Method for realizing agent memory system development through computing power of intelligent computing center

Abstract

The invention provides a method for realizing agent memory system development through computing power of an intelligent computing center. The method comprises the following steps: S1, a memory system development agent receives task information provided by a user and used for constructing a target agent memory system; s2, determining a development demand of the memory system according to the task information in combination with a large language model method and a user; s3, determining a design scheme of the memory system according to the development requirements in combination with a large language model method; s4, generating codes of the memory system according to the development requirements and the design scheme in combination with a large language model method; s5, deploying a debugging environment and codes of a debugging memory system in combination with a large language model method; when the debugging result indicates that the memory system does not meet the debugging requirement, the design scheme and codes of the memory system are checked in combination with the large language model, the step S3 is executed if the design scheme needs to be modified, and the step S4 is executed if the codes need to be modified until the debugging requirement is met; and S6, feeding back result information to the user through the interactive interface.

Description

Technical Field

[0001] The present invention relates to the technical fields of intelligent computing centers, intelligent computing centers and computing power infrastructure, and particularly relates to a method for developing an intelligent agent memory system through the computing power of an intelligent computing center. Background Art

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

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

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

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

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

[0007] An "intelligent agent" is an agent that can perceive the environment and take actions to achieve specific goals. It can be software, hardware, or a system, and has autonomy, adaptability, and interaction capabilities. An intelligent agent perceives changes in the environment (such as through sensors or data input), makes judgments and decisions based on the knowledge and algorithms it has learned, and then executes actions to affect the environment or achieve a predetermined goal. Intelligent agents are widely used in the field of artificial intelligence, commonly found in automation systems, robots, virtual assistants, and game characters, and their core lies in the ability to learn independently and evolve continuously to better complete tasks and adapt to complex environments.

[0008] Currently, the cost of manually developing a complete memory system for an intelligent agent is extremely high, resulting in extremely limited investment in expanding and enriching the knowledge extraction methods of the intelligent agent and constructing complex memory patterns. This has led to obvious shortcomings in the breadth, depth, and flexibility of the intelligent agent's knowledge processing, restricting its performance improvement and application expansion. Summary of the Invention

[0009] The present invention provides a method for developing an intelligent agent memory system through the computing power of an intelligent computing center, which is used to solve the problems of high cost in manually developing an intelligent agent memory system, low investment in the development of the intelligent agent's memory system, and limited performance of the intelligent agent in knowledge processing.

[0010] To solve the above technical problems, the present invention is implemented as follows:

[0011] In a first aspect, the present invention provides a method for developing an intelligent agent memory system through the computing power of an intelligent computing center, including:

[0012] Step S1: The memory system development intelligent agent receives task information for constructing the memory system of the target intelligent agent provided by the user through the interaction interface;

[0013] Step S2: The memory system development intelligent agent determines the development requirements of the memory system in combination with the large language model method and the user according to the task information;

[0014] Step S3: The memory system development intelligent agent determines the design scheme of the memory system in combination with the large language model method according to the development requirements;

[0015] Step S4: The memory system development intelligent agent generates the code of the memory system in combination with the large language model method according to the development requirements and the design scheme;

[0016] Step S5: The memory system development intelligent agent deploys the debugging environment of the memory system and debugs the code of the memory system in combination with the large language model method; when the debugging result indicates that the memory system does not meet the debugging requirements, the memory system development intelligent agent checks the design scheme and code of the memory system in combination with the large language model method. If the design scheme needs to be modified, return to step S3. If the code needs to be modified, return to step S4 until the debugging result indicates that the memory system meets the debugging requirements;

[0017] Step S6: The memory system development intelligent agent feeds back result information to the user through the interaction interface.

[0018] Optionally, the development requirements of the memory system include at least one of the following: memory mechanism, memory content, memory format, extraction timing, performance requirements, security and privacy protection.

[0019] Optionally, step S2 includes:

[0020] Step S21: The memory system development agent communicates with the user at least once through the interaction interface based on the large language model according to the task information.

[0021] Step S22: Determine the development requirements of the memory system according to the task information and the communication content with the user, in combination with the large language model method.

[0022] Optionally, step S3 includes:

[0023] Step S31: The memory system development agent obtains learning results by using at least one of the following learning methods in combination with the large language model method according to the development requirements of the memory system: searching for relevant materials, viewing and learning open source code, learning the existing code and documents of the target agent, and consulting data. Determine the design scheme of the memory system according to the learning results.

[0024] Optionally, step S3 includes:

[0025] Step S32: If the memory system development agent finds that the design of the memory system does not meet the debugging requirements during the debugging process, adjust the design scheme of the memory system by using at least one of the following adjustment methods in combination with the large language model method:

[0026] The first adjustment method is to optimize the design scheme of the memory system;

[0027] The second adjustment method is to adopt other design schemes;

[0028] The third adjustment method is to integrate multiple memory mechanisms.

[0029] Optionally, step S4 includes:

[0030] Step S41: The memory system development agent generates the code of the memory system and the debugging log printing code according to the development requirements and design scheme of the memory system. The debugging log printing code is used to record the key information during the operation of the memory system. The key information includes at least one of the following: function call, variable value change, operation result, and exception situation;

[0031] Step S5 includes:

[0032] Step S51: The memory system development agent runs the code of the memory system and the debug log printing code in combination with the large language model method. According to the key information recorded by the debug log printing code, it checks whether the design scheme and code of the memory system are running properly and meet the development requirements in combination with the large language model method.

[0033] Optionally, step S5 includes at least one of the following sub-steps:

[0034] Step S52: When the number of debugging times reaches the first preset threshold, or the number of modifications to the design scheme reaches the second preset threshold, or the number of modifications to the code of the memory system reaches the third preset threshold, and the code of the memory system still cannot meet the debugging requirements, the memory system development agent terminates the development task of the memory system and feedbacks the termination reason and related situation to the user through the interaction interface;

[0035] Step S53: During the debugging process, the memory system development agent records the running log, intermediate data, and debugging results in combination with the large language model method and archives them for preservation;

[0036] Step S54: The memory system development agent deploys a debug environment with resetability or deploys multiple different versions of the debug environment in combination with the large language model method to meet the debugging requirements of different versions;

[0037] Step S55: The memory system development agent expands the test cases in combination with the large language model method.

[0038] In a second aspect, the present invention provides a device for realizing the development of an agent memory system through the computing power of an intelligent computing center, including:

[0039] A receiving module, configured to receive task information for constructing a target agent memory system provided by a user through an interaction interface;

[0040] A requirement determination module, configured to determine the development requirements of the memory system in combination with the large language model method and the user according to the task information;

[0041] A design module, configured to determine the design scheme of the memory system in combination with the large language model method according to the development requirements;

[0042] An encoding module, configured to generate the code of the memory system in combination with the large language model method according to the development requirements and the design scheme;

[0043] A debugging module, which is used to deploy the debugging environment of the memory system in combination with the large language model method and debug the code of the memory system; when the debugging result indicates that the memory system does not meet the debugging requirements, check the design scheme and code of the memory system in combination with the large language model method. If the design scheme needs to be modified, trigger the design module to continue working. If the code needs to be modified, trigger the coding module to continue working until the debugging result indicates that the memory system meets the debugging requirements;

[0044] A feedback module, which is used to feedback result information to the user through an interactive interface.

[0045] In a third aspect, the present invention provides an electronic device, including: a processor, a memory, and a program stored on the memory and executable on the processor. When the program is executed by the processor, it implements the steps of the method for developing an intelligent agent memory system through the computing power of an intelligent computing center as described in the first aspect above.

[0046] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the method for developing an intelligent agent memory system through the computing power of an intelligent computing center as described in the first aspect above.

[0047] In a fifth aspect, the present invention provides a computer program product, including computer instructions. When the computer instructions are executed by a processor, they implement the steps of the method for developing an intelligent agent memory system through the computing power of an intelligent computing center as described in the first aspect above.

[0048] In the present invention, through the memory system development intelligent agent running on the intelligent computing center, it is possible to communicate with the user and confirm clear and detailed development requirements, automatically complete the development of the memory system of the target intelligent agent, and replace the process of manually developing the code of the memory system, thereby effectively saving development costs. Moreover, the intelligent computing center can provide sufficient computing power resources, greatly improving the development efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] By reading the detailed description of the preferred embodiments below, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0050] Figure 1 It is a flowchart of the method for developing an intelligent agent memory system through the computing power of an intelligent computing center of the present invention;

[0051] Figure 2Schematic diagram of the structure of the device for developing an intelligent agent memory system through the computing power of an intelligent computing center according to the present invention;

[0052] Figure 3 Schematic diagram of the structure of the electronic device according to the present invention. Detailed implementation manners

[0053] Next, the technical solutions in the present invention will be clearly and completely described in conjunction with the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0054] First, the technical terms related to the present invention will be briefly described below.

[0055] The "computing power" referred to in the present invention means: the ability of a computer device or a computing / data center to process information, the ability of computer hardware and software to cooperate to jointly execute a certain computing requirement, the computing ability to process information data and output a target result, a new type of productive force integrating information computing power, network carrying capacity, and data storage capacity, and mainly providing services to society through computing power infrastructure.

[0056] The "computational power" (Computational Power, CP) referred to in the present invention means: the ability of a data center server to process data and output results, a comprehensive index for measuring the computing ability of a data center, including general computing ability, supercomputing ability, and intelligent computing ability. The commonly used measurement unit is the number of floating-point operations per second (FLOPS, 1EFLOPS = 10^18 FLOPS), and the larger the value, the stronger the comprehensive computing ability. It is estimated that 1EFLOPS is approximately the computing power output of 5 Tianhe 2A or 500,000 mainstream server CPUs or 2 million mainstream laptops. The calculation formula is: CP = CP_general + CP_intelligent + CP_super

[0057] The "carrying capacity" (Network Power, NP) referred to in the present invention means: the manifestation of the data transmission ability of computing power facilities, a comprehensive ability including network architecture, network bandwidth, transmission delay, intelligent management and scheduling, etc., involving network transmission inside and between data centers, and a comprehensive index for measuring network transmission scheduling ability.

[0058] The "Storage Power (SP)" described in the present invention refers to the comprehensive ability of a data center in four aspects: data storage capacity, performance, security and reliability, and green and low-carbon. It is a comprehensive indicator for measuring the data storage capacity of a data center, including external storage devices such as storage arrays and built-in storage devices of servers. The commonly used measurement unit for storage capacity is the exabyte (EB, 1EB = 2^60 bytes), the commonly used measurement unit for performance is the number of read and write operations per second per unit capacity (IOPS / TB, Input / Output Operations Per Second / TB), and the disaster recovery ratio is an important manifestation of security and reliability.

[0059] The "computing power infrastructure" described in the present invention refers to a new type of information infrastructure that integrates information computing power, network carrying capacity, and data storage power, and can realize the centralized computing, storage, transmission, and application of information.

[0060] The "new type of information infrastructure" described in the present invention mainly includes network infrastructures such as 5G networks, fiber broadband networks, backbone networks, international communication networks, and satellite Internet, computing power infrastructures such as data centers, general computing power centers, intelligent computing centers, and supercomputing centers, and new technology facilities such as artificial intelligence, blockchain, and quantum computing.

[0061] The "computing power" described in the present invention includes general computing power, intelligent computing power, and super computing power.

[0062] The "general computing power" described 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.

[0063] The "intelligent computing power" described in the present invention refers to a computing platform that is scaled for various artificial intelligence innovation applications based on dedicated chips such as GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), and ASIC (Application Specific Integrated Circuit), such as natural language processing and machine vision.

[0064] The "super computing power" described in the present invention mainly 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 processes extremely complex or data-intensive problems through a dedicated operating system. It is mainly used for computing in cutting-edge scientific fields, such as planetary simulation, drug molecule design, and gene analysis.

[0065] The "Intelligent Computing Center" described in the present invention refers to a facility that uses large-scale heterogeneous computing power resources, including general computing power (CPU) and intelligent computing power (GPU, FPGA, ASIC, etc.), and mainly provides the required computing power, data, and algorithms for artificial intelligence applications (such as scenarios like artificial intelligence deep learning model development, model training, and model inference). The intelligent computing center covers facilities, hardware, and software, and can provide full-stack capabilities from the underlying computing power to the top-level application enablement.

[0066] The "Intelligent Computing Center" described in the present invention includes, but is not limited to, the "Intelligent Computing Center".

[0067] The "Intelligent Computing Center" described in the present invention, namely the artificial intelligence computing center, is a type of computing power infrastructure based on artificial intelligence theory, adopting an artificial intelligence computing architecture, and providing computing power services, data services, and algorithm services required for artificial intelligence applications.

[0068] The "Computing Power Center" described in the present invention refers to a facility mainly composed of infrastructure such as wind, fire, water, and electricity and IT software and hardware devices, and having computing power, carrying capacity, and storage capacity, including general data centers, intelligent computing centers, supercomputing centers, etc.

[0069] The "Supercomputing Center" described in the present invention refers to the supercomputing data center, which is a data center based on supercomputers or large-scale computing clusters, and can provide functions such as large-scale computing, storage, and network services, and is widely used in application scenarios such as aerospace, national defense, oil exploration, climate modeling, and genome sequencing.

[0070] The "Computing Power Resources" described in the present invention refers to technologies and facilities with information computing, transmission, storage, and application capabilities required for the development of the digital society, including but not limited to computing resources such as CPU and GPU, network resources such as switches and routers, storage resources such as storage arrays and distributed storage, security resources such as firewalls and intrusion detection systems, and support and guarantee resources such as wind, fire, water, and electricity.

[0071] The "Large Language Model" described in the present invention refers to the large language model (LLM), which is a language model with a relatively large number of parameters, aiming to understand and generate human language, trained with a large amount of text data, and can perform a wide range of tasks including text summarization, translation, sentiment analysis, etc.

[0072] The "memory system" described in the present invention refers to a functional module in the target intelligent agent, which is mainly used to perform the following functions: realizing the memory function of the intelligent agent, including recording the following information: historical conversation information, decision-making process information, execution feedback information, generated design documents or code information, as well as existing materials and learned and summarized knowledge; when the target intelligent agent executes a task, extracting relevant memories or knowledge to assist in decision-making and complete the task.

[0073] The "method of combining large language models" described in the present invention refers to organizing input information and designing precise prompt words to call large language models to generate various content forms, including thinking, responses, decisions, tool selection, code, and documents, etc. This method supports the intelligent agent to perform tasks such as reasoning and thinking, design, conversation with users, generating solution designs, and code generation, and combines with the code framework to execute instructions such as tool method calls and code program executions, achieving the completion of specific tasks such as document access, information search, file processing, and script running.

[0074] To solve the problems of high cost in manually developing the memory system of an intelligent agent and low investment in the development of the memory system of the intelligent agent, resulting in limited performance of the intelligent agent in knowledge processing, please refer to Figure 1 ., the present invention provides a method for developing the memory system of an intelligent agent through the computing power of an intelligent computing center. This method can also be called a method for developing the memory system of an intelligent agent through the computing power of an intelligent computing center to achieve the development goal of the memory system of the target intelligent agent. This method includes:

[0075] Step S1: The memory system development intelligent agent receives the task information for constructing the memory system of the target intelligent agent provided by the user through the interaction interface;

[0076] The "memory system development intelligent agent" described in the present invention has the ability to develop the memory system for other intelligent agents by combining the method of large language models.

[0077] In the present invention, the memory system development intelligent agent can be an intelligent agent specifically used to develop the memory system of the target intelligent agent, or a development intelligent agent used to develop the entire target intelligent agent. The memory system development intelligent agent can be one intelligent agent or multiple intelligent agents. The multiple intelligent agents form a multi-intelligent agent collaboration system. Each intelligent agent has specific capabilities, and the multiple intelligent agents can collaborate to complete the task of developing the memory system for the target intelligent agent.

[0078] In the present invention, the target intelligent agent can be an intelligent agent used to complete a specified task, such as a data analysis intelligent agent, a large language model training (also known as fine-tuning the large language model) intelligent agent, etc.

[0079] In the present invention, the task information may include the development requirements and debugging requirements of the memory system of the target intelligent agent.

[0080] Step S2: The memory system development intelligent agent determines the development requirements of the memory system in combination with the large language model method and the user according to the task information;

[0081] Optionally, the development requirements may be to develop a new intelligent agent and develop a memory system for the new intelligent agent, or to develop a memory system for an existing intelligent agent or upgrade the memory system of an existing intelligent agent.

[0082] Optionally, the development requirements may include at least one of the following: memory mechanism, memory content, memory format, extraction timing, performance requirements, security and privacy protection.

[0083] Most existing intelligent agents operate relying on limited memory mechanisms, which greatly restricts their ability development. For example, some intelligent agents only support the Retrieval-Augmented Generation (RAG) technology or long short-term memory function, and do not implement a more complex and complete memory system. A relatively single knowledge extraction method often makes it difficult to fully explore and utilize the potential value of knowledge and cannot meet the requirements of intelligent agent knowledge application capabilities for complex tasks and diverse scenarios.

[0084] In the present invention, the memory system development intelligent agent can support the development of a memory system with more diverse memory mechanisms.

[0085] In the present invention, the memory mechanism includes at least one of the following: memory access, multi-round memory, long short-term memory, working memory, memory compression, Retrieval-Augmented Generation (RAG), cross-modal memory, knowledge graph memory, associative memory, parallel / serial memory extraction, distributed memory access, memory cleaning and archiving, hybrid mode memory, inference retrieval.

[0086] Among them, memory access: refers to the process in which the intelligent agent supports reading or writing of stored memory content to ensure that information can be effectively saved and retrieved.

[0087] Multi-round memory: refers to the ability of the intelligent agent to remember context information in multi-round conversations or interactions and use this information for more relevant responses or operations in subsequent rounds.

[0088] Long short-term memory: Short-term memory is used to store temporary information in the current task or conversation and is usually cleared after the task is completed. Long-term memory is used to store important and persistent information that can be retrieved and used across tasks or over a long time.

[0089] Working Memory: A mechanism for temporarily storing and processing information, which supports an agent to operate and reason about information in real time during the current task, similar to the short-term memory of humans when thinking.

[0090] Memory Compression: By extracting key information or patterns, the long and verbose memory content is compressed into a smaller storage form to save storage space and improve retrieval efficiency.

[0091] Retrieval-Augmented Generation (RAG): A mechanism that combines memory retrieval and generation capabilities. When an agent generates content, it dynamically retrieves information from relevant memories or knowledge bases to improve the accuracy and relevance of the generated content.

[0092] Cross-Modal Memory: Supports the storage and retrieval of multiple data modalities (such as text, images, audio, etc.), and can establish associations between different modalities and perform comprehensive analysis.

[0093] Knowledge Graph Memory: A memory mechanism based on a knowledge graph that stores information in the form of entities and relationships, supporting complex semantic queries and reasoning.

[0094] Parallel / Serial Memory Retrieval: Parallel memory retrieval refers to retrieving information simultaneously from multiple memory modules or stores to improve retrieval speed. Serial memory retrieval retrieves memory content step by step in sequence and is suitable for complex tasks that require context dependence.

[0095] Distributed Memory Access: Distributes memories across multiple storage nodes or modules, and realizes efficient access and management through distributed computing, suitable for large-scale memory systems.

[0096] Memory Cleaning and Archiving: Cleans up outdated or no-longer-needed memory content, and at the same time archives important but infrequently used memories to long-term storage to optimize storage resources.

[0097] Hybrid-Mode Memory: Hybrid-mode memory is a comprehensive memory system that combines multiple memory mechanisms (such as long short-term memory, working memory, knowledge graph memory, etc.). It can dynamically select and switch different memory modes according to task requirements to achieve efficient information storage, retrieval, and utilization.

[0098] Associative Memory: Associative memory is a mechanism for retrieving memories by triggering relevant information or concepts. It can automatically associate with related content according to the current input (such as keywords, context, or modal features) and extract it from memory, thus forming a more comprehensive understanding or reasoning.

[0099] Inference Retrieval: Inference retrieval is a type of information retrieval method based on logical reasoning. It first deeply analyzes and understands the input question. Based on the nature, goal, and domain knowledge involved in the question, it infers the key information and relevant content that needs to be extracted to answer the question. Then, using these inference results, it retrieves relevant memories in the stored memory system according to specific retrieval strategies.

[0100] The definitions of other contents in the development requirements are described below.

[0101] Memory content can include at least one of the following: the user's historical interaction records, context information, relevant information in the external knowledge base, dynamically generated intermediate results, data provided by the user (such as files, pictures, audio, etc.), the inference chains generated by the system, etc. These memory contents can be customized according to specific application scenarios to meet different functional requirements.

[0102] Memory format can include storage in at least one of the forms of structured data, text fragments, image features, knowledge graph nodes, embedding vectors, etc., for efficient retrieval and use.

[0103] The extraction timing, that is, the timing of extracting memories, can include at least one of the following: extracting during intent recognition, extracting during the execution of specific tasks (such as when writing code), extracting during the thought chain reasoning process (such as when making action decisions), extracting during context switching (such as when the user switches sessions), extracting during initialization or state update (such as when the user starts a session).

[0104] Performance requirements can include at least one of the system's response speed, memory access efficiency, concurrent processing ability, resource occupancy, stability, etc.

[0105] Security and privacy protection are important considerations in the design of the memory system, especially in scenarios involving the user's sensitive information or privacy data.

[0106] In the present invention, the memory system development agent determines the development requirements of the memory system in combination with the large language model method and the user according to the task information. Specifically, the memory system development agent constructs prompt words and invokes the large language model according to the task information to generate thoughts, responses, decisions, tool selections, code, etc., to support the memory system development agent to complete tasks such as reasoning and thinking, conversing with the user, and code generation. And in combination with the code framework, it realizes the execution of instructions such as tool method calls and code program executions, and realizes conversing with the user through single-round or multi-round conversations and determines the development requirements.

[0107] Determine the development requirements of the memory system according to the task information. The memory system development agent inputs the prompt into the large language model and obtains the development requirements output by the large language model.

[0108] Step S3: The memory system development agent determines the design solution of the memory system according to the development requirements and in combination with the large language model method;

[0109] The design solution of the present invention may include: the design solution of the system architecture of the memory system, the design solutions of each module of the memory system, the detailed design document, the interface design document, the database design document, etc.

[0110] The design solution of the system architecture refers to the functional modules included in the memory system and the data flow between the modules.

[0111] The design solution of each module refers to the functions of each module and the implementation methods of the functions, etc.

[0112] The detailed design document refers to the document that describes in detail the function implementation of each module in the memory system, including the logical flow of the module, algorithm design, data processing method, exception handling mechanism, etc.

[0113] The interface design document refers to the document that describes the interfaces for interaction between each module in the memory system or with external systems, including the function definition of the interface, input and output parameters, call method, data format, communication protocol, and error code description, etc.

[0114] The database design document refers to the document that designs the data storage structure involved in the memory system, including the table structure of the database, field definition, data type, primary key and foreign key relationship, index design, stored procedure, and trigger, etc.

[0115] In the present invention, the memory system development agent determines the design solution of the memory system according to the development requirements and in combination with the large language model method. Specifically, the memory system development agent constructs a prompt according to the development requirements and calls the large language model to generate the design solution of the memory system. The memory system development agent obtains the design solution output by the large language model by inputting the prompt into the large language model and supports the completion of the design task of the memory system.

[0116] Step S4: The memory system development agent generates the code of the memory system according to the development requirements and design solution and in combination with the large language model method;

[0117] In the present invention, the memory system development agent generates the code of the memory system according to the development requirements and design scheme, specifically: the memory system development agent constructs a prompt and invokes the large language model according to the development requirements and design scheme to generate the code of the memory system. The memory system development agent inputs the prompt into the large language model, obtains the code output by the large language model, and supports the completion of the code generation task of the memory system.

[0118] Step S5: The memory system development agent deploys the debugging environment of the memory system and debugs the code of the memory system in combination with the large language model method; when the debugging result indicates that the memory system does not meet the debugging requirements, the memory system development agent checks the design scheme and code of the memory system in combination with the large language model method. If the design scheme needs to be modified, return to step S3. If the code needs to be modified, return to step S4 until the debugging result indicates that the memory system meets the debugging requirements.

[0119] Step S6: The memory system development agent feeds back result information to the user through the interaction interface.

[0120] Among them, the fed-back result information may include at least one of the following: information about the code of the memory system, debugging information of the memory system, and a conclusion on whether the memory system is developed successfully or unsuccessfully;

[0121] Among them, the conclusion of unsuccessful development may be the conclusion after multiple debuggings and the number of debuggings exceeds the failure threshold number.

[0122] The information about the code of the memory system may be the code itself or the storage location of the code, etc.

[0123] The debugging information may include at least one of the following: running logs, intermediate data, debugging results, etc.

[0124] In the present invention, the memory system development agent deploys the debugging environment of the memory system and debugs the code of the memory system in combination with the large language model method, specifically: the memory system development agent constructs a prompt and invokes the large language model according to the debugging requirements to generate content such as thinking, reply, decision-making, tool selection, and code, and supports the memory system development agent to complete tasks such as reasoning and thinking, code generation, and tool selection. Combining with the code framework, the memory system development agent can implement instructions such as tool method calls and code program executions to complete specific tasks such as environment deployment, debugging and startup of the target agent, and running log collection, so as to support the debugging work of the memory system.

[0125] The memory system development agent combines with the large language model method to check the design scheme and code of the memory system. Specifically, the memory system development agent constructs prompt words based on the debugging results and invokes the large language model. The prompt words include the design scheme, code, and debugging results of the memory system, and are used to prompt the large language model to check the design scheme and code. The memory system development agent inputs the prompt words into the large language model and obtains the inspection results output by the large language model. The inspection results can indicate errors or defects in the design scheme or code, thus supporting the optimization and improvement of the memory system.

[0126] It should be noted that when it is necessary to modify the design scheme and return to step S3, in the new step S3, the memory system development agent re-determines the design scheme of the memory system according to the development requirements and the inspection results, in combination with the large language model method. Specifically, the memory system development agent constructs prompt words according to the development requirements and the inspection results and invokes the large language model to generate the design scheme of the memory system. The memory system development agent inputs the prompt words into the large language model and obtains the new design scheme output by the large language model.

[0127] It should be noted that when it is necessary to modify the code and return to step S4, in the new step S4, the memory system development agent regenerates the code of the memory system according to the development requirements, design scheme, and inspection results, in combination with the large language model method. Specifically, the memory system development agent constructs prompt words according to the development requirements, design scheme, and inspection results and invokes the large language model to generate the code of the memory system. The memory system development agent inputs the prompt words into the large language model, obtains the code output by the large language model, and supports the completion of the code generation task of the memory system.

[0128] In the present invention, the memory system development agent running in the intelligent computing center can communicate with the user to confirm clear and detailed development requirements, automatically complete the development of the memory system of the target agent, and replace the process of manually developing the code of the memory system, thereby effectively saving development costs. Moreover, the intelligent computing center can provide sufficient computing power resources, greatly improving the development efficiency.

[0129] In some embodiments, optionally, step S2 includes:

[0130] Step S21: The memory system development agent communicates with the user at least once through the interaction interface based on the task information and the large language model.

[0131] After the memory system development agent receives the task information sent by the user through the interaction interface, it can construct a prompt word according to the task information. The prompt word is used to prompt the large language model to analyze whether the development requirements and / or test requirements of the task information are complete, and whether it is necessary to confirm more detailed development requirements and / or test requirements with the user. The memory system development agent inputs the prompt word into the large language model and obtains the analysis result output by the large language model. The analysis result includes the content of the development requirements and / or test requirements to be determined. The memory system development agent displays the content of the development requirements and / or test requirements to be determined to the user based on the interaction interface.

[0132] When the user provides new development requirements and / or test requirements through the interaction interface, the memory system development agent can construct a prompt word according to the new development requirements and / or test requirements. The prompt word is used to prompt the large language model to analyze whether the development requirements and / or test requirements are complete, and whether it is necessary to confirm more detailed development requirements and / or test requirements with the user. The memory system development agent inputs the prompt word into the large language model and obtains the analysis result output by the large language model. The analysis result includes the content of the development requirements and / or test requirements to be determined. The memory system development agent displays the content of the development requirements and / or test requirements to be determined to the user based on the interaction interface.

[0133] The above process of determining development requirements and / or test requirements in combination with the large language model method can be carried out multiple times to obtain more detailed development requirements and / or test requirements.

[0134] Step S22: Determine the development requirements of the memory system according to the task information and the communication content with the user, in combination with the large language model method.

[0135] Specifically, the memory system development agent constructs a prompt word according to the task information and calls the large language model to generate thinking, replies, decisions, tool selections, code, etc., to support the memory system development agent to complete tasks such as reasoning and thinking, communicating with the user, and code generation. In combination with the code framework, the execution of instructions such as tool method calls and code program executions is realized, and the communication with the user is realized through single-round or multi-round conversations to determine the development requirements.

[0136] In the embodiment of the present invention, the development agent communicates with the user at least once to confirm the detailed development requirements and / or debugging requirements of the memory system, so as to ensure a comprehensive and accurate understanding of the functions, performance, compatibility, etc. of the memory system expected by the user, and lay a solid foundation for subsequent design and development work.

[0137] In the embodiment of the present invention, optionally, the step S3 includes:

[0138] Step S31: According to the development requirements of the memory system, the memory system development agent obtains learning results by using at least one of the following learning methods in combination with the large language model method: searching for relevant materials, viewing and learning open-source code, learning the existing code and documents of the target agent, and consulting data. According to the learning results, the design scheme of the memory system is determined.

[0139] The learning method is instructed by the large language model to the memory system development agent according to the development requirements, and the memory system development agent obtains the learning results.

[0140] Specifically, the memory system development agent can generate a prompt word according to the development requirements and the learning results. The prompt word is used to prompt the large language model to determine the design scheme of the memory system according to the development requirements and the learning results. The memory system development agent inputs the prompt word into the large language model and obtains the design scheme output by the large language model.

[0141] Through the above learning methods, different memory schemes can be deeply understood, and the learning results will be used as important references to provide diverse ideas and methods for the design of the memory system, and then a more demand-compliant and complete memory system can be designed.

[0142] In some embodiments, optionally, step S3 includes:

[0143] Step S32: If the memory system development agent finds during the debugging process that the design of the memory system does not meet the debugging requirements, at least one of the following adjustment methods is used to adjust the design scheme of the memory system in combination with the large language model method:

[0144] The first adjustment method is to optimize the design scheme of the memory system;

[0145] The second adjustment method is to adopt other design schemes;

[0146] The third adjustment method is to integrate multiple memory mechanisms.

[0147] The adjustment method can be instructed by the development agent to the large language model through a prompt word.

[0148] In some embodiments, optionally, step S4 includes:

[0149] Step S41: The memory system development agent generates the code of the memory system and the debugging log printing code according to the development requirements and design scheme of the memory system, in combination with the large language model method. The debugging log printing code is used to record the key information during the operation of the memory system, and the key information includes at least one of the following: function call, variable value change, operation result, and exception situation;

[0150] Step S5 includes:

[0151] Step S51: The memory system development agent runs the code of the memory system and the debugging log printing code in combination with the large language model method. According to the key information recorded by the debugging log printing code, in combination with the large language model method, it checks whether the design scheme and code of the memory system are running properly and meet the development requirements.

[0152] In the present invention, optionally, the memory system development agent can set different levels of operation logs (such as FATAL, ERROR, WARNING, INFO, DEBUG), and when finally delivered to the user, it can output the operation logs of the specified level (such as outputting the logs of FATAL, ERROR, WARNING levels). After the code debugging is completed, the memory system development agent can remove or disable the debugging log printing code.

[0153] In some embodiments, optionally, Step S5 includes at least one of the following sub-steps:

[0154] Step S52: When the number of debugging times reaches the first preset threshold, or the number of modifications to the design scheme reaches the second preset threshold, or the number of modifications to the code of the memory system reaches the third preset threshold, and the code of the memory system still cannot meet the debugging requirements, the memory system development agent terminates the development task of the memory system and feedbacks the termination reason and related situations to the user through the interaction interface;

[0155] Through this step, excessive waste of computing power resources can be avoided. Feedback the termination reason and related situations to the user so that the user can adjust the development strategy in a timely manner.

[0156] Step S53: During the debugging process, the memory system development agent records the operation logs, intermediate data, and debugging results in combination with the large language model method, and archives and saves them;

[0157] Especially, the error messages and related codes in the operation logs are archived key points, so as to analyze problems and summarize experience later, providing strong support for subsequent development and optimization.

[0158] Step S54: The memory system development agent combines the large language model method to deploy a resettable debugging environment or multiple different versions of the debugging environment to meet the debugging requirements of different versions.

[0159] A resettable debugging environment means that the data imported during debugging can be reset.

[0160] Through this step, it is possible to adapt to the debugging requirements and scenarios of different versions, and improve the efficiency and accuracy of the debugging work.

[0161] Step S55: The memory system development agent combines the large language model method to expand test cases.

[0162] In the present invention, the development agent has the ability to expand test cases by itself in combination with the large language model method. Combining user requirements and actual application scenarios, it generates diverse test cases, such as not only test cases for one scenario, but also test cases for the memory mechanism and test cases for extraction timing. The development agent can also classify test cases to clearly distinguish necessary test cases and optional test cases. Necessary test cases are those that must be executed to ensure the normal operation of the basic functions of the memory system and meet the core requirements, and these cases cover the key functions of the system. Optional test cases are for testing some non-core functions, special scenarios, or edge cases.

[0163] In the present invention, after step S5, it may further include:

[0164] Step S7: The memory system development agent combines the large language model method to evaluate the entire development process and the final memory system. Optionally, it can analyze and evaluate from multiple aspects such as requirement satisfaction, technical implementation, and performance indicators to determine whether the task of developing the memory system is completed.

[0165] As mentioned above, the memory system development agent may include multiple agents. In some embodiments, optionally, the multiple agents include at least one of the following agents: requirement confirmation agent, search agent, learning agent, memory system design agent, coding agent, debugging agent, archiving agent, deployment agent, evaluation agent, etc.

[0166] Among them, the requirement confirmation agent is responsible for in-depth interaction with users. By means of natural language processing and other technologies, it accurately understands the development requirements and / or debugging requirements proposed by users, asks questions and clarifies ambiguous or unclear development requirements and / or debugging requirements, transforms user requirements into clear and executable requirement documents, and communicates and confirms with other agents to ensure that all agents have the same understanding of the requirements. It should be noted that the requirement determination agent needs to combine the large language model method to understand the development requirements and / or debugging requirements proposed by users.

[0167] The search agent uses a search engine to search for materials related to the development project, such as technical materials, open-source code, industry cases, etc. It will screen and sort the information retrieved and extract valuable content to provide to other agents for reference to assist the smooth progress of the development work. It should be noted that the search agent needs to combine the large language model method to screen and sort the information retrieved and extract valuable content.

[0168] The learning agent can automatically learn new technical knowledge and development methods. By learning project documents, existing project source codes, and relevant technical materials retrieved, it organizes the knowledge useful for the current development memory system for use in the design and development of the memory system. It should be noted that the learning agent needs to combine the large language model method to learn new technical knowledge and development methods and organize the knowledge useful for the current development memory system.

[0169] The memory system design agent, based on the requirement document provided by the requirement confirmation agent, combines the existing project materials, existing project codes, relevant knowledge retrieved, and the relevant knowledge of the memory system learned, to conduct system architecture design, module design, etc. It should be noted that the memory system design agent needs to combine the large language model method to conduct system architecture design, module design, etc.

[0170] The coding agent writes code according to the design plan provided by the memory system design agent, using the corresponding programming language and development tools. It should be noted that the coding agent needs to combine the large language model method to write code using the corresponding programming language and development tools.

[0171] The debugging agent is responsible for debugging the code written by the coding agent. By running test cases and analyzing error logs, etc., it searches for and locates errors and defects in the code. It should be noted that the debugging agent needs to combine the large language model method to debug the code written by the coding agent.

[0172] The archiving agent, after the project development is completed, sorts and archives all relevant documents and materials of the project, including requirement documents, design plans, code, test reports, etc. It stores them according to certain specifications and standards, facilitating subsequent queries and reuse, and also providing an important basis for project maintenance and upgrade. It should be noted that the archiving agent needs to combine large language model methods to sort and archive all relevant documents and materials of the project.

[0173] The deployment agent is responsible for deploying the developed memory system to the debugging environment, including server configuration, software installation, database migration, etc. It should be noted that the deployment agent needs to combine large language model methods to deploy the debugging environment.

[0174] The evaluation agent evaluates the entire development process and the final memory system, analyzes and evaluates from multiple aspects such as requirement satisfaction, technical implementation, and performance indicators, and determines whether the task of developing the memory system is completed. It should be noted that the evaluation agent needs to combine large language model methods to evaluate the development process and the final memory system.

[0175] Next, in combination with specific application scenarios, an example of the method for developing an agent memory system through the computing power of an intelligent computing center according to the present invention will be described.

[0176] Example 1

[0177] The method for developing an agent memory system through the computing power of an intelligent computing center in this embodiment includes the following steps:

[0178] Step S1: The development agent receives the task information for constructing the target agent memory system provided by the user through the interaction interface;

[0179] The content provided by the user through the interaction interface may include the following:

[0180] User: Help me add a RAG-based memory system to the data analysis agent, requiring that the historical department report PPT can be used as knowledge. When the data analysis agent performs data analysis, the agent can extract the historical analysis report as a reference and then perform the analysis. The PPT is placed in the file storage path / mnt / share / data_analysis / reports / ppt / .

[0181] Among them, the "historical department report PPT" is the memory content in the above development requirements,

[0182] "RAG" is the memory mechanism in the above development requirements, and "when the data analysis agent performs data analysis, the agent can extract the historical analysis report as a reference" is the extraction timing in the above development requirements.

[0183] Step S2: The memory system development agent determines the development requirements of the memory system in combination with the large language model method and the user according to the task information;

[0184] The development agent confirms the requirements with the user through the interaction interface as follows:

[0185] Development agent: Wait a moment. I need to search for relevant conceptual knowledge of RAG to ensure that the design solution can make full use of the characteristics of RAG.

[0186] User: Okay, you can search first.

[0187] [The development agent retrieves relevant information on "RAG", "Memory System RAG", and "Agent RAG" through the search engine tool method of the development agent]

[0188] Development agent: I have found the relevant concepts of RAG as a whole: RAG (Retrieval-Augmented Generation) is a technical framework that combines retrieval and generation. It retrieves relevant information from an external knowledge base and combines the retrieved content with a generation model to generate more accurate and contextually relevant answers. The core advantage of RAG is the ability to dynamically utilize external knowledge to improve the accuracy and reliability of the generation results.

[0189] Development agent: Now I have a clearer understanding of the characteristics of RAG. Next, I will check the data situation in the / mnt / share / data_analysis / reports / ppt / path.

[0190] [The development agent checks the PPT file storage structure and samples in the path through the file reading, writing, and code generation tools of the development agent]

[0191] Development agent: The check is complete. The PPT files in the path are stored in directories divided by department structure. For example, PPT files related to the finance department are stored in the / mnt / share / data_analysis / reports / ppt / finance / path, and there are subdirectories further divided by year. The PPT for the 2024 financial report is placed in / mnt / share / data_analysis / reports / ppt / finance / 2024 / . The content of these PPTs can be directly used for extraction and processing. Next, we can continue to discuss the specific implementation details.

[0192] Development agent: Wait a moment. I will now check the code and relevant documents of the data analysis agent to confirm how to integrate the RAG framework with the existing system.

[0193] [The development agent, through tool methods such as file reading / writing and code generation of the development agent, views the source code and documentation of the target agent project]

[0194] Development agent: A preliminary understanding of the project situation is as follows:

[0195] ```

[0196] Main functions

[0197] ● The data analysis agent mainly provides data analysis capabilities. Users put forward data analysis requirements through the dialog box function, and the agent analyzes according to the requirements and returns the results.

[0198] ● Support analysis based on data uploaded by users and generate corresponding analysis results.

[0199] ● Support multi-turn conversations and retain the history of the last 5 conversations

[0200] Core technologies

[0201] The data analysis agent adopts the classic B / S (Browser / Server) architecture, uses Python as the core development language, and carefully builds the backend service with the FastAPI framework. During operation, the agent calls the Llama large language model, combines prompt engineering, realizes the ability of tool selection and code generation based on the Chain-of-Thought, and constructs the intelligent core part with the React (Reasoning+Acting) mode. At the same time, the agent has integrated a user permission management module internally, supporting data isolation at the user level. Currently, the system only supports the context memory of multi-turn conversations and does not have the support of other memory mechanisms.

[0202] Core code structure

[0203]

[0204]

[0205] ```

[0206] Development agent: Please tell me some specific requirements. What is the main goal of this memory system?

[0207] User: The main goal is to enable the agent to remember the historical departmental report content and, when conducting data analysis, generate more valuable analysis results by combining these historical analysis reports.

[0208] Developing an agent: Understood. Then how should these PPT files be processed? Should the content be directly retrieved, or does it need to be converted into a certain structured form?

[0209] User: The content of the PPT needs to be extracted and stored in the form of embedded vectors for efficient retrieval and matching later.

[0210] Developing an agent: Okay, I've noted it down. Regarding the matching method during retrieval, how do you want the system to select the reference historical reports? Is it based on keyword matching, or does it need to consider context relevance?

[0211] User: I hope to match based on the vector similarity between the question and the PPT to achieve efficient retrieval.

[0212] Developing an agent: Understood. Then when generating the analysis report, is multi - language support required? For example, do we need to support both Chinese and English analysis reports simultaneously?

[0213] User: Currently, it's mainly Chinese.

[0214] Developing an agent: Okay, I've noted it down. To ensure the accuracy and practicality of the memory system, do you have some specific test cases or scenarios? For example, the historical report content of a certain department and a new analysis task to see if the system can correctly match and generate valuable analysis results.

[0215] User: Yes. For example, in the historical report of the finance department, there is an analysis report on quarterly income and expenditure, and the new analysis task is to predict the income trend for the next quarter. The system needs to be able to extract the relevant historical reports and generate a predictive analysis in combination with the current data.

[0216] Developing an agent: Understood. For the case of "predicting the income trend for the next quarter" in the finance department, during this process, the memory system of the agent does not simply extract data, but needs in - depth thinking and reasoning. First, based on the task requirements, it infers that it needs to extract data such as "last month's income and expenditure analysis data", and then it is likely to match with "finance_ <year> Q <n>Documents such as "_Income and Expenditure Analysis.pptx". And during subsequent analysis, comprehensive analysis will be carried out by leveraging the data extracted from these documents in combination with the current data. I plan to use this as a test case. I wonder if it meets our test requirements for the memory system?

[0217] User: No problem.

[0218] Developing the agent: Okay, I have recorded it. To summarize, your requirements include:

[0219] ```

[0220] - Goal: Let the agent remember the historical departmental report content and generate more valuable analysis results in combination with the historical analysis report.

[0221] - PPT processing: Extract the content and store it in the form of embedded vectors for efficient retrieval and matching.

[0222] - Retrieval method: Infer the content to be extracted based on the question and use vector similarity to retrieve relevant PPTs.

[0223] - Security: The content is isolated by department and stored encrypted to ensure data security.

[0224] - Test case: When a user from the finance department asks "Predict the revenue trend for the next quarter", it should be able to match to the finance _ <year> Q <n>Analyze using _Income and Expenditure Analysis.pptx

[0225] ```

[0226] Is there anything else that needs to be added?

[0227] User: No.

[0228] Agent: Okay, I will start developing the RAG memory system for the data analysis agent later.

[0229] User: Okay.

[0230] [The agent saves the user requirement document through the agent's file reading and writing and other tool methods]

[0231] Step S3: The memory system development agent determines the design scheme of the memory system according to the development requirements and in combination with the large language model method;

[0232] In this embodiment, the development agent starts to design the memory system according to the requirements proposed by the user in step S2 and in combination with the file structure viewed under the / mnt / share / data_analysis / reports / ppt / path.

[0233] The following is the design scheme feedback by the development agent on the interaction interface:

[0234] Agent: According to your requirements, I have designed a RAG-based memory system and identified the files that need to be newly created and modified. The following is the specific scheme:

[0235] ```

[0236] 1. Overall Design

[0237] Memory mechanism: Adopt an inference retrieval mechanism. When performing an extraction operation, first, based on the problem content, infer the knowledge that may need to be extracted through logical analysis. Subsequently, apply the RAG (Retrieval-Augmented Generation) concept and introduce a signal recognition link for the required historical information extraction during the React process. Before officially conducting data analysis, retrieve relevant PPTs and then perform data analysis.

[0238] Memory format: Adopt vectorized storage. Convert the PPT content into vectors through an embedding model for efficient retrieval and matching.

[0239] Memory content: Store existing data analysis PPT materials in a vectorized format.

[0240] Extraction timing: When performing data analysis, give priority to retrieving historical knowledge materials.

[0241] Privacy Protection: Achieve department-level isolation to ensure data security.

[0242] 2. System Architecture

[0243] ● PPT Content Processing Module: Extract PPT content and convert it into embedded vectors.

[0244] ● Embedded Storage and Retrieval Module: Store the embeddings in a vector database and support context-related retrieval.

[0245] ● RAG Tool Method Module: Combine the retrieved historical reports and the current task to generate analysis results.

[0246] ● Permission and Isolation Module: Store and retrieve data in isolation by department to ensure data security.

[0247] 3. Module Design and File Operations

[0248] 1. PPT Content Processing Module:

[0249] ○ Function: Extract PPT content and generate embedded vectors.

[0250] ○ New File:

[0251] ■ ppt_processor.py: Responsible for PPT content extraction and preprocessing.

[0252] ■ embedding_generator.py: Call the embedding model to generate vectors.

[0253] ○ Modify File:

[0254] ■ file_handler.py: Add the processing logic for PPT files and call ppt_processor.py and embedding_generator.py.

[0255] 2. Embedded Storage and Retrieval Module:

[0256] ○ Function: Store embedded vectors and support context-related retrieval.

[0257] ○ New File:

[0258] ■ vector_store.py: Encapsulate the vector database operation to access FAISS and add an index for each department.

[0259] 3. RAG Tool Method Module:

[0260] ○ Function: Combine historical reports and the current task to generate analysis results.

[0261] ○ Modify File:

[0262] ■tools.py: Add tool methods for extracting knowledge content required for reasoning and corresponding knowledge extraction methods in this file to support an efficient knowledge acquisition process.

[0263] 4. Permission and Isolation Module:

[0264] ○ Function: Store and retrieve data isolated by department to ensure data security.

[0265] ○ Modified Files:

[0266] ■dbutil.py: Add the acquisition of user department ID and name.

[0267] 5. Initialization:

[0268] ○ Function: Initialize data.

[0269] ○ Newly Created Files:

[0270] ■init_ppts_job.py: Store the existing ppt documents in the vector database according to the department structure of the directory structure.

[0271] 3. Summary of File Operations

[0272] ● Newly Created Files:

[0273] ○ppt_processor.py: Extract PPT content.

[0274] ○embedding_generator.py: Generate embeddings.

[0275] ○vector_store.py: Operate on the vector database.

[0276] ○init_ppts_job.py: Import existing data.

[0277] ● Modified Files:

[0278] ○file_handler.py: Add PPT file processing logic.

[0279] ○tools.py: Add the extraction of knowledge content required for reasoning and the corresponding knowledge extraction methods.

[0280] ○dbutil.py: Add the acquisition of user department ID and name.

[0281] ```

[0282] Step S4: The memory system development agent generates the code of the memory system according to the development requirements and design scheme, combined with the large language model method;

[0283] In the embodiment of the present invention, the memory system development agent can save historical code versions.

[0284] Specifically, the development agent can generate corresponding code and corresponding file names by combining the existing code with the large language model method according to the development requirements and design scheme.

[0285] Step S5: The memory system development agent combines the large language model method to deploy the debugging environment of the memory system and debug the code of the memory system; when the debugging result indicates that the memory system does not meet the debugging requirements, the memory system development agent combines the large language model method to check the design scheme and code of the memory system. If the design scheme needs to be modified, return to step S3. If the code needs to be modified, return to step S4 until the debugging result indicates that the memory system meets the debugging requirements.

[0286] In this embodiment, the memory system development agent performs embodiment verification and calls the test_agent.py method. Input the question "Predict the revenue trend for the next quarter" and the user ID for verification. Check the log information during the running process to ensure that there are no error reports and ensure that the latest financial_ is matched. <year> Q <n>The document of "_Income and Expenditure Analysis.pptx".

[0287] Step S6: The memory system development agent feeds back result information to the user through the interaction interface.

[0288] Please refer to Figure 2 , the present invention also provides a device 10 for developing an agent memory system through the computing power of an intelligent computing center, including:

[0289] A receiving module 11, configured to receive task information for constructing a target agent memory system provided by the user through the interaction interface;

[0290] A requirement determination module 12, configured to determine the development requirements of the memory system in combination with the large language model method and the user according to the task information;

[0291] A design module 13, configured to determine the design scheme of the memory system in combination with the large language model method according to the development requirements;

[0292] An encoding module 14, configured to generate code for the memory system in combination with the large language model method according to the development requirements and the design scheme;

[0293] A debugging module 15, configured to deploy a debugging environment for the memory system and debug the code of the memory system in combination with the large language model method; when the debugging result indicates that the memory system does not meet the debugging requirements, the memory system development agent checks the design scheme and code of the memory system in combination with the large language model method. If the design scheme needs to be modified, trigger the design module to continue working. If the code needs to be modified, trigger the encoding module to continue working until the debugging result indicates that the memory system meets the debugging requirements;

[0294] A feedback module 16, configured to feed back result information to the user through the interaction interface.

[0295] Optionally, the development requirements of the memory system include at least one of the following: memory mechanism, memory content, memory format, extraction timing, performance requirements, security and privacy protection.

[0296] Optionally, the requirement determination module 12 is configured to communicate with the user through the interaction interface at least once based on the large language model according to the task information; determine the development requirements of the memory system in combination with the large language model method according to the task information and the communication content with the user.

[0297] Optionally, the design module 13 is configured to obtain learning outcomes by using at least one of the following learning methods in combination with the large language model method according to the development requirements of the memory system: searching for relevant materials, viewing and learning open-source code, learning the existing code and documentation of the target agent, and consulting data, and determining the design scheme of the memory system according to the learning outcomes.

[0298] Optionally, if it is found during the debugging process of the memory system development agent that the design of the memory system does not meet the debugging requirements, the design module 13 is configured to adjust the design scheme of the memory system by using at least one of the following adjustment methods in combination with the large language model method:

[0299] The first adjustment method is to optimize the design scheme of the memory system;

[0300] The second adjustment method is to adopt other design schemes;

[0301] The third adjustment method is to integrate multiple memory mechanisms.

[0302] Optionally, the coding module 14 is configured to generate the code of the memory system and the debugging log printing code in combination with the large language model method according to the development requirements and design scheme of the memory system. The debugging log printing code is used to record the key information during the operation of the memory system, and the key information includes at least one of the following: function call, variable value change, operation result, and exception situation;

[0303] The debugging module 15 is configured to run the code of the memory system and the debugging log printing code in combination with the large language model method, and check whether the design scheme and code of the memory system are running properly and meet the development requirements in combination with the large language model method according to the key information recorded by the debugging log printing code.

[0304] Optionally, the debugging module 15 includes at least one of the following sub-modules:

[0305] The first debugging sub-module is configured to terminate the development task of the memory system and feedback the termination reason and related situation to the user through the interaction interface when the number of debugging times reaches the first preset threshold, or the number of modifications to the design scheme reaches the second preset threshold, or the number of modifications to the code of the memory system reaches the third preset threshold, and the code of the memory system still cannot meet the debugging requirements;

[0306] The second debugging sub-module is configured to record the operation log, intermediate data, and debugging results during the debugging process and archive and save them;

[0307] The third debugging sub-module is used to deploy a resettable debugging environment or multiple different versions of debugging environments in combination with the large language model method to meet different versions of debugging requirements;

[0308] The fourth debugging sub-module is used to expand test cases in combination with the large language model method.

[0309] 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 on the memory 22 and executable on the processor 21. When the computer program is executed by the processor 21, it implements each process of the above method embodiment for developing an agent memory system through the computing power of an intelligent computing center, and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0310] The present invention also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements each process of the above method embodiment for developing an agent memory system through the computing power of an intelligent computing center, and can achieve the same technical effect. To avoid repetition, it will not be elaborated here. Among them, the computer-readable storage medium is, for example, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc, etc.

[0311] The embodiment of the present application also provides a computer program product, including computer instructions. When the computer instructions are executed by a processor, they implement each process of the above Figure 1 shown method embodiment for developing an agent memory system through the computing power of an intelligent computing center, and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0312] It should be noted that in this article, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the process, method, article or device including that element.

[0313] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0314] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the purpose of the present invention and the scope protected by the claims, and all of them belong to the protection scope of the present invention.< / n> < / year> < / n> < / year> < / n> < / year>

Claims

1. A method for developing an intelligent agent memory system through the computing power of an intelligent computing center, characterized in that, Including: Step S1: The memory system development agent receives task information for constructing the memory system of the target agent provided by the user through the interaction interface; Step S2: The memory system development agent determines the development requirements of the memory system in combination with the large language model method and the user according to the task information; Step S3: The memory system development agent determines the design scheme of the memory system in combination with the large language model method according to the development requirements; Step S4: The memory system development agent generates the code of the memory system in combination with the large language model method according to the development requirements and the design scheme; Step S5: The memory system development agent deploys the debugging environment of the memory system and debugs the code of the memory system in combination with the large language model method; When the debugging result indicates that the memory system does not meet the debugging requirements, the memory system development agent checks the design scheme and code of the memory system in combination with the large language model method. If the design scheme needs to be modified, return to Step S3. If the code needs to be modified, return to Step S4 until the debugging result indicates that the memory system meets the debugging requirements; Step S6: The memory system development agent feeds back the result information to the user through the interaction interface.

2. The method according to claim 1, characterized in that, The development requirements of the memory system include at least one of the following: memory mechanism, memory content, memory format, extraction timing, performance requirements, security and privacy protection.

3. The method according to claim 1, wherein The said Step S2 includes: Step S21: The memory system development agent communicates with the user through the interaction interface at least once in combination with the large language model method according to the task information; Step S22: Determine the development requirements of the memory system in combination with the large language model method according to the task information and the communication content with the user.

4. The method according to claim 1, characterized in that The said Step S3 includes: Step S31: The memory system development agent obtains learning results by using at least one of the following learning methods in combination with the large language model method according to the development requirements of the memory system: searching for relevant materials, viewing and learning open source code, learning the existing code and documents of the target agent, and consulting data. According to the learning results, determine the design scheme of the memory system.

5. The method according to claim 1, characterized in that The said Step S3 includes: Step S32: If the memory system development agent finds that the design of the memory system does not meet the debugging requirements during the debugging process, adjust the design scheme of the memory system by using at least one of the following adjustment methods in combination with the large language model method: The first adjustment method is to optimize the design scheme of the memory system; The second adjustment method is to adopt other design schemes; The third adjustment method is to integrate multiple memory mechanisms.

6. According to the method described in claim 1, characterized in that: The said Step S4 includes: Step S41: The memory system development agent generates the code of the memory system and the debug log printing code according to the development requirements and design scheme of the memory system, in combination with the large language model method. The debug log printing code is used to record the key information during the operation of the memory system, and the key information includes at least one of the following: function call, variable value change, operation result, and exception situation; The step S5 includes: Step S51: The memory system development agent runs the code of the memory system and the debug log printing code in combination with the large language model method, and checks whether the design scheme and code of the memory system are running properly and meet the development requirements according to the key information recorded by the debug log printing code, in combination with the large language model method.

7. The method according to claim 1, characterized in that, The step S5 includes at least one of the following sub-steps: Step S52: When the number of debugging times reaches the first preset threshold, or the number of modifications to the design scheme reaches the second preset threshold, or the number of modifications to the code of the memory system reaches the third preset threshold, and the code of the memory system still cannot meet the debugging requirements after that, the memory system development agent terminates the development task of the memory system and feedbacks the termination reason and related situation to the user through the interaction interface; Step S53: During the debugging process, the memory system development agent records the running log, intermediate data, and debugging results in combination with the large language model method, and archives and saves them; Step S54: The memory system development agent deploys a debug environment with resetability or deploys multiple different versions of debug environments in combination with the large language model method to meet the debug requirements of different versions; Step S55: The memory system development agent expands the test cases in combination with the large language model method.

8. A device for developing an intelligent agent memory system through the computing power of an intelligent computing center, characterized in that, including: A receiving module, configured to receive the task information for constructing the memory system of the target intelligent agent provided by the user through the interaction interface; A requirement determination module, configured to determine the development requirements of the memory system in combination with the large language model method and the user according to the task information; A design module, configured to determine the design scheme of the memory system in combination with the large language model method according to the development requirements; An encoding module, configured to generate the code of the memory system in combination with the large language model method according to the development requirements and design scheme; A debugging module, configured to deploy the debug environment of the memory system and debug the code of the memory system in combination with the large language model method; When the debugging result indicates that the memory system does not meet the debugging requirements, check the design scheme and code of the memory system in combination with the large language model method. If the design scheme needs to be modified, trigger the design module to continue working. If the code needs to be modified, trigger the encoding module to continue working until the debugging result indicates that the memory system meets the debugging requirements; A feedback module, configured to feedback the result information to the user through the interaction interface.

9. An electronic device, characterized in that, including: A processor, a memory, and a program stored on the memory and executable on the processor, the program, when executed by the processor, implementing the steps of the method for developing an agent memory system through the computing power of an intelligent computing center as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and the computer program, when executed by a processor, implements the steps of the method for developing an agent memory system through the computing power of an intelligent computing center as described in any one of claims 1 to 7.

11. A computer program product, characterized in that, It includes computer instructions, and the computer instructions, when executed by a processor, implement the steps of the method for developing an agent memory system through the computing power of an intelligent computing center as described in any one of claims 1 to 7.

Images