agent, which is mainly used to perform the following functions: realize the memory function of the intelligent agent, including recording the following information: historical dialogue information, decision-making process information, execution feedback information, generated design documents or code information, as well as existing materials and learned knowledge; when the intelligent agent performs a task, it extracts relevant memory or knowledge to assist decision-making and complete the task.

[0074] The "tool function" described in the present invention can also be called a "tool method", which is an executable function method called by the intelligent agent to complete a specific task; it is an executable function method selected and called by the intelligent agent through intelligent decision-making based on the task context; it aims to use the large language model to intelligently construct parameters and dynamically call them to complete the task sub-goals.

[0075] The "thinking framework" described in this article refers to the logical structure and methodology that supports intelligent agents in task processing, reasoning, decision-making, and interaction. It provides AI agents with a systematic way of thinking, enabling them to efficiently complete complex tasks. Large language model (LLM)-based agents typically require a clear thinking framework to integrate model capabilities, external tools, and data sources to achieve specific functions.

[0076] The "demand module" described in the present invention can also be called a demand confirmation module or a demand confirmation function. The intelligent agent needs to interact with the user through the demand module to determine the user's task requirements, and then perform the task based on the determined task requirements.

[0077] The "large language model-based approach" described in this invention involves organizing input information and designing precise prompts, then invoking the large language model to generate a variety of content forms, including reflection, response, decision-making, tool selection, code, and documentation. This approach supports intelligent agents in tasks such as reasoning, design, dialogue with users, solution generation, and code generation. Combined with a code framework, it enables the execution of instructions such as tool method invocation and code program execution, enabling the completion of specific tasks such as document review, data search, file processing, and script execution.

[0078] To solve the problem of high cost and low efficiency of the existing artificial development of intelligent agent function modules, please refer to Figure 1 The present invention provides a method for realizing self-evolution of intelligent agents through computing power on an intelligent computing center cloud platform, the method comprising:

[0079] Step S1: An embedded agent embedded in a self-evolving agent determines that a target functional module needs to be developed for the self-evolving agent, and determines the development requirements of the target functional module. The self-evolving agent is the target agent in which a development agent is embedded, and the embedded development agent is the embedded agent.

[0080] The self-evolving agent in the present invention can be an agent used to complete a specified task, such as a mathematical processing agent, a data analysis agent, etc.

[0081] The embedded intelligent agent in the present invention has the ability to develop functional modules for the self-evolving intelligent agent by combining the large language model method.

[0082] The embedded intelligent agent in the present invention can be a single intelligent agent or a plurality of intelligent agents, wherein the plurality of intelligent agents form a multi-agent collaborative system, each of the intelligent agents has specific capabilities, and the plurality of intelligent agents can collaborate to complete the task of developing functional modules for the self-evolving intelligent agent.

[0083] In the present invention, the development demand may be to develop a new functional module for the self-evolving intelligent body, or to upgrade the existing functional modules of the self-evolving intelligent body.

[0084] In the present invention, optionally, the target function module includes at least one of the following: a reflection system, a memory system, a tool function, a thinking framework, a demand module and a learning system.

[0085] In the present invention, optionally, different functional modules have different corresponding development requirements, and the common development requirements in different functional modules may include at least one of the following: test cases, acceptance criteria, output requirements, technical requirements, performance requirements, security and privacy, etc.

[0086] The following is an explanation of each of the above development requirements.

[0087] A test case is an example used to test whether the target functional module meets the acceptance criteria.

[0088] Acceptance criteria are the basis for evaluating whether the target functional module meets development requirements.

[0089] Output requirements refer to the requirements for the content or format of the output of the target functional module to facilitate understanding and analysis.

[0090] Technical requirements refer to the specific technical implementation requirements for the target functional module, such as algorithm support, computing power requirements, and compatible development environments. For example, whether the target functional module needs to support multimodal data processing (such as text, images, and audio) and whether it needs to integrate with existing systems. Clarifying technical requirements helps ensure the feasibility and applicability of the target functional module.

[0091] Performance requirements refer to specific requirements for the performance of the target functional module, including inference speed, accuracy, resource consumption or scalability.

[0092] Security and privacy refer to the target module's ability to protect data security and privacy during use. For example, whether the target module can prevent sensitive data leaks, support data encryption, and comply with relevant laws and regulations (such as GDPR and CCPA) must be considered. Security and privacy protection are crucial requirements that cannot be ignored in practical applications of the framework, especially in scenarios involving user data or sensitive information.

[0093] When the target functional module is a reflection system, the development requirements may further include reflection capability requirements of the reflection system, where the reflection capability requirements of the reflection system refer to whether the reflection system has online reflection and / or offline reflection capabilities.

[0094] The "online reflection" mentioned in the present invention refers to the real-time reflection, correction and experience summary of the intelligent agent in the process of performing user tasks to improve the accuracy and quality of the completion of this task.

[0095] "Offline reflection" in this context means regularly reflecting on and summarizing past task execution processes through historical logs, and saving these lessons learned to a knowledge base or model parameters. The lessons learned from offline reflection are then applied to future tasks to improve the accuracy and quality of subsequent tasks.

[0096] When the target functional module is a reflection system, the acceptance criteria may refer to whether the execution process of the self-evolving intelligent agent includes a reflection process when running the test case. If the execution process includes the reflection process, it indicates that the reflection system development is successful; if the execution process does not include the reflection process, it indicates that the reflection system development has failed.

[0097] When the target functional module is a memory system, the development requirements may further include at least one of the following: memory mechanism, memory content, memory format, and extraction timing.

[0098] The memory mechanism includes at least one of the following: memory access, multi-round memory, long-term and short-term memory, working memory, memory compression, retrieval-augmented generation (RAG), cross-modal memory, knowledge graph memory, associative memory, parallel / serial memory retrieval, distributed memory access, memory cleaning and archiving, mixed-mode memory, and reasoning retrieval.

[0099] Memory content can include at least one of the following: historical user interaction records, contextual information, relevant information from external knowledge bases, dynamically generated intermediate results, user-provided data (such as files, images, and audio), and system-generated reasoning chains. This memory content can be customized based on specific application scenarios to meet different functional requirements.

[0100] The memory format may include storage in at least one of the following forms: structured data, text fragments, image features, knowledge graph nodes, embedded vectors, etc., to facilitate efficient retrieval and use.

[0101] The retrieval timing, that is, the timing of memory retrieval, can include at least one of the following: retrieval during intent recognition, retrieval during the execution of a specific task (such as when writing code), retrieval during the thought chain reasoning process (such as when making action decisions), retrieval during context switching (such as when the user switches sessions), and retrieval during initialization or status update (such as when the user starts a session).

[0102] When the target functional module is a thinking framework, the development requirements may further include at least one of the following: a thinking pattern, a visual explanation.

[0103] The thinking mode may include at least one of the following: ReAct (Reason + Act), Chain-of-Thought (CoT), Tree-of-Thought (ToT), Graph-of-Thought (GoT), Plan-and-Resolve, Reflexion (Reflection Framework), Cycle-of-Thought (CoC), Diverge-Converge Thought (DCT), Analogical Thought (AT), Multi-Perspective Thought (MPT), Program of Thoughts (PoT), Stepwise Reflection, Multi-Agent Debate, Attention-Guided Thinking, Intuitive Leap, Check List, Multi-Modal Thinking, Parallel Thinking, Fusion Framework, and Adaptive Framework.

[0104] The term "visual explanation" refers to the fact that users may need a visual description of the working principles of a thinking model to more intuitively understand its operation. For example, chain thinking can be compared to "building dominoes," tree thinking to "exploring branches in a decision tree," and loop thinking to "a process of repeated trial and error and optimization." Visual explanations help lower the barrier to understanding and enable users to quickly grasp the core logic and application methods of the framework.

[0105] When the target functional module is a requirement module, the development requirement may further include at least one of the following: interactive function requirement, position requirement in the workflow of the self-evolving intelligent agent, requirement confirmation point requirement, etc.

[0106] The interactive function requirement refers to whether the self-evolving intelligent agent supports multi-round dialogue functions when receiving user questions to confirm the needs of specific questions.

[0107] The position requirement in the workflow of the self-evolving intelligent agent refers to the requirement to confirm the position in the entire workflow of the self-evolving intelligent agent, for example, between "receiving data analysis instructions" and "generating data analysis code".

[0108] Demand confirmation point requirements refer to which demand confirmation points the demand module of the self-evolving intelligent agent needs to confirm with the user. For example, the demand confirmation points may include: core goals, data confirmation, analysis dimensions, chart types, delivery forms, etc.

[0109] When the target functional module is a learning system, the development requirements may also include at least one of the following: the input method of learning materials, whether it is necessary to prompt users to provide learning points during learning, whether users are required to provide knowledge tags during learning, the learning ability required by the learning system, the learning depth, whether it is necessary to provide the ability to visualize the learned knowledge, the application links of the learning system in the self-evolving intelligent body, etc.

[0110] The input method of learning materials refers to the way in which the materials to be learned are provided to the learning system, for example, in the form of files, or through dialogue in an interactive interface.

[0111] Is it necessary to prompt the user to provide learning focus during learning? For example, when providing the code to be learned to the learning system, the learning system can be prompted to focus on which functions in the code to learn.

[0112] Learning capabilities may include at least one of the following: knowledge learning, knowledge archiving, knowledge extraction, knowledge matching, knowledge application, knowledge practice, knowledge display, knowledge encoding, knowledge maintenance, knowledge version maintenance, multi-document knowledge learning, large document knowledge learning, and parallel learning.

[0113] Learning depth: refers to the ability to learn at different depths, which can include at least one of the following: simple archiving, summary summary, concept summary, methodological summary, and learning by analogy.

[0114] The application link of the learning system in the intelligent agent to be developed refers to the position of the learning system in the entire workflow of the self-evolving intelligent agent.

[0115] Step S2: The embedded agent determines the design scheme of the target functional module according to the development requirements and in combination with the large language model method;

[0116] In the present invention, the embedded intelligent agent determines the design scheme of the target functional module based on the development requirements and the large language model method. Specifically, it can be: the embedded intelligent agent constructs prompt words and calls the large language model according to the development requirements to complete the task of generating the design scheme of the target functional module.

[0117] In the present invention, the embedded intelligent agent determines the design scheme of the target functional module based on the development requirements and combined with the large language model method, which may specifically include: the embedded intelligent agent analyzes the development requirements, determines the design considerations and design implementation strategies, and determines the functional architecture of the target functional module and functional modifications to the self-evolving intelligent agent based on the implementation strategies.

[0118] When the target functional module is a reflective system, the design solution may include at least one of the following:

[0119] Adding a reflection experience storage and extraction module, the reflection experience storage and extraction module is used to store the experience obtained through reflection; and extract the stored experience during online reflection;

[0120] Adding an online reflection link to the thinking chain of the self-evolving intelligent agent, so as to reflect on the answers given by the self-evolving intelligent agent during the dialogue with the user;

[0121] Add an offline reflection module, which is used to reflect on the execution process of the historical tasks of the self-evolving intelligent agent to gain experience;

[0122] Added a save module for historical conversations;

[0123] An experience confirmation module is added, which is used to manually confirm the experience obtained through reflection.

[0124] Among them, the experience stored in the reflection experience storage and extraction module can include the experience obtained during online reflection (in this case, the self-evolving intelligent agent needs to have online reflection capabilities), and can also include the experience obtained during offline reflection (in this case, the self-evolving intelligent agent needs to have offline reflection capabilities).

[0125] Among them, adding an offline reflection module means being able to reflect on the execution process of the historical tasks of the self-evolving intelligent agent offline.

[0126] The purpose of adding a save module for saving historical conversations is to enable offline reflection and historical conversation query. In the present invention, the learning history can also be recorded and saved in a database to avoid repeated reflection on historical conversations.

[0127] The purpose of adding an experience verification module is to manually verify the experience gained through reflection. Only after manual confirmation can it be stored in the database to avoid storing inaccurate experience. If the experience is not verified, it can be reflected and summarized again until the experience summary is consistent with the benchmark or reaches the upper limit of the failure threshold.

[0128] In the present invention, optionally, the reflective experience storage and extraction module includes at least one of the following:

[0129] A RAG (Retrieval-augmented Generation) storage and retrieval module stores reflected questions, answers, and experiences gained from reflection in vector form in a vector database. Furthermore, when extracting experiences through online reflection, experience extraction is performed using vector similarity matching and large language model recognition.

[0130] A storage and retrieval module based on an experience model, wherein the experience model is obtained by fine-tuning the large language model of the self-evolving intelligent agent using the experience obtained through reflection; when extracting experience through online reflection, the self-evolving intelligent agent uses the experience model to extract experience and uses the large language model to identify the extracted experience.

[0131] The experience stored in the RAG storage and retrieval module can be either online or offline. When extracting experience through online reflection, vector similarity matching can be performed directly using the current task's questions or using the answers to the current task's questions.

[0132] In the present invention, large language model identification refers to using the large language model of the target intelligent agent to identify the extracted experience, and determining whether to use the experience after identification.

[0133] In the present invention, optionally, the reflection includes at least one of the following:

[0134] Search engine-based reflection refers to searching for answers to questions through search engines during reflection, and reflecting and gaining experience based on the searched answers; if standard questions and answers, or similar questions and answers can be searched, experience can be learned and summarized based on the searched questions and answers.

[0135] Reflection based on a large language model refers to the reflection performed by the large language model and the experience gained during reflection.

[0136] In the present invention, the above-mentioned reflection can be online reflection or offline reflection, that is, online reflection can be performed using a search engine-based reflection and / or a large language model-based reflection method, and offline reflection can also be performed using a search engine-based reflection and / or a large language model-based reflection method.

[0137] Since the reflection based on the large language model has no reference to the real answer, it is necessary to try to generate answers from different problem-solving methods and compare different problem-solving methods. In the present invention, optionally, the reflection based on the large language model includes: using the large language model to answer the problem from different angles, obtaining multiple answers, and comparing the multiple answers, determining the correct answer based on the comparison results, and reflecting based on the correct answer to gain experience. Among them, "comparing the multiple answers and determining the correct answer based on the comparison results" can be: obtaining an answer that appears more often as a preliminary benchmark answer, comparing the differences between other answers and the benchmark answer, including the problem-solving process and the answer, giving a difference analysis and possible error analysis, and checking the error analysis points on each answer to verify whether the answer is correct. The experience points finally obtained can also be verified by manual verification.

[0138] In some embodiments, the correct answer can also be obtained from the conversation context. For example, the user and the target intelligent agent have multiple rounds of questions and answers, and the correct answer information is found from the multiple rounds of questions and answers by finding the identifier given by the user. The correct answer is used as the standard answer to reflect on other rounds of questions and answers to gain experience.

[0139] When the target functional module is a memory system, the design solution may include at least one of the following: a design solution for the system architecture of the memory system, a design solution for each module of the memory system, etc. The system architecture design solution refers to the functional modules of the memory system and the data flow between the modules. The design solution for each module refers to the functions of each module and how the functions are implemented.

[0140] When the target functional module is a thinking framework, the design scheme may include at least one of the following: a design scheme for the system architecture of the thinking framework, a design scheme for each module of the thinking framework, and a nested logic design. The nested logic design is the nesting and organizational structure between the various thinking modes that the target intelligent agent needs to support. The design scheme for the system architecture refers to the functional modules included in the thinking framework and the data flow between the modules. The design scheme for each module refers to the functions of each module and the way to implement the functions, etc.

[0141] When the target functional module is a requirement module, the design scheme may include at least one of the following: design of the interactive function of the requirement module, design of the position of the requirement module in the workflow of the intelligent body to be developed, design of the requirement confirmation point, design of adding the requirement confirmation point to the prompt words of the work steps after the requirement confirmation, etc.

[0142] When the target functional module is a learning system, the design solution may include at least one of the following:

[0143] Modifying the task processing module of the self-evolving agent so that the task processing module can identify whether the user task is a learning task;

[0144] Modify the upload module of the self-evolving agent to support uploading and archiving of learning materials;

[0145] Added learning system;

[0146] Modifying the task processing module in the self-evolving agent so that the task processing module can call the knowledge extraction module of the learning system and integrate the knowledge learned by the learning system into the prompt words of the task processing module;

[0147] Added practice module.

[0148] Step S3: The embedded agent generates the code of the target functional module according to the development requirements and design plan in combination with the large language model method;

[0149] In the present invention, the embedded intelligent agent generates the code of the target functional module based on the development requirements and design plan in combination with the large language model method. Specifically, it can be: the embedded intelligent agent constructs prompt words and calls the large language model based on the development requirements and design plan to complete the task of generating the code of the target functional module.

[0150] Step S4: The embedded agent deploys the operating environment of the target functional module in combination with the large language model method, and debugs the code of the target functional module based on the operating environment; when the debugging result indicates that the target functional module does not meet the development requirements, the design scheme and code of the target functional module are checked in combination with the large language model method, and when the design scheme needs to be modified, the process returns to step S2; when the code needs to be modified, the process returns to step S3, until the debugging result indicates that the target functional module meets the development requirements or the number of debugging times exceeds a preset threshold;

[0151] In the present invention, the embedded intelligent agent deploys the operating environment of the target functional module in combination with the large language model method, and debugs the code of the target functional module based on the operating environment. Specifically, it can be: the embedded intelligent agent constructs prompt words and calls the large language model according to the development requirements, design plan and the code to complete the tasks of operating environment deployment and code debugging.

[0152] The embedded intelligent agent combines the large language model method to check the design plan and code of the target functional module. Specifically, the embedded intelligent agent constructs prompt words and calls the large language model based on the design plan, code and debugging results of the target functional module to complete the task of checking the design plan and code.

[0153] It should be noted that if the design plan needs to be modified and the process returns to step S2, in the new step S2, the embedded intelligent agent needs to redefine the design plan of the target functional module based on the development requirements, the design plan and the debugging results, combined with the large language model method.

[0154] It should be noted that if the code needs to be modified and the process returns to step S3, in the new step S3, the embedded agent regenerates the code of the target functional module based on the development requirements, code and debugging results in combination with the large language model method.

[0155] Step S5: the embedded agent feeds back the development result information of the target functional module to the user through the interactive interface of the self-evolving agent;

[0156] In the present invention, the feedback development result information may include at least one of the following: code information of the target functional module, debugging information of the target functional module, applicable scenario analysis of the target functional module, optimization suggestions for the target functional module, and conclusion on the success or failure of the development of the target functional module.

[0157] The conclusion of development failure may be a conclusion after multiple debugging and the number of debugging times exceeds a preset threshold.

[0158] The code information of the target functional module may be the code itself or the storage location of the code.

[0159] The debugging information may include at least one of the following: operation logs, intermediate data, debugging results, etc.

[0160] Step S6: The embedded agent brings the target functional module online.

[0161] In the present invention, a self-evolving intelligent agent with self-evolving ability is formed by embedding a development intelligent agent in the target intelligent agent. The development intelligent agent embedded in the self-evolving intelligent agent can confirm whether the target functional module is developed for the self-evolving intelligent agent, and determine the development requirements of the target functional module, and automatically complete the development, debugging and online launch of the target functional module, replacing the process of manually developing the target functional module, realizing the self-evolution of the intelligent agent, thereby effectively saving development costs, and the intelligent computing center cloud platform can provide sufficient computing power resources for the self-evolution of the intelligent agent, greatly improving the efficiency of the self-evolution of the development intelligent agent.

[0162] In the present invention, optionally, before step S1, the following steps may be further included:

[0163] Step S01: The development agent receives the task information embedded in the target agent given by the user through the interaction interface of the development agent;

[0164] In the present invention, the task information may include at least one of the following: the name of the target intelligent agent, functional description, current version, operating environment, interface documentation, existing functions (or capabilities), core code structure, etc.

[0165] The development agent in the present invention can be one agent or multiple agents. The multiple agents form a multi-agent collaborative system. Each of the agents has specific capabilities. The multiple agents can collaborate to complete the task of embedding the target agent and subsequently developing functional modules for the target agent.

[0166] In the present invention, optionally, the multiple agents in the development agent include at least one of the following manufacturing agents: integration agent, tool function development agent, memory system development agent, thinking framework development agent, demand confirmation development agent, learning system development agent, reflection system development agent, etc.

[0167] Each of the manufacturing agents can include at least one of the following basic agents: a demand confirmation agent, a framework design agent, a search agent, a learning agent, a coding agent, a debugging agent, an archiving agent, and a deployment agent. Alternatively, some or all of the multiple agents can share one or more of the basic agents.

[0168] The requirement confirmation agent is responsible for accurately understanding and clarifying development requirements and converting them into executable design documents;

[0169] The framework design agent is responsible for designing the overall architecture and sub-function modules of the target function module;

[0170] The search agent is responsible for finding relevant technical information and best practices to provide reference for design;

[0171] The learning agent summarizes the searched information and extracts key technologies and implementation methods;

[0172] The coding agent converts the design plan into code implementation;

[0173] The debugging agent is responsible for testing and fixing problems and recording error information during the debugging process;

[0174] The archiving agent archives and manages documents, codes and logs during the development process to ensure data traceability;

[0175] The deployment agent is responsible for deploying the developed agent to the target operating environment and monitoring the operating status.

[0176] The embedded agent is a replica of the above-mentioned development agent, and its structure is the same as that of the above-mentioned development agent, so it will not be described in detail.

[0177] Step S02: The development agent automatically identifies the relevant information of the target agent, and embeds itself into the target agent according to the relevant information of the target agent to form the self-evolving agent.

[0178] In the present invention, embedding the development agent into the target agent may include: the development agent copies its own code, and copies the copied code into the target agent.

[0179] In the present invention, embedding the development agent into the target agent may further include at least one of the following:

[0180] The dialogue interface of the target intelligent agent is modified, and an online dialogue monitoring module is added to the target intelligent agent. The online dialogue monitoring module is used to send the execution process of the current task of the target intelligent agent to the embedded intelligent agent at the same time. The embedded intelligent agent reflects on the execution process of the current task and determines the development requirements.

[0181] An offline dialogue monitoring module is added to the target agent. The offline dialogue monitoring module is used to reflect on the execution process of the target agent's historical tasks and determine development needs.

[0182] The "online" mentioned in the present invention refers to the process of the intelligent agent performing user tasks.

[0183] The "offline" mentioned in the present invention refers to when the agent is not performing a task.

[0184] The "execution process" in this context refers to the entire process from when an agent receives a user's command, interprets it, continuously decides on its next action, executes the corresponding action, generates a response, reflects on it, makes corrections, and finally responds to the user. This execution process can include the conversation between the user and the target agent, as well as the target agent's thought process.

[0185] In the present invention, the development agent automatically embeds itself into the target agent to form a self-evolving agent. There is no need to manually develop a development agent for the target agent, thereby effectively saving development costs.

[0186] In the present invention, optionally, step S02 includes:

[0187] Step S021: The development agent combines the large language model method to read the documents and codes of the target agent, and determines the relevant information of the target agent based on the reading results. The relevant information includes at least one of the following: architecture, interface, dependent model, extensible functional module, and functional module that the target agent already has.

[0188] In the present invention, optionally, step S1 includes:

[0189] Step S11: The embedded agent determines the need to develop a target functional module for the self-evolving agent and determines the development requirements of the target functional module based on the user's requirements for functional upgrade, expansion or optimization of the self-evolving agent; or, the embedded agent reflects on the current task or historical task of the self-evolving agent, determines the need to develop a target functional module for the self-evolving agent based on the reflection results, and determines the development requirements of the target functional module.

[0190] In the present invention, optionally, the embedded intelligent agent determines the need to develop a target functional module for the self-evolving intelligent agent based on the user's requirements for functional upgrade, expansion or optimization of the self-evolving intelligent agent, and determines the development requirements of the target functional module. Specifically, it can be: the embedded intelligent agent constructs prompt words and calls a large language model to generate replies or questions, etc., to complete the task of communicating with the user, and obtains development requirements based on the content of the dialogue.

[0191] In the present invention, the embedded intelligent agent can communicate with the user for at least one round based on the large language model. According to the content of the communication with the user, combined with the large language model method, it is determined that a target functional module needs to be developed for the self-evolving intelligent agent, and the development requirements of the target functional module are determined.

[0192] In the present invention, the embedded intelligent agent conducts at least one round of communication with the user to confirm the detailed development requirements of the target functional module, ensure a comprehensive and accurate understanding of the functions, performance, compatibility and other aspects of the functional module expected by the user, and lay a solid foundation for subsequent design and development work.

[0193] In the present invention, the embedded agent reflects on the current or past tasks of the self-evolving agent, determines based on the reflection results that a target functional module needs to be developed for the self-evolving agent, and determines the development requirements for the target functional module. Specifically, the embedded agent may identify and organize development requirements from the conversations about the current or past tasks of the self-evolving agent. Organizing the development requirements may include converting the identified development requirements into requirement descriptions in a desired format.

[0194] In the present invention, optionally, step S1 includes:

[0195] Step S12: The embedded intelligent agent compares the development requirements with the historical development requirements in the database, and determines whether it is necessary to continue developing the target functional module based on the comparison results, wherein the database records the historical development requirements and the status of the historical development requirements, and the status includes at least one of the following: collected, confirmed, developed, online, and abandoned.

[0196] In the present invention, optionally, determining whether to continue developing the target functional module according to the comparison result may be:

[0197] If the comparison result indicates that there is a historical development requirement that is the same as or similar to the development requirement in the database, and the status of the historical development requirement is developed, launched, or abandoned, there is no need to continue developing the target functional module.

[0198] If the comparison result indicates that there is no historical development requirement identical or similar to the development requirement in the database, the target functional module can be continued to be developed.

[0199] By comparing the development requirements with the historical development requirements in the database and removing duplicates, demand explosion or repeated development can be avoided.

[0200] In the present invention, optionally, step S1 may further include:

[0201] Step S13: The embedded agent provides a development requirement display interface for displaying the historical development requirements in the database and the status of the historical development requirements for users to view.

[0202] In the present invention, optionally, step S1 may further include:

[0203] Step S14: The embedded agent provides a confirmation module for the administrator to confirm whether to develop the target functional module, and the development requirements determined by the embedded agent to avoid uncontrollable modifications.

[0204] In the present invention, optionally, step S6 may include:

[0205] Step S61: The embedded agent uses a grayscale release method to launch the target functional module. The grayscale release method is to open the self-evolving agent with the target functional module to some users, and then officially launch it when there are no large-scale errors.

[0206] In the present invention, optionally, step S6 may include:

[0207] Step S61: The embedded intelligent body provides an administrator review module, and the administrator reviews the developed target functional module before launching it online.

[0208] The following examples illustrate how the embedded agent determines whether it is necessary to develop a target functional module for the self-evolving agent, how to determine the development requirements of the target functional module, and how to determine the design plan based on the development requirements.

[0209] First embodiment:

[0210] In this embodiment, the embedded agent of the question-answering robot (self-evolving agent) upgrades the thinking framework (target function module) of the question-answering robot.

[0211] The following content shows the historical conversations between users and the Q&A robot, and the process by which the embedded intelligent agent of the Q&A robot identifies development requirements in the historical conversations.

[0212] 1. Demand Identification

[0213] In this step, the embedded intelligent agent reflects on the historical tasks of the question-answering robot, determines whether it is necessary to develop a target functional module for the question-answering robot, and determines the development requirements of the target functional module.

[0214] <promp>

[0215] # The conversation history is as follows

[0216] **User**: The data analysis code you wrote often omitted the null value check for each field, resulting in low code availability. Can you perform a null value analysis on each field before data analysis in the future?

[0217] **Q&A Bot**: You raise an important question. Indeed, omitting null checks is a common problem in data analysis and can lead to less reliable code.

[0218] When I analyze data in the future, I will make sure to perform a null value analysis on each field, including:

[0219] Check the number and proportion of missing values ​​in each column;

[0220] Visualize the distribution of missing values;

[0221] Provide treatment suggestions based on the analysis results.

[0222] The following is a simple null value analysis template that you can refer to in future data analysis:

[0223] ```Python

[0224] # Null value analysis

[0225] # 1. Check the missing value overview

[0226] missing_values ​​= df.isnull().sum()

[0227] missing_percentage = (missing_values ​​ / len(df)) * 100

[0228] missing_df = pd.DataFrame({'missing value number': missing_values,

[0229] 'Missing value percentage (%)': missing_percentage})

[0230] missing_df = missing_df[missing_df['Number of missing values']>0].sort_values('Number of missing values', ascending=False)

[0231] # 2. Visualize missing values

[0232] # Visualization of missing value distribution for important fields

[0233] # 3. Suggestions

[0234] # Provide filling or deletion suggestions based on missing conditions

[0235] ```

[0236] **User**: You are answering the wrong question. I don’t want you to write code for me. I hope you can have such a processing mechanism in the future.

[0237] **Q&A Robot**: You're right, I misunderstood your intention. I understand that you want me to proactively include null value checks in future data analysis, rather than asking for code examples.

[0238] I will keep this in mind and make sure that when I do future data analysis:

[0239] Automatically perform null value analysis on each field;

[0240] Report missing values;

[0241] Provide reasonable treatment suggestions;

[0242] Thank you for your feedback, which will help me provide more reliable data analysis services.

[0243] # Your Task

[0244] You are a question-and-answer robot. The above is a conversation between you and the user. Did the user raise any requirements for you, or imply any requirements that need to be improved? Please provide them in XML format.

[0245] #Format parameters

[0246] <thought> ...

[0247] < / thought>

[0248] <result>

[0249] yes or no

[0250] < / result>

[0251] <content> ...

[0252] < / content>

[0253] in:

[0254] - thought: your thoughts

[0255] - result: can only be yes or no

[0256] - content: is your summary of the requirements. You only need to provide content when the result is yes. Otherwise, you do not need to provide content.

[0257] Please start.

[0258]

[0259] <response>

[0260] <thought>

[0261] Users have pointed out that Q&A robots often omit checks for null values ​​in their data analysis, reducing the usability of the code. They also hope that the robot will automatically incorporate null value checks in future data processing. This indicates that users have higher expectations for the reliability of the robot's data analysis. At the same time, users emphasize that they do not require separate code, but rather want such processing logic built into the robot. This is an implicit demand for improving the quality and practicality of data analysis services and ensuring that every step of analysis is based on reliable data.

[0262] < / thought>

[0263] <result> yes< / result>

[0264] <content>

[0265] The user's requirement is that the Q&A robot should automatically include a mechanism for checking for null values ​​in fields when providing data analysis related content. This includes:

[0266] 1. Automatically check the number and proportion of missing values ​​for each field.

[0267] 2. Visualize the distribution of null values ​​(if applicable).

[0268] 3. Based on the analysis results, provide reasonable suggestions for handling missing data.

[0269] This demand indicates that users expect robots to have more rigorous and practical professional capabilities when outputting data analysis to improve the usability of the code and the reliability of the analysis conclusions.

[0270] < / content>

[0271] < / response>

[0272] The following content is the process of organizing the development requirements identified by the embedded intelligent agent of the question-answering robot.

[0273] 2. Requirements organization

[0274] <prompt>

[0275] # The following information has been collected:

[0276] The user's requirement is that the Q&A robot should automatically include a mechanism for checking for null values ​​in fields when providing data analysis related content. This includes:

[0277] 1. Automatically check the number and proportion of missing values ​​for each field.

[0278] 2. Visualize the distribution of null values ​​(if applicable).

[0279] 3. Based on the analysis results, provide reasonable suggestions for handling missing data.

[0280] This demand indicates that users expect robots to have more rigorous and practical professional capabilities when outputting data analysis to improve the usability of the code and the reliability of the analysis conclusions.

[0281] # Your Task

[0282] Please convert it into a specific demand description. Please provide it in XML format.

[0283] #Format reference

[0284] <thought> ...< / thought>

[0285] <content> ...< / content>

[0286] in:

[0287] - thought: your thoughts

[0288] - Content: This is your summary of the requirements, which will be handed over to designers and developers for design. Please start.

[0289] < / prompt>

[0290] <response>

[0291] <thought> Users explicitly requested that data analysis content include a mechanism for checking and handling null values ​​in fields. This would include three specific aspects: checking the number and proportion of missing values, visualizing the distribution of null values, and providing action suggestions based on the analysis results. These requirements needed to be clearly and accurately translated into a format to ensure that designers and developers clearly understood the intended functionality.< / thought>

[0292] <content>When processing data analysis related content, the following requirements regarding the field null value check processing mechanism must be implemented:

[0293] Automatically performs a check on the number and proportion of missing values ​​for each field.

[0294] If applicable, visualize the distribution of null values.

[0295] Based on the analysis results of missing values, reasonable suggestions for handling missing data are given.< / content>

[0296] < / response>

[0297] The following content describes the process by which the embedded intelligent agent of the question-answering robot determines the design plan based on the sorted development requirements.

[0298] 3. Design

[0299] <prompt> Sample prompt words omitted here< / prompt>

[0300] <response>

[0301] <thought>

[0302] Demand Analysis:

[0303] The agent needs to automatically check for null values ​​in data analysis scenarios;

[0304] The existing CoT / ReAct reasoning process needs to be modified;

[0305] Need to integrate code execution capabilities;

[0306] The null check needs to be triggered at the start of the analysis.

[0307] Design considerations:

[0308] How to identify the beginning of a data analysis task;

[0309] How to seamlessly integrate null value checks without interrupting the analysis flow;

[0310] How to provide valuable null value handling suggestions;

[0311] How to embed this functionality into the CoT / ReAct framework.

[0312] Implementation strategy:

[0313] Design trigger mechanisms to identify data analysis scenarios;

[0314] Create standardized null value analysis template code;

[0315] Modify the reasoning process to insert null value checks at appropriate locations;

[0316] Ensure that analysis results are visualized and provide targeted recommendations.

[0317] < / thought>

[0318] <content>

[0319] 1. Null value check function architecture

[0320] Generate data analysis code request identification → whether data has been uploaded [Yes → Automatic null value check → Generate null value check result] → Continue to generate code.

[0321] 2. Function modification

[0322] Added Python script running to support running the null value analysis code and observing the results.

[0323] < / content>

[0324] <detail>

[0325] 1. Modify the CoT / ReAct and ReAct process framework code to support the expansion of tool method functions, and add Python method execution tool method, data analysis code request recognition tool method, automatic null value check tool method, and null value check result generation tool method.

[0326] 2. Modify the Prompt of CoT / ReAct to support the selection of the above tools and methods, and add the relevant prompt of "Generate data analysis code request recognition".

[0327] < / detail>

[0328] < / response>

[0329] Second embodiment:

[0330] In this embodiment, the embedded agent of the question-answering robot (self-evolving agent) adds a memory system (target function module) to the question-answering robot.

[0331] The following content describes the process of users identifying development requirements in historical conversations for the question-answering robot and the embedded intelligent agent of the question-answering robot.

[0332] 1. Demand Identification

[0333] In this step, the embedded intelligent agent reflects on the historical tasks of the question-answering robot, determines whether it is necessary to develop a target functional module for the question-answering robot, and determines the development requirements of the target functional module.

[0334] <prompt>

[0335] # The conversation history is as follows

[0336] **User**: Do you remember the math problem I mentioned 10 days ago?

[0337] **Q&A Bot**: I can't automatically remember the specific math question you asked me 10 days ago. If you need to review a previous math question, please describe the relevant information to me so that I can answer or recall the relevant content based on the clues you provide.

[0338] # Your Task

[0339] You are a question-and-answer robot. The above is a conversation between you and the user. Did the user raise any requirements for you, or imply any requirements that need to be improved? Please provide them in XML format.

[0340] # Format reference

[0341] <thought> ...

[0342] < / thought>

[0343] <result>

[0344] yes or no

[0345] < / result>

[0346] <content> ...

[0347] < / content>

[0348] in:

[0349] - thought: your thoughts;

[0350] - result: can only be yes or no;

[0351] - content: This is your summary of the requirements. You only need to provide content when the result is yes. Otherwise, you do not need to provide content.

[0352] Please start.

[0353] < / prompt>

[0354] <response>

[0355] <thought>The user asks whether he remembers the math problem he asked 10 days ago. Since he cannot automatically remember the specific math problem he asked 10 days ago, the user's request exceeds his ability.

[0356] < / thought>

[0357] <result> yes< / result>

[0358] <content>The user's requirement is to let the robot remember the math problems asked 10 days ago, but the robot cannot automatically remember the specific math problems asked 10 days ago, so this requirement is beyond the robot's capabilities.

[0359] < / content>

[0360] < / response>

[0361] The following content is the process of organizing the development requirements identified by the embedded intelligent agent of the question-answering robot.

[0362] 2. Requirements organization

[0363] <prompt>

[0364] # The following information has been collected:

[0365] The user's requirement is to let the robot remember the math problems asked 10 days ago, but the robot cannot automatically remember the specific math problems asked 10 days ago, so this requirement is beyond the robot's capabilities.

[0366] # Your Task

[0367] Please convert it into a specific demand description. Please provide it in XML format.

[0368] # Format reference

[0369] <thought> ...< / thought>

[0370] <content> ...< / content>

[0371] in:

[0372] - thought: your thoughts

[0373] - Content: This is your summary of the requirements, which will be handed over to designers and developers for design. Please start.

[0374] < / prompt>

[0375] <response>

[0376] <thought>

[0377] The core user requirement is for the robot to be able to retrieve and match information across 10 days, focusing on storing memories and actively retrieving them during subsequent interactions. The original requirement stated that "the robot cannot automatically remember questions from 10 days ago," indicating that the current system lacks this fundamental mechanism. Therefore, the requirement should be simplified, focusing on building a functional module that supports a memory period of at least 10 days, allowing the system to match and reference historical memories when answering questions.

[0378] It is important to emphasize that:

[0379] - Memory retention should be done automatically or semi-automatically (e.g. based on context);

[0380] - When answering new questions, the system should be able to retrieve and utilize relevant memories from up to 10 days in the past;

[0381] - This is an enhanced memory mechanism to make responses more relevant and personal.

[0382] Translating this into a requirement description that can be understood by design and development is as follows:

[0383] < / thought>

[0384] <content>

[0385] Requirement name: Memory storage and retrieval function

[0386] Functional goals:

[0387] Provide users with a memory period of at least 10 days, during which the system can extract and use the user's historical questions or conversation content when answering questions, and can recall questions asked in the past.

[0388] < / content>

[0389] < / response>

[0390] The following content describes the process by which the embedded intelligent agent of the question-answering robot determines the design plan based on the sorted development requirements.

[0391] 3. Design

[0392] <prompt> Sample prompt words omitted here< / prompt>

[0393] <response>

[0394] <thought>

[0395] Demand Analysis:

[0396] The agent needs to provide at least 10 days of user conversation memory;

[0397] Need to be able to extract and use historical conversation content when answering questions;

[0398] Need to be able to recall questions users have asked in the past;

[0399] Memory continuity needs to be maintained across different conversation sessions.

[0400] Design considerations:

[0401] How to efficiently store large amounts of conversation history;

[0402] How to quickly retrieve relevant historical information;

[0403] How to manage long-term memory within a limited context window;

[0404] How to distinguish important information from non-important information;

[0405] how to handle memory conflicts and updates;

[0406] How to preserve memories while ensuring privacy.

[0407] Implementation strategy:

[0408] Designing a hierarchical memory architecture to distinguish between short-term and long-term memory;

[0409] Implement vectorized storage to support semantic retrieval;

[0410] Create a memory importance scoring mechanism;

[0411] Design memory compression and summary generation algorithms;

[0412] Establish memory retrieval and activation mechanisms;

[0413] Implement memory periodic update and elimination strategies.

[0414] < / thought>

[0415] <content>

[0416] 1. Memory Storage and Retrieval Architecture

[0417] User conversation → memory processing → memory storage → memory retrieval → memory integration → response generation.

[0418] 2. Memory system component design

[0419] Memory storage layer:

[0420] Short-term memory: the complete conversation history of the current session;

[0421] Medium-term memory: Conversation summaries and key information from the last 7 days;

[0422] Long-term memory: 10-30 days of important information vector storage.

[0423] Memory processing mechanism:

[0424] Conversation summary generation: automatically extract key information points;

[0425] Entity recognition: identifying key entities such as names, places, and times;

[0426] Importance rating: based on factors such as user sentiment and repeated inquiries;

[0427] Memory Retrieval System:

[0428] Semantic similarity search: retrieve relevant history based on the current question;

[0429] Time decay mechanism: recent memories are given higher weight;

[0430] Contextual relevance analysis: Identifying thematic continuity across conversations.

[0431] 3. Memory Consolidation Process

[0432] User asks a question → analyzes the current question → retrieves relevant historical memory → evaluates memory relevance → integrates into the current context → generates a response.

[0433] 4. Memory Management Strategies

[0434] Memory compression: Regularly condense detailed conversations into summaries of key information;

[0435] Memory updating: updating when new information conflicts with existing memories;

[0436] Memory elimination: non-important memories that exceed the retention period are automatically eliminated;

[0437] Memory Activation: Automatically activate historical memory when relevant topics appear.

[0438] < / content>

[0439] <detail>

[0440] Memory Storage:

[0441] Vector database: uses vectors to store user historical memories and uses RAG to recall memories;

[0442] Graph database storage: uses an associative graph database, which contains timestamps, conversation content, summaries, entity labels, and importance scores;

[0443] Hierarchical storage structure: active memory layer and archive memory layer.

[0444] System Integration:

[0445] Modify the dialogue management module to perform memory retrieval before each user asks a question;

[0446] integrating retrieved historical memories during the context-building phase;

[0447] Update the memory store after the response is generated;

[0448] Implement regular memory maintenance tasks (compression, update, elimination);

[0449] Performance optimization:

[0450] Implement a memory cache mechanism to reduce retrieval latency;

[0451] Design batch memory updates to reduce computing resource consumption;

[0452] Create memory index to optimize retrieval efficiency;

[0453] < / detail>

[0454] < / response>

[0455] Third embodiment:

[0456] In this embodiment, the embedded agent of the question-answering robot (self-evolving agent) adds interactive functions (target function module) to the question-answering robot.

[0457] The following content shows the historical conversations between users and the Q&A robot, and the process by which the embedded intelligent agent of the Q&A robot identifies development requirements in the historical conversations.

[0458] 1. Demand Identification

[0459] In this step, the embedded intelligent agent reflects on the historical tasks of the question-answering robot, determines whether it is necessary to develop a target functional module for the question-answering robot, and determines the development requirements of the target functional module.

[0460] <prompt>

[0461] # The conversation history is as follows

[0462] **User**: Could you please provide a download link? I would like to download the document of each conversation in markdown format.

[0463] **Q&A Bot**: Currently, there's no official button or link to download every conversation as a document. However, there are several common workarounds to save the conversation as a document.

[0464] # Your Task

[0465] You are a question-and-answer robot. The above is a conversation between you and the user. Did the user raise any requirements for you, or imply any requirements that need to be improved? Please provide them in XML format.

[0466] # Format reference

[0467] <thought> ...

[0468] < / thought>

[0469] <result>

[0470] yes or no

[0471] < / result>

[0472] <content> ...

[0473] < / content>

[0474] in:

[0475] - thought: your thoughts;

[0476] - result: can only be yes or no;

[0477] - content: This is your summary of the requirements. You only need to provide content when the result is yes. Otherwise, you do not need to provide content.

[0478] Please start.

[0479] < / prompt>

[0480] <response>

[0481] <thought>

[0482] The user's need is to obtain a download link for each conversation as a markdown format document. Currently, the Q&A robot does not have this function, and the user's need cannot be directly met, which is beyond the current capabilities.

[0483] < / thought>

[0484] <result> yes< / result>

[0485] <content>

[0486] The user requested a download link for each conversation as a markdown document, but the Q&A robot currently does not provide an official button or link for this function, which cannot meet the user's needs.< / content>

[0487] < / response>

[0488] The following content is the process of organizing the development requirements identified by the embedded intelligent agent of the question-answering robot.

[0489] 2. Requirements organization

[0490] <prompt>

[0491] # The following information has been collected:

[0492] A user requested a link to download each conversation as a Markdown document. However, the Q&A bot currently doesn't offer an official button or link for this feature, which doesn't meet the user's needs. The above is a summary of user requests.

[0493] # Your Task

[0494] Please convert it into a specific demand description. Please provide it in XML format.

[0495] # Format reference

[0496] <thought> ...< / thought>

[0497] <content> ...< / content>

[0498] in:

[0499] - thought: your thoughts;

[0500] - Content: This is your summary of the requirements, which will be handed over to designers and developers for design. Please start.

[0501] < / prompt>

[0502] <response>

[0503] <thought>

[0504] A user requested a download link for each conversation document, presented in Markdown format. Since the Q&A bot doesn't have an official button or link to fulfill this requirement, we need to clarify this requirement and share it with designers and developers for feature design. This should clarify that the user's desired functionality is to obtain a download link, and that this functionality is currently unavailable due to the lack of a button or link.

[0505] < / thought>

[0506] <content>

[0507] Currently, users have a need to obtain a download link for a document saved in markdown format for each conversation. However, the existing functions of the Q&A robot do not have an official button or link that directly provides this download link. We now need to design and develop a function in the Q&A robot that allows users to obtain a download link for a document saved in markdown format for each conversation, so that users can download the document of each conversation presented in markdown format through the corresponding button or link.

[0508] < / content>

[0509] < / response>

[0510] The following content describes the process by which the embedded intelligent agent of the question-answering robot determines the design plan based on the sorted development requirements.

[0511] 3. Design

[0512] <prompt> Sample prompt words omitted here< / prompt>

[0513] <response>

[0514] <thought>

[0515] Demand Analysis:

[0516] Users need to obtain conversation documents saved in Markdown format;

[0517] The current system does not have an official function that directly provides a download link;

[0518] A solution to add download functionality needs to be designed and developed;

[0519] You need to ensure that the exported document format complies with the Markdown specification.

[0520] Design considerations:

[0521] How to integrate download functionality naturally into the existing UI;

[0522] How to correctly convert the conversation content into Markdown format;

[0523] How to handle special elements in the dialog (code blocks, images, tables, etc.);

[0524] How to generate and manage download links;

[0525] How to ensure the security of download functionality and user privacy;

[0526] How to handle the export requirements of long conversations.

[0527] Implementation strategy:

[0528] Design UI components to add obvious export buttons;

[0529] Create a converter for conversation content to Markdown;

[0530] Implement temporary file generation and link management mechanism;

[0531] Design file naming and organization structure;

[0532] Ensure the performance and reliability of export functionality;

[0533] Provides export options and customization capabilities.

[0534] < / thought>

[0535] <content>

[0536] 1.Markdown export function architecture

[0537] User requests export → Conversation content collection → Markdown conversion → File generation → Download link provided → User downloads.

[0538] 2. Functional component design

[0539] UI interaction components;

[0540] Add "Export to Markdown" button at the top / bottom of the dialogue interface;

[0541] Export option pop-up window (you can select export range and format options);

[0542] Download progress indicator;

[0543] Markdown conversion engine;

[0544] Conversation structure parser: distinguishes between user and AI messages;

[0545] Content format converter: processing rich text, code blocks, images, etc.;

[0546] Metadata Adder: Add timestamp, session ID and other information.

[0547] Document Management System:

[0548] Temporary file generator: create .md file;

[0549] Download Link Generator: Generate secure temporary download URLs;

[0550] File cleaning mechanism: Regularly clean up expired files.

[0551] 3. Markdown conversion rules

[0552] User message format: ## user (timestamp)\n\nUser message content;

[0553] AI reply format: ## Assistant (timestamp)\n\nAssistant reply content;

[0554] Code block processing: Use the language name and three backticks (```) to wrap the code content;

[0555] Image processing: convert to  format;

[0556] Table processing: convert to Markdown table syntax;

[0557] Mathematical formulas: Preserve LaTeX formatting, using $ and $$ wrappers.

[0558] 4. File Organization

[0559] File naming format: conversation_YYYYMMDD_HHMMSS.md.

[0560] The file header contains: session summary, date, and participant information.

[0561] Directory structure: Organized by conversation topic and date.

[0562] < / content>

[0563] <detail>

[0564] Front-end implementation:

[0565] Added export button component to the dialogue interface;

[0566] Implemented export option pop-up window and configuration interface;

[0567] Create download progress indicators and status feedback;

[0568] Implement client-side Markdown preview function.

[0569] Backend implementation:

[0570] Develop conversation content collection API endpoints;

[0571] Implement Markdown conversion service;

[0572] Create a temporary file storage and management system;

[0573] Develop a secure file download link generator;

[0574] Implement file expiration and cleanup mechanism.

[0575] Security and Privacy:

[0576] Implement access control and verification of download links;

[0577] Ensure secure storage of temporary files;

[0578] Added user privacy data desensitization option;

[0579] Implements logging of download operations.

[0580] Performance optimization:

[0581] Implement background asynchronous export processing;

[0582] Added export task queue management;

[0583] Optimize file generation and download speed.

[0584] < / detail>

[0585] Please refer to Figure 2 The present invention also provides a system 10 for realizing self-evolution of an intelligent agent through computing power on an intelligent computing center cloud platform, comprising a self-evolving intelligent agent 11, wherein the self-evolving intelligent agent 11 is a target intelligent agent embedded with a development intelligent agent, and the embedded development intelligent agent is an embedded intelligent agent 111;

[0586] The embedded agent 111 includes:

[0587] A requirement confirmation module 1111 is used to determine whether a target functional module needs to be developed for the self-evolving intelligent agent and to determine the development requirements of the target functional module;

[0588] A design module 1112 is configured to determine a design solution for the target functional module based on the development requirements and in combination with a large language model method;

[0589] Coding module 1113, used to generate the code of the target functional module according to the development requirements and design plan in combination with the large language model method;

[0590] The debugging module 1114 is used to deploy the operating environment of the target functional module in combination with the large language model method, and debug the code of the target functional module based on the operating environment; when the debugging result indicates that the target functional module does not meet the development requirements, the design scheme and code of the target functional module are checked in combination with the large language model method. When the design scheme needs to be modified, the design module 1112 is triggered to continue working, and when the code needs to be modified, the coding module 1113 is triggered to continue working, until the debugging result indicates that the target functional module meets the development requirements or the number of debugging times exceeds a preset threshold;

[0591] Feedback module 1115, used to provide feedback to the user on the development result information of the target functional module through the interactive interface of the self-evolving intelligent agent;

[0592] The online module 1116 is used to bring the target functional module online.

[0593] In the present invention, a self-evolving intelligent agent with self-evolving ability is formed by embedding a development intelligent agent in the target intelligent agent. The development intelligent agent embedded in the self-evolving intelligent agent can confirm whether the target functional module is developed for the self-evolving intelligent agent, and determine the development requirements of the target functional module, and automatically complete the development, debugging and online launch of the target functional module, replacing the process of manually developing the target functional module, realizing the self-evolution of the intelligent agent, thereby effectively saving development costs, and the intelligent computing center cloud platform can provide sufficient computing power resources for the self-evolution of the intelligent agent, greatly improving the efficiency of the self-evolution of the development intelligent agent.

[0594] Optionally, the system 10 for realizing self-evolution of intelligent agents through computing power on the intelligent computing center cloud platform further includes:

[0595] The development agent is used to receive task information embedded in the target agent given by the user through the interactive interface of the development agent; automatically identify the relevant information of the target agent, and embed itself into the target agent according to the relevant information of the target agent to form the self-evolving agent.

[0596] Optionally, the development agent is used to combine the large language model method to read the documents and codes of the target agent, and determine the relevant information of the target agent based on the reading results, and the relevant information includes at least one of the following: architecture, interface, dependent model, extensible functional module, and functional module that the target agent already has.

[0597] Optionally, the demand confirmation module 1111 is used to determine the need to develop a target functional module for the self-evolving intelligent agent and determine the development requirements of the target functional module based on the user's demand for functional upgrade, expansion or optimization of the self-evolving intelligent agent; or, the embedded intelligent agent reflects on the current task or historical task of the self-evolving intelligent agent, determines the need to develop a target functional module for the self-evolving intelligent agent based on the reflection results, and determines the development requirements of the target functional module.

[0598] Optionally, the demand confirmation module 1111 is used to compare the development requirements with the historical development requirements in the database, and determine whether it is necessary to continue developing the target functional module based on the comparison results, wherein the database records the historical development requirements and the status of the historical development requirements, and the status includes at least one of the following: collected, confirmed, developed, launched, and abandoned.

[0599] Optionally, the target function module includes at least one of the following: a reflection system, a memory system, a tool function, a thinking framework, a demand module and a learning system.

[0600] Please refer to Figure 3 The present invention also provides an electronic device 20, including a processor 21, a memory 22, and a computer program stored in the memory 22 and executable on the processor 21. When the computer program is executed by the processor 21, the various processes of the embodiment of the method for realizing the self-evolution of an intelligent agent through computing power on the cloud platform of the above-mentioned intelligent computing center are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described here.

[0601] The present invention also provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the various steps of the aforementioned method for implementing intelligent agent self-evolution through computing power on an intelligent computing center cloud platform, achieving the same technical effects. To avoid repetition, the description is omitted here. The computer-readable storage medium may be, for example, a read-only memory (ROM), random access memory (RAM), a magnetic disk, or an optical disk.

[0602] The present invention also provides a computer program product, comprising computer instructions, which, when executed by a processor, implement the above Figure 1 The intelligent computing center cloud platform shown implements each process of the method embodiment for realizing self-evolution of intelligent agents through computing power, and can achieve the same technical effect. To avoid repetition, it will not be described here.

[0603] It should be noted that, in this document, the terms "comprise," "include," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a..." does not preclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0604] Through the above description of the embodiments, those skilled in the art will clearly understand that the methods of the above embodiments can be implemented using software plus the necessary general-purpose hardware platform. Of course, hardware can also be used, but in many cases the former is the more preferred implementation method. Based on this understanding, the technical solution of the present invention, or the portion 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, or optical disk) and includes a number of instructions for enabling a terminal (such as a mobile phone, computer, server, air conditioner, or network device) to execute the methods described in the various embodiments of the present invention.

[0605] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are protected by the present invention.< / response> < / promp>

Claims

1. A method for realizing self-evolution of intelligent agents through computing power on an intelligent computing center cloud platform, characterized in that: include: Step S1: An embedded agent embedded in a self-evolving agent determines that a target functional module needs to be developed for the self-evolving agent, and determines the development requirements of the target functional module. The self-evolving agent is the target agent in which a development agent is embedded, and the embedded development agent is the embedded agent. Step S2: The embedded agent determines the design scheme of the target functional module according to the development requirements and in combination with the large language model method; Step S3: The embedded agent generates the code of the target functional module according to the development requirements and design plan in combination with the large language model method; Step S4: the embedded agent deploys the operating environment of the target functional module in combination with the large language model method, and debugs the code of the target functional module based on the operating environment; When the debugging result indicates that the target functional module does not meet the development requirements, the design scheme and code of the target functional module are checked in combination with the large language model method. If the design scheme needs to be modified, the process returns to step S2; if the code needs to be modified, the process returns to step S3 until the debugging result indicates that the target functional module meets the development requirements or the number of debugging times exceeds a preset threshold. Step S5: the embedded agent feeds back the development result information of the target functional module to the user through the interactive interface of the self-evolving agent; Step S6: the embedded agent brings the target functional module online; Wherein, before step S1, the method further includes: Step S01: The development agent receives the task information embedded in the target agent given by the user through the interaction interface of the development agent; Step S02: the development agent automatically identifies the relevant information of the target agent, and embeds itself into the target agent according to the relevant information of the target agent to form the self-evolving agent; Wherein, the step S02 includes: Step S021: The development agent uses a large language model method to read the documents and code of the target agent, and determines relevant information of the target agent based on the reading results, wherein the relevant information includes at least one of the following: architecture, interface, dependent model, extensible functional modules, and functional modules already possessed by the target agent; Wherein, the step S1 includes: Step S11: The embedded agent determines that a target functional module needs to be developed for the self-evolving agent based on a user's request for functional upgrade, expansion, or optimization of the self-evolving agent, and determines the development requirements of the target functional module; alternatively, the embedded agent reflects on the current or historical tasks of the self-evolving agent, and determines that a target functional module needs to be developed for the self-evolving agent based on the reflection results, and determines the development requirements of the target functional module; Step S12: The embedded agent compares the development requirements with historical development requirements in a database, and determines whether to continue developing the target functional module based on the comparison result, wherein the database records historical development requirements and the status of the historical development requirements, and the status includes at least one of the following: collected, confirmed, developed, launched, and abandoned; Step S13: The embedded agent provides a development requirement display interface for displaying historical development requirements in the database and the status of the historical development requirements; Step S14: The embedded agent provides a confirmation module for the administrator to confirm whether to develop the target functional module, as well as the development requirements determined by the embedded agent.

2. The method according to claim 1, characterized in that The target function module includes at least one of the following: a reflection system, a memory system, a tool function, a thinking framework, a demand module and a learning system.

3. A system for realizing self-evolution of intelligent agents through computing power on an intelligent computing center cloud platform, characterized in that: Including a self-evolving intelligent agent, the self-evolving intelligent agent is a target intelligent agent embedded with a development intelligent agent, and the embedded development intelligent agent is an embedded intelligent agent; The embedded agent includes: A requirement confirmation module is used to determine the need to develop a target functional module for the self-evolving intelligent agent and determine the development requirements of the target functional module; A design module is used to determine the design scheme of the target functional module based on the development requirements and in combination with the large language model method; An encoding module, configured to generate the code of the target functional module according to the development requirements and design plan in combination with a large language model method; a debugging module for deploying the operating environment of the target functional module in combination with the large language model method, and debugging the code of the target functional module based on the operating environment; when the debugging result indicates that the target functional module does not meet the development requirements, checking the design scheme and code of the target functional module in combination with the large language model method; when the design scheme needs to be modified, triggering the design module to continue working; when the code needs to be modified, triggering the encoding module to continue working, until the debugging result indicates that the target functional module meets the development requirements or the number of debugging times exceeds a preset threshold; A feedback module, configured to provide feedback to a user on the development result of the target functional module via an interactive interface of the self-evolving intelligent agent; An online module, used to bring the target functional module online; The system for realizing the self-evolution of intelligent agents through computing power on the intelligent computing center cloud platform also includes: A development agent is configured to receive task information provided by a user and embedded in the target agent through the interaction interface of the development agent; automatically identify relevant information of the target agent, and embed itself into the target agent based on the relevant information of the target agent to form the self-evolving agent; The development agent is configured to read the documents and code of the target agent in combination with the large language model method, and determine relevant information of the target agent based on the reading results, wherein the relevant information includes at least one of the following: architecture, interface, dependent model, extensible functional modules, and functional modules already possessed by the target agent; The demand confirmation module is further configured to determine the need to develop a target functional module for the self-evolving agent and determine the development requirements of the target functional module based on the user's demand for functional upgrade, expansion, or optimization of the self-evolving agent; or, the embedded agent reflects on the current or historical tasks of the self-evolving agent, and determines the need to develop a target functional module for the self-evolving agent based on the reflection results, and determines the development requirements of the target functional module; The demand confirmation module is further configured to compare the development demand with historical development demands in a database, and determine whether it is necessary to continue developing the target functional module based on the comparison result, wherein the database records historical development demands and the status of the historical development demands, and the status includes at least one of the following: collected, confirmed, developed, launched, and abandoned; The demand confirmation module is further configured to provide a development demand display interface for displaying historical development demands in the database and the status of the historical development demands; The demand confirmation module is also used to provide a confirmation module for the administrator to confirm whether to develop the target functional module and the development requirements determined by the embedded intelligent agent.

4. An electronic device, characterized in that: include: A processor, a memory, and a program stored in the memory and executable on the processor, wherein when the program is executed by the processor, the steps of the method for realizing self-evolution of intelligent agents through computing power on an intelligent computing center cloud platform as described in any one of claims 1 to 2 are implemented.

5. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for realizing self-evolution of intelligent agents through computing power by an intelligent computing center cloud platform as described in any one of claims 1 to 2.

6. A computer program product, characterized in that It includes computer instructions, which, when executed by a processor, implement the steps of the method for realizing self-evolution of intelligent agents through computing power by an intelligent computing center cloud platform as described in any one of claims 1 to 2.

Images