Quantum calculation assistance system and information generation method

Abstract

The invention discloses a quantum computing auxiliary system based on a large language model and an information generation method. The system comprises a long-term memory module, a short-term memory module, a cue word generation module, a large language model and a function chain. The long-term memory module can respond to the obtained input information and obtain target quantum information associated with the input information from a pre-stored quantum computing knowledge base. The short-term memory module can acquire historical session information associated with input information in response to the input information. And the cue word generation module can generate a target cue word according to the input information, the target quantum information and the historical session information. The large language model can process the target cue word to obtain initial output information, or call the function chain to process the target cue word to obtain the initial output information. Therefore, by combining the large language model and the quantum computing knowledge base, enough effective information is provided, and the input information reply preciseness is ensured.

Description

Technical Field

[0001] The present application relates to the field of quantum computing, and more specifically, to a quantum computing auxiliary system and information generation method based on a large language model. Background Art

[0002] Artificial intelligence technology based on large-scale language models can quickly and conveniently respond to users' questions and has become an important way for users to learn new knowledge. However, when applying artificial intelligence technology to the field of quantum computing, it faces challenges such as insufficient effective information and insufficient rigor. Summary of the invention

[0003] The present application provides a quantum computing auxiliary system and information generation method based on a large language model.

[0004] The embodiment of the present application provides a quantum computing auxiliary system based on a large language model, the quantum computing auxiliary system includes a long-term memory module, a short-term memory module, a prompt word generation module, a large language model and a function chain;

[0005] The long-term memory module is pre-stored with a quantum computing knowledge base, and the long-term memory module is configured to obtain target quantum information associated with the input information from the quantum computing knowledge base in response to the obtained input information;

[0006] The short-term memory module is configured to obtain historical session information associated with the input information in response to the input information;

[0007] The prompt word generation module is configured to generate a target prompt word according to the input information, the target quantum information and the historical conversation information;

[0008] The large language model is configured to process the target prompt word to obtain initial output information, or to call the function chain to process the target prompt word to obtain the initial output information.

[0009] Thus, the present application provides a quantum computing auxiliary system based on a large language model, and the quantum computing auxiliary system includes a long-term memory module, a short-term memory module, a prompt word generation module, a large language model and a function chain. Among them, the quantum computing knowledge base in the long-term memory module stores a large amount of quantum computing related information, so that the quantum computing auxiliary system can quickly retrieve and provide knowledge related to the input information. The short-term memory module can better understand the user's intention and context by storing and utilizing historical session information, thereby providing more accurate answers and suggestions. The prompt word generation module can generate target prompt words according to input information, target quantum information and historical session information, and the target prompt words can guide the generation of more relevant and accurate outputs. The large language model can independently process the target prompt words generated by the prompt word generation module, and can also call the function chain to perform more complex processing on the target prompt words to adapt to different application scenarios and needs. In this way, in the field of quantum computing, the quantum computing auxiliary system provided by the present application provides enough effective information in the field of quantum computing by combining a large language model and a quantum computing knowledge base, ensuring the rigor of the reply to the user's input information, and promoting the popularization and progress of quantum computing technology.

[0010] In some embodiments, the quantum computing knowledge base includes a plurality of quantum information, each of the quantum information has a corresponding information index, and the long-term memory module is configured as follows:

[0011] Processing the input information according to a preset retrieval technology to obtain an input information index;

[0012] Respectively obtaining the similarity between the input information index and each of the information indexes;

[0013] The target quantum information is determined according to a preset similarity threshold and the similarity.

[0014] In this way, the pre-stored quantum computing database includes multiple quantum information, each quantum information has a corresponding information index. The long-term memory module can process the input information according to the preset retrieval technology to obtain the input information index. Then, the long-term memory module can obtain the similarity of the input information index and each information index respectively. Finally, the long-term memory module can determine the target quantum information according to the preset similarity threshold and similarity. In this way, the long-term memory module can effectively retrieve knowledge related to the input information from the quantum computing database, thereby supporting the quantum computing auxiliary system to provide accurate and useful information to users.

[0015] In some embodiments, the short-term memory module is configured to:

[0016] The target output information and the input information are stored to update the historical session information, wherein the target output information is obtained according to the initial output information.

[0017] In this way, the short-term memory module can store the target output information and input information to update the historical conversation information. In this way, the short-term memory module can achieve the effect of multiple rounds of dialogue with the user by storing and updating the historical conversation information, supporting the quantum computing-assisted system to provide a more coherent, efficient and personalized interactive experience.

[0018] In some embodiments, the short-term memory module is configured to:

[0019] Based on a preset storage text length threshold, the updated historical conversation information is summarized and refined.

[0020] In this way, based on the preset storage text length threshold, the short-term memory module can summarize and refine the updated historical session information. In this way, taking into account factors such as the system's storage capacity, processing speed, and user experience, a preset storage text length threshold is set. Based on the summary and refinement processing of the preset storage text length threshold, the short-term memory module can effectively manage historical session information, balance the use of storage space and the quality of interactive experience, and ensure that the system can quickly retrieve and use relevant information.

[0021] In some embodiments, the prompt word generation module is configured with preset function chain description information, and the prompt word generation module is configured to:

[0022] The input information, the target quantum information, the preset function chain description information and the historical session information are spliced ​​to obtain a target spliced ​​body;

[0023] The target prompt word is obtained according to the target splicing body.

[0024] In this way, the prompt word generation module can splice the input information, the target quantum information, the preset function chain description information configured in the prompt word generation module and the historical session information to obtain the target splice body, and the preset function chain description information describes the functions and operations that different function chains can perform. Then, the prompt word generation module can obtain the target prompt word according to the target splice body. In this way, through this target prompt word generation method based on multi-source information splicing, the prompt word generation module can provide a richer and more specific guidance to the large language model or function chain, thereby improving the accuracy and relevance of the output information.

[0025] In certain embodiments, the functional chain includes a quantum circuit generation chain, a network search chain, a visualization chain, and a quantum circuit execution chain;

[0026] The quantum circuit generation chain is configured to generate a quantum circuit according to the target prompt word;

[0027] The network search chain is configured to extract keywords associated with the target prompt word according to the target prompt word, so as to call a network interface to perform retrieval according to the keyword to obtain retrieval data;

[0028] The visualization chain is configured to perform visualization drawing according to the target prompt word to obtain visualization data;

[0029] The quantum circuit execution chain is configured to execute the quantum circuit in the target prompt word to obtain quantum circuit output data.

[0030] In this way, the functional chain includes a quantum circuit generation chain, a network search chain, a visualization chain, and a quantum circuit execution chain. Among them, the quantum circuit generation chain can generate a quantum circuit according to the target prompt word. The network search chain can extract keywords associated with the target prompt word according to the target prompt word, so as to call the network interface for retrieval according to the keywords and obtain retrieval data. The visualization chain can perform visualization drawing according to the target prompt word to obtain visualization data. The quantum circuit execution chain can execute the quantum circuit in the target prompt word to obtain quantum circuit output data. In this way, different functional chains focus on different tasks. Through the collaborative work of these functional chains, the quantum computing auxiliary system can provide comprehensive quantum computing auxiliary services and provide efficient and accurate quantum computing solutions.

[0031] In certain embodiments, the quantum computing-assisted system further comprises an output polishing chain;

[0032] The output polishing chain is configured to process the input initial output information in response to the input initial output information to obtain target output information.

[0033] Thus, the quantum computing assistance system also includes an output polishing chain, which can respond to the input initial output information, process the initial output information, and obtain target output information. In this way, the initial output information obtained by processing the large language model is suitable for machine reading, and the output polishing chain can process the initial output information to obtain target output information suitable for user reading, ensuring that the information provided by the quantum computing assistance system is both accurate and easy for users to understand and apply, thereby enhancing the overall value and practicality of the system.

[0034] In certain embodiments, the quantum computing knowledge base is obtained based on the following steps:

[0035] Storing multiple collected quantum information;

[0036] Each of the quantum information is processed based on a preset retrieval technology to generate multiple information indexes to obtain the quantum computing knowledge base, wherein each of the quantum information corresponds to one of the information indexes.

[0037] In this way, the collected multiple quantum information are first stored. Then, each quantum information is processed based on the preset retrieval technology to generate multiple information indexes to obtain a quantum computing knowledge base, where each quantum information corresponds to an information index. In this way, the constructed quantum computing knowledge base can effectively organize and manage knowledge related to quantum computing and provide users with fast and accurate information retrieval.

[0038] The embodiment of the present application provides an information generation method, the information generation method is based on the quantum computing auxiliary system as described above, and the method includes:

[0039] In response to the acquired input information, acquiring target quantum information associated with the input information from the quantum computing knowledge base;

[0040] Acquire historical session information associated with the input information;

[0041] Generate a target prompt word according to the input information, the target quantum information and the historical conversation information;

[0042] The target prompt word is processed to obtain initial output information, or the function chain is called to process the target prompt word to obtain the initial output information.

[0043] In this way, in response to the input information obtained, the quantum computing auxiliary system obtains the target quantum information associated with the input information from the quantum computing knowledge base. Next, the quantum computing auxiliary system obtains the historical session information associated with the input information. Then, the quantum computing auxiliary system generates the target prompt word according to the input information, the target quantum information and the historical session information. Finally, the quantum computing auxiliary system processes the target prompt word to obtain the initial output information, or calls the function chain to process the target prompt word to obtain the initial output information. In this way, the quantum computing auxiliary system provides professional knowledge and information based on the field of quantum computing by combining the rich resources of the quantum computing knowledge base and the intelligent processing capabilities of the large language model, and supports users to solve problems and execute tasks efficiently.

[0044] In some embodiments, in response to the acquired input information, acquiring target quantum information associated with the input information from the quantum computing knowledge base includes:

[0045] Processing the input information according to a preset retrieval technology to obtain an input information index;

[0046] Respectively obtaining the similarity between the input information index and each of the information indexes;

[0047] The target quantum information is determined according to a preset similarity threshold and the similarity.

[0048] In this way, the quantum computing auxiliary system processes the input information according to the preset retrieval technology to obtain the input information index. Then, the quantum computing auxiliary system obtains the similarity between the input information index and each of the information indexes. Finally, the quantum computing auxiliary system determines the target quantum information according to the preset similarity threshold and the similarity. In this way, through the similarity-based retrieval method, the quantum computing auxiliary system can effectively find the target quantum information related to the user input information from the quantum computing knowledge base, thereby providing professional knowledge and information based on the field of quantum computing, and supporting users to solve problems and perform tasks efficiently.

[0049] In certain embodiments, the method further comprises:

[0050] The target output information and the input information are stored to update the historical session information, wherein the target output information is obtained according to the initial output information.

[0051] In this way, the quantum computing assisted system stores the target output information and the input information to update the historical session information. In this way, by storing and updating the historical session information, the quantum computing assisted system can provide a more coherent and efficient interactive experience.

[0052] In certain embodiments, the method further comprises:

[0053] Based on a preset storage text length threshold, the updated historical conversation information is summarized and refined.

[0054] In this way, based on the preset storage text length threshold, the quantum computing assisted system summarizes and refines the updated historical session information. In this way, the threshold-based summary and refinement process helps the quantum computing assisted system to maintain efficiency and practicality when processing large amounts of historical data.

[0055] In some embodiments, generating a target prompt word according to the input information, the target quantum information and the historical session information includes:

[0056] The input information, the target quantum information, the preset function chain description information and the historical session information are spliced ​​to obtain a target spliced ​​body;

[0057] The target prompt word is obtained according to the target splicing body.

[0058] In this way, the quantum computing auxiliary system splices the input information, target quantum information, preset function chain description information and historical session information to obtain the target splice. Then, the quantum computing auxiliary system obtains the target prompt word based on the target splice. In this way, through this target prompt word generation method based on multi-source information splicing, the prompt word generation module can provide a richer and more specific guidance to the large language model or function chain, thereby improving the accuracy and relevance of the output information.

[0059] In certain embodiments, the method further comprises:

[0060] In response to the inputted initial output information, the initial output information is processed to obtain target output information.

[0061] In this way, in response to the input initial output information, the quantum computing auxiliary system processes the initial output information to obtain the target output information. In this way, the quantum computing auxiliary system can provide clear, accurate and easy-to-understand target output information.

[0062] Additional aspects and advantages of the embodiments of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the embodiments of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:

[0064] Figure 1 is a schematic diagram of the structure of a quantum computing auxiliary system in certain embodiments of the present application;

[0065] Figure 2 is a schematic diagram of the architecture of a quantum computing auxiliary system according to certain embodiments of the present application;

[0066] Figure 3 is a schematic diagram of the workflow of the long-term memory module of certain embodiments of the present application;

[0067] Figure 4 is a schematic diagram of the workflow of the short-term memory module of certain embodiments of the present application;

[0068] Figure 5 It is a schematic diagram of the workflow of the functional chain of certain embodiments of the present application;

[0069] Figure 6 It is one of the flowcharts of the information generation method of certain embodiments of the present application;

[0070] Figure 7 This is the second flow chart of the information generation method of certain implementation modes of the present application;

[0071] Figure 8 This is the third flow chart of the information generation method of certain implementation modes of the present application;

[0072] Fig. 9 This is the fourth flow chart of the information generation method of certain implementation modes of the present application;

[0073] Fig.10 This is the fifth flow chart of the information generation method of certain implementation modes of the present application;

[0074] Fig.11 This is the sixth flow chart of the information generation method of certain implementation methods of the present application. DETAILED DESCRIPTION

[0075] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals represent the same or similar elements or elements having the same or similar functions from beginning to end. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the embodiments of the present application, and cannot be understood as limiting the embodiments of the present application.

[0076] Artificial intelligence technology based on large-scale language models has become an important way for users to acquire new knowledge in terms of quickly and conveniently answering user questions and providing information. This technology can understand user questions through natural language processing and machine learning algorithms, and retrieve and generate relevant answers from large amounts of data. Users can obtain information on a variety of topics, from daily knowledge to in-depth discussions in professional fields, by simply asking questions.

[0077] However, quantum computing is a new computing paradigm that uses the superposition and entanglement of quantum bits (qubits) to perform calculations based on the principles of quantum mechanics. As a highly specialized and technology-intensive field, quantum computing has a knowledge system and application scenarios that are significantly different from traditional computer science and information processing. When applying artificial intelligence technology to the field of quantum computing, the training of large models is often not expected to be timely, and it is impossible to obtain enough effective information in the field of quantum computing.

[0078] In addition, the field of quantum computing requires very high rigor and accuracy. Quantum computing involves complex physical principles and mathematical models, and its results often require rigorous verification and experimental support. When processing this information, artificial intelligence systems need to ensure the rigor and accuracy of their output to avoid misleading users.

[0079] Based on the above questions, please refer to Figure 1The embodiment of the present application provides a quantum computing auxiliary system 100 based on a large language model 40. The quantum computing auxiliary system 100 includes a long-term memory module 10, a short-term memory module 20, a prompt word generation module 30, a large language model 40 and a function chain 50.

[0080] The long-term memory module 10 is pre-stored with a quantum computing knowledge base, and the long-term memory module 10 is configured to obtain target quantum information associated with the input information from the quantum computing knowledge base in response to the acquired input information;

[0081] The short-term memory module 20 is configured to obtain historical session information associated with the input information in response to the input information;

[0082] The prompt word generation module 30 is configured to generate a target prompt word according to the input information, the target quantum information and the historical conversation information;

[0083] The large language model 40 is configured to process the target prompt word to obtain initial output information, or to call the function chain 50 to process the target prompt word to obtain initial output information.

[0084] Specifically, the large language model 40 refers to a large-scale language model (LLM), which is a deep learning model in the field of artificial intelligence and an advanced language processing system based on artificial intelligence technology that can understand and generate natural language text. This type of model is trained on massive texts, has a huge parameter scale, has multi-task capabilities, and has strong context understanding capabilities, so that it can complete new tasks based on a small number of example samples or even zero samples.

[0085] The quantum computing knowledge base can be seen as a system dedicated to storing and retrieving information related to quantum computing. The quantum computing knowledge base includes a large amount of text, data, algorithms, research papers, hardware technical documents, etc., covering all aspects of quantum computing, such as quantum algorithms, quantum programming languages, quantum hardware technology, quantum encryption, quantum simulation, etc.

[0086] Quantum circuit is a concept in quantum computing that is used to describe the order of operations between quantum bits and the way quantum gates are connected. A quantum circuit consists of a series of quantum gates that act on quantum bits in a certain order to implement specific quantum algorithms or operations.

[0087] Quantum gate is the basic operation unit of quantum computing, which can transform the state of quantum bits and realize the processing of quantum information. The function of quantum gate is to create entanglement between quantum bits or change the superposition state of quantum bits.

[0088] Input information can be regarded as text information converted from data information such as text information, voice information, and image information input by a user or other input device.

[0089] Target quantum information refers to the information in the quantum computing field in the quantum computing knowledge base that is most relevant to the information input by the user (or other input device), and can answer the user's questions or meet the user's needs.

[0090] Historical session information refers to the records of the interaction between the user (or other input device) and the quantum computing assistance system 100 over a period of time. These records usually include questions asked by the user, answers generated by the quantum computing assistance system 100, and the like.

[0091] The target prompt word refers to the input text provided to the large language model 40, which is intended to guide the large language model 40 to generate a specific output or perform a specific task. In the implementation mode of the present application, prompt word engineering is required to improve the quality of the answer of the large language model 40 on a specific task. Prompt word engineering refers to the process of designing and optimizing prompt words. Common optimization methods include: adding additional contextual information, giving several examples for guidance, using fixed templates, etc.

[0092] The initial output information refers to the output information output by the quantum computing auxiliary system 100 that can be understood by the computer device. Only after subsequent operations can it become the target output information that can be understood by the user (or other input device).

[0093] See also Figure 2 , the system architecture diagram of the quantum computing auxiliary system 100 is shown in the figure. Among them, the quantum computing auxiliary system 100 includes an agent, which is the core of the quantum computing auxiliary system 100. An agent refers to an autonomous software entity that can use a large model to interact with the external environment and perform specific tasks. It can make decisions based on the information received, plan actions and execute them, and ultimately solve complex problems in a specific field. In the implementation mode of the present application, the agent will independently decide whether to use the function chain 50 and which function chain 50 to use based on the problem and the available function chain 50 (Chain), thereby realizing the disassembly of user problems and the arrangement of tasks, and finally completing the user's complex problems in the field of quantum computing.

[0094] The agent includes many function chains 50 (Chain) for implementing different functions, which can also be called chains. A chain refers to a sequence of operations or components used to answer a specific question or perform a specific, specific operation, accompanied by a clear description of the chain's function. Multiple chains can be coordinated and combined by the agent to work together to build more complex artificial intelligence applications.

[0095] In the quantum computing auxiliary system 100, the next layer of the function chain 50 is the large language model 40, the long-term memory module 10 and the short-term memory module 20. The large language model 40 outputs text purely according to the input prompt words, while the long-term memory module 10 and the short-term memory module 20 can realize the storage function of the quantum computing auxiliary system 100. The long-term memory module 10 is responsible for storing large sections of quantum computing related documents and materials for the large language model 40 to use as background information for answers. The short-term memory module 20 is responsible for accessing the historical chat records of each conversation, which is used to support the multi-round conversation capability of the large language model 40 and the user (or other input device).

[0096] In this way, the quantum computing auxiliary system 100 makes full use of the contextual understanding ability of the large language model 40 and the task scheduling ability of the intelligent agent, realizes the effect of multiple rounds of dialogue with the user (or other input device) through the short-term memory module 20, and provides the user (or other input device) with high-precision replies in vertical fields in combination with the long-term memory module 10. This quantum computing auxiliary system 100 combined with the long-term memory module 10 can significantly improve the accuracy of answering questions related to the field of quantum computing, and effectively alleviate the probability of the large model outputting wrong answers. In addition, this quantum computing auxiliary system 100 combined with the short-term memory module 20 can significantly improve the user question and answer experience in the same session. This quantum computing auxiliary system 100 can also greatly enrich the functions of the quantum computing auxiliary system 100 and support users to complete more specific tasks by combining various specific quantum computing function chains 50, such as online query of information, generation of quantum circuits, and checking quantum circuit grammar.

[0097] In addition, the quantum computing auxiliary system 100 also has a series of auxiliary functions, including a log module and a monitoring module.

[0098] In summary, the present application provides a quantum computing auxiliary system 100 based on a large language model 40, and the quantum computing auxiliary system 100 includes a long-term memory module 10, a short-term memory module 20, a prompt word generation module 30, a large language model 40 and a function chain 50. Among them, the quantum computing knowledge base in the long-term memory module 10 stores a large amount of quantum computing related information, so that the quantum computing auxiliary system 100 can quickly retrieve and provide knowledge related to the input information. The short-term memory module 20 can better understand the user's intention and context by storing and utilizing historical session information, thereby providing more accurate answers and suggestions. The prompt word generation module 30 can generate a target prompt word based on the input information, the target quantum information and the historical session information, and the target prompt word can guide the generation of more relevant and accurate outputs. The large language model 40 can independently process the target prompt word generated by the prompt word generation module 30, and can also call the function chain 50 to perform more complex processing on the target prompt word to adapt to different application scenarios and needs. In this way, in the field of quantum computing, the quantum computing auxiliary system 100 provided by the present application provides sufficient effective information in the field of quantum computing by combining the large language model 40 and the quantum computing knowledge base, ensures the rigor of responding to user input information, and promotes the popularization and advancement of quantum computing technology.

[0099] In some embodiments, the quantum computing knowledge base includes a plurality of quantum information, each quantum information has a corresponding information index, and the long-term memory module 10 is configured as follows:

[0100] Process the input information according to a preset retrieval technology to obtain an input information index;

[0101] Obtain the input information index and the similarity of each information index respectively;

[0102] The target quantum information is determined according to a preset similarity threshold and similarity.

[0103] Specifically, the preset retrieval technology refers to the use of mathematical models and algorithms to convert text information into operable and comparable vector representations, thereby achieving efficient retrieval and information extraction. In the implementation of the present application, a vector model is used as the preset retrieval technology.

[0104] See also Figure 3In the implementation mode of the present application, the pre-stored quantum computing database includes multiple quantum information, and each quantum information has a corresponding information index. The long-term memory module 10 can process the input information according to the preset retrieval technology to obtain the input information index. Then, the long-term memory module 10 can obtain the similarity of the input information index and each information index respectively. Finally, the long-term memory module 10 can determine the target quantum information according to the preset similarity threshold and the similarity, that is, by comparing the various similarities obtained and the preset similarity threshold, the quantum information in the quantum computing database whose similarity is greater than the preset similarity threshold is the target quantum information.

[0105] In this way, the long-term memory module 10 can effectively retrieve knowledge related to the input information from the quantum computing database, thereby supporting the quantum computing auxiliary system 100 to provide accurate and useful information to the user.

[0106] See also Figure 4 In some embodiments, the short-term memory module 20 is configured to store the target output information and the input information to update the historical session information.

[0107] Specifically, the short-term memory module 20 can store the target output information and the input information to update the historical session information. Each time the quantum computing auxiliary system 100 processes the input information input by the user (or other input device) to obtain the target output information, the short-term memory module 20 will combine the historical session information previously stored in the short-term memory module 20 with the input information and the target output information of this time. Subsequently, the short-term memory module 20 will store the historical session information obtained after the combination.

[0108] In this way, the short-term memory module 20 achieves the effect of multiple rounds of dialogue with the user by storing and updating historical conversation information, supporting the quantum computing assisted system 100 to provide a more coherent, efficient and personalized interactive experience.

[0109] Please refer to Figure 4 In some embodiments, the short-term memory module 20 is configured to summarize and refine the updated historical conversation information based on a preset storage text length threshold.

[0110] Specifically, the preset storage text length threshold refers to a value preset in the short-term memory module 20 and is used to limit the maximum length of the stored text data.

[0111] In the quantum computing auxiliary system 100, since the length of the text in the large language model 40 is usually limited, in order to avoid text exceeding the limit, it is generally necessary to add a preset storage text length threshold to determine whether the text exceeds the limit. The preset storage text length threshold depends on the text length that the large language model 40 can receive. If the session record exceeds the preset storage text length threshold, the short-term memory module 20 will summarize and refine the historical session information, and update the shortened text to the historical session information as the historical session information. It should be noted that the historical session information that is usually summarized and refined is the historical session information recorded earlier.

[0112] In this way, taking into account factors such as the system's storage capacity, processing speed, and user experience, a preset storage text length threshold is set. Based on the summary and refinement of the preset storage text length threshold, the short-term memory module 20 can effectively manage historical session information, balance the use of storage space and the quality of the interactive experience, while ensuring that the system can quickly retrieve and utilize relevant information.

[0113] Please refer to Figure 1 In some embodiments, the prompt word generation module 30 is configured with preset function chain description information, and the prompt word generation module 30 is configured as follows:

[0114] The input information, the target quantum information, the preset function chain description information and the historical session information are spliced ​​to obtain the target spliced ​​body;

[0115] The target prompt word is obtained according to the target splice.

[0116] Specifically, the preset function chain description information refers to the description of the functions of different function chains. For example, the quantum circuit generation chain: generates a quantum circuit according to the target prompt word. The network search chain: extracts keywords associated with the target prompt word according to the target prompt word, and calls the network interface to search according to the keywords to obtain the search data. The visualization chain: performs visualization drawing according to the target prompt word to obtain visualization data. The quantum circuit execution chain: executes the quantum circuit in the target prompt word to obtain the quantum circuit output data.

[0117] The prompt word generation module 30 can splice the input information, the target quantum information, the preset function chain description information configured in the prompt word generation module 30 and the historical session information to obtain the target splice, and the preset function chain description information describes the functions and operations that can be performed by different function chains 50. Then, the prompt word generation module 30 can obtain the target prompt word according to the target splice.

[0118] In this way, through this target prompt word generation method based on multi-source information splicing, the prompt word generation module 30 can provide a richer and more specific guidance to the large language model 40 or the function chain 50, thereby improving the accuracy and relevance of the output information.

[0119] In certain embodiments, the functional chain 50 includes a quantum circuit generation chain, a network search chain, a visualization chain, and a quantum circuit execution chain;

[0120] The quantum circuit generation chain is configured to generate a quantum circuit according to a target prompt word;

[0121] The network search chain is configured to extract keywords associated with the target prompt word according to the target prompt word, so as to call the network interface to perform retrieval according to the keyword and obtain retrieval data;

[0122] The visualization chain is configured to perform visualization drawing according to the target prompt word to obtain visualization data;

[0123] The quantum circuit execution chain is configured to execute the quantum circuit in the target prompt word and obtain quantum circuit output data.

[0124] Specifically, the function chain 50 includes a quantum circuit generation chain, a network search chain, a visualization chain, and a quantum circuit execution chain. The quantum circuit generation chain can generate a quantum circuit according to a target prompt word. The network search chain can extract keywords associated with the target prompt word according to the target prompt word, so as to call a network interface for retrieval according to the keywords and obtain retrieval data. The visualization chain can perform visualization drawing according to the target prompt word and obtain visualization data. The quantum circuit execution chain can execute the quantum circuit in the target prompt word and obtain quantum circuit output data.

[0125] See also Figure 5 , Figure 5 It is a schematic diagram of the structure of the function chain 50. Among them, the input information of the function chain 50 may be the input information or the target output information of the previous function chain 50, etc. The quantum computing auxiliary system 100 will splice the input information of this function chain 50 with the interface description of the current function chain 50, obtain the complete target prompt word and send it to the large language model 40 to construct a request to call the relevant interface, so as to obtain the final result. Whether a function chain 50 needs to be used is determined by the large language model 40 of the intelligent agent.

[0126] In this way, different functional chains 50 focus on different tasks, and through the coordinated work of these functional chains 50, the quantum computing assistance system 100 can provide comprehensive quantum computing assistance services and provide efficient and accurate quantum computing solutions.

[0127] Please refer to Figure 1, in certain embodiments, the quantum computing auxiliary system 100 further includes an output polishing chain 60;

[0128] The output polishing chain 60 is configured to process the input initial output information in response to the input initial output information to obtain target output information.

[0129] Specifically, the quantum computing auxiliary system 100 further includes an output polishing chain 60, which can process the input initial output information in response to the input initial output information to obtain target output information. Since the initial output information directly obtained by the quantum computing auxiliary system 100 is usually only suitable for machine reading, the initial output information needs to be processed to obtain the target output information for user reading.

[0130] In this way, the initial output information processed by the large language model 40 is suitable for machine reading, and the output polishing chain 60 can process the initial output information to obtain target output information suitable for user reading, ensuring that the information provided by the quantum computing auxiliary system 100 is both accurate and easy for users to understand and apply, thereby enhancing the overall value and practicality of the system.

[0131] In certain embodiments, the quantum computing knowledge base is obtained based on the following steps:

[0132] Storing multiple collected quantum information;

[0133] Each quantum information is processed separately based on a preset retrieval technology to generate multiple information indexes to obtain a quantum computing knowledge base, wherein each quantum information corresponds to an information index.

[0134] Specifically, quantum information refers to various types of information related to quantum computing, such as quantum algorithms, quantum programming languages, quantum hardware technology, quantum encryption, and quantum simulation.

[0135] First, the collected multiple quantum information are stored. Then, each quantum information is processed based on the vector model to generate multiple information indexes to obtain a quantum computing knowledge base, wherein each quantum information corresponds to an information index.

[0136] In this way, the constructed quantum computing knowledge base can effectively organize and manage knowledge related to quantum computing and provide users with fast and accurate information retrieval.

[0137] See also Figure 6 The embodiment of the present application provides an information generation method, which is based on the quantum computing auxiliary system 100 as described above, and includes:

[0138] 01: In response to the acquired input information, acquiring target quantum information associated with the input information from a quantum computing knowledge base;

[0139] 02: Get the historical session information associated with the input information;

[0140] 03: Generate target prompt words based on input information, target quantum information and historical conversation information;

[0141] 04: Process the target prompt word to obtain initial output information, or call the function chain to process the target prompt word to obtain initial output information.

[0142] The embodiment of the present application also provides a computer device, including a memory and a processor. The information generation method of the embodiment of the present application can be implemented by the computer device of the embodiment of the present application. Specifically, a computer program is stored in the memory, and the processor is used to obtain the target quantum information associated with the input information from the quantum computing knowledge base in response to the input information obtained. And obtain the historical session information associated with the input information. The processor is also used to generate a target prompt word based on the input information, the target quantum information and the historical session information. And process the target prompt word to obtain initial output information, or call the function chain 50 to process the target prompt word to obtain initial output information.

[0143] The embodiment of the present application also provides an information generating device. The information generating method of the embodiment of the present application can be implemented by the information generating device of the embodiment of the present application. Specifically, the information generating device includes an acquisition module, a generation module and a processing module. The acquisition module is used to obtain the target quantum information associated with the input information from the quantum computing knowledge base in response to the acquired input information. The acquisition module is also used to obtain the historical session information associated with the input information. The generation module is used to generate the target prompt word according to the input information, the target quantum information and the historical session information. The processing module is used to process the target prompt word to obtain the initial output information, or call the function chain 50 to process the target prompt word to obtain the initial output information.

[0144] Specifically, in response to the input information obtained, the quantum computing auxiliary system 100 obtains the target quantum information associated with the input information from the quantum computing knowledge base. Next, the quantum computing auxiliary system 100 obtains the historical session information associated with the input information. Then, the quantum computing auxiliary system 100 generates the target prompt word according to the input information, the target quantum information and the historical session information. Finally, the quantum computing auxiliary system 100 processes the target prompt word to obtain the initial output information, or calls the function chain 50 to process the target prompt word to obtain the initial output information.

[0145] In this way, the quantum computing assistance system 100 provides professional knowledge and information based on the field of quantum computing by combining the rich resources of the quantum computing knowledge base and the intelligent processing capabilities of the large language model 40, supporting users in efficient problem solving and task execution.

[0146] See also Figure 7 In some embodiments, step 01 (in response to the acquired input information, acquiring target quantum information associated with the input information from a quantum computing knowledge base) includes:

[0147] 011: Process the input information according to the preset retrieval technology to obtain the input information index;

[0148] 012: Obtain the input information index and the similarity of each information index respectively;

[0149] 013: Determine the target quantum information according to the preset similarity threshold and similarity.

[0150] In some embodiments, the processing module is used to process the input information according to a preset retrieval technology to obtain an input information index. The acquisition module is used to obtain the similarity of the input information index and each information index respectively. The determination module is used to determine the target quantum information according to a preset similarity threshold and similarity.

[0151] In some embodiments, the processor is further configured to process the input information according to a preset search technology to obtain an input information index, obtain the similarity between the input information index and each information index, and determine the target quantum information according to a preset similarity threshold and the similarity.

[0152] Specifically, the quantum computing auxiliary system 100 processes the input information according to the preset retrieval technology to obtain the input information index. Then, the quantum computing auxiliary system 100 obtains the similarity of the input information index and each information index respectively. Finally, the quantum computing auxiliary system 100 determines the target quantum information according to the preset similarity threshold and the similarity.

[0153] In this way, through the similarity-based retrieval method, the quantum computing auxiliary system 100 can effectively find the target quantum information related to the user input information from the quantum computing knowledge base, thereby providing professional knowledge and information based on the field of quantum computing, supporting users to solve problems and execute tasks efficiently.

[0154] See also Figure 8 In some embodiments, the method further comprises:

[0155] 05: Store the target output information and input information to update the historical session information.

[0156] In some implementations, the processing module is further configured to store the target output information and the input information to update the historical session information.

[0157] In some implementations, the processor is further configured to store the target output information and the input information to update the historical session information.

[0158] Specifically, the quantum computing auxiliary system 100 stores the target output information and the input information to update the historical session information.

[0159] In this way, by storing and updating historical session information, the quantum computing-assisted system 100 can provide a more coherent and efficient interactive experience.

[0160] See also Fig. 9 In some embodiments, the method further comprises:

[0161] 06: Based on the preset storage text length threshold, the updated historical session information is summarized and refined.

[0162] In some implementations, the processing module is further configured to summarize and refine the updated historical conversation information based on a preset storage text length threshold.

[0163] In some implementations, the processor is further configured to summarize and refine the updated historical conversation information based on a preset storage text length threshold.

[0164] Specifically, based on a preset storage text length threshold, the quantum computing auxiliary system 100 summarizes and refines the updated historical session information.

[0165] In this way, the threshold-based summary and refinement process helps the quantum computing auxiliary system 100 to remain efficient and practical when processing a large amount of historical data.

[0166] See also Fig.10 In some embodiments, step 03 (generating a target prompt word according to input information, target quantum information and historical session information) includes:

[0167] 031: splicing the input information, target quantum information, preset function chain description information and historical session information to obtain the target splicing body;

[0168] 032: Get the target prompt word based on the target splice.

[0169] In some embodiments, the processing module is used to perform splicing processing on the input information, the target quantum information, the preset function chain description information and the historical session information to obtain the target splicing body. The processing module is also used to obtain the target prompt word according to the target splicing body.

[0170] In some embodiments, the processor is further used to perform splicing processing on the input information, the target quantum information, the preset function chain description information and the historical session information to obtain a target splicing body, and to obtain a target prompt word according to the target splicing body.

[0171] Specifically, the quantum computing assistance system 100 performs splicing processing on the input information, the target quantum information, the preset function chain description information and the historical session information to obtain the target splicing body. Next, the quantum computing assistance system 100 obtains the target prompt word according to the target splicing body.

[0172] In this way, through this target prompt word generation method based on multi-source information splicing, the prompt word generation module 30 can provide a richer and more specific guidance to the large language model 40 or the function chain 50, thereby improving the accuracy and relevance of the output information.

[0173] See also Fig.11 In some embodiments, the method further comprises:

[0174] 07: In response to the input initial output information, the initial output information is processed to obtain target output information.

[0175] In some embodiments, the processing module is used to process the initial output information in response to the input initial output information to obtain the target output information.

[0176] In some embodiments, the processor is used to process the initial output information in response to the input initial output information to obtain the target output information.

[0177] Specifically, in response to the input initial output information, the quantum computing auxiliary system 100 processes the initial output information to obtain target output information.

[0178] In this way, the quantum computing-assisted system 100 can provide clear, accurate and easy-to-understand target output information.

[0179] The following is an example to illustrate the above information generation method. The input information is "Please search for a brief usage of QCIS quantum computing syntax" or "Please generate a GHZ quantum circuit for preparing quantum bits."

[0180] At the initial stage of multiple rounds of interaction related to quantum computing, the user makes a request of “please search for a brief usage of QCIS quantum computing grammar”. Since the conversation has just started, the short-term memory module 20 does not store the historical conversation records, so the answer process relies on the long-term memory module 10 to generate the prompt word.

[0181] The long-term memory module 10 in the quantum computing auxiliary system 100 vectorizes the request "Please search for a brief usage of QCIS quantum computing syntax" by using a vector model, and then calculates the similarity with the index vector in the quantum computing knowledge base, extracts the text fragments with a similarity greater than a preset similarity threshold, and splices them with the user's original input "Please search for a brief usage of QCIS quantum computing syntax" and the preset function chain 50 description information, and processes them in the prompt word generation module 30 to obtain the first target prompt word for input into the large language model 40. After reading the first target prompt word, the large language model 40 determines that the network search chain needs to be used, constructs a call request according to the first target prompt word, and calls the call interface of the network search chain to call the network search chain to search for relevant web pages to obtain the first initial output information. The output polishing chain 60 then processes this first initial output information to obtain the first target output information that the user can read, that is, the sorted QCIS usage instructions and web page links. The quantum computing auxiliary system 100 finally provides the sorted QCIS usage instructions and web page links. Then, the short-term memory module 20 will store the input information "Please search for the brief usage of QCIS quantum computing syntax" and the first target output information.

[0182] Next, the user makes a request of "please generate a GHZ quantum circuit for preparing quantum bits". At this time, the system will combine the long-term memory module 10 and the short-term memory module 20 to generate a prompt word. The quantum computing auxiliary system 100 also uses the vector model to vectorize "please generate a GHZ quantum circuit for preparing quantum bits" and obtains the target quantum information related to "please generate a GHZ quantum circuit for preparing quantum bits" in the quantum computing knowledge base.

[0183] The quantum computing auxiliary system 100 splices "Please generate a quantum bit GHZ quantum circuit", the target quantum information, the description information of the preset function chain 50 and the historical conversation information to obtain the target splice. Then, the quantum computing auxiliary system 100 obtains the target prompt word according to the target splice, and inputs the target prompt word into the large language model 40. The large language model 40 obtains the complete conversation history information, and starts to call the quantum circuit generation chain, generates the required quantum circuit and displays it to the user. Through the analysis of the short-term memory module 20, the quantum computing auxiliary system 100 understands that the previous communication focuses on the QCIS quantum programming language, so when generating quantum circuits, the QCIS language format will be used first.

[0184] Similarly, the quantum computing auxiliary system 100 will record the information of this session, first combine the information of this session with the historical session information, and then determine whether the length of the combined text exceeds the preset storage text length threshold. When the length of the combined text exceeds the preset storage text length threshold, the early historical session records are summarized and refined, and the shortened text is updated to the historical session as session information. If the length of the combined text does not exceed the preset storage text length threshold, no processing is required, and the short-term memory module 20 directly stores the combined text.

[0185] The present application also provides a computer-readable storage medium containing a computer program. When the computer program is executed by one or more processors, the one or more processors execute the method of the present application.

[0186] It is understood that a computer program includes computer program code. The computer program code may be in source code form, object code form, executable file or some intermediate form. Computer readable storage media may include: any entity or device capable of carrying computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), and software distribution medium.

[0187] In the description of this specification, the descriptions with reference to the terms "specifically", "further", "particularly", "understandably", etc. are intended to mean that the specific features, structures, materials or characteristics described in conjunction with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms are not intended to refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, unless they are contradictory.

[0188] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0189] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A quantum computing auxiliary system based on a large language model, characterized in that: The quantum computing auxiliary system includes a long-term memory module, a short-term memory module, a prompt word generation module, a large language model and a function chain; The long-term memory module is pre-stored with a quantum computing knowledge base, and the long-term memory module is configured to obtain target quantum information associated with the input information from the quantum computing knowledge base in response to the obtained input information; The short-term memory module is configured to obtain historical session information associated with the input information in response to the input information; The prompt word generation module is configured to generate a target prompt word according to the input information, the target quantum information and the historical conversation information; The large language model is configured to process the target prompt word to obtain initial output information, or to call the function chain to process the target prompt word to obtain the initial output information.

2. The quantum computing auxiliary system according to claim 1, characterized in that: The quantum computing knowledge base includes a plurality of quantum information, each of which has a corresponding information index, and the long-term memory module is configured as follows: Processing the input information according to a preset retrieval technology to obtain an input information index; Respectively obtaining the similarity between the input information index and each of the information indexes; The target quantum information is determined according to a preset similarity threshold and the similarity.

3. The quantum computing auxiliary system according to claim 1, characterized in that: The short-term memory module is configured to: The target output information and the input information are stored to update the historical session information, wherein the target output information is obtained according to the initial output information.

4. The quantum computing auxiliary system according to claim 3, characterized in that: The short-term memory module is configured to: Based on a preset storage text length threshold, the updated historical conversation information is summarized and refined.

5. The quantum computing auxiliary system according to claim 3, characterized in that: The prompt word generation module is configured with preset function chain description information, and the prompt word generation module is configured as follows: The input information, the target quantum information, the preset function chain description information and the historical session information are spliced ​​to obtain a target spliced ​​body; The target prompt word is obtained according to the target splicing body.

6. The quantum computing auxiliary system according to claim 1, characterized in that: The functional chain includes a quantum circuit generation chain, a network search chain, a visualization chain and a quantum circuit execution chain; The quantum circuit generation chain is configured to generate a quantum circuit according to the target prompt word; The network search chain is configured to extract keywords associated with the target prompt word according to the target prompt word, so as to call a network interface to perform retrieval according to the keyword to obtain retrieval data; The visualization chain is configured to perform visualization drawing according to the target prompt word to obtain visualization data; The quantum circuit execution chain is configured to execute the quantum circuit in the target prompt word to obtain quantum circuit output data.

7. The quantum computing auxiliary system according to claim 6, characterized in that: The quantum computing auxiliary system also includes an output polishing chain; The output polishing chain is configured to process the input initial output information in response to the input initial output information to obtain target output information.

8. The quantum computing auxiliary system according to claim 1, characterized in that: The quantum computing knowledge base is obtained based on the following steps: Storing multiple collected quantum information; Each of the quantum information is processed based on a preset retrieval technology to generate multiple information indexes to obtain the quantum computing knowledge base, wherein each of the quantum information corresponds to one of the information indexes.

9. A method for generating information, characterized in that: The information generation method is based on the quantum computing auxiliary system according to any one of claims 1 to 8, and the method comprises: In response to the acquired input information, acquiring target quantum information associated with the input information from the quantum computing knowledge base; Acquire historical session information associated with the input information; Generate a target prompt word according to the input information, the target quantum information and the historical conversation information; The target prompt word is processed to obtain initial output information, or the function chain is called to process the target prompt word to obtain the initial output information.

10. The information generating method according to claim 9, characterized in that: The step of acquiring target quantum information associated with the input information from the quantum computing knowledge base in response to the acquired input information includes: Processing the input information according to a preset retrieval technology to obtain an input information index; Respectively obtaining the similarity between the input information index and each of the information indexes; The target quantum information is determined according to a preset similarity threshold and the similarity.

11. The information generating method according to claim 9, characterized in that: The method further comprises: The target output information and the input information are stored to update the historical session information, wherein the target output information is obtained according to the initial output information.

12. The information generating method according to claim 11, characterized in that: The method further comprises: Based on a preset storage text length threshold, the updated historical conversation information is summarized and refined.

13. The information generating method according to claim 9, characterized in that: The generating of the target prompt word according to the input information, the target quantum information and the historical conversation information comprises: The input information, the target quantum information, the preset function chain description information and the historical session information are spliced ​​to obtain a target spliced ​​body; The target prompt word is obtained according to the target splicing body.

14. The information generating method according to claim 9, characterized in that: The method further comprises: In response to the inputted initial output information, the initial output information is processed to obtain target output information.

Images