Method and device for forming intelligent agent team in intelligent computing center for universal computing power

Abstract

The present invention provides a method and device for forming an intelligent computing center agent team for inclusive computing power, which relates to the technical field of computing power infrastructure. The method is applied to a multi-agent collaboration system deployed in an intelligent computing center, and the multi-agent collaboration system includes: a dialogue interface and multiple agents, each agent has specific capabilities, and multiple agents can form a team and collaborate to complete a specified task in a dialogue communication manner in the dialogue interface, and the dialogue interface is used to display the process of executing the specified task, the communication process between multiple agents, and part or all of the decision-making process of the agent; the method includes: receiving a first specified task sent by a user through an interactive terminal; in response to the first specified task, obtaining a fitness evaluation standard corresponding to the first specified task, and, according to the fitness evaluation standard, using a Monte Carlo algorithm to form a target team that can complete the first specified task. The present invention realizes the inclusive application of computing power.

Description

Technical Field

[0001] Embodiments of the present invention relate to the technical field of computing power infrastructure, and in particular to a method and device for forming an intelligent computing center intelligent agent team for universal computing power. Background Art

[0002] With the development of artificial intelligence (AI) and computing power, the concept of intelligent computing centers has emerged. These centers utilize large-scale, heterogeneous computing resources, including both general-purpose and intelligent computing power, to provide the computing power, data, and algorithms required for AI applications (such as deep learning model development, model training, and model inference). Intelligent computing centers encompass facilities, hardware, and software, providing a full stack of capabilities, from bottom-level computing power to top-level application enablement.

[0003] An agent is an agent that can perceive its environment and take actions to achieve specific goals. It can be software, hardware, or a system, possessing autonomy, adaptability, and interactivity. An agent perceives changes in its environment (e.g., through sensors or data input), makes judgments and decisions based on learned knowledge and algorithms, and then executes actions to influence the environment or achieve a predetermined goal. Agents are widely used in the field of artificial intelligence, often found in automated systems, robots, virtual assistants, and game characters. Their core capability lies in their ability to autonomously learn and continuously evolve to better complete tasks and adapt to complex environments.

[0004] The existing team formation methods, even if they consume a lot of computing power, are still unable to efficiently form a multi-agent team to complete a given task. This also results in a large amount of computing power resources being wasted in the intelligent computing center, resulting in low utilization of the computing power resources corresponding to the agents, making it difficult to achieve universal application of computing power. Summary of the Invention

[0005] The embodiments of the present invention provide a method and device for forming an intelligent agent team in an intelligent computing center for universal computing power, so as to solve the problem that the existing team formation method is difficult to efficiently form a multi-agent team to complete a given task according to the given task even if a large amount of computing power is consumed, and a large amount of computing power resources of the intelligent computing center are wasted, resulting in low utilization of computing power resources corresponding to the agents, and difficulty in realizing the universal application of computing power.

[0006] In order to solve the above-mentioned technical problems, the present invention is achieved as follows:

[0007] In a first aspect, embodiments of the present invention provide a method for forming an agent team in an intelligent computing center for inclusive computing power, which is applied to a multi-agent collaboration system deployed in the intelligent computing center. The multi-agent collaboration system includes a dialogue interface and multiple agents, each of which has specific capabilities. The multiple agents can form a team and collaborate to complete a specified task through dialogue communication in the dialogue interface. The dialogue interface is used to display the process of executing the specified task, the communication process between the multiple agents, and part or all of the decision-making process of the agents.

[0008] The team building method includes:

[0009] Step S1: receiving a first designated task sent by a user through an interactive terminal;

[0010] Step S2: In response to the first designated task, a fitness evaluation standard corresponding to the first designated task is obtained, and based on the fitness evaluation standard, a Monte Carlo algorithm is used to form a target team that can complete the first designated task.

[0011] Optionally, step S2 includes:

[0012] Step S201: Determine an agent that is adapted to the first designated task as a root node in the Monte Carlo algorithm; perform capability requirement analysis on the first designated task based on the capability of the agent corresponding to the root node to obtain a capability requirement analysis result; and use the agent corresponding to the root node as a base agent;

[0013] Step S202: According to the capability requirement analysis result, the basic agent is expanded to obtain multiple candidate agents of the next layer node below the level of the basic agent;

[0014] Step S203: Evaluate the fitness of the sub-teams composed of the basic agent and each of the candidate agents of the next-layer node according to the fitness evaluation standard, and determine the candidate agent of the sub-team with the highest fitness among all the sub-teams as the agent of the next-layer node;

[0015] Step S204: determining whether the first team composed of the agents of all current nodes can complete the first designated task;

[0016] Step S205: If the first team can complete the first designated task, the first team is used as the target team;

[0017] Step S206: If the first team cannot complete the first designated task, the agent of the current lowest layer node is used as the new basic agent, and the process returns to step S202 until the first team can complete the first designated task.

[0018] Optionally, step S204 includes:

[0019] Step S2041: Verifying whether the fitness of the first team is greater than or equal to a preset fitness threshold according to the fitness evaluation standard, and obtaining a verification result;

[0020] Step S2042: If the verification result indicates that the fitness of the first team is greater than or equal to the preset fitness threshold, determining that the first team is capable of completing the first designated task;

[0021] Step S2043: If the verification result indicates that the fitness of the first team is less than the preset fitness threshold, it is determined that the first team is unable to complete the first designated task.

[0022] Optionally, each of the intelligent agents includes:

[0023] A data module, configured to store a description of relevant information of the agent, wherein the relevant information includes capabilities of the agent;

[0024] Communication module, used for dialogue and communication with other agents and users;

[0025] The decision-making module is used to make decisions and take corresponding actions on tasks given by other intelligent agents or users based on dialogue exchanges with other intelligent agents or users.

[0026] Optionally, the step S2 then includes:

[0027] Step S3: Sending a prompt message to the interactive terminal, wherein the prompt message prompts the user to instruct each of the intelligent agents in the target team to collaboratively complete the first designated task through dialogue communication.

[0028] In a second aspect, an embodiment of the present invention provides an intelligent computing center agent team formation device for inclusive computing power, which is applied to a multi-agent collaboration system deployed in an intelligent computing center. The multi-agent collaboration system includes: a dialogue interface and multiple agents, each of which has specific capabilities. The multiple agents can form a team and collaborate to complete a specified task through dialogue communication in the dialogue interface. The dialogue interface is used to display the process of executing the specified task, the communication process between the multiple agents, and part or all of the decision-making process of the agents.

[0029] The team forming device comprises:

[0030] A receiving module, configured to receive a first designated task sent by a user through an interactive terminal;

[0031] An execution module is configured to, in response to the first designated task, obtain a fitness evaluation standard corresponding to the first designated task, and, based on the fitness evaluation standard, adopt a Monte Carlo algorithm to form a target team capable of completing the first designated task.

[0032] Optionally, the execution module includes:

[0033] a first execution submodule, configured to determine an agent adapted to the first designated task as a root node under the Monte Carlo algorithm; perform capability requirement analysis on the first designated task based on the capability of the agent corresponding to the root node to obtain a capability requirement analysis result; and use the agent corresponding to the root node as a base agent;

[0034] A second execution submodule is configured to expand the basic agent according to the capability requirement analysis result to obtain multiple candidate agents at the next layer of nodes below the level where the basic agent is located;

[0035] A third execution submodule is configured to evaluate the fitness of the basic agent and each of the candidate agents of the next-layer node according to the fitness evaluation standard, and determine the candidate agent of the sub-team with the highest fitness among all the sub-teams as the agent of the next-layer node;

[0036] a fourth execution submodule, configured to determine whether the first team composed of the agents of all current nodes can complete the first specified task;

[0037] A fifth execution submodule, configured to use the first team as the target team if the first team can complete the first designated task;

[0038] The sixth execution submodule is used to use the agent of the current lowest-level node as the new basic agent if the first team cannot complete the first designated task, and return to the second execution submodule until the first team can complete the first designated task.

[0039] In a third aspect, an embodiment of the present invention provides an electronic device comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor. When the program or instruction is executed by the processor, the steps in the method for forming an intelligent computing center intelligent agent team for inclusive computing power as described in any one of the first aspects are implemented.

[0040] In a fourth aspect, an embodiment of the present invention provides a readable storage medium storing a program or instruction, which, when executed by a processor, implements the steps in the method for forming an intelligent computing center intelligent agent team for inclusive computing power as described in any one of the first aspects.

[0041] In a fifth aspect, an embodiment of the present invention provides a computer program product comprising computer instructions, which, when executed by a processor, implement the steps of the method for forming an intelligent computing center intelligent agent team for inclusive computing power as described in any one of the first aspects.

[0042] In an embodiment of the present invention, through step S1: receiving a first designated task sent by a user through an interactive terminal; step S2: in response to the first designated task, obtaining a fitness evaluation standard corresponding to the first designated task, and, based on the fitness evaluation standard, using a Monte Carlo algorithm to form a target team that can complete the first designated task, the embodiment of the present invention can quickly form a target team based on the fitness evaluation standard using a Monte Carlo algorithm, and the efficiency of team formation is high. The embodiment of the present invention is applied to a multi-agent collaborative system deployed in an intelligent computing center, and the computing power resources of the intelligent computing center are called during the execution of each step. Based on the present invention, a team can be formed efficiently, the utilization rate of the computing power resources of the intelligent computing center is high, and the universal application of computing power is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0044] Figure 1 Schematic diagram of a dialogue interface of a multi-agent collaboration system according to an embodiment of the present invention;

[0045] Figure 2 This is a flow chart of a method for forming an intelligent computing center agent team for inclusive computing power according to an embodiment of the present invention;

[0046] Figure 3 This is the second flow chart of the method for forming an intelligent computing center agent team for inclusive computing power according to an embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of the Monte Carlo tree structure formed during the target team formation process;

[0048] Figure 5This is a functional block diagram of a device for forming an intelligent computing center agent team for inclusive computing power according to an embodiment of the present invention;

[0049] Figure 6 This is a principle block diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0051] The terms "first", "second", etc. in the embodiments of the present invention are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way are interchangeable where appropriate, so that the embodiments of the embodiments of the present invention can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "or" in the embodiments of the present invention represents at least one of the connected objects. For example, "A or B" covers three options, namely, Option 1: including A but not including B; Option 2: including B but not including A; Option 3: including both A and B. The character " / " generally indicates that the objects associated before and after are in an "or" relationship.

[0052] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0053] It should be noted that the collection, collection, updating, analysis, processing, use, transmission, and storage of personal information involved in the technical solutions disclosed in the embodiments of this invention all comply with relevant laws and regulations, are used for legitimate purposes, and do not violate public order and good morals. Necessary measures are taken with respect to personal information to prevent unauthorized access to personal information data and maintain personal information security and network security.

[0054] The following first briefly describes the technical terms involved in the embodiments of the present invention.

[0055] The "computing power" described in the embodiments of the present invention is the ability of computer equipment or computing / data centers to process information. It is the ability of computer hardware and software to work together to perform certain computing requirements. It is the computing power to achieve target result output by processing information data. It is a new type of productivity that integrates information computing power, network carrying capacity, and data storage capacity. It mainly provides services to society through computing power infrastructure.

[0056] The "computational power" (CP) described in the embodiments of this invention refers to the ability of data center servers to process data and output results. It is a comprehensive indicator of data center computing capabilities, encompassing general computing power, supercomputing power, and intelligent computing power. The commonly used unit of measurement is floating point operations per second (FLOPS), with 1 EFLOPS equal to 10^18 FLOPS. A higher EFLOPS represents greater overall computing power. It is estimated that 1 EFLOPS is equivalent to the computing power of approximately 5 Tianhe-2A processors, 500,000 mainstream server CPUs, or 2 million mainstream laptops. The calculation formula is: CP = CP General + CP Intelligent + CP Super.

[0057] The "Network Power" (NP) described in the embodiments of this invention represents the data transmission capabilities of computing facilities, encompassing comprehensive capabilities such as network architecture, network bandwidth, transmission latency, and intelligent management and scheduling. NP involves network transmission within and between data centers and is a comprehensive indicator of network transmission and scheduling capabilities. In this embodiment of this invention, NP is measured using graphics memory bandwidth.

[0058] The "Storage Power" (SP) described in the embodiments of this invention refers to the comprehensive capabilities of a data center in four areas: data storage capacity, performance, security and reliability, and environmental friendliness and low-carbon development. It is a comprehensive indicator for measuring a data center's data storage capabilities, encompassing both external storage devices such as storage arrays and internal server storage. Storage capacity is commonly measured in exabytes (EB, 1EB = 2^60 bytes), while performance is commonly measured in IOPS / TB (Input / Output Operations Per Second). Disaster recovery ratio is a key indicator of security and reliability.

[0059] The "computing power infrastructure" described in the embodiments of the present invention is a new type of information infrastructure that integrates information computing power, network carrying capacity, and data storage capacity. It can realize the centralized calculation, storage, transmission and application of information, and presents the characteristics of diversity and ubiquity, intelligence and agility, security and reliability, green and low-carbon. It is of great significance to promote industrial transformation and upgrading, enable my country's scientific and technological innovation, meet people's needs for a better life, and achieve high-efficiency social governance.

[0060] The “computing power” described in the embodiments of the present invention includes general computing power, intelligent computing power and super computing power.

[0061] The "general computing power" described in the embodiments of 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.

[0062] The "intelligent computing power" described in the embodiments of the present invention is a computing platform for large-scale deployment of dedicated chips such as GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), and ASIC (Application Specific Integrated Circuit) for various innovative artificial intelligence applications, such as natural language processing (NLP) and machine vision.

[0063] The "super computing power" described in the embodiments of the present invention is mainly the computing power provided by high-performance computing clusters such as supercomputers. It utilizes the centralized computing resources of multiple computer systems working in parallel and uses a dedicated operating system to handle extremely complex or data-intensive problems. It is mainly used for calculations in cutting-edge scientific fields, such as planetary simulation, drug molecule design, genetic analysis, etc.

[0064] The "intelligent computing center" described in the embodiments of this invention refers to a facility that primarily provides the computing power, data, and algorithms required for AI applications (such as AI deep learning model development, model training, and model inference) by utilizing large-scale heterogeneous computing resources, including general-purpose computing power (CPUs (Central Processing Units)) and intelligent computing power (GPUs (Graphics Processing Units), FPGAs (Field Programmable Gate Arrays), and ASICs (Application Specific Integrated Circuits). This facility encompasses facilities, hardware, and software, providing a full stack of capabilities, from bottom-level computing power to top-level application enablement.

[0065] The "computing resources" mentioned in the embodiments of the present invention refer to technologies and facilities with information computing, transmission, storage and application capabilities required for the development of a digital society, including but not limited to computing resources such as CPU (Central Processing Unit), GPU (Graphics Processing Unit), network resources such as switches and routers, storage resources such as storage arrays and distributed storage, security resources such as firewalls and intrusion detection systems, and supporting and guarantee resources such as wind, fire, water and electricity.

[0066] The "inclusive computing power" described in the embodiments of the present invention refers to providing appropriate and effective computing power services at an affordable cost to all social classes and groups that have computing power service needs based on the requirements of equal opportunity and the principle of commercial sustainability.

[0067] The agent described in the embodiments of this invention is an agent capable of perceiving its environment and taking actions to achieve specific goals. It can be software, hardware, or a system, and possesses autonomy, adaptability, and interactivity. An agent perceives changes in its environment (e.g., through sensors or data input), makes judgments and decisions based on learned knowledge and algorithms, and then executes actions to influence the environment or achieve predetermined goals. Agents are widely used in the field of artificial intelligence, often found in automated systems, robots, virtual assistants, and game characters. Their core capability lies in their ability to autonomously learn and continuously evolve to better complete tasks and adapt to complex environments.

[0068] The "large language model" described in the embodiments of the present invention refers to a large-scale language model (LLM), which is a language model with a large parameter scale. It is designed to understand and generate human language. It is trained with a large amount of text data and can perform a wide range of tasks including text summarization, translation, sentiment analysis, etc.

[0069] An embodiment of the present invention provides a method for forming an intelligent computing center agent team for inclusive computing power, which is applied to a multi-agent collaboration system deployed in the intelligent computing center. The multi-agent collaboration system includes a dialogue interface and multiple agents, each of which has specific capabilities. The multiple agents can form a team and collaborate to complete a specified task through dialogue communication in the dialogue interface. The dialogue interface is used to display the process of executing the specified task, the communication process between the multiple agents, and part or all of the decision-making process of the agents.

[0070] See also Figure 1 As shown, Figure 1 This is a schematic diagram of the dialogue interface of the multi-agent collaboration system according to an embodiment of the present invention. Figure 1As can be seen from the figure, the multi-agent collaboration system in this embodiment includes multiple agents, including: NANA (project manager), Hu (PM assistant), Wang (operation and maintenance engineer), Doudou (data engineer), Li (large language model training engineer), Guo (large language model evaluation engineer), among which each agent is built on the basis of a large language model, and each agent has specific capabilities. For example, "Hu (PM assistant)" can be responsible for confirming necessary parameters and recording parameters for reference by other users. Each agent can communicate with other agents and users (i.e. humans, such as Figure 1 The dialogue interface can display the communication process between multiple agents during the execution of a specified task, as well as part or all of the decision-making process of the agents.

[0071] See also Figure 2 As shown, Figure 2 This is a flow chart of a method for forming an intelligent computing center agent team for inclusive computing power according to an embodiment of the present invention. The team forming method includes:

[0072] Step S1: receiving a first designated task sent by a user through an interactive terminal;

[0073] Step S2: In response to the first designated task, a fitness evaluation standard corresponding to the first designated task is obtained, and based on the fitness evaluation standard, a Monte Carlo algorithm is used to form a target team that can complete the first designated task.

[0074] In an embodiment of the present invention, the first designated task is varied. Exemplarily, the first designated task may be to train a large language model for allocating computing resources, and the target team formed may include: an intelligent agent capable of collecting training data, an intelligent agent capable of calling the computing resources required for the training model from an intelligent computing center, and an intelligent agent capable of using actual data to verify the training results. The above first designated tasks are only exemplary and not restrictive. The first designated tasks are varied and are determined by the user according to his or her own needs. They cannot be listed here exhaustively. As long as the specific first designated tasks are identified by ordinary technicians in this field under the conception of the present invention, they are all within the protection of the present invention.

[0075] The Monte Carlo algorithm, also known as Monte Carlo Tree Search (MCTS), is a search algorithm for decision-making processes that combines random sampling with a tree structure to find the optimal action strategy in complex decision spaces. MCTS can handle large state spaces and does not require a complete tree structure, making it suitable for environments with incomplete or complex information.

[0076] In an embodiment of the present invention, the target team is a team assembled using a Monte Carlo algorithm that meets the fitness requirements under the fitness evaluation criteria; that is, meeting the fitness requirements under the fitness evaluation criteria indicates that the team is capable of completing the first designated task. The target team is composed of a combination of multiple agents, and all agents in the target team are capable of collaboratively completing the first designated task. In some optional embodiments, the fitness requirement is a preset fitness threshold.

[0077] It should be noted that for a specific first designated task, the capabilities and number of agents that can be configured to complete the first designated task are not unique, meaning that multiple teams can be formed, and there are multiple possibilities for forming teams. In step S2, this embodiment of the present invention can rapidly form a team capable of completing the first designated task using a Monte Carlo algorithm based on the fitness evaluation criteria as the target team, without considering the multiple possibilities for forming a team, resulting in high team formation efficiency.

[0078] The embodiment of the present invention is applied to a multi-agent collaborative system deployed in an intelligent computing center. The execution process of each step of the embodiment of the present invention calls upon the abundant computing resources of the intelligent computing center, thereby improving the execution efficiency of each step.

[0079] In an embodiment of the present invention, through step S1: receiving a first designated task sent by a user through an interactive terminal; step S2: in response to the first designated task, obtaining a fitness evaluation standard corresponding to the first designated task, and, based on the fitness evaluation standard, using a Monte Carlo algorithm to form a target team that can complete the first designated task, the embodiment of the present invention can quickly form a target team based on the fitness evaluation standard using a Monte Carlo algorithm, and the efficiency of team formation is high. The embodiment of the present invention is applied to a multi-agent collaborative system deployed in an intelligent computing center, and the computing power resources of the intelligent computing center are called during the execution of each step. Based on the present invention, a team can be formed efficiently, the utilization rate of the computing power resources of the intelligent computing center is high, and the universal application of computing power is realized.

[0080] In some embodiments of the present invention, optionally, see Figure 3 As shown, Figure 3 This is a second flow chart of the method for forming an intelligent computing center agent team for inclusive computing power according to an embodiment of the present invention. Step S2 includes:

[0081] Step S201: determining an agent that is adapted to a first designated task as a root node under the Monte Carlo algorithm; performing capability requirement analysis on the first designated task based on the capability of the agent corresponding to the root node to obtain a capability requirement analysis result; and taking the agent corresponding to the root node as a base agent;

[0082] Step S202: According to the capability requirement analysis result, the basic agent is expanded to obtain multiple candidate agents of the next layer node below the level of the basic agent;

[0083] Step S203: Evaluate the fitness of the sub-teams composed of the basic agent and each candidate agent of the next-layer node according to the fitness evaluation standard, and determine the candidate agent of the sub-team with the highest fitness among all sub-teams as the agent of the next-layer node;

[0084] Step S204: Determine whether the first team composed of all the agents of the current nodes can complete the first designated task;

[0085] Step S205: If the first team can complete the first designated task, the first team is used as the target team;

[0086] Step S206: If the first team cannot complete the first designated task, the agent of the current lowest layer node is used as the new basic agent, and the process returns to step S202 until the first team can complete the first designated task.

[0087] In an embodiment of the present invention, the multi-agent collaboration system determines an agent that is adapted to the first designated task from all agents based on its own prior knowledge as the agent corresponding to the root node under the Monte Carlo algorithm.

[0088] It should be noted that the expansion performed by the base agent is derived using a Monte Carlo algorithm under the fitness evaluation criteria. The multiple candidate agents for the next-level nodes obtained each time are agents with capabilities not possessed by the base agent and included in the capability requirement analysis results, and / or agents with capabilities that can assist the base agent in achieving its own capabilities. For example, if the base agent's capability is data cleaning, the candidate agent's capability for the next-level node can be data sorting, which can assist the base agent in achieving data cleaning.

[0089] See also Figure 4 As shown, Figure 4 This is a schematic diagram of the Monte Carlo tree structure formed during the target team formation process. Figure 4Each circle in the figure represents an agent with specific capabilities. Agent A is an agent determined by the multi-agent collaborative system to be adapted to the first designated task, that is, the agent corresponding to the root node under the Monte Carlo algorithm. Taking the first designated task of training a large language model for image recognition as an example, the capability of agent A is training sample labeling. Based on the capabilities of the agent corresponding to the root node, the capability requirements of the first designated task are analyzed to obtain a capability requirement analysis result. In other words, the task of training a large language model for image recognition is analyzed to obtain the capabilities required to complete the task (i.e., capability requirement analysis results), such as training sample labeling, supervised training, data cleaning, and data collection. Furthermore, when expanding with agent A as the basic agent, the multiple candidate agents (e.g., B1, B2, C1, and D3) each possess at least one capability in the capability requirement analysis result that agent A does not possess, and / or the capabilities possessed can assist agent A in achieving training sample labeling. Based on the fitness evaluation criteria, the fitness of the four sub-teams A-B1, A-B2, A-C1, and A-D3 is evaluated. Sub-team A-B2 has the highest fitness, so its candidate agent, B2, becomes the agent for the next node. Furthermore, the agents from all the current nodes form the first team, Team A-B2, and are evaluated to see if Team A-B2 can complete the first assigned task.

[0090] The evaluation results indicate that the A-B2 team is unable to complete the first designated task. Therefore, agent B2 is used as the base agent for the next layer of expansion, resulting in two candidate agents, C2 and D4. According to the fitness evaluation criteria, the fitness of the two sub-teams B2-C2 and B2-D4 is evaluated. Among them, the fitness of the B2-D4 sub-team is the highest. Therefore, the candidate agent D4 of the B2-D4 sub-team is the agent of the next layer of nodes. Furthermore, the agents of all the current nodes form the first team, namely the A-B2-D4 team. The A-B2-D4 team is evaluated to see if it can complete the first designated task. The evaluation results indicate that the A-B2-D4 team can complete the first designated task, and the A-B2-D4 team is determined to be the target team.

[0091] It can be understood that as the next layer is expanded according to step S202, the capabilities of the next layer of intelligent agents become more and more refined. Therefore, in some embodiments, when the target team performs the first designated task, the actual output order of the intelligent agents is from the lowest layer of intelligent agents to the intelligent agent corresponding to the root node.

[0092] In the embodiment of the present invention, through the above steps S201 to S206, the embodiment of the present invention can efficiently and quickly determine the target team and improve the utilization efficiency of computing resources.

[0093] In some embodiments of the present invention, optionally, step S204 includes:

[0094] Step S2041: Verify whether the first team's fitness is greater than or equal to a preset fitness threshold according to the fitness evaluation standard, and obtain a verification result;

[0095] Step S2042: If the verification result indicates that the fitness of the first team is greater than or equal to a preset fitness threshold, it is determined that the first team is capable of completing the first designated task;

[0096] Step S2043: If the verification result indicates that the fitness of the first team is less than a preset fitness threshold, it is determined that the first team is unable to complete the first designated task;

[0097] The adaptation threshold is determined by the user according to his actual needs. It can be understood that the higher the adaptation threshold, the more adapted the target team is to the first designated task, and the more confident the target team is in completing the first designated task.

[0098] In some embodiments of the present invention, optionally, each agent includes:

[0099] The data module is used to store the description of the relevant information of the intelligent agent, including the capabilities of the intelligent agent;

[0100] Communication module, used for dialogue and communication with other agents and users;

[0101] The decision-making module is used to make decisions and take corresponding actions on tasks given by other intelligent agents or users based on dialogues with other intelligent agents or users.

[0102] The description of the relevant information of the agent in the data module can also be referred to as an introduction to the agent, or a definition of the agent, which is used to describe what kind of agent the agent is. The relevant information includes at least the capabilities of the agent, for example, an agent's capability is to train the initial large language model based on given training data and the initial large language model. In some embodiments, the relevant information may optionally further include the name of the agent, for example, the name of the agent is "NANA (project manager)".

[0103] The communication module can communicate with other intelligent agents and users (i.e., humans) through natural language processing.

[0104] It should be noted that "other agents" refers to any agent other than itself in the multi-agent collaborative system.

[0105] In an embodiment of the present invention, the above-mentioned intelligent agent includes: a data module for storing a description of relevant information of the intelligent agent, the relevant information including the capabilities of the intelligent agent; a communication module for conducting dialogues and exchanges with other intelligent agents and users; a decision-making module for making decisions and taking corresponding actions on tasks given by other intelligent agents or users based on dialogues and exchanges with other intelligent agents or users, to ensure that the intelligent agents in the target team formed by the multi-agent collaborative system can collaborate with each other until the first specified task is completed.

[0106] In some embodiments of the present invention, optionally, step S2 then includes:

[0107] Step S3: Send a prompt message to the interactive end, where the prompt message prompts the user to instruct the various agents in the target team to collaboratively complete the first designated task through dialogue communication.

[0108] In an embodiment of the present invention, through the above-mentioned step S3: a prompt message is sent to the interactive end, and the prompt message prompts the user to instruct the various intelligent entities in the target team to collaboratively complete the first specified task through dialogue communication, which can assist the user to use a convenient dialogue communication method to issue instructions to the target team and improve the user experience.

[0109] An embodiment of the present invention provides an intelligent computing center agent team formation device for inclusive computing power, which is applied to a multi-agent collaboration system deployed in the intelligent computing center. The multi-agent collaboration system includes: a dialogue interface and multiple agents, each of which has specific capabilities. The multiple agents can form a team and collaborate to complete a specified task through dialogue communication in the dialogue interface. The dialogue interface is used to display the process of executing the specified task, the communication process between the multiple agents, and part or all of the decision-making process of the agents.

[0110] See also Figure 5 As shown, Figure 5 This is a block diagram of the principle of an intelligent computing center agent team building device for inclusive computing power according to an embodiment of the present invention. The team building device 50 includes:

[0111] A receiving module 51 is configured to receive a first designated task sent by a user via an interactive terminal;

[0112] The execution module 52 is configured to obtain, in response to the first designated task, a fitness evaluation standard corresponding to the first designated task, and, based on the fitness evaluation standard, use a Monte Carlo algorithm to form a target team capable of completing the first designated task.

[0113] In some embodiments of the present invention, optionally, the execution module 52 includes:

[0114] a first execution submodule, configured to determine an agent adapted to the first designated task as a root node under the Monte Carlo algorithm; perform capability requirement analysis on the first designated task based on the capability of the agent corresponding to the root node to obtain a capability requirement analysis result; and use the agent corresponding to the root node as a base agent;

[0115] A second execution submodule is configured to expand the basic agent according to the capability requirement analysis result to obtain multiple candidate agents at the next layer of nodes below the level where the basic agent is located;

[0116] A third execution submodule is configured to evaluate the fitness of the basic agent and each of the candidate agents of the next-layer node according to the fitness evaluation standard, and determine the candidate agent of the sub-team with the highest fitness among all the sub-teams as the agent of the next-layer node;

[0117] a fourth execution submodule, configured to determine whether the first team composed of the agents of all current nodes can complete the first specified task;

[0118] A fifth execution submodule, configured to use the first team as the target team if the first team can complete the first designated task;

[0119] The sixth execution submodule is used to use the agent of the current lowest-level node as the new basic agent if the first team cannot complete the first designated task, and return to the second execution submodule until the first team can complete the first designated task.

[0120] In some embodiments of the present invention, optionally,

[0121] The fourth execution module is further configured to verify, according to the fitness evaluation standard, whether the fitness of the first team is greater than or equal to a preset fitness threshold, and obtain a verification result;

[0122] The fourth execution module is further configured to determine that the first team is capable of completing the first designated task if the verification result indicates that the fitness of the first team is greater than or equal to the preset fitness threshold;

[0123] The fourth execution module is further configured to determine that the first team is unable to complete the first designated task if the verification result indicates that the fitness of the first team is less than the preset fitness threshold;

[0124] In some embodiments of the present invention, optionally, each of the intelligent agents includes:

[0125] A data module, configured to store a description of relevant information of the agent, wherein the relevant information includes capabilities of the agent;

[0126] Communication module, used for dialogue and communication with other agents and users;

[0127] The decision-making module is used to make decisions and take corresponding actions on tasks given by other intelligent agents or users based on dialogue exchanges with other intelligent agents or users.

[0128] In some embodiments of the present invention, optionally, the team building device 50 further includes:

[0129] A prompt module is used to send a prompt message to the interactive end, where the prompt message prompts the user to instruct each of the intelligent agents in the target team to collaboratively complete the first specified task through dialogue communication.

[0130] The intelligent computing center intelligent agent team formation device for universal computing power provided in the embodiment of the present application can achieve Figures 1 to 4 The various processes implemented by the method embodiment achieve the same technical effect and are not described here again to avoid repetition.

[0131] The embodiment of the present invention provides an electronic device 60, see Figure 6 As shown, Figure 6 This is a principle block diagram of an electronic device 60 according to an embodiment of the present invention, including a processor 61, a memory 62, and a program or instruction stored in the memory 62 and executable on the processor 61. When the program or instruction is executed by the processor, any step in the method for forming an intelligent computing center intelligent agent team for universal computing power according to the present invention is implemented.

[0132] An embodiment of the present invention provides a readable storage medium, which stores programs or instructions. When the program or instructions are executed by a processor, the various processes of the embodiment of the method for forming an intelligent computing center intelligent agent team for universal computing power, such as any of the above-mentioned items, are implemented, and the same technical effect can be achieved. To avoid repetition, they will not be repeated here.

[0133] The readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk. In some examples, the readable storage medium may be a non-transitory readable storage medium.

[0134] An embodiment of the present invention also provides a computer program product, including computer instructions. When the computer instructions are executed by a processor, the various processes of any of the above-mentioned embodiments of the method for forming an intelligent computing center intelligent agent team for universal computing power are implemented, and the same technical effect can be achieved. To avoid repetition, they will not be repeated here.

[0135] The embodiments of the present invention are described above in conjunction with the accompanying drawings, but the present invention is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present invention, ordinary technicians in this field can also make many forms without departing from the scope of protection of the present invention and the claims, all of which are protected by the present invention.

Claims

1. A method for forming an intelligent agent team in an intelligent computing center for universal computing power, characterized in that: A multi-agent collaboration system deployed in an intelligent computing center, the multi-agent collaboration system comprising: a dialogue interface and multiple agents, each of the agents having specific capabilities, the multiple agents being able to form a team and collaborate to complete a specified task in a dialogue communication manner in the dialogue interface, the dialogue interface being used to display the process of executing the specified task, the communication process between the multiple agents, and part or all of the decision-making process of the agents; The team building method includes: Step S1: receiving a first designated task sent by a user through an interactive terminal; Step S2: in response to the first designated task, obtaining a fitness evaluation standard corresponding to the first designated task, and using a Monte Carlo algorithm to form a target team capable of completing the first designated task according to the fitness evaluation standard; The step S2 comprises: Step S201: determining an agent adapted to the first designated task as a root node under the Monte Carlo algorithm; performing capability requirement analysis on the first designated task according to the capability of the agent corresponding to the root node to obtain a capability requirement analysis result; and taking the agent corresponding to the root node as the base agent; Step S202: According to the capability requirement analysis result, the basic agent is expanded to obtain multiple candidate agents of the next layer node below the level where the basic agent is located; Step S203: According to the fitness evaluation standard, the fitness of the sub-teams composed of the basic agent and each of the candidate agents of the next layer node is evaluated, and the candidate agent of the sub-team with the highest fitness among all the sub-teams is determined as the agent of the next layer node; Step S204: determining whether the first team composed of the agents of all current nodes can complete the first designated task; Step S205: If the first team can complete the first designated task, the first team is used as the target team; Step S206: If the first team cannot complete the first designated task, the agent at the current bottom-level node is used as the new basic agent, and the process returns to step S202 until the first team can complete the first designated task.

2. The method for forming an intelligent computing center agent team for universal computing power according to claim 1 is characterized in that: The step S204 includes: Step S2041: according to the fitness evaluation standard, verify whether the fitness of the first team is greater than or equal to a preset fitness threshold, and obtain a verification result; Step S2042: If the verification result indicates that the fitness of the first team is greater than or equal to the preset fitness threshold, determining that the first team is able to complete the first designated task; Step S2043: If the verification result indicates that the fitness of the first team is less than the preset fitness threshold, it is determined that the first team is unable to complete the first designated task.

3. The method for forming an intelligent computing center agent team for universal computing power according to claim 1 is characterized in that: Each of the intelligent agents comprises: A data module, used to store descriptions of relevant information of the agent, wherein the relevant information includes capabilities of the agent; The communication module is used to communicate with other agents and users; The decision-making module is used to make decisions and take corresponding actions on tasks given by other intelligent agents or users based on dialogue exchanges with other intelligent agents or users.

4. The method for forming an intelligent computing center agent team for universal computing power according to claim 3 is characterized in that: The step S2 then comprises: Step S3: Sending a prompt message to the interactive end, wherein the prompt message prompts the user to instruct each of the intelligent agents in the target team to collaboratively complete the first designated task through dialogue communication.

5. A device for forming an intelligent computing center agent team for universal computing power, characterized in that: A multi-agent collaboration system deployed in an intelligent computing center, the multi-agent collaboration system comprising: a dialogue interface and multiple agents, each of the agents having specific capabilities, the multiple agents being able to form a team and collaborate to complete a specified task in a dialogue communication manner in the dialogue interface, the dialogue interface being used to display the process of executing the specified task, the communication process between the multiple agents, and part or all of the decision-making process of the agents; The team building device comprises: A receiving module, used for receiving a first designated task sent by a user through an interactive terminal; an execution module, configured to, in response to the first designated task, obtain a fitness evaluation standard corresponding to the first designated task, and, based on the fitness evaluation standard, use a Monte Carlo algorithm to form a target team capable of completing the first designated task; The execution module comprises: The first execution submodule is used to determine an agent adapted to the first specified task as a root node under the Monte Carlo algorithm; perform capability requirement analysis on the first specified task according to the capability of the agent corresponding to the root node to obtain a capability requirement analysis result; and use the agent corresponding to the root node as a base agent; A second execution submodule is used to expand the basic agent according to the capability requirement analysis result to obtain multiple candidate agents of the next layer node below the level where the basic agent is located; The third execution submodule is used to evaluate the fitness of the sub-teams composed of the basic agent and each of the candidate agents of the next layer node according to the fitness evaluation standard, and determine the candidate agent of the sub-team with the highest fitness among all the sub-teams as the agent of the next layer node; A fourth execution submodule, used to determine whether the first team composed of the agents of all current nodes can complete the first specified task; A fifth execution submodule, configured to use the first team as the target team if the first team can complete the first designated task; The sixth execution submodule is used to use the agent of the current bottom-level node as the new basic agent if the first team cannot complete the first designated task, and return to the second execution submodule until the first team can complete the first designated task.

6. An electronic device, characterized in that: It includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps in the method for forming an intelligent computing center intelligent agent team for inclusive computing power as described in any one of claims 1 to 4.

7. A readable storage medium, characterized in that: The readable storage medium stores programs or instructions, which, when executed by the processor, implement the steps in the method for forming an intelligent computing center intelligent agent team for inclusive computing power as described in any one of claims 1 to 4.

8. A computer program product, characterized in that It includes computer instructions, which, when executed by a processor, implement the steps of the method for forming an intelligent computing center intelligent agent team for universal computing power as described in any one of claims 1 to 4.

Images