Information processing method and device based on large language model, equipment and storage medium

Abstract

The invention discloses an information processing method, device and equipment based on a large language model and a storage medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: responding to a first chat message sent by a first user terminal or an intelligent group establishment request, inputting the first chat message to the large language model to obtain a first reply message, or connecting the large language model to the intelligent group and sending a first chat message to the large model; sending a first reply message to the first user terminal, or sending the first reply message to the intelligent group established by the first user terminal; in response to a display request sent by the first user terminal, the chat message and the reply message are published to a display interface, so that a second user terminal obtains the chat message and the reply message from the display interface, the reply message at least comprises a first reply message, and the chat message at least comprises a first chat message; and in response to the comment content sent by the second user terminal for the chat message and the reply message, publishing the comment content to the display interface.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to an information processing method, apparatus, device and storage medium based on a large language model. Background Art

[0002] As more and more AI-generated content continues to grow, the supervision of the authenticity and accuracy of Internet content information will face new challenges.

[0003] Usually, users can communicate with a large language model by inputting messages into the large language model. In this process, it is often a dialogue between the user and the large language model. The user can only obtain the information provided by the large language model, which makes the information that the user can obtain in the process of communicating with the large language model very limited. Summary of the invention

[0004] The purpose of this application is to provide an information processing method, apparatus, device and storage medium based on a large language model.

[0005] To achieve the above objectives, this application provides the following solutions:

[0006] In a first aspect, the present application provides an information processing method based on a large language model, comprising:

[0007] In response to at least one first chat message sent by a first user terminal or a request to establish an intelligent group, input at least one first chat message to at least one large language model to obtain at least one first reply message, or connect at least one large language model to the intelligent group and send at least one first chat message to the large language model;

[0008] Sending at least one first reply message to the first user terminal, or sending at least one first reply message to the smart group established by the first user terminal;

[0009] In response to a display request sent by the first user terminal, publishing the chat message and the reply message to a display interface, so that at least one second user terminal obtains the chat message and the reply message from the display interface, the reply message at least including the first reply message, and the chat message at least including the first chat message;

[0010] In response to the comment content sent by at least one of the second user terminals to the chat message and the reply message, the comment content is published to the display interface, or the smart group joined by at least one of the second users.

[0011] Optionally, the information processing method based on the large language model further includes:

[0012] Inputting a first chat message sent by at least one of the first user terminals into at least two large language models connected to the smart group to obtain a first reply message;

[0013] Sending the first reply message to at least one of the first user terminals in the smart group, so that the smart group interface on the first user terminal displays the first reply message;

[0014] In response to a second reply message sent by the first user terminal to the first reply message, or a second reply message sent to a first reply message of at least one large language model, or a second reply message sent to a first reply message of all large language models, the second reply message is input to at least one large language model to obtain a third reply message, and the third reply message is sent to the smart group interface of the first user terminal for display.

[0015] Optionally, the information processing method based on the large language model further includes:

[0016] Inputting a first chat message sent by at least one of the first user terminals into a first large language model or at least one first large language model corresponding to the smart group to which the first user terminal belongs, to obtain a first reply message;

[0017] Inputting the first reply message into the second largest language model or the second largest language model corresponding to the smart group to obtain a comment message;

[0018] The comment message is sent to the first user terminal or at least two first user terminals included in the smart group, so that the first user terminal displays the comment message on a display interface, or the first user terminal displays the comment message on a smart group interface.

[0019] Optionally, the information processing method based on the large language model further includes:

[0020] Sending a comment message or a reply message given by at least one first language model, or a comment message or a reply message given by at least one user terminal, to a second language model, so that the second language model analyzes the comment message or the reply message to determine whether the comment message or the reply message meets the output requirement;

[0021] The comment message or reply message that does not meet the output requirement is updated in the database.

[0022] Optionally, the step of sending a comment message or a reply message given by at least one first language model, or a comment message or a reply message given by at least one user terminal, to a second language model so that the second language model analyzes the comment message or the reply message to determine whether the comment message or the reply message meets the output requirement includes:

[0023] Send at least one comment message or reply message given by the first largest language model or at least one comment message or reply message given by a user terminal, as well as the number of likes for the user terminal, to the second largest language model, so that the second largest language model analyzes the comment information or reply message and the number of likes to determine whether the comment message or reply message meets the output requirements.

[0024] Optionally, the information processing method based on the large language model further includes:

[0025] Sending a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message;

[0026] Sending the first reply message of the second largest language model to the first largest language model, so that the first largest language model comprehensively analyzes the first reply message of the second largest language model and the first reply message of the first largest language model, and gives a first comprehensive reply message;

[0027] Sending the first reply message of the first largest language model to the second largest language model, so that the second largest language model comprehensively analyzes the first reply message of the first largest language model and the first reply message of the second largest language model to give a second comprehensive reply message;

[0028] When the similarity between the first comprehensive reply message and the second comprehensive reply message is greater than or equal to a similarity threshold, sending the first comprehensive reply message or the second comprehensive reply message to at least one of the user terminals for display;

[0029] When the similarity between the first comprehensive reply message and the second comprehensive reply message is less than a similarity threshold, the first reply message of the first large language model, the first reply message of the second large language model, the first comprehensive reply message and the second comprehensive reply message are updated into the database.

[0030] Optionally, the information processing method based on the large language model further includes:

[0031] Sending a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message;

[0032] sending the first reply message of the second largest language model to the first largest language model, so that the first largest language model performs a comprehensive analysis on the first reply message of the second largest language model and the first reply message of the first largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content;

[0033] When the first reply message of the first largest language model and the first reply message of the second largest language model contain error content, obtaining a first comprehensive reply message output by the first largest language model that does not contain error content;

[0034] sending the first reply message of the first largest language model to the second largest language model, so that the second largest language model performs a comprehensive analysis on the first reply message of the first largest language model and the first reply message of the second largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content;

[0035] In the case that the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content, a second comprehensive reply message output by the second largest language model that does not contain erroneous content is obtained.

[0036] Optionally, the information processing method based on the large language model further includes:

[0037] Sending the first comprehensive reply message and the second comprehensive reply message to the one user terminal;

[0038] In response to the continuing processing request sent by the user terminal, continue the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message;

[0039] Alternatively, when the first largest language model or the second largest language model confirms that the first comprehensive reply message or the second comprehensive reply message contains erroneous content, the first comprehensive reply message and the second comprehensive reply message are used to continue the comprehensive analysis process:

[0040] The step of continuing the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message comprises:

[0041] Inputting the first comprehensive reply message and the second comprehensive reply message into the first large language model and the second large language model respectively, updating the first comprehensive reply message using the output data of the first large language model, and updating the second comprehensive reply message using the output data of the second large language model;

[0042] When the first largest language model or the second largest language model analyzes and confirms that the first comprehensive reply message or the second comprehensive reply message contains erroneous content, the comprehensive analysis process is continued using the first comprehensive reply message and the second comprehensive reply message until the first largest language model and the second largest language model analyze and confirm that the first comprehensive reply message and the second comprehensive reply message do not contain erroneous content.

[0043] Optionally, the information processing method based on the large language model further includes:

[0044] The first reply message of the first large language model containing erroneous content, the first reply message of the second large language model, the first comprehensive reply message, and the second comprehensive reply message are updated into the large language model database.

[0045] Optionally, the information processing method based on the large language model further includes:

[0046] When at least two large language models have different comment messages or reply messages for the same chat information, the comment messages or reply messages are compared and analyzed to obtain comment messages or reply messages that meet the output requirements.

[0047] Optionally, the information processing method based on the large language model further includes:

[0048] A like button is set in the dialogue interface between the user and the large language model, the intelligent group interface, the display interface, and the comment interface, and the number of likes is displayed;

[0049] Using the large language model to obtain the number of likes and the conversation information of the likes, and performing processing and analysis in combination with the conversation information of the likes;

[0050] When the number of likes is greater than or equal to the likes number threshold, the conversation information of the likes is updated in the database as information recognized by the user.

[0051] Optionally, the information processing method based on the large language model further includes:

[0052] Copying the chat message content between the user terminal and the large language model to a sharing link interface, generating a sharing link image, or generating a sharing link QR code, or generating a sharing picture including the chat message content and the sharing link QR code;

[0053] Sending the shared link image or shared link QR code to a user group associated with the user terminal, or sending a shared image with the chat message content and the shared link QR code to the user terminal, so that the shared image with the chat message content and the shared link QR code is displayed on the content display interface of the user terminal;

[0054] In response to a query request sent by a user terminal in the user group for the shared link image or the shared link QR code, sending the chat message content to the user terminals in the user group;

[0055] Or, in response to a query request from another terminal regarding the QR code of the shared link in the shared picture, the chat message content is sent to the other terminal.

[0056] Optionally, the information processing method based on the large language model further includes:

[0057] In response to the reply content carrying the indication information sent by the first user terminal, determining the large language model identifier and / or the user identifier carried in the indication information;

[0058] The reply content is sent to the large language model corresponding to the large language model identifier, so that the large language model replies to the reply content, and / or prompt information carrying the reply content is sent to the second user terminal corresponding to the user identifier.

[0059] In a second aspect, the present application provides an information processing device based on a large language model, comprising:

[0060] a processing module, configured to, in response to at least one first chat message sent by a first user terminal or a request to establish an intelligent group, input at least one first chat message into at least one large language model to obtain at least one first reply message, or connect at least one large language model to the intelligent group and send at least one first chat message to the large language model;

[0061] Sending at least one first reply message to the first user terminal, or sending at least one first reply message to the smart group established by the first user terminal;

[0062] a publishing module, configured to publish the chat message and the reply message to a display interface in response to a display request sent by the first user terminal, so that at least one second user terminal obtains the chat message and the reply message from the display interface, the reply message at least including the first reply message, and the chat message at least including the first chat message;

[0063] In response to the comment content sent by at least one of the second user terminals to the chat message and the reply message, the comment content is published to the display interface, or the smart group joined by at least one of the second users.

[0064] Optionally, the processing module is further used to:

[0065] Inputting a first chat message sent by at least one of the first user terminals into at least two large language models connected to the smart group to obtain a first reply message;

[0066] Sending the first reply message to at least one of the first user terminals in the smart group, so that the smart group interface on the first user terminal displays the first reply message;

[0067] In response to a second reply message sent by the first user terminal to the first reply message, or a second reply message sent to a first reply message of at least one large language model, or a second reply message sent to a first reply message of all large language models, the second reply message is input to at least one large language model to obtain a third reply message, and the third reply message is sent to the smart group interface of the first user terminal for display.

[0068] Optionally, the processing module is further used to:

[0069] Inputting a first chat message sent by at least one of the first user terminals into a first large language model or at least one first large language model corresponding to the smart group to which the first user terminal belongs, to obtain a first reply message;

[0070] Inputting the first reply message into the second largest language model or the second largest language model corresponding to the smart group to obtain a comment message;

[0071] The comment message is sent to the first user terminal or at least two first user terminals included in the smart group, so that the first user terminal displays the comment message on a display interface, or the first user terminal displays the comment message on a smart group interface.

[0072] Optionally, the processing module is further used to:

[0073] Sending a comment message or a reply message given by at least one first language model, or a comment message or a reply message given by at least one user terminal, to a second language model, so that the second language model analyzes the comment message or the reply message to determine whether the comment message or the reply message meets the output requirement;

[0074] The comment message or reply message that does not meet the output requirement is updated in the database.

[0075] Optionally, the processing module is further used to:

[0076] Send at least one comment message or reply message given by the first largest language model or at least one comment message or reply message given by a user terminal, as well as the number of likes for the user terminal, to the second largest language model, so that the second largest language model analyzes the comment information or reply message and the number of likes to determine whether the comment message or reply message meets the output requirements.

[0077] Optionally, the processing module is further used to:

[0078] Sending a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message;

[0079] Sending the first reply message of the second largest language model to the first largest language model, so that the first largest language model comprehensively analyzes the first reply message of the second largest language model and the first reply message of the first largest language model, and gives a first comprehensive reply message;

[0080] Sending the first reply message of the first largest language model to the second largest language model, so that the second largest language model comprehensively analyzes the first reply message of the first largest language model and the first reply message of the second largest language model to give a second comprehensive reply message;

[0081] When the similarity between the first comprehensive reply message and the second comprehensive reply message is greater than or equal to a similarity threshold, sending the first comprehensive reply message or the second comprehensive reply message to at least one of the user terminals for display;

[0082] When the similarity between the first comprehensive reply message and the second comprehensive reply message is less than a similarity threshold, the first reply message of the first large language model, the first reply message of the second large language model, the first comprehensive reply message and the second comprehensive reply message are updated into the database.

[0083] Optionally, the processing module is further used to:

[0084] Sending a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message;

[0085] sending the first reply message of the second largest language model to the first largest language model, so that the first largest language model performs a comprehensive analysis on the first reply message of the second largest language model and the first reply message of the first largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content;

[0086] When the first reply message of the first largest language model and the first reply message of the second largest language model contain error content, obtaining a first comprehensive reply message output by the first largest language model that does not contain error content;

[0087] sending the first reply message of the first largest language model to the second largest language model, so that the second largest language model performs a comprehensive analysis on the first reply message of the first largest language model and the first reply message of the second largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content;

[0088] In the case that the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content, a second comprehensive reply message output by the second largest language model that does not contain erroneous content is obtained.

[0089] Optionally, the processing module is further used to:

[0090] Sending the first comprehensive reply message and the second comprehensive reply message to the one user terminal;

[0091] In response to the continuing processing request sent by the user terminal, continue the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message;

[0092] Alternatively, when the first largest language model or the second largest language model confirms that the first comprehensive reply message or the second comprehensive reply message contains erroneous content, the first comprehensive reply message and the second comprehensive reply message are used to continue the comprehensive analysis process:

[0093] The step of continuing the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message comprises:

[0094] Inputting the first comprehensive reply message and the second comprehensive reply message into the first large language model and the second large language model respectively, updating the first comprehensive reply message using the output data of the first large language model, and updating the second comprehensive reply message using the output data of the second large language model;

[0095] When the first largest language model or the second largest language model analyzes and confirms that the first comprehensive reply message or the second comprehensive reply message contains erroneous content, the comprehensive analysis process is continued using the first comprehensive reply message and the second comprehensive reply message until the first largest language model and the second largest language model analyze and confirm that the first comprehensive reply message and the second comprehensive reply message do not contain erroneous content.

[0096] Optionally, the processing module is further used to:

[0097] The first reply message of the first large language model containing erroneous content, the first reply message of the second large language model, the first comprehensive reply message, and the second comprehensive reply message are updated into the large language model database.

[0098] Optionally, the processing module is further used to:

[0099] When at least two large language models have different comment messages or reply messages for the same chat information, the comment messages or reply messages are compared and analyzed to obtain comment messages or reply messages that meet the output requirements.

[0100] Optionally, the processing module is further used to:

[0101] A like button is set in the dialogue interface between the user and the large language model, the intelligent group interface, the display interface, and the comment interface, and the number of likes is displayed;

[0102] Using the large language model to obtain the number of likes and the conversation information of the likes, and performing processing and analysis in combination with the conversation information of the likes;

[0103] When the number of likes is greater than or equal to the likes number threshold, the conversation information of the likes is updated in the database as information recognized by the user.

[0104] Optionally, the processing module is further used to:

[0105] Copying the chat message content between the user terminal and the large language model to a sharing link interface, generating a sharing link image, or generating a sharing link QR code, or generating a sharing picture including the chat message content and the sharing link QR code;

[0106] Sending the shared link image or shared link QR code to a user group associated with the user terminal, or sending a shared image with the chat message content and the shared link QR code to the user terminal, so that the shared image with the chat message content and the shared link QR code is displayed on the content display interface of the user terminal;

[0107] In response to a query request sent by a user terminal in the user group for the shared link image or the shared link QR code, sending the chat message content to the user terminals in the user group;

[0108] Or, in response to a query request from another terminal regarding the QR code of the shared link in the shared picture, the chat message content is sent to the other terminal.

[0109] Optionally, the processing module is further used to:

[0110] In response to the reply content carrying the indication information sent by the first user terminal, determining the large language model identifier and / or the user identifier carried in the indication information;

[0111] The reply content is sent to the large language model corresponding to the large language model identifier, so that the large language model replies to the reply content, and / or prompt information carrying the reply content is sent to the second user terminal corresponding to the user identifier.

[0112] In a third aspect, an embodiment of the present application provides an information processing system based on a large language model, the information processing system based on a large language model comprising: a server and a user terminal, the user terminal being connected to the server in communication via an API interface provided by the server;

[0113] The user terminal comprises:

[0114] A comment input module, used to receive a comment message input by a user for at least one chat content in the display interface, and send the comment message to a source end of the chat content, wherein the chat content includes at least one of a chat message, a reply message and a comment message between a large language model and a user terminal, and the source device includes a large language model or a user terminal;

[0115] A target selection module is used to display a selection interface including a large language model identifier and a user identifier in response to a preset operation of a user on a target selection button or in response to an @ symbol input by a user in a message input box; in response to a user's selection operation on at least one target large language model identifier and / or target user identifier in the selection interface, send a comment message input by the user to a large language model corresponding to the target large language model identifier and / or a user terminal corresponding to the user identifier;

[0116] The server comprises:

[0117] a processing module, configured to, in response to at least one first chat message sent by a first user terminal or a request to establish an intelligent group, input at least one first chat message into at least one large language model to obtain at least one first reply message, or connect at least one large language model to the intelligent group and send at least one first chat message to the large model;

[0118] Sending at least one first reply message to the first user terminal, or sending at least one first reply message to the smart group established by the first user terminal;

[0119] a publishing module, configured to publish the chat message and the reply message to a display interface in response to a display request sent by the first user terminal, so that at least one second user terminal obtains the chat message and the reply message from the display interface, the reply message at least including the first reply message, and the chat message at least including the first chat message;

[0120] In response to the comment content sent by at least one of the second user terminals to the chat message and the reply message, the comment content is published to the display interface, or the smart group joined by at least one of the second users.

[0121] Optionally, the server further includes:

[0122] The security filtering module is used to perform a security audit on the output content of the large language model; when the output content fails the security audit, a prompt message indicating that the audit has failed is sent to the user terminal, and the large language model is instructed to regenerate the output content.

[0123] Optionally, the server further includes:

[0124] The database is used to store and manage user information, display content source data, comment messages, like information, forwarding records, and historical conversation content between users and the large language model.

[0125] In a fourth aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of any one of the above-mentioned information processing methods based on a large language model.

[0126] In a fifth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any one of the above-mentioned information processing methods based on a large language model.

[0127] In a sixth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of any one of the above-mentioned information processing methods based on a large language model.

[0128] According to the specific embodiments provided in this application, this application discloses the following technical effects:

[0129] The present application provides an information processing method, apparatus, device and storage medium based on a large language model, which publishes the chat content between a first user terminal and the large language model to a display platform or an intelligent group, so that other users can participate in the comment process of the chat content in the display platform or the intelligent group through a second terminal, thereby enabling users to obtain various comments of various users on the chat content of the large language model, thereby improving the richness of the comments on the chat content of the large language model.

[0130] This application solves the problem of authenticity and accuracy of AIGC content information:

[0131] As AI generates more and more content, the supervision of the authenticity and accuracy of Internet content information will face new challenges. This application enables users to participate in the comments on AIGC, which is content generated by AI, and the large language model also participates in the conversations, chats, messages, replies, comments, and replies to comments with users, and even multiple large language models have conversations, chats, messages, replies, comments, and replies to comments. The reply content of other large language models is analyzed and summarized, and incorrect reply content is corrected. It can also obtain the user's evaluation of the content of the large language model from the user's reply, comment and like data on the reply content of the large language model, correct its erroneous content and update it to the database, so that users can better distinguish right from wrong in the conversations, chats, and comments between large language models and between large language models, and correct errors in time to reduce or avoid the generation of false and untrue content.

[0132] This application solves the problem that a single large language model is not good at certain aspects of content:

[0133] Since the content data and quantity of training samples of different large language models are not exactly the same, each large language model has its own characteristics and strengths. Multiple large language models participating in answering questions, replying to messages and replying to comments in the same chat interface will make up for the deficiencies or errors of the large language models and can effectively improve the comprehensiveness, authenticity and accuracy of AIGC content information.

[0134] In addition, this application also solves the shortcomings of traditional Internet media content such as user-generated content (UGC), professionally generated content (PGC) and professional user-generated content (PUGC). The participation of the large language model makes the content of AIGC richer, and the interactivity between the large language model and users is better, further enhancing the interest. BRIEF DESCRIPTION OF THE DRAWINGS

[0135] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0136] Figure 1 A flowchart of an information processing method based on a large language model provided in one embodiment of the present application;

[0137] Figure 2 A schematic diagram of the structure of an information processing system based on a large language model provided in one embodiment of the present application;

[0138] Figure 3 One of the interface schematic diagrams of an information processing method based on a large language model provided in one embodiment of the present application;

[0139] Figure 4 A second interface diagram of an information processing method based on a large language model provided in an embodiment of the present application;

[0140] Figure 5 A third interface diagram of an information processing method based on a large language model provided in an embodiment of the present application;

[0141] Figure 6 A fourth interface diagram of an information processing method based on a large language model provided in an embodiment of the present application;

[0142] Figure 7 A fifth interface diagram of an information processing method based on a large language model provided in an embodiment of the present application;

[0143] Figure 8A sixth interface diagram of an information processing method based on a large language model provided in an embodiment of the present application;

[0144] Fig. 9 A seventh interface diagram of an information processing method based on a large language model provided in an embodiment of the present application;

[0145] Fig.10 A schematic diagram of functional modules of an information processing device based on a large language model provided in one embodiment of the present application;

[0146] Fig.11 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0147] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0148] like Figure 1 As shown, some embodiments of the present application provide an information processing method based on a large language model, which includes the following steps 101 to 104. Among them:

[0149] Step 101, in response to at least one first chat message sent by a first user terminal or a request to establish an intelligent group, at least one of the first chat messages is input into at least one large language model to obtain at least one first reply message, or at least one large language model is connected to the intelligent group and at least one of the first chat messages is sent to the large language model.

[0150] Step 102: Send at least one of the first reply messages to the first user terminal, or send at least one of the first reply messages to the smart group established by the first user terminal.

[0151] Step 103, in response to the display request sent by the first user terminal, publish the chat message and the reply message to the display interface, so that at least one second user terminal obtains the chat message and the reply message from the display interface, the reply message includes at least the first reply message, and the chat message includes at least the first chat message.

[0152] Step 104, in response to the comment content sent by at least one of the second user terminals to the chat message and the reply message, the comment content is published to the display interface, or to the smart group joined by at least one of the second users.

[0153] It should be noted that the execution subject of some embodiments of the present application is an electronic device deployed with a large language model, which can be a physical server or a cloud server, so that the server can provide service data based on the large language model to the user terminal. The user terminal is a terminal device installed with a client with a large language model. The client for the large language model dialogue function provided in some embodiments of the present application can be deployed on the user terminal in the form of a mini program, an APP or a web page. The relevant module components can be deployed to the user terminal or the server according to the specific situation of the mini program, the APP or the web page. The specific deployment is based on the actual product situation and demand. Only one of the preferred solutions is given in the embodiments of the present application. The difference between the first user terminal and the second user terminal is that the first user terminal is a terminal device used by users who initiate chats between large language models, and the second user terminal is a terminal device used by users who participate in commenting and interacting with the chat content. Among them, a large language model refers to a machine model with language interaction capabilities, such as LLM (Large Language Model). This model uses a deep learning model trained with a large amount of text data to generate natural language text or understand the meaning of language text. The large language model can handle a variety of natural language tasks, such as text classification, question answering, and dialogue, and is an important path to artificial intelligence. At present, the large language model uses a Transformer architecture and pre-training objectives (such as Language Modeling) similar to the small model. The difference from the small model is the increase in model size, training data, and computing resources.

[0154] In an embodiment of the present application, a user opens an intelligent dialogue interface on a first user terminal, and after inputting a first chat message in the input field of the intelligent dialogue interface, the first user terminal will send the first chat message to the system server, and the system server will input the received first chat message into the large language model, and the large language model will output a first reply message for the first chat message, and the system server will send the first reply message to the first user terminal, and the first user terminal will display the first reply message in the display interface. After viewing the first reply message in the intelligent dialogue interface, the user clicks the display function button in the intelligent dialogue interface, and the first user terminal publishes the chat message and reply message exchanged between the user and the large language model to the display interface. The display interface is used to display the content of the dialogue message in the intelligent dialogue interface selected by the user terminal, and the user can turn on or off the display function by clicking the display button in the intelligent dialogue interface in the user terminal. In the display interface of the first user terminal and / or the second user terminal, a comment content is input for the chat content between the user and the large language model. The content of the comment can be in text form or in image form, or in text plus image. It can be input according to actual needs, and is not limited here. The first user terminal and / or the second user terminal will send the comment content to the system server, and the system server will then publish the comment content to the display interface for display to each user terminal.

[0155] Furthermore, the system also includes an intelligent group, which can display the chat messages and reply messages between the large language model and the user terminal in the intelligent group. The intelligent group can include at least two user terminals and be connected to at least one large language model, or the intelligent group includes at least one user terminal and is connected to at least two large language models. The user terminal in the intelligent group can send messages and reply messages to the intelligent group by requesting the system. The system can send the messages and reply messages generated by the large language model and the user in response to the messages and reply messages to the intelligent group dialogue interface for display.

[0156] The embodiment of the present application publishes the chat content between the first user terminal and the large language model to a display platform or smart group, allowing other users to participate in the comment process of the chat content in the display platform or smart group through a second terminal, thereby allowing users to obtain various comments of various users on the chat content of the large language model, thereby improving the richness of the comments on the chat content of the large language model.

[0157] Optionally, the information processing method based on the large language model further includes:

[0158] Step 201: input a first chat message sent by at least one of the first user terminals into at least two large language models connected to the smart group to obtain a first reply message.

[0159] Step 202: Send the first reply message to at least one of the first user terminals in the smart group, so that the smart group interface on the first user terminal displays the first reply message.

[0160] Step 203, in response to the second reply message sent by the first user terminal to the first reply message, or the second reply message sent to the first reply message of at least one large language model, or the second reply message sent to the first reply message of all large language models, the second reply message is input to at least one large language model to obtain a third reply message, and the third reply message is sent to the smart group interface of the first user terminal for display.

[0161] In an embodiment of the present application, at least two different large language models are connected to the smart group. The system can input the first chat message into each large language model respectively to obtain the first reply message output by each large language model. Then the system can send part or all of the first reply messages output by multiple large language models to some or all of the first user terminals in the smart group, and the user can view the first reply message through the smart group interface of the first user terminal.

[0162] Then, the user can enter a second reply message for part or all of the first reply message in the smart group interface of the first user terminal, and the first user terminal will send the second reply message to the online system. The system will input the second reply messages of several users into part or all of the large language models connected to the smart group to obtain a third reply message, and then send the obtained third reply message to the first user terminal for the user to query the third reply message in the smart group interface of the first user terminal. Of course, the user can also continue to enter the third reply message in the smart group interface of the first user terminal to chat with the large language model connected to the smart group.

[0163] Optionally, the information processing method based on the large language model further includes:

[0164] Step 301: input a first chat message sent by at least one of the first user terminals into a first language model or at least one first language model corresponding to the smart group to which the first user terminal belongs to obtain a first reply message.

[0165] Step 302: input the first reply message into the second largest language model or the second largest language model corresponding to the smart group to obtain a comment message.

[0166] Step 303: Send the comment message to the first user terminal or at least two first user terminals included in the smart group, so that the first user terminal displays the comment message on the display interface, or the first user terminal displays the comment message on the smart group interface.

[0167] In the embodiment of the present application, in the scenario of the comment function or the smart group function, no matter whether the user terminal is a user terminal that initiates the comment or a user terminal that participates in the comment, no matter whether the user is a smart group user or multiple independent users, the user can conduct the above-mentioned chat dialogue through the user terminal with the large language model chat interface before the comment function is initiated or with the large language model in the comment interface after the comment function is initiated. The user can also conduct the above-mentioned chat dialogue with the large language model in the smart group interface. The above-mentioned chat dialogue technical solution can be summarized as follows: the user can initiate a first chat message to multiple large language models, for example, let's discuss topic A (A is the specific content of the topic) together now, please everyone (at least refer to the major language models, if there are other users in the group or comment area, it is also included)), and multiple large language models give the first reply after receiving the chat discussion topic (make the first reply to the discussion topic respectively).

[0168] In an achievable embodiment of the present application, a user initiates a first chat message to multiple large language models, i.e., the discussion topic includes text content for mutual evaluation of content that allows the large language models to reply or comment on each other. In response to the first chat message including the text content for mutual evaluation of content, first, multiple large language models give first replies respectively, and then multiple large language models analyze based on these first replies. When one large language model sees that the first reply given by another large language model is incorrect or objectionable, it can give a second reply that it believes to be correct to express its own opinion, or if it sees that the first reply is correct or very interesting, it can also give an agreeing or similar second reply.

[0169] In an achievable embodiment of the present application, a large model content mutual evaluation button control is set in the comment function or smart group function interface. Clicking the button control can activate the large model content mutual evaluation function, send the preset content mutual evaluation information to multiple large language models, and the large language model responds to the content mutual evaluation information and makes a second reply to the first reply given by another large language model, which can also achieve the effect of the above-mentioned first solution. The content mutual evaluation button control can be clicked to activate when the user needs mutual evaluation between large models, or it can be clicked to activate in advance, so that the function of this solution can be used directly at the beginning of this conversation.

[0170] Optionally, the information processing method based on the large language model further includes:

[0171] Step 401, sending a comment message or reply message given by at least one first language model, or a comment message or reply message given by at least one user terminal to a second language model, so that the second language model analyzes the comment message or reply message to determine whether the comment message or reply message meets the output requirements.

[0172] Step 402: updating the comment messages or reply messages that do not meet the output requirements in the database.

[0173] In an embodiment of the present application, the system may submit the reply message or comment message output by the large language model to other large language models for analysis to determine whether the content of the reply message or comment message is correct or relatively correct, so that the large language models can supervise and correct the output content of each other. If a comment message or reply message that does not meet the output requirements appears, it can be updated to the database connected to the large language model to ensure that the large language model can adjust its own output with reference to the content in the database.

[0174] Optionally, the step 401 includes: sending a comment message or reply message given by at least one first largest language model or a comment message or reply message given by at least one user terminal, and the number of likes for the user terminal to the second largest language model, so that the second largest language model analyzes the comment information or reply message and the number of likes to determine whether the comment message or reply message meets the output requirements.

[0175] In the embodiment of the present application, considering that the number of likes can usually reflect the user's recognition of the output content of the large language model, when the reply message or comment message output by the large language model is input into other large language models for evaluation, the number of likes for the input reply message or comment message can be input into other large language models at the same time, so that when other large language models analyze the comment message or reply message, they can refer to the number of likes of the user for the comment message or reply message, that is, refer to the user's recognition of the comment message or reply message to make a judgment. It can be understood that the higher the number of likes, the higher the user's recognition of the comment message or reply message, and vice versa. The large language model can learn the comment message or reply message with a large number of likes.

[0176] Optionally, the information processing method based on the large language model further includes:

[0177] Step 501: Send a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message.

[0178] Step 502: Send the first reply message of the second largest language model to the first largest language model, so that the first largest language model comprehensively analyzes the first reply message of the second largest language model and the first reply message of the first largest language model to give a first comprehensive reply message.

[0179] Step 503: Send the first reply message of the first largest language model to the second largest language model, so that the second largest language model comprehensively analyzes the first reply message of the first largest language model and the first reply message of the second largest language model to give a second comprehensive reply message.

[0180] Step 504: When the similarity between the first comprehensive reply message and the second comprehensive reply message is greater than or equal to a similarity threshold, the first comprehensive reply message or the second comprehensive reply message is sent to at least one of the user terminals for display.

[0181] Step 505, when the similarity between the first comprehensive reply message and the second comprehensive reply message is less than a similarity threshold, the first reply message of the first large language model, the first reply message of the second large language model, the first comprehensive reply message and the second comprehensive reply message are updated into the database.

[0182] In an embodiment of the present application, the same chat message input by the user can be simultaneously input into multiple large language models, for example, the first large language model and the second large language model. Different large language models will output respective reply messages for the chat message, that is, the first large language model will output the first reply message of the first large language model, and the second language model will output the first reply message of the second large language model. At this time, in order to judge the rationality of the two first reply messages, the reply message output by the large language model can be input into other large language models, so that the other large language models re-output the chat information based on the input reply message to obtain a comprehensive reply message, that is, the first large language model will output the first comprehensive reply message obtained by analyzing the first reply message of the first large language model and the first reply message of the second large language model, and the second large language model will output the first reply message of the first large language model and the second comprehensive reply message obtained by analyzing the first reply message of the second large language model.

[0183] Then the first comprehensive reply message and the second comprehensive reply message can be compared. If the similarity between the comprehensive reply messages is greater than or equal to the similarity threshold, it is determined that the contents of the first comprehensive reply message and the second comprehensive reply message are consistent, and it is determined that these comprehensive reply messages have met the output requirements. Any comprehensive reply message and the second comprehensive reply message can be selected and sent to the user terminal. If the similarity between the comprehensive reply messages is less than the similarity threshold, it is determined that both the first comprehensive reply message and the second comprehensive reply message do not meet the output requirements. The first reply message of the first large language model, the first reply message of the second large language model, the first comprehensive reply message and the second comprehensive reply message generated previously are updated to the database for the large language model to learn as a negative sample.

[0184] Optionally, the information processing method based on the large language model further includes:

[0185] Step 601: Send a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message.

[0186] Step 602: Send the first reply message of the second largest language model to the first largest language model, so that the first largest language model performs a comprehensive analysis on the first reply message of the second largest language model and the first reply message of the first largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content;

[0187] Step 603: When the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content, obtain a first comprehensive reply message output by the first largest language model that does not contain erroneous content.

[0188] Step 604: Send the first reply message of the first largest language model to the second largest language model, so that the second largest language model performs a comprehensive analysis on the first reply message of the first largest language model and the first reply message of the second largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content.

[0189] Step 605: When the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content, obtain a second comprehensive reply message output by the second largest language model that does not contain erroneous content.

[0190] Step 601 is similar to step 501 and will not be described again here.

[0191] For steps 602 to 605, the large language model can combine the reply messages of other large language models to determine whether there is erroneous content in its own and other large language models' first reply messages. The erroneous content may be content that does not match the chat message input by the user, or may violate the output content rules, such as containing illegal words, sensitive words, etc. When the large language model determines that there is erroneous content in itself and other large language models, it can regenerate a comprehensive reply message generated for the chat message or comment message input by the user in combination with the determined erroneous content. The comprehensive reply message will not contain the erroneous content identified by the large language model, thereby ensuring the standardization of the output of the large language model.

[0192] Optionally, the information processing method based on the large language model further includes:

[0193] Step 701: Send the first comprehensive reply message and the second comprehensive reply message to the one user terminal;

[0194] Step 702: In response to the request for continuing processing sent by the user terminal, continue the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message; or, when the first large language model or the second large language model confirms that the first comprehensive reply message or the second comprehensive reply message contains error content, continue the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message;

[0195] Step 703: input the first comprehensive reply message and the second comprehensive reply message into the first large language model and the second large language model respectively, update the first comprehensive reply message using the output data of the first large language model, and update the second comprehensive reply message using the output data of the second large language model;

[0196] Step 704, when the first largest language model or the second largest language model analyzes and confirms that the first comprehensive reply message or the second comprehensive reply message contains erroneous content, the comprehensive analysis process is continued using the first comprehensive reply message and the second comprehensive reply message until the first largest language model and the second largest language model analyze and confirm that the first comprehensive reply message and the second comprehensive reply message do not contain erroneous content.

[0197] In an embodiment of the present application, it is assumed that there are two large language models, large language model A and large language model B. The first comprehensive reply message A described below is equivalent to the first comprehensive reply message in the above embodiment of the present application, and the first comprehensive reply message B is equivalent to the second comprehensive reply message in the above embodiment of the present application, that is, the message marked with A is the output information of the large language model A, and the message marked with B is the output information of the large language model B.

[0198] A chat message of at least one user terminal may be sent to large language model A and large language model B, so that large language model A outputs a first reply message A for the chat message, and large language model B outputs a first reply message B for the chat message;

[0199] Sending the first reply message B to the large language model A, so that the large language model A performs a comprehensive analysis on the first reply message B and the first reply message A, analyzes and determines whether there are erroneous contents in the first reply message A and the first reply message B, and if so, finds out the errors, and gives a first comprehensive reply message A that the large language model A considers to be correct;

[0200] The first reply message A is sent to the large language model B, so that the large language model B performs a comprehensive analysis on the first reply message A and the first reply message B, analyzes and determines whether there are erroneous content in the first reply message A and the first reply message B, and if so, finds out the errors and gives a first comprehensive reply message B that the large language model B considers to be correct; then the first comprehensive reply message A and the first comprehensive reply message B can be sent to at least one user terminal.

[0201] Displaying the first comprehensive reply message A and the first comprehensive reply message B on at least one user terminal, and making a request whether to continue processing according to the comprehensive reply message;

[0202] Or, when at least one of the large language model A and the large language model B determines through analysis and comparison that the first comprehensive reply message A or the first comprehensive reply message B has erroneous content, a request to continue processing is made;

[0203] In response to the request to continue processing, the first comprehensive reply message A is sent to the large language model B, so that the large language model B performs a comprehensive analysis on the first comprehensive reply message A in combination with the first comprehensive reply message B, and gives a second comprehensive reply message B that the large language model B considers to be correct; the first comprehensive reply message B is sent to the large language model A, so that the large language model A performs a comprehensive analysis on the first comprehensive reply message B in combination with the first comprehensive reply message A, and gives a second comprehensive reply message A that the large language model A considers to be correct;

[0204] Analyze and determine whether the second comprehensive reply message B or the second comprehensive reply message A has incorrect content. If so, give a third comprehensive reply message B that the large language model B considers to be correct and a third comprehensive reply message A that the large language model A considers to be correct respectively. Similarly, there may also be a fourth comprehensive reply message and a fifth comprehensive reply message, and so on, until the given comprehensive reply message is determined to be correct by both the large language model A and the large language model B.

[0205] Optionally, the information processing method based on the large language model also includes: updating the first reply message of the first large language model containing erroneous content, the first reply message of the second large language model, the first comprehensive reply message and the second comprehensive reply message into the large language model database.

[0206] In an embodiment of the present application, the first reply message of the first large language model with erroneous content, the first reply message of the second large language model, the first comprehensive reply message, and the second comprehensive reply message can all be updated and stored in the large language model database for use by the large language model for learning, thereby reducing the probability of the large language model outputting erroneous content and improving the accuracy of the large language model's replies.

[0207] Optionally, the information processing method based on the large language model also includes: when at least two large language models have different comment messages or reply messages for the same chat information, comparing and analyzing the comment messages or reply messages to obtain comment messages or reply messages that meet the output requirements.

[0208] In an achievable embodiment of the present application, a user client sends a chat message with the same question to two or more large language models; the two or more large language models respond to the chat message and respectively give two or more comments or reply messages, which are sent to the user client for display; when the user finds on his client that the comments or reply messages given by the two or more large language models are completely different or partially different, he clicks the analysis and processing button to send the two or more comments or reply messages to the two or more large language models together; each large language model analyzes and compares the two or more comments or reply messages, selects or re-gives a correct or relatively correct comment or reply message, and sends it to the user client for display; or, each large language model analyzes and compares the two or more comments or reply messages, selects or re-gives a correct or relatively correct comment or reply message, and when the comment or reply message screened out by the analysis and comparison reaches a preset accuracy threshold, the comment or reply message is determined and sent to the user client for display.

[0209] In an achievable embodiment of the present application, a user client sends a chat message with the same question to two or more large language models; the two or more large language models respond to the chat message and give two or more comments or reply messages respectively; the two or more comments or reply messages are sent together to the two or more large language models, and the respective large language models analyze and compare the two or more comments or reply messages, select or re-give correct or relatively correct comments or reply messages, and send them to the user client for display; or, the two or more comments or reply messages are sent together to the two or more large language models. Large language models, each large language model analyzes and compares the two or more comments or reply messages, gives the analysis and comparison results between the two or more comments or reply messages, and sends them to the user client for display; or, the two or more comments or reply messages are sent to the two or more large language models together, and the respective large language models analyze and compare the two or more comments or reply messages, select or re-give correct or relatively correct comments or reply messages therefrom, and when the comments or reply messages screened out by the analysis and comparison reach a preset accuracy threshold, the comments or reply messages are determined and sent to the user client for display.

[0210] Optionally, the information processing method based on the large language model further includes:

[0211] Step 801, setting a like button on the dialogue interface between the user and the large language model, the intelligent group interface, the display interface, and the comment interface, and displaying the number of likes;

[0212] Step 802, using the large language model to obtain the number of likes and the conversation information of the likes, and perform processing and analysis in combination with the conversation information of the likes;

[0213] Step 803, when the number of likes is greater than or equal to the threshold number of likes, the conversation information of the likes is updated in the database as information recognized by the user.

[0214] In an embodiment of the present application, an interface for displaying chat content between a large language model and a user, such as a dialogue interface between a user and a large language model, an intelligent group interface, a display interface, a comment interface, etc., can be provided with a button for liking the chat content. The user can like the chat content through the button based on the degree of recognition of the displayed chat content, thereby accumulating the number of likes for the chat content. For chat content with a number of likes greater than or equal to a threshold number of likes, the dialogue information can be updated to the database for the large language model to learn from it as a positive sample. The likes number threshold can be a system default setting or a user-defined setting. It can be set according to actual needs and is not limited here.

[0215] Optionally, the information processing method based on the large language model further includes:

[0216] Step 901, copy the chat message content between the user terminal and the large language model to the sharing link interface, generate a sharing link image, or generate a sharing link QR code, or generate a sharing picture containing the chat message content and the sharing link QR code.

[0217] Step 902, sending the shared link image or shared link QR code to the user group associated with the user terminal, or sending the shared image with the chat message content and the shared link QR code to the user terminal, so that the user terminal content display interface displays the shared image with the chat message content and the shared link QR code.

[0218] Step 902: In response to a query request sent by a user terminal in the user group regarding the shared link image or shared link QR code, the chat message content is sent to the user terminals in the user group; or, in response to a query request sent by other terminals regarding the shared link QR code in the shared image, the chat message content is sent to the other terminals.

[0219] In the embodiment of the present application, a shared link image refers to an image that contains part or all of the chat message content, such as a chat message, a reply message, a comment message, etc., a shared link QR code refers to a QR code that can be linked to the platform display interface, and a shared image refers to an image that contains both a shared link QR code and chat message content.

[0220] The user can send the sharing link image or sharing link QR code to a friend or group, or send the sharing picture with the chat message content and the sharing link QR code to the content display interface; the friend or the group user can click the sharing link image or long press to identify the QR code, open the sharing link interface, and view the chat message content; if the user has not registered as a user of this system, click the registration button to register; or, the friend or group user clicks the sharing link image or long press to identify the QR code, opens the sharing link interface, and views the chat content; if the group user and the user are not friends, click the add friend button to become friends; if the sharing link is a sharing link of a group with at least one large language model, click the add group button to join the group chat.

[0221] Optionally, in an achievable embodiment of the present application, in response to a smart group display request sent by the first user terminal, the smart group information is published to a display interface, so that at least a second user terminal obtains the smart group information from the display interface and sends a request to join the group, and responds to the request to join the group and approves it, and adds the second user to the smart group; or, the first user terminal sends an invitation to join the group to the second user terminal, and responds to the invitation to join the group and approves it, and adds the second user to the smart group.

[0222] The smart group information display request sent by the first user terminal publishes the smart group to the display interface, and also includes selecting only the chat messages and reply messages between the first user and the large language model, or selecting or simultaneously selecting the chat messages and reply messages that the smart group user agrees to display, on the smart group interface of the first user terminal. When the first user selects the chat messages and reply messages between the smart group user and the large language model on the smart group interface, the first user sends a display request to the selected user, and in response to the display request, the display request is displayed on the smart group user terminal, and the smart group user clicks the consent display control. In response to the consent display request, the chat messages and reply messages between the smart group user selected by the first user and the large language model are displayed on the smart group information display interface.

[0223] The information processing system of the present application also includes an intelligent group information selection module for running and processing the selection and display of the above-mentioned intelligent group information.

[0224] Optionally, the information processing method based on the large language model further includes:

[0225] Step 1001, in response to a reply content carrying indication information sent by a first user terminal, determining a large language model identifier and / or a user identifier carried in the indication information;

[0226] Step 1002: Send the reply content to the large language model corresponding to the large language model identifier, so that the large language model replies to the reply content, and / or sends prompt information of the reply content to the second user terminal corresponding to the user identifier.

[0227] In an embodiment of the present application, the indication information may be a specific symbol such as "@", and the user may indicate the user or the large language model by adding indication information + user representation in the input content, or indication information + large language model identifier, such as @first large language model, @user 1, @user 2, etc.

[0228] For user identifiers or large language model identifiers with indication information, the system will regard them as objects that need to be replied to. For large language model identifiers, the reply can be made by sending the reply message input by the user to the large language model corresponding to the large language model identifier. For user identifiers, the reply message can be sent to the user terminal corresponding to the user identifier, prompting the user of the user terminal to reply to the reply message.

[0229] Reference Figure 2 , showing an information processing system based on a large language model provided in an embodiment of the present application, comprising: a server 1 and a user terminal 2, wherein the user terminal 1 is connected to the server 2 in communication via an API interface provided by the server 2;

[0230] The user terminal 2 comprises:

[0231] The comment input module 21 is used to receive a comment message input by a user for at least one chat content in the display interface, and send the comment message to the source end of the chat content, wherein the chat content includes at least one of a chat message, a reply message and a comment message between a large language model and a user terminal, and the source device includes a large language model or a user terminal;

[0232] The target selection module 22 is used to display a selection interface including a large language model identifier and a user identifier in response to a preset operation of a user on a target selection button or in response to an @ symbol input by a user in a message input box; in response to a selection operation of a user on at least one target large language model identifier and / or a target user identifier in the selection interface, send the comment message input by the user to the large language model corresponding to the target large language model identifier and / or the user terminal corresponding to the user identifier;

[0233] The server 1 comprises:

[0234] The processing module 11 is used to respond to at least one first chat message sent by the first user terminal or a request to establish an intelligent group, input at least one first chat message to at least one large language model to obtain at least one first reply message, or connect at least one large language model to the intelligent group and send at least one first chat message to the large model;

[0235] Sending at least one first reply message to the first user terminal, or sending at least one first reply message to the smart group established by the first user terminal;

[0236] a publishing module 12, configured to publish the chat message and the reply message to a display interface in response to a display request sent by the first user terminal, so that at least one second user terminal obtains the chat message and the reply message from the display interface, the reply message at least including the first reply message, and the chat message at least including the first chat message;

[0237] In response to the comment content sent by at least one of the second user terminals to the chat message and the reply message, the comment content is published to the display interface, or the smart group joined by at least one of the second users.

[0238] In the embodiment of the present application, the functions of each module in the server 1 are similar to the steps in an information processing method based on a large language model provided in some of the above embodiments, and will not be repeated here.

[0239] For user terminals, refer to Figure 3 , the user terminal interface is provided with a target selection button with an "@" symbol, of course, the target selection button can also be in other styles, which is only an exemplary description here, and can be set according to actual needs, and is not limited here. The user can open, for example, Figure 4 The selection interface includes a user identifier and a large language model identifier. The user identifier can be in the form of a user nickname and a user avatar. The large language model identifier can be in the form of a large language model logo and a large language model name. The user can click on the user identifier and the large language model identifier in the selection interface to check. After the selection is completed, refer to Figure 5 , the user terminal interface will add an indicator (such as @) and the selected user identifier and / or large language model identifier in the input field, such as Figure 5In the example, the user can click the send button in the user terminal interface to send the chat message in the input field to the user terminal corresponding to the user identifier or the large language model corresponding to the large language model identifier. The user can also enter the @ symbol in the message input box through the input method, and the target selection module responds to the input and opens the Figure 4 The selection interface is used to implement the above selection of the target large language model or the target user.

[0240] Reference Figure 6 The application interface may also include a "hot conversation" function option. Users can click on this function option to input a request to open the hot conversation display interface. The hot conversation display will display the conversation content between the user and the large language model posted on the display platform. These conversation contents can be sorted from high to low according to the number of views, or according to the user's browsing habits, the conversation content that conforms to the user's browsing habits can be placed in the front position. Of course, the sorting method of hot conversations can also be based on other rules, which can be set according to actual needs and are not limited here.

[0241] Reference Figure 7 , users can also enter comments on the conversation content posted on the display platform in the conversation display interface, and then the comment messages will be posted to the display platform. The large language model can also reply to the user's comment messages. Similarly, the reply messages of the large language model will also be posted on the display platform.

[0242] Reference Figure 8 , a target shortcut key can be set in the comment interface of the conversation content, that is, the "@" style key on the right side of the comment input box in the figure. Users can use the target shortcut key to open Figure 4 Select the destination to which the input content is sent from the display list. The target shortcut key is the same as Figure 3 The function types of the target shortcut keys in are not described here.

[0243] Reference Fig. 9 Users can choose to share the conversation content with the large language model to the display platform, to other users, or to the smart group. The shared content can be sent in the form of a sharing page. Users can view the conversation content by clicking on the sharing page. Before sharing, users can edit the sharing title and sharing cover of the sharing page. After confirming that the sharing title and sharing cover are correct, users can click the "Share" button to send the sharing page of the conversation content to the display platform, user or smart group.

[0244] Optionally, the server 1 further includes:

[0245] The security filtering module 13 is used to perform a security audit on the output content of the large language model; when the output content fails the security audit, a prompt message indicating that the audit has failed is sent to the user terminal, and the large language model is instructed to regenerate the output content.

[0246] In an embodiment of the present application, the security filtering module will perform a content security review on the output content output by the large language model, mainly checking the sensitive words and illegal words in the output content that do not meet the output content requirements. If the output content passes the security review, the output content can be sent to the user terminal or published to the display platform. If the output content fails the security review, the output content is directly intercepted and a prompt message is sent to the user terminal indicating that the output content has not passed the security review.

[0247] Optionally, the server 1 further includes:

[0248] Database 14 is used to store and manage user information, display content source data, comment messages, like information, forwarding records, and historical conversation content between users and the large language model.

[0249] In an embodiment of the present application, the database can store various interactive contents between the large language model and the user terminal, as well as content records such as likes, forwardings, and comments of users on the display platform on the published content, and support functions such as user management, message content management, smart group management, forwarding management, large model dialogue management, and security review.

[0250] Based on the same inventive concept, the embodiment of the present application also provides an information processing device based on a large language model for implementing the information processing method based on a large language model involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the one or more embodiments of the information processing device based on a large language model provided below can refer to the limitations of the information processing method based on a large language model above, and will not be repeated here.

[0251] In an exemplary embodiment, Fig.10 As shown, an information processing device 110 based on a large language model is provided, comprising:

[0252] a processing module, configured to, in response to at least one first chat message sent by a first user terminal or a request to establish an intelligent group, input at least one first chat message into at least one large language model to obtain at least one first reply message, or connect at least one large language model to the intelligent group and send at least one first chat message to the large language model;

[0253] Sending at least one first reply message to the first user terminal, or sending at least one first reply message to the smart group established by the first user terminal;

[0254] A publishing module 1102 is used to publish the chat message and the reply message to a display interface in response to a display request sent by the first user terminal, so that at least one second user terminal obtains the chat message and the reply message from the display interface, the reply message at least includes the first reply message, and the chat message at least includes the first chat message;

[0255] In response to the comment content sent by at least one of the second user terminals to the chat message and the reply message, the comment content is published to the display interface, or the smart group joined by at least one of the second users.

[0256] Optionally, the processing module 1101 is further configured to:

[0257] Inputting a first chat message sent by at least one of the first user terminals into at least two large language models connected to the smart group to obtain a first reply message;

[0258] Sending the first reply message to at least one of the first user terminals in the smart group, so that the smart group interface on the first user terminal displays the first reply message;

[0259] In response to a second reply message sent by the first user terminal to the first reply message, or a second reply message sent to a first reply message of at least one large language model, or a second reply message sent to a first reply message of all large language models, the second reply message is input to at least one large language model to obtain a third reply message, and the third reply message is sent to the smart group interface of the first user terminal for display.

[0260] Optionally, the processing module 1101 is further configured to:

[0261] Inputting a first chat message sent by at least one of the first user terminals into a first large language model or at least one first large language model corresponding to the smart group to which the first user terminal belongs, to obtain a first reply message;

[0262] Inputting the first reply message into the second largest language model or the second largest language model corresponding to the smart group to obtain a comment message;

[0263] The comment message is sent to the first user terminal or at least two first user terminals included in the smart group, so that the first user terminal displays the comment message on a display interface, or the first user terminal displays the comment message on a smart group interface.

[0264] Optionally, the processing module 1101 is further configured to:

[0265] Sending a comment message or a reply message given by at least one first language model, or a comment message or a reply message given by at least one user terminal, to a second language model, so that the second language model analyzes the comment message or the reply message to determine whether the comment message or the reply message meets the output requirement;

[0266] The comment message or reply message that does not meet the output requirement is updated in the database.

[0267] Optionally, the processing module 1101 is further configured to:

[0268] Send at least one comment message or reply message given by the first largest language model or at least one comment message or reply message given by a user terminal, as well as the number of likes for the user terminal, to the second largest language model, so that the second largest language model analyzes the comment information or reply message and the number of likes to determine whether the comment message or reply message meets the output requirements.

[0269] Optionally, the processing module 1101 is further configured to:

[0270] Sending a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message;

[0271] Sending the first reply message of the second largest language model to the first largest language model, so that the first largest language model comprehensively analyzes the first reply message of the second largest language model and the first reply message of the first largest language model, and gives a first comprehensive reply message;

[0272] Sending the first reply message of the first largest language model to the second largest language model, so that the second largest language model comprehensively analyzes the first reply message of the first largest language model and the first reply message of the second largest language model to give a second comprehensive reply message;

[0273] When the similarity between the first comprehensive reply message and the second comprehensive reply message is greater than or equal to a similarity threshold, sending the first comprehensive reply message or the second comprehensive reply message to at least one of the user terminals for display;

[0274] When the similarity between the first comprehensive reply message and the second comprehensive reply message is less than a similarity threshold, the first reply message of the first large language model, the first reply message of the second large language model, the first comprehensive reply message and the second comprehensive reply message are updated into the database.

[0275] Optionally, the processing module 1101 is further configured to:

[0276] Sending a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message;

[0277] sending the first reply message of the second largest language model to the first largest language model, so that the first largest language model performs a comprehensive analysis on the first reply message of the second largest language model and the first reply message of the first largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content;

[0278] When the first reply message of the first largest language model and the first reply message of the second largest language model contain error content, obtaining a first comprehensive reply message output by the first largest language model that does not contain error content;

[0279] sending the first reply message of the first largest language model to the second largest language model, so that the second largest language model performs a comprehensive analysis on the first reply message of the first largest language model and the first reply message of the second largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content;

[0280] In the case that the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content, a second comprehensive reply message output by the second largest language model that does not contain erroneous content is obtained.

[0281] Optionally, the processing module 1101 is further configured to:

[0282] Sending the first comprehensive reply message and the second comprehensive reply message to the one user terminal;

[0283] In response to the continuing processing request sent by the user terminal, continue the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message;

[0284] Alternatively, when the first largest language model or the second largest language model confirms that the first comprehensive reply message or the second comprehensive reply message contains erroneous content, the first comprehensive reply message and the second comprehensive reply message are used to continue the comprehensive analysis process:

[0285] The step of continuing the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message comprises:

[0286] Inputting the first comprehensive reply message and the second comprehensive reply message into the first large language model and the second large language model respectively, updating the first comprehensive reply message using the output data of the first large language model, and updating the second comprehensive reply message using the output data of the second large language model;

[0287] When the first largest language model or the second largest language model analyzes and confirms that the first comprehensive reply message or the second comprehensive reply message contains erroneous content, the comprehensive analysis process is continued using the first comprehensive reply message and the second comprehensive reply message until the first largest language model and the second largest language model analyze and confirm that the first comprehensive reply message and the second comprehensive reply message do not contain erroneous content.

[0288] Optionally, the processing module 1101 is further configured to:

[0289] The first reply message of the first large language model containing erroneous content, the first reply message of the second large language model, the first comprehensive reply message, and the second comprehensive reply message are updated into the large language model database.

[0290] Optionally, the processing module 1101 is further configured to:

[0291] When at least two large language models have different comment messages or reply messages for the same chat information, the comment messages or reply messages are compared and analyzed to obtain comment messages or reply messages that meet the output requirements.

[0292] Optionally, the processing module 1101 is further configured to:

[0293] A like button is set in the dialogue interface between the user and the large language model, the intelligent group interface, the display interface, and the comment interface, and the number of likes is displayed;

[0294] Using the large language model to obtain the number of likes and the conversation information of the likes, and performing processing and analysis in combination with the conversation information of the likes;

[0295] When the number of likes is greater than or equal to the likes number threshold, the conversation information of the likes is updated in the database as information recognized by the user.

[0296] Optionally, the processing module 1101 is further configured to:

[0297] Copying the chat message content between the user terminal and the large language model to a sharing link interface, generating a sharing link image, or generating a sharing link QR code, or generating a sharing picture including the chat message content and the sharing link QR code;

[0298] Sending the shared link image or shared link QR code to a user group associated with the user terminal, or sending a shared image with the chat message content and the shared link QR code to the user terminal, so that the shared image with the chat message content and the shared link QR code is displayed on the content display interface of the user terminal;

[0299] In response to a query request sent by a user terminal in the user group for the shared link image or the shared link QR code, sending the chat message content to the user terminals in the user group;

[0300] Or, in response to a query request from another terminal regarding the QR code of the shared link in the shared picture, the chat message content is sent to the other terminal.

[0301] Optionally, the processing module 1101 is further configured to:

[0302] In response to the reply content carrying the indication information sent by the first user terminal, determining the large language model identifier and / or the user identifier carried in the indication information;

[0303] The reply content is sent to the large language model corresponding to the large language model identifier, so that the large language model replies to the reply content, and / or prompt information carrying the reply content is sent to the second user terminal corresponding to the user identifier.

[0304] The embodiment of the present application publishes the chat content between the first user terminal and the large language model to a display platform or smart group, allowing other users to participate in the comment process of the chat content in the display platform or smart group through a second terminal, thereby allowing users to obtain various comments of various users on the chat content of the large language model, thereby improving the richness of the comments on the chat content of the large language model.

[0305] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Fig.11 As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store information interaction data. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, an information processing method based on a large language model is implemented.

[0306] Those skilled in the art will understand that Fig.11 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0307] In an exemplary embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above-mentioned method embodiments when executing the computer program.

[0308] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0309] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0310] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0311] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0312] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0313] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0314] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. An information processing method based on a large language model, characterized in that: The information processing method based on the large language model includes: In response to at least one first chat message sent by a first user terminal or a request to establish an intelligent group, input at least one first chat message to at least one large language model to obtain at least one first reply message, or connect at least one large language model to the intelligent group and send at least one first chat message to the large language model; Sending at least one first reply message to the first user terminal, or sending at least one first reply message to the smart group established by the first user terminal; In response to a display request sent by the first user terminal, publishing the chat message and the reply message to a display interface, so that at least one second user terminal obtains the chat message and the reply message from the display interface, the reply message at least including the first reply message, and the chat message at least including the first chat message; In response to the comment content sent by at least one of the second user terminals to the chat message and the reply message, the comment content is published to the display interface, or the smart group joined by at least one of the second users.

2. The information processing method based on a large language model according to claim 1, characterized in that: The information processing method based on the large language model also includes: Inputting a first chat message sent by at least one of the first user terminals into at least two large language models connected to the smart group to obtain a first reply message; Sending the first reply message to at least one of the first user terminals in the smart group, so that the smart group interface on the first user terminal displays the first reply message; In response to a second reply message sent by the first user terminal to the first reply message, or a second reply message sent to a first reply message of at least one large language model, or a second reply message sent to a first reply message of all large language models, the second reply message is input to at least one large language model to obtain a third reply message, and the third reply message is sent to the smart group interface of the first user terminal for display.

3. The information processing method based on a large language model according to claim 1, characterized in that: The information processing method based on the large language model also includes: Inputting a first chat message sent by at least one of the first user terminals into a first large language model or at least one first large language model corresponding to the smart group to which the first user terminal belongs, to obtain a first reply message; Inputting the first reply message into the second largest language model or the second largest language model corresponding to the smart group to obtain a comment message; The comment message is sent to the first user terminal or at least two first user terminals included in the smart group, so that the first user terminal displays the comment message on a display interface, or the first user terminal displays the comment message on a smart group interface.

4. The information processing method based on a large language model according to any one of claims 1 to 3, characterized in that: The information processing method based on the large language model also includes: Sending a comment message or a reply message given by at least one first language model, or a comment message or a reply message given by at least one user terminal, to a second language model, so that the second language model analyzes the comment message or the reply message to determine whether the comment message or the reply message meets the output requirement; The comment message or reply message that does not meet the output requirement is updated in the database.

5. The information processing method based on a large language model according to claim 4, characterized in that: The step of sending a comment message or a reply message given by at least one first language model, or a comment message or a reply message given by at least one user terminal to a second language model, so that the second language model analyzes the comment message or the reply message, and determines whether the comment message or the reply message meets the output requirement, includes: Send at least one comment message or reply message given by the first largest language model or at least one comment message or reply message given by a user terminal, as well as the number of likes for the user terminal, to the second largest language model, so that the second largest language model analyzes the comment information or reply message and the number of likes to determine whether the comment message or reply message meets the output requirements.

6. The information processing method based on a large language model according to claim 1, characterized in that: The information processing method based on the large language model also includes: Sending a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message; Sending the first reply message of the second largest language model to the first largest language model, so that the first largest language model comprehensively analyzes the first reply message of the second largest language model and the first reply message of the first largest language model to give a first comprehensive reply message; Sending the first reply message of the first largest language model to the second largest language model, so that the second largest language model comprehensively analyzes the first reply message of the first largest language model and the first reply message of the second largest language model to give a second comprehensive reply message; When the similarity between the first comprehensive reply message and the second comprehensive reply message is greater than or equal to a similarity threshold, sending the first comprehensive reply message or the second comprehensive reply message to at least one of the user terminals for display; When the similarity between the first comprehensive reply message and the second comprehensive reply message is less than a similarity threshold, the first reply message of the first large language model, the first reply message of the second large language model, the first comprehensive reply message and the second comprehensive reply message are updated into the database.

7. The information processing method based on a large language model according to claim 1, characterized in that: The information processing method based on the large language model also includes: Sending a chat message of at least one user terminal to at least a first large language model and a second large language model, so that the first large language model outputs a first reply message of the first large language model for the chat message, and the second large language model outputs a first reply message of the second large language model for the chat message; sending the first reply message of the second largest language model to the first largest language model, so that the first largest language model performs a comprehensive analysis on the first reply message of the second largest language model and the first reply message of the first largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content; When the first reply message of the first largest language model and the first reply message of the second largest language model contain error content, obtaining a first comprehensive reply message output by the first largest language model that does not contain error content; sending the first reply message of the first largest language model to the second largest language model, so that the second largest language model performs a comprehensive analysis on the first reply message of the first largest language model and the first reply message of the second largest language model, and analyzes and determines whether the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content; In the case that the first reply message of the first largest language model and the first reply message of the second largest language model contain erroneous content, a second comprehensive reply message output by the second largest language model that does not contain erroneous content is obtained.

8. The information processing method based on a large language model according to claim 7, characterized in that: The information processing method based on the large language model further includes: Sending the first comprehensive reply message and the second comprehensive reply message to the one user terminal; In response to the continuing processing request sent by the user terminal, continue the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message; Alternatively, when the first largest language model or the second largest language model confirms that the first comprehensive reply message or the second comprehensive reply message contains erroneous content, the first comprehensive reply message and the second comprehensive reply message are used to continue the comprehensive analysis process; The step of continuing the comprehensive analysis process using the first comprehensive reply message and the second comprehensive reply message comprises: Inputting the first comprehensive reply message and the second comprehensive reply message into the first large language model and the second large language model respectively, updating the first comprehensive reply message using the output data of the first large language model, and updating the second comprehensive reply message using the output data of the second large language model; When the first largest language model or the second largest language model analyzes and confirms that the first comprehensive reply message or the second comprehensive reply message contains erroneous content, the comprehensive analysis process is continued using the first comprehensive reply message and the second comprehensive reply message until the first largest language model and the second largest language model analyze and confirm that the first comprehensive reply message and the second comprehensive reply message do not contain erroneous content.

9. The information processing method based on a large language model according to any one of claims 6 to 8, characterized in that: The information processing method based on the large language model also includes: The first reply message of the first large language model containing erroneous content, the first reply message of the second large language model, the first comprehensive reply message, and the second comprehensive reply message are updated into the large language model database.

10. The information processing method based on a large language model according to any one of claims 1 to 3, characterized in that: The information processing method based on the large language model also includes: When at least two large language models have different comment messages or reply messages for the same chat information, the comment messages or reply messages are compared and analyzed to obtain comment messages or reply messages that meet the output requirements.

11. The information processing method based on a large language model according to claim 1, characterized in that: The information processing method based on the large language model also includes: A like button is set in the dialogue interface between the user and the large language model, the intelligent group interface, the display interface, and the comment interface, and the number of likes is displayed; Using the large language model to obtain the number of likes and the conversation information of the likes, and performing processing and analysis in combination with the conversation information of the likes; When the number of likes is greater than or equal to the like number threshold, the conversation information of the likes is updated in the database as information recognized by the user.

12. The information processing method based on a large language model according to claim 1, characterized in that: The information processing method based on the large language model also includes: Copying the chat message content between the user terminal and the large language model to a sharing link interface, generating a sharing link image, or generating a sharing link QR code, or generating a sharing picture including the chat message content and the sharing link QR code; Sending the shared link image or shared link QR code to a user group associated with the user terminal, or sending a shared image with the chat message content and the shared link QR code to the user terminal, so that the shared image with the chat message content and the shared link QR code is displayed on the content display interface of the user terminal; In response to a query request sent by a user terminal in the user group for the shared link image or the shared link QR code, sending the chat message content to the user terminals in the user group; Or, in response to a query request from another terminal regarding the sending of a shared link QR code in the shared picture, the chat message content is sent to the other terminal.

13. The information processing method based on a large language model according to claim 1, characterized in that: The information processing method based on the large language model also includes: In response to the reply content carrying the indication information sent by the first user terminal, determining the large language model identifier and / or the user identifier carried in the indication information; The reply content is sent to the large language model corresponding to the large language model identifier, so that the large language model replies to the reply content, and / or prompt information carrying the reply content is sent to the second user terminal corresponding to the user identifier.

14. An information processing device based on a large language model, characterized in that: The information processing device based on the large language model includes: a processing module, configured to, in response to at least one first chat message sent by a first user terminal or a request to establish an intelligent group, input at least one first chat message into at least one large language model to obtain at least one first reply message, or connect at least one large language model to the intelligent group and send at least one first chat message to the large language model; Sending at least one first reply message to the first user terminal, or sending at least one first reply message to the smart group established by the first user terminal; a publishing module, configured to publish the chat message and the reply message to a display interface in response to a display request sent by the first user terminal, so that at least one second user terminal obtains the chat message and the reply message from the display interface, the reply message at least including the first reply message, and the chat message at least including the first chat message; In response to the comment content sent by at least one of the second user terminals to the chat message and the reply message, the comment content is published to the display interface, or the smart group joined by at least one of the second users.

15. An information processing system based on a large language model, characterized in that: The information processing system based on the large language model includes: a server and a user terminal, wherein the user terminal is connected to the server through an API interface provided by the server; The user terminal comprises: A comment input module, used to receive a comment message input by a user for at least one chat content in the display interface, and send the comment message to a source end of the chat content, wherein the chat content includes at least one of a chat message, a reply message and a comment message between a large language model and a user terminal, and the source device includes a large language model or a user terminal; A target selection module is used to display a selection interface including a large language model identifier and a user identifier in response to a preset operation of a user on a target selection button or in response to an @ symbol input by a user in a message input box; in response to a user's selection operation on at least one target large language model identifier and / or target user identifier in the selection interface, send a comment message input by the user to a large language model corresponding to the target large language model identifier and / or a user terminal corresponding to the user identifier; The server comprises: a processing module, configured to, in response to at least one first chat message sent by a first user terminal or a request to establish an intelligent group, input at least one first chat message into at least one large language model to obtain at least one first reply message, or connect at least one large language model to the intelligent group and send at least one first chat message to the large model; Sending at least one first reply message to the first user terminal, or sending at least one first reply message to the smart group established by the first user terminal; a publishing module, configured to publish the chat message and the reply message to a display interface in response to a display request sent by the first user terminal, so that at least one second user terminal obtains the chat message and the reply message from the display interface, the reply message at least including the first reply message, and the chat message at least including the first chat message; In response to the comment content sent by at least one of the second user terminals to the chat message and the reply message, the comment content is published to the display interface, or the smart group joined by at least one of the second users.

16. The information processing system of the large language model according to claim 15, characterized in that: The server also includes: The security filtering module is used to perform a security audit on the output content of the large language model; when the output content fails the security audit, a prompt message indicating that the audit has failed is sent to the user terminal, and the large language model is instructed to regenerate the output content.

17. The information processing system of the large language model according to claim 15, characterized in that: The server also includes: The database is used to store and manage user information, display content source data, comment messages, like information, forwarding records, and historical conversation content between users and the large language model.

18. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the information processing method based on a large language model described in any one of claims 1 to 13.

19. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the information processing method based on a large language model described in any one of claims 1 to 13 are implemented.

Images