Chip type selection method, framework and circuit design method based on large language model

Abstract

The invention provides a chip type selection method based on a large language model, a framework and a circuit design method. The method comprises the following steps: SA1, obtaining interaction information to be processed; sA2, performing information extraction on the to-be-processed interaction information to obtain initial demand information; sA3, analyzing whether the initial demand information comprises an abnormal demand or not; sA4, if yes, feeding back the abnormal demand and the occurrence reason of the abnormal demand to the user, and meanwhile, interacting with the user to regenerate interaction information so as to execute the SA2 according to the newly generated interaction information; sA5, if not, demand analysis and standardization processing are carried out according to the initial demand information, and chip product demand information with data item names at least including product classification is obtained based on the initial demand information and output. According to the method, the interaction function is optimized, and the accuracy of large language model conclusion output is improved.

Description

Technical Field

[0001] The present invention relates to the field of computers, and in particular to a chip selection method, framework and circuit design method based on a large language model. Background Art

[0002] With the rapid development of artificial intelligence technology, large language models have begun to show their great potential in many fields. In the field of chip design, large language models can assist in automating multiple links in the chip design process through natural language processing and generation technology, thereby improving design efficiency and quality.

[0003] In the process of automated chip design, the first step is to use a large language model to interactively collect users' chip product requirements. However, due to the lack of corresponding design, on the one hand, it is difficult for technical personnel to express the requirements in a standardized and accurate manner, so that the description content, regardless of its length, has problems of lack of direction or specificity, and the accuracy of the conclusion output of the large language model is unsatisfactory. On the other hand, the chip product requirements provided by the front end usually have problems such as uncertainty, fuzzy presentation, and missing key weight information, which leads to the standardization of information transmission and missing key information when the information is further processed, which also affects the accuracy of the conclusion output of the large language model. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a chip selection method, a chip selection framework and a circuit design method based on a large language model, which solves the problem that the existing large language model-driven automated chip design has defects in interactive functions and reduces the accuracy of the large language model conclusion output.

[0005] According to an embodiment of the present invention, a first aspect provides a chip selection method based on a large language model, comprising: SA1. Obtain interaction information to be processed, where the interaction information to be processed is all interaction information generated by chip selection in a human-machine collaborative mode; SA2. Extracting information from the interaction information to be processed to obtain initial demand information; SA3. Analyze whether the initial demand information includes abnormal demand; the abnormal demand indicates that the initial demand information contains at least one of unreasonable demand expression, ambiguous expression, and irrelevant expression; SA4. If yes, the abnormal demand and the reason for the abnormal demand are fed back to the user, and interaction with the user is performed to regenerate interaction information, so as to execute SA2 according to the newly generated interaction information; SA5. If not, perform demand analysis and standardization processing according to the initial demand information to obtain chip product demand information whose data item names at least include product classification based on the initial demand information and output it; Among them, the first basic agent is supervised and fine-tuned by a first mixed data set including a proper noun data set, a product classification data set, a multi-round dialogue data set and a general data set to obtain an interactive agent to implement SA1 to SA5.

[0006] Optionally, the proper noun data set includes K noun data contents based on K proper nouns; The k-th noun data content includes a proper noun QA pair consisting of a correct explanation of the k-th proper noun and a question sentence based on the correct explanation of the k-th proper noun, and a system prompt word based on the proper noun QA pair; Wherein, k is a positive integer less than or equal to K, and K is a positive integer.

[0007] Optionally, the product classification data set includes N category data contents based on M primary chip categories; If the n-th category data content belongs to the m-th primary chip category, the n-th category data content includes a secondary chip category subordinate to the m-th primary chip category and a product classification QA pair consisting of a question sentence based on the secondary chip category subordinate to the m-th primary chip category, and a system prompt word based on the product classification QA pair; Wherein, M and N are positive integers, n is a positive integer less than or equal to N, and m is a positive integer less than or equal to M.

[0008] Optionally, the multi-round dialogue data set includes U multi-round dialogue data contents based on L multi-round dialogues; If the u-th multi-round dialogue data content belongs to the l-th multi-round dialogue, the u-th multi-round dialogue data content includes a multi-round dialogue QA pair consisting of a simulated question in the l-th multi-round dialogue and an answer content based on the simulated question in the l-th multi-round dialogue, a system prompt word based on the multi-round dialogue QA pair, and a historical dialogue set; Among them, L and U are positive integers, l is a positive integer less than or equal to L, u is a positive integer less than or equal to U, and in the l-th multi-round dialogue, the simulated question is generated by simulating the user's question to the randomly selected chip product manual; based on the reply content of the simulated question in the l-th multi-round dialogue, it is generated by the randomly selected chip product manual.

[0009] Optionally, in the first multi-round dialogue, the simulated question is also obtained by randomly drawing from a preset question database, and the answer content of the simulated question randomly drawn from the preset question database includes the content of the randomly drawn chip product manual; The preset question database includes a plurality of preset questions, and each preset question includes at least one of an unreasonable demand expression, a vague expression, and an irrelevant expression.

[0010] A second aspect provides a chip selection method based on a large language model, including: SB1. Obtain chip product demand information provided by an interactive agent, wherein the interactive agent executes the above-mentioned chip selection method based on a large language model; SB2, the selection agent acts as a generation module, combines the pre-set retrieval module, rearrangement module and local knowledge base, performs chip selection based on the chip product demand information, and outputs chip recommendation information; The SB2 includes: The chip product demand information is compared with the chip product data stored in the local knowledge base by the retrieval module to obtain a plurality of related documents and send them to the rearrangement module; The reordering module sorts the relevance of the multiple related documents, and sends the sorting result to the selection agent; in the sorting result, the related documents with higher relevance have smaller sorting numbers, and the related documents with sorting numbers less than the preset number value are strongly related documents; All strongly related documents are analyzed according to the chip product demand information, and chip recommendation information is output.

[0011] Optionally, construct a retrieval triple dataset to perform supervised contrastive learning training on the basic embedding model, and after the training is completed, obtain the retrieval module; Among them, the constructed retrieval ternary data set includes retrieval query elements, retrieval positive sample elements and retrieval negative sample elements; the retrieval query elements are extracted from the total demand information data; the retrieval positive sample elements are unique chip product data selected in the local knowledge base to reply to the retrieval query elements; the retrieval negative sample elements are multiple chip product data randomly extracted from the local knowledge base.

[0012] Optionally, a rearrangement ternary data set is constructed to perform supervised contrastive learning training on the basic rearrangement model, and after the training is completed, the rearrangement module is obtained; Among them, the constructed rearranged ternary data set includes rearranged query elements, rearranged positive sample elements and rearranged negative sample elements; the rearranged query element is a simulated requirement description, and the data size of the simulated requirement description is less than a preset number of bytes; the rearranged positive sample element is a unique correct answer written based on the simulated requirement description; the rearranged negative sample element is multiple incorrect answers written based on the simulated requirement description, and the similarity between the incorrect answers and the unique correct answer is greater than a preset similarity.

[0013] Optionally, any chip product data stored in the local knowledge base includes a document description of product features, a document description of product application examples, a document description of key parameters, and a document description of complete information.

[0014] Optionally, the second basic agent is trained in two stages by continuing pre-training and supervised fine-tuning training to obtain a selected agent to implement SB1 to SB2; In the continued pre-training, multi-dimensional chip text data is obtained, and text pre-processing is performed on the multi-dimensional chip text data to obtain a plurality of corpora, and the second basic agent is trained by the plurality of corpora; In the supervised fine-tuning training, the pre-trained second basic agent is continued through training with a second mixed data set including a proper noun data set, a product classification data set, a chip selection data set and a general data set to obtain a selection agent.

[0015] Optionally, the chip selection data set includes P chip matching data; The p-th chip matching data includes a chip matching QA pair consisting of an analog chip requirement description and recommended information satisfying the analog chip requirement description, and a system prompt word based on the chip matching QA pair; Wherein, the simulation chip requirement description is generated by randomly selecting a chip product manual; The recommended information that meets the analog chip requirement description is the chip product model and chip product introduction recorded in the randomly selected chip product manual.

[0016] A third aspect provides a chip selection framework based on a large language model, comprising an interactive agent driven by a large language model, a selection agent, and an evaluation agent; the interactive agent executes the chip selection method based on the large language model as described above; the selection agent executes the chip selection method based on the large language model as described above; the evaluation agent executes the following steps: SC1. Obtain chip recommendation information; The chip recommendation information is the result of chip selection based on chip product demand information by using a selection agent as a generation module, combining a pre-set retrieval module, a rearrangement module and a local knowledge base; SC2. Conducting confidence evaluation on the chip recommendation information; SC3. If the confidence level obtained according to SC2 is the highest level, the chip recommendation information is sent to the user. Optionally, in SC2, when the confidence level of the chip recommendation information is evaluated, evaluation content is also generated.

[0017] Optionally, it also includes: SC4. If the confidence level obtained according to SC2 is not the highest level, the evaluation content corresponding to the confidence level obtained according to SC2 is sent to the selection agent so that the selection agent changes the chip recommendation information.

[0018] Optionally, an evaluation agent is designed through prompt word engineering, and SC1 to SC3 are implemented through the evaluation agent.

[0019] A fourth aspect provides a circuit design method, comprising: SD1. Obtain a circuit design request, where the circuit design request includes interaction information describing at least one circuit module; wherein the interaction information of different circuit modules is different; SD2, decomposing the circuit design request into a plurality of circuit module design tasks and an execution order of the plurality of circuit module design tasks; SD3. When executing the target circuit module design task, the chip selection framework as described above is called; the selection framework is used to perform demand analysis according to the circuit design request to obtain chip product demand information, and to perform chip selection according to the chip product demand information to obtain chip recommendation information, and to perform confidence evaluation on the chip recommendation information; SD4. When the confidence level of the chip recommendation information is the highest level, a target chip is obtained based on the chip recommendation information, and a simulation circuit or a test circuit is built using the target chip for verification according to the requirements of the target circuit module design task; SD5. When the verification is passed, the target chip is the chip used to execute the target circuit module design task; SD6. Execute each circuit module design task according to SD3 to SD5 to complete the circuit design.

[0020] A fifth aspect provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the chip selection method based on the large language model as described above, or the chip selection method based on the large language model as described above, or the circuit design method as described above is implemented.

[0021] The sixth aspect provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the chip selection method based on the large language model as described above, or the chip selection method based on the large language model as described above, or the circuit design method as described above.

[0022] The seventh aspect provides a computer program product, including a computer program, which, when executed by a processor, implements the chip selection method based on the large language model as described above, or the chip selection method based on the large language model as described above, or the circuit design method as described above.

[0023] The chip selection method based on a large language model provided by the present invention continuously interacts with the user in a human-computer collaborative mode, guides the user to select a chip, extracts information from the interaction information to be processed, and thereby collects the user's demand for chip products, that is, initial demand information. The initial demand information is also analyzed to prevent abnormal demands such as unreasonable demand expressions, vague expressions, and irrelevant expressions from affecting subsequent information processing, thereby ensuring the availability of the output chip product demand information. In addition, the initial demand information is also standardized to improve the directionality of the output chip product demand information, and ultimately improve the accuracy of the large language model conclusion output. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of the implementation flow of the chip selection method based on a large language model according to an embodiment of the present invention; Figure 2 A schematic diagram of the implementation flow of another chip selection method based on a large language model according to an embodiment of the present invention; Figure 3 A schematic diagram of the composition structure of a chip selection framework based on a large language model according to an embodiment of the present invention; Figure 4 The execution steps to be implemented by the evaluation agent of the embodiment of the present invention; Figure 5 A schematic diagram of an implementation flow of a circuit system design method according to an embodiment of the present invention; Figure 6 Schematic diagram of the structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0026] like Figure 1As shown, the embodiment of the present invention proposes a chip selection method based on a large language model, so that the currently controlled intelligent agent, that is, the interactive intelligent agent, continuously interacts with the user to obtain the user's demand for chip products and extract information. At the same time, the function of abnormal demand identification is provided to ensure the availability of the output chip product demand information.

[0027] The chip selection method based on the large language model includes but is not limited to the following steps: SA1. Obtain interaction information to be processed, where the interaction information to be processed is all interaction information generated by chip selection in a human-machine collaborative mode; SA2. Extracting information from the interaction information to be processed to obtain initial demand information; SA3. Analyze whether the initial demand information includes abnormal demand; the abnormal demand indicates that the initial demand information contains at least one of unreasonable demand expression, ambiguous expression, and irrelevant expression; SA4. If yes, the abnormal demand and the reason for the abnormal demand are fed back to the user, and interaction with the user is performed to regenerate interaction information, so as to execute SA2 according to the newly generated interaction information; SA5. If not, perform demand analysis and standardization processing according to the initial demand information to obtain chip product demand information whose data item names at least include product classification based on the initial demand information and output it.

[0028] In the above step SA1, in the human-machine collaborative mode, the interactive agent of the embodiment of the present invention guides the user to select a chip. The detailed process may be: the user and the interactive agent perform multiple rounds of interaction to provide information about chip selection in the form of voice, text, etc., such as chip signals, chip parameters, chip functions, etc. Based on this, the interactive information to be processed includes but is not limited to voice information, text information, etc. provided by the user.

[0029] For the above steps SA2, SA3 and SA5, the embodiment of the present invention provides a specific data set to train the interactive intelligent agent, so that it has the functions of extracting information from the interactive information to be processed, identifying abnormal demands, performing demand analysis on the initial demand information that does not include abnormal demands, and data standardization.

[0030] For the above step SA3, the abnormal demand analyzed in step SA3 indicates that the initial demand information contains unreasonable demand expressions, vague expressions, irrelevant expressions, etc. In specific applications, the abnormal demand affects the chip product demand information output by step SA5, and ultimately leads to the failure of the selection agent to select the chip or inaccurate chip selection results.

[0031] For the above-mentioned step SA5, the chip product demand information output by step SA5 is finally output to the selection intelligent body. The chip product demand information is standardized data, and the data item name includes at least product classification, and exemplarily, the data item name includes primary product category, secondary product category, product application, and product key parameters.

[0032] In an embodiment of the present invention, the basic agent is trained before the above-mentioned step SA1 to obtain an interactive agent capable of implementing the functions shown in steps SA1 to SA5, which is: supervised fine-tuning training is performed on the first basic agent through a first mixed data set including a proper noun data set, a product classification data set, a multi-round dialogue data set and a general data set.

[0033] It should be noted that the proper noun dataset and product classification dataset here are used to enable the interactive intelligent agent to better understand and solve problems in the field of electronic information, especially to have the ability to select suitable chip products for each circuit module in board-level circuit design. The multi-round dialogue dataset is used to enable the final interactive intelligent agent to continuously interact with the user, and the dialogue does not deviate from the field of electronic information.

[0034] For the above-mentioned proper noun data set, in an embodiment of the present invention, the proper noun data set includes K noun data contents based on K proper nouns, wherein the K proper nouns all belong to the field of electronic information. Exemplarily, the k-th noun data content includes a proper noun QA pair consisting of a correct interpretation of the k-th proper noun and a question sentence based on the correct interpretation of the k-th proper noun, and a system prompt word based on the proper noun QA pair.

[0035] Among them, the system prompt words based on the proper noun QA pair are used to instruct the interactive intelligent agent to extract information from the interactive information to be processed based on the correspondence rules between the correct interpretation of the proper nouns exemplified by the proper noun data set and the question sentences of the correct interpretation of the proper nouns.

[0036] In specific applications, Indicates the correct interpretation of the kth proper noun, using Indicates the correct interpretation of any proper noun, i.e. If the question statement and It forms a "question-answer" pair, that is, a proper noun QA pair, and then uses Indicates the system prompt words of the proper noun QA pair. Based on this, the proper noun dataset of the embodiment of the present invention can be expressed as:

[0037] in, For the proper noun dataset The number of data in =K.

[0038] For the above-mentioned product classification data set, in an embodiment of the present invention, the product classification data set includes N category data contents based on M primary chip categories; illustratively, if the nth category data content belongs to the mth primary chip category, then the nth category data content includes a product classification QA pair consisting of a secondary chip category subordinate to the mth primary chip category and a question statement based on the secondary chip category subordinate to the mth primary chip category, and a system prompt word based on the product classification QA pair. Among them, the system prompt word based on the product classification QA pair is used to instruct the interactive intelligent agent to classify the primary chip category and the secondary chip category based on a certain primary chip category based on the initial demand information, using the corresponding rule between the secondary chip category of the mth primary chip category and the question statement based on the secondary chip category of the mth primary chip category as exemplified by the product classification data set.

[0039] It should be noted that in the above-mentioned product classification data set, N categories of data content are divided into M first-level chip categories for storage. In one embodiment, 16 first-level chip categories are provided, namely amplifiers, audio, battery management IC, clock and timing, data converters, DLP products, interfaces, isolators, logic and voltage conversion, microcontrollers and processors, motor drivers, power management, RF and microwave, sensors, switches and multiplexers, and wireless connectors. Then, the M first-level chip categories are divided into second-level chip categories, wherein the relationship between the second-level chip category and the first-level chip category is a category subordination relationship, and each first-level chip category includes multiple category data contents based on its second-level chip category. For example, the sensor is a first-level chip category, which includes photoelectric sensors, infrared sensors, etc., and the photoelectric sensors and infrared sensors are the category data contents of the second-level chip category. For a category data content in the product classification dataset, assuming that this category data content represents photoelectric sensors, then this category data content includes photoelectric sensors (secondary chip category) belonging to sensors (first-level chip category), and product classification QA pairs composed of question sentences based on sensors (first-level chip category) and photoelectric sensors (secondary chip category), and also includes system prompt words based on the product classification QA pairs.

[0040] In specific applications, Represents the relationship between the primary chip category and the secondary chip category, that is, the secondary chip category belonging to the mth primary chip category, for each Designed question sentences ,but and A "question-answer" pair is formed, that is, a product classification QA pair, and then Indicates the system prompt words of the product classification QA pair. Based on this, the product classification data set of the embodiment of the present invention can be expressed as:

[0041] in, Product classification dataset The number of data in =N.

[0042] For the above-mentioned multi-round dialogue data set, in an embodiment of the present invention, the multi-round dialogue data set includes U multi-round dialogue data contents based on L multi-round dialogues; illustratively, if the u-th multi-round dialogue data content belongs to the l-th multi-round dialogue, then the u-th multi-round dialogue data content includes a multi-round dialogue QA pair consisting of a simulated question in the l-th multi-round dialogue and an answer content based on the simulated question in the l-th multi-round dialogue, a system prompt word based on the multi-round dialogue QA pair, and a historical dialogue set.

[0043] The simulated user simulates the questioning ideas of a natural person to the chip product manual. Exemplarily, in the first multi-round dialogue, the simulated question is generated by simulating the user's questioning to the randomly selected chip product manual; based on the reply content of the simulated question in the first multi-round dialogue, it is generated by the randomly selected chip product manual. In addition, the interactive intelligent agent in the embodiment of the present invention also realizes the analysis of abnormal requirements. Therefore, in the first multi-round dialogue, the simulated question is also obtained by randomly extracting from the preset question database. For the simulated question randomly extracted from the preset question database, its reply content includes the content of the randomly extracted chip product manual; wherein, the preset question database includes multiple preset questions, and each preset question includes at least one of unreasonable demand expression, fuzzy expression, and irrelevant expression. In specific applications, Q i | 多轮对话 Indicates the simulated question in the l-th multi-round dialogue, with A i | 多轮对话 Indicates that based on Q i | 多轮对话 The answer content is Q i | 多轮对话 and A i | 多轮对话 It constitutes a "question-answer" pair, that is, a multi-round dialogue QA pair, and then uses S| 多轮对话 Indicates the system prompt words of multi-round dialogue QA pairs. In addition, the historical dialogue set H is also used i | 多轮对话 , auxiliary training. Based on this, the multi-round dialogue data set of the embodiment of the present invention is expressed as:

[0044] in For multi-round dialogue dataset The number of data in =U.

[0045] It should be noted that, in the embodiment of the present invention, the multi-round dialogue data set actually includes a data set constructed based on the questioning ideas of natural persons to the chip product manual, and also includes a data set constructed based on the questioning ideas of natural persons raising abnormal requirements, that is, a preset question database. Then, the system prompt words based on the multi-round dialogue QA pair are used to instruct the interactive intelligent agent to analyze the abnormal requirements of the initial demand information based on the correspondence rules between the simulated questions and the reply content based on the simulated questions as exemplified by the multi-round dialogue data set, and at the same time guide the situation where the initial demand information includes abnormal requirements, that is, interact with the user to regenerate interactive information, so as to supplement or correct the previous abnormal requirements according to the newly generated interactive information.

[0046] like Figure 2 As shown, the embodiment of the present invention also proposes a chip selection method based on a large language model. After the SA5 of the above embodiment outputs the chip product demand information to the selection agent, the currently controlled agent, i.e., the selection agent, selects chips according to the chip product demand information and outputs chip recommendation information. At the same time, a retrieval enhancement technology is provided to improve the chip selection ability of the selection agent. Different from the existing board-level circuit design, the automatic recommendation is based on market value, user selection rate and other data. If multiple similar chip products appear, the user is prompted to make an independent selection. Figure 2 The chip selection method based on the large language model includes but is not limited to the following steps: SB1. Obtain chip product demand information provided by an interactive agent, and the interactive agent executes the chip selection method of the large language model; SB2, the selection agent acts as a generation module, combines the pre-set retrieval module, rearrangement module and local knowledge base, performs chip selection based on the chip product demand information, and outputs chip recommendation information; The SB2 includes: The chip product demand information is compared with the chip product data stored in the local knowledge base by the retrieval module to obtain a plurality of related documents and send them to the rearrangement module; The reordering module sorts the relevance of the multiple related documents, and sends the sorting result to the selection agent; in the sorting result, the related documents with higher relevance have smaller sorting numbers, and the related documents with sorting numbers less than the preset number value are strongly related documents; All strongly related documents are analyzed according to the chip product demand information, and chip recommendation information is output.

[0047] The chip product demand information in step SB1 is provided by the interactive agent. In detail, the interactive agent will Figure 1 The chip selection method based on the large language model shown in the figure performs chip selection with the user, and then generates interaction information to be processed in this continuous interactive process of chip selection. Finally, the above-mentioned chip product demand information is obtained after performing information extraction, demand analysis and other steps based on the interaction information to be processed.

[0048] The execution subjects of the above steps SB1 and SB2 are both selection agents, and according to the above step SB2, the embodiment of the present invention provides a chip selection method based on a large language model. Not only is the chip selection performed through the selection agent, but also a retrieval enhancement technology with the selection agent as the generation module is designed, thereby realizing the integration of the large language model and retrieval enhancement generation. In this way, the selection ability of the chip selection is improved, and the optimal solution can be output even when there are multiple similar chip products.

[0049] The embodiments of the present invention respectively provide detailed descriptions of the retrieval module, the rearrangement module, the local knowledge base and the selection agent.

[0050] First, for the retrieval module, in one embodiment, a retrieval ternary data set is constructed to perform supervised contrastive learning training on the basic embedding model. After the training is completed, the retrieval module is obtained. In a specific application, the ternary data set contains three elements: query Q, positive sample P, and negative sample N. Among them, Q is the input question, P is the correct answer or the standard answer, and N is the negative wrong answer. In a better implementation method, the retrieval ternary data set constructed by the embodiment of the present invention includes a retrieval query element, a retrieval positive sample element, and a retrieval negative sample element; the retrieval query element is extracted from the total demand information data; the retrieval positive sample element is the unique chip product data selected in the local knowledge base to reply to the retrieval query element; the retrieval negative sample element is a plurality of chip product data randomly extracted from the local knowledge base. It should be noted that the total demand information data refers to all demand information output in the interactive intelligent agent within the historical time period, and it is used here as the data source of the retrieval query element in the retrieval unit data set.

[0051] In specific applications, Represents a search query element, using Represents the retrieval of positive sample elements, using Indicates that negative sample elements will be retrieved. Based on this, the retrieval ternary data set of the embodiment of the present invention can be expressed as:

[0052] in For this triplet dataset The number of data in .

[0053] Secondly, for the rearrangement module, in one embodiment, a rearranged ternary data set is constructed to perform supervised contrastive learning training on the basic rearrangement model. After the training is completed, the rearrangement module is obtained. In a specific application, the ternary data set contains three elements: query Q, positive sample P, and negative sample N. Among them, Q is the input question, P is the correct answer or the standard answer, and N is a negative wrong answer. In a better implementation method, the rearranged ternary data set constructed by the embodiment of the present invention is more concise and precise than the above-mentioned retrieval ternary data set, including a rearranged query element, a rearranged positive sample element, and a rearranged negative sample element; the rearranged query element is a simulated demand description, and the data size of the simulated demand description is less than a preset number of bytes; the rearranged positive sample element is the only correct answer written based on the simulated demand description; the rearranged negative sample element is a plurality of wrong answers written based on the simulated demand description, and the similarity between the wrong answer and the only correct answer is greater than a preset value. It should be noted that the constraint condition that the data size of the simulation requirement description is less than the preset number of bytes is set to indicate that the rearranged ternary data set is more concise and precise than the retrieval ternary data set previously used to train the embedding model; the constraint condition that the similarity between the wrong answer and the only correct answer is set to be greater than a preset value is used to enhance the performance of the rearrangement model, so that it can more accurately distinguish between strongly correlated related documents and weakly correlated related documents, which is conducive to the selection agent finding the optimal solution when there are multiple similar chip products.

[0054] In specific applications, Indicates rearrangement of query elements, Represents the rearrangement of positive sample elements, using Represents the rearranged negative sample element. Based on this, the rearranged triplet data set of the embodiment of the present invention can be expressed as:

[0055] in For this triplet dataset The number of data in .

[0056] Then, for the local knowledge base, in one embodiment, any chip product data stored in the local knowledge base includes a document description of product features, a document description of product application examples, a document description of key parameters, and a document description of complete information. In a specific application, the chip product data can be obtained by extracting information from multiple chip product manuals and then stored in the local knowledge base. New chip product manuals can also be obtained regularly to extract information based on the new chip product manuals to obtain new chip product data, and then updated to the local knowledge base.

[0057] Finally, for the selection agent, in one embodiment, to train the second basic agent, the following is achieved: Figure 2 The chip selection method based on the large language model shown, especially the selection agent in step SB2. The detailed implementation steps include: The second basic agent is trained in two stages by continuing pre-training and supervised fine-tuning training to obtain a selection agent to analyze all strongly related documents according to the chip product demand information and output chip recommendation information; In the continued pre-training, multi-dimensional chip text data is obtained, and text pre-processing is performed on the multi-dimensional chip text data to obtain a plurality of corpora, and the second basic agent is trained by the plurality of corpora; In the supervised fine-tuning training, the pre-trained second basic agent is continued through training with a second mixed data set including a proper noun data set, a product classification data set, a chip selection data set and a general data set to obtain a selection agent.

[0058] It should be noted that multi-dimensional chip text data includes but is not limited to books related to electronic science and technology, industry standard documents, research reports, web forums, public guides from well-known companies in the industry and other materials, and text preprocessing methods include but are not limited to text content extraction, regular expression refinement, noise reduction optimization and manual refinement. Among them, the continued pre-training based on multi-dimensional chip text data is aimed at enabling the selection agent to better understand and solve problems in the field of electronic information. The proper noun dataset and product classification dataset used here have the same data content as the proper noun dataset and product classification dataset shown in the first mixed dataset above. The chip selection dataset is used to train the chip selection ability of the selection agent, that is, to find chip products that match the target demand information.

[0059] Different from the first mixed data set mentioned above, the chip selection data set in the second mixed data set includes P chip matching data; exemplarily, the pth chip matching data includes a chip matching QA pair consisting of a preset chip requirement description and recommended information that meets the preset chip requirement description, and a system prompt word based on the chip matching QA pair; wherein, the preset chip requirement description is generated by a randomly selected chip product manual; and the recommended information that meets the preset chip requirement description is the chip product model and chip product introduction recorded in the randomly selected chip product manual.

[0060] In specific applications, Indicates the preset chip requirement description, using Express satisfaction If the recommended information and Construct a chip matching QA pair, and then use System prompt words indicating chip matching QA pair. Based on this, the chip selection data set of the embodiment of the present invention can be expressed as:

[0061] in Dataset selection for chip The number of data in =P.

[0062] Among them, the system prompt words based on the chip matching QA pair are used to instruct the selection agent to use the correspondence rules between the preset chip requirement description and the recommended information that meets the preset chip requirement description as exemplified by the chip matching data of the multi-round dialogue data set, select the chip based on the chip product requirement information, and output the chip recommendation information.

[0063] like Figure 3 As shown, the embodiment of the present invention proposes a chip selection frame 30 based on a large language model, including an interactive agent 31 driven by a large language model, a selection agent 32 and an evaluation agent 33; wherein the interactive agent 31 executes as follows Figure 1 The chip selection method based on the large language model shown in the figure, the selection agent 32 performs the following steps: Figure 2 The chip selection method based on the large language model is shown.

[0064] like Figure 4 As shown, the evaluation agent 33 performs the following steps to compare the demand information output by the interactive agent with the recommended information output by the selection agent, so as to determine whether the chip recommendation information based on the target chip output by the selection agent to the evaluation agent meets the requirements of the interactive agent, and gives a confidence level evaluation and basis: SC1. Obtain chip recommendation information; The chip recommendation information is the result of chip selection based on chip product demand information by using a selection agent as a generation module, combining a pre-set retrieval module, a rearrangement module and a local knowledge base; SC2. Conducting confidence evaluation on the chip recommendation information; SC3. If the confidence level obtained according to SC2 is the highest level, the chip recommendation information is sent to the user.

[0065] For the above step SC2, when the embodiment of the present invention performs confidence evaluation on the recommendation information based on the target chip, evaluation content is also generated. Based on this, after the above step SC2 performs confidence evaluation on the chip recommendation information, confidence levels and evaluation content are generated. Exemplarily, the confidence levels are sorted from high to low, including: Fully meet the needs > Partially meet the needs > Partially meet the needs > No needs are met.

[0066] Exemplarily, the evaluation content of the confidence level of fully meeting the demand is: the recommended information based on the target chip fully meets the target demand information. The evaluation content of the confidence level of partially meeting the demand is: the recommended information based on the target chip partially meets the target demand information. The evaluation content of the confidence level of slightly meeting the demand is: only a small amount of the recommended information based on the target chip meets the user's demand. The evaluation content of the confidence level of not meeting the demand is: the recommended information based on the target chip does not meet the user's demand at all or the selection agent generates irrelevant content.

[0067] In another embodiment of the present invention, based on Figure 4 The steps SC1 to SC3 performed by the evaluation agent 33 shown in the figure also propose a processing method when the confidence level is not the highest level, including: SC4. If the confidence level obtained according to SC2 is not the highest level, the evaluation content corresponding to the confidence level obtained according to SC2 is sent to the selection agent so that the selection agent changes the chip recommendation information.

[0068] It should be noted that for the selection agent, if Figure 2 The chip selection method based on the large language model shown may also include: SB3: If the received confidence level is not the highest level, the evaluation content corresponding to the current confidence level is generated, the chip selection result in SB2 is changed, and the changed chip recommendation information is output.

[0069] Through the above steps, the selection agent re-selects the chip according to the feedback of the evaluation agent, that is, the evaluation content corresponding to the current confidence level, thereby effectively optimizing the accuracy of chip selection.

[0070] For the above steps SC1 to SC3, and the above step SC4, the embodiment of the present invention designs an evaluation agent through prompt word engineering to implement the above chip selection method based on a large language model, especially for implementing step SC2. Among them, the prompt word engineering is designed into three parts: task-specific prompt words, evaluation references, and evaluation precautions. The task-specific prompt words are intended to allow the evaluation agent to understand its own confidence evaluation task, the evaluation reference includes examples and instructions for four level evaluations, and the evaluation precautions include the requirements for the evaluation agent's answers so that the evaluation agent can give accurate level evaluations and basis.

[0071] Based on the above detailed description of the interactive agent 31, the selection agent 32 and the evaluation agent 33, the basic principle of the chip selection framework 20 based on the large language model in the embodiment of the present invention is: The interactive agent continuously interacts with the user, extracts the user's chip product requirements after judging that the user's demand for chip products is feasible, and transmits the results to the selection agent. The selection agent conducts a comprehensive analysis based on the chip product requirements input by the interactive agent, and finally inputs the proposed recommended chip product model and related introduction as the result to the evaluation agent. The evaluation agent compares the user requirements extracted by the interactive agent information with the chip products proposed to be recommended by the selection agent to determine whether the latter fully meets the requirements of the former, and gives a confidence level evaluation and basis. If the confidence level evaluation given by the evaluation agent is not the highest level, the basis given by the evaluation agent is re-input into the selection agent to change the proposed recommended chip product. If the confidence level evaluation given by the evaluation agent is the highest level, the final recommended chip product and the basis and related product introduction parameters are input to the user.

[0072] Based on the above-mentioned chip selection framework 20 based on the large language model, the embodiment of the present invention obtains the interaction information to be processed through all the interaction information during the interaction between the interactive agent and the user and information extraction, and then performs demand analysis on the interaction information to be processed, and outputs the data required by the selection agent for chip selection, that is, the demand information. The selection agent acts as a generation module, which combines the pre-set retrieval module, rearrangement module and local knowledge base to perform chip selection based on demand information, thereby realizing the integration of large language model and retrieval enhanced generation. In this way, the selection ability can be improved, and the optimal solution can be output even when there are multiple similar chip products. In addition, the confidence of the recommended information is evaluated by the evaluation agent on the result of the chip selection of the selection agent, that is, the reliability of the recommended information is guaranteed to meet user expectations.

[0073] like Figure 5 As shown, an embodiment of the present invention further provides a circuit design method, comprising: SD1. Obtain a circuit design request, where the circuit design request includes interaction information describing at least one circuit module; wherein the interaction information of different circuit modules is different; SD2, decomposing the circuit design request into a plurality of circuit module design tasks and an execution order of the plurality of circuit module design tasks; SD3. When executing the target circuit module design task, the chip selection framework as described above is called; the selection framework is used to perform demand analysis according to the circuit design request to obtain chip product demand information, and to perform chip selection according to the chip product demand information to obtain chip recommendation information, and to perform confidence evaluation on the chip recommendation information; SD4. When the confidence level of the chip recommendation information is the highest level, a target chip is obtained based on the chip recommendation information, and a simulation circuit or a test circuit is built using the target chip for verification according to the requirements of the target circuit module design task; SD5. When the verification is passed, the target chip is the chip used to execute the target circuit module design task; SD6. Execute each circuit module design task according to SD3 to SD5 to complete the circuit design.

[0074] Through the above steps SD1 to SD6, the embodiment of the present invention disassembles the circuit design request to obtain all circuit modules required to complete the circuit design request and the design order of each circuit module, and for each circuit module, chip selection is performed through the above chip selection framework, thereby simplifying the existing board-level circuit design process, realizing chip selection for each module of the automated circuit system, and improving the design efficiency and accuracy of the board-level circuit.

[0075] Figure 6 A schematic diagram of the structure of an electronic device 60 that can be used to implement an embodiment of the present invention regarding a chip selection method based on a large language model, and a circuit design method is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0076] like Figure 6 As shown, the electronic device 60 includes at least one processor 61, and a memory connected to the at least one processor 61, such as a read-only memory (ROM) 62, a random access memory (RAM) 63, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 61 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 62 or the computer program loaded from the storage unit 68 to the random access memory (RAM) 63. In the RAM 63, various programs and data required for the operation of the electronic device 60 can also be stored. The processor 61, the ROM 62, and the RAM 63 are connected to each other via a bus 64. An input / output (I / O) interface 65 is also connected to the bus 64.

[0077] A number of components in the electronic device 60 are connected to the I / O interface 65, including: an input unit 66, such as a keyboard, a mouse, etc.; an output unit 67, such as various types of displays, speakers, etc.; a storage unit 68, such as a disk, an optical disk, etc.; and a communication unit 69, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 69 allows the electronic device 60 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0078] The processor 61 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 61 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 61 executes the various methods and processes described above, such as the optimization method based on the gun-ball linkage monitoring system.

[0079] In some embodiments, the data processing method may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 68. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 60 via the ROM 62 and / or the communication unit 69. When the computer program is loaded into the RAM 63 and executed by the processor 61, one or more steps of the data processing method described above may be performed. Alternatively, in other embodiments, the processor 61 may be configured to perform the data processing method in any other suitable manner (e.g., by means of firmware).

[0080] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0081] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0082] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, device, or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0083] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0084] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0085] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on respective computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in a cloud computing service system.

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A chip selection method based on a large language model, characterized in that: include: SA1. Obtain interaction information to be processed, where the interaction information to be processed is all interaction information generated by chip selection in a human-machine collaborative mode; SA2. Extracting information from the interaction information to be processed to obtain initial demand information; SA3. Analyze whether the initial demand information includes abnormal demand; the abnormal demand indicates that the initial demand information contains at least one of unreasonable demand expression, ambiguous expression, and irrelevant expression; SA4. If yes, the abnormal demand and the reason for the abnormal demand are fed back to the user, and interaction with the user is performed to regenerate interaction information, so as to execute SA2 according to the newly generated interaction information; SA5. If not, perform demand analysis and standardization processing according to the initial demand information to obtain chip product demand information whose data item names at least include product classification based on the initial demand information and output it; Among them, the first basic agent is supervised and fine-tuned by a first mixed data set including a proper noun data set, a product classification data set, a multi-round dialogue data set and a general data set to obtain an interactive agent to implement SA1 to SA5.

2. The chip selection method based on a large language model according to claim 1, characterized in that: The proper noun data set includes K noun data contents based on K proper nouns; The k-th noun data content includes a proper noun QA pair consisting of a correct explanation of the k-th proper noun and a question sentence based on the correct explanation of the k-th proper noun, and a system prompt word based on the proper noun QA pair; Wherein, k is a positive integer less than or equal to K, and K is a positive integer.

3. The chip selection method based on a large language model according to claim 1, characterized in that: The product classification data set includes N category data contents based on M primary chip categories; If the n-th category data content belongs to the m-th primary chip category, the n-th category data content includes a secondary chip category subordinate to the m-th primary chip category and a product classification QA pair consisting of a question sentence based on the secondary chip category subordinate to the m-th primary chip category, and a system prompt word based on the product classification QA pair; Wherein, M and N are positive integers, n is a positive integer less than or equal to N, and m is a positive integer less than or equal to M.

4. The chip selection method based on a large language model according to claim 1, characterized in that: The multi-round dialogue data set includes U multi-round dialogue data contents based on L multi-round dialogues; If the u-th multi-round dialogue data content belongs to the l-th multi-round dialogue, the u-th multi-round dialogue data content includes a multi-round dialogue QA pair consisting of a simulated question in the l-th multi-round dialogue and an answer content based on the simulated question in the l-th multi-round dialogue, a system prompt word based on the multi-round dialogue QA pair, and a historical dialogue set; Among them, L and U are positive integers, l is a positive integer less than or equal to L, u is a positive integer less than or equal to U, and in the l-th multi-round dialogue, the simulated question is generated by simulating the user's question to the randomly selected chip product manual; based on the reply content of the simulated question in the l-th multi-round dialogue, it is generated by the randomly selected chip product manual.

5. The chip selection method based on a large language model as claimed in claim 4, characterized in that: In the first multi-round dialogue, the simulated question is also obtained by randomly selecting from a preset question database, and the answer content of the simulated question randomly selected from the preset question database includes the content of the randomly selected chip product manual; The preset question database includes a plurality of preset questions, and each preset question includes at least one of an unreasonable demand expression, a vague expression, and an irrelevant expression.

6. A chip selection method based on a large language model, applied to an interactive agent, wherein the interactive agent executes the chip selection method based on a large language model as claimed in any one of claims 1 to 5, characterized in that: include: SB1. Obtain chip product demand information provided by the interactive agent; SB2, the selection agent acts as a generation module, combines the pre-set retrieval module, rearrangement module and local knowledge base, performs chip selection based on the chip product demand information, and outputs chip recommendation information; The SB2 includes: The chip product demand information is compared with the chip product data stored in the local knowledge base by the retrieval module to obtain a plurality of related documents and send them to the rearrangement module; The reordering module sorts the relevance of the multiple related documents, and sends the sorting result to the selection agent; in the sorting result, the related documents with higher relevance have smaller sorting numbers, and the related documents with sorting numbers less than the preset number value are strongly related documents; All strongly related documents are analyzed according to the chip product demand information, and chip recommendation information is output.

7. The chip selection method based on a large language model according to claim 6, characterized in that: Constructing a retrieval triple dataset to perform supervised contrastive learning training on the basic embedding model, and obtaining the retrieval module after the training is completed; Among them, the constructed retrieval ternary data set includes retrieval query elements, retrieval positive sample elements and retrieval negative sample elements; the retrieval query elements are extracted from the total demand information data; the retrieval positive sample elements are unique chip product data selected in the local knowledge base to reply to the retrieval query elements; the retrieval negative sample elements are multiple chip product data randomly extracted from the local knowledge base.

8. The chip selection method based on a large language model according to claim 6, characterized in that: Constructing a rearrangement ternary data set to perform supervised contrastive learning training on the basic rearrangement model, and obtaining the rearrangement module after the training is completed; Among them, the constructed rearranged ternary data set includes rearranged query elements, rearranged positive sample elements and rearranged negative sample elements; the rearranged query element is a simulated requirement description, and the data size of the simulated requirement description is less than a preset number of bytes; the rearranged positive sample element is a unique correct answer written based on the simulated requirement description; the rearranged negative sample element is multiple incorrect answers written based on the simulated requirement description, and the similarity between the incorrect answers and the unique correct answer is greater than a preset similarity.

9. The chip selection method based on a large language model according to claim 6, characterized in that: Any chip product data stored in the local knowledge base includes document descriptions of product features, product application examples, key parameters, and complete information.

10. The chip selection method based on a large language model according to claim 6, characterized in that: The second basic agent is trained in two stages by continuing pre-training and supervised fine-tuning training to obtain a selected agent to implement SB1 to SB2; In the continued pre-training, multi-dimensional chip text data is obtained, and text pre-processing is performed on the multi-dimensional chip text data to obtain a plurality of corpora, and the second basic agent is trained by the plurality of corpora; In the supervised fine-tuning training, the pre-trained second basic agent is continued through training with a second mixed data set including a proper noun data set, a product classification data set, a chip selection data set and a general data set to obtain a selection agent.

11. The chip selection method based on a large language model according to claim 10, characterized in that: The chip selection data set includes P chip matching data; The p-th chip matching data includes a chip matching QA pair consisting of an analog chip requirement description and recommended information satisfying the analog chip requirement description, and a system prompt word based on the chip matching QA pair; Wherein, p is a positive integer less than or equal to P, P is a positive integer, and the simulation chip requirement description is generated by randomly selecting a chip product manual; The recommended information that meets the analog chip requirement description is the chip product model and chip product introduction recorded in the randomly selected chip product manual.

12. A chip selection framework based on a large language model, characterized in that: The invention comprises an interactive agent, a selection agent and an evaluation agent driven by a large language model; the interactive agent executes the chip selection method based on the large language model as described in any one of claims 1 to 5; the selection agent executes the chip selection method based on the large language model as described in any one of claims 6 to 11; the evaluation agent executes the following steps: SC1. Obtain chip recommendation information; The chip recommendation information is the result of chip selection based on chip product demand information by using a selection agent as a generation module, combining a pre-set retrieval module, a rearrangement module and a local knowledge base; SC2. Conducting confidence evaluation on the chip recommendation information; SC3. If the confidence level obtained according to SC2 is the highest level, the chip recommendation information is sent to the user.

13. The chip selection framework according to claim 12, characterized in that: In the SC2, when the confidence evaluation of the chip recommendation information is performed, evaluation content is also generated.

14. The chip selection framework according to claim 13, characterized in that: Also includes: SC4. If the confidence level obtained according to SC2 is not the highest level, the evaluation content corresponding to the confidence level obtained according to SC2 is sent to the selection agent so that the selection agent changes the chip recommendation information.

15. The chip selection framework according to claim 12, characterized in that: An evaluation agent is designed through prompt word engineering, and SC1 to SC3 are implemented through the evaluation agent.

16. A circuit design method, characterized in that: include: SD1. Obtain a circuit design request, wherein the circuit design request includes interaction information describing at least one circuit module; wherein the interaction information of different circuit modules is different; SD2, decomposing the circuit design request into a plurality of circuit module design tasks and an execution order of the plurality of circuit module design tasks; SD3. When executing the target circuit module design task, calling the chip selection framework according to any one of claims 12 to 15; the selection framework is used to perform demand analysis according to the circuit design request to obtain chip product demand information, and perform chip selection according to the chip product demand information to obtain chip recommendation information, and perform confidence evaluation on the chip recommendation information; SD4. When the confidence level of the chip recommendation information is the highest level, a target chip is obtained based on the chip recommendation information, and a simulation circuit or a test circuit is built using the target chip for verification according to the requirements of the target circuit module design task; SD5. When the verification is passed, the target chip is the chip used to execute the target circuit module design task; SD6. Execute each circuit module design task according to SD3 to SD5 to complete the circuit design.

17. An electronic device, characterized in that: It comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the chip selection method based on a large language model as described in any one of claims 1 to 5, or the chip selection method based on a large language model as described in any one of claims 6 to 11, or the circuit design method as described in claim 16.

18. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, it implements the chip selection method based on a large language model as described in any one of claims 1 to 5, or the chip selection method based on a large language model as described in any one of claims 6 to 11, or the circuit design method as described in claim 16.

19. A computer program product comprising a computer program, characterized in that When executed by a processor, the computer program implements the chip selection method based on a large language model as described in any one of claims 1 to 5, or the chip selection method based on a large language model as described in any one of claims 6 to 11, or the circuit design method as described in claim 16.

Images