CN117370985A - Vulnerability environment automatic building method based on large language model - Google Patents
Vulnerability environment automatic building method based on large language model Download PDFInfo
- Publication number
- CN117370985A CN117370985A CN202311318082.XA CN202311318082A CN117370985A CN 117370985 A CN117370985 A CN 117370985A CN 202311318082 A CN202311318082 A CN 202311318082A CN 117370985 A CN117370985 A CN 117370985A
- Authority
- CN
- China
- Prior art keywords
- vulnerability
- model
- environment
- baichuan
- information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
- G06F21/50—Monitoring users, programs or devices to maintain the integrity of platforms, e.g. of processors, firmware or operating systems
- G06F21/57—Certifying or maintaining trusted computer platforms, e.g. secure boots or power-downs, version controls, system software checks, secure updates or assessing vulnerabilities
- G06F21/577—Assessing vulnerabilities and evaluating computer system security
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/951—Indexing; Web crawling techniques
Landscapes
- Engineering & Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Computer Security & Cryptography (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Data Mining & Analysis (AREA)
- Stored Programmes (AREA)
Abstract
The invention provides a vulnerability environment automatic building method based on a large language model, which comprises the following steps: crawling the latest vulnerability information, processing the latest vulnerability information, storing the latest vulnerability information into a vector database, vectorizing user input, recalling the vulnerability environment construction related information through a vector similarity comparison algorithm, and generating a promt according to a promt template to serve as the input of the trimmed baichuan-7B model; generating a vulnerability environment construction strategy and related instructions according to the input promt; decomposing the vulnerability environment building strategy into a task chain through langchain-agents, executing a control instruction, judging whether the vulnerability environment building strategy is successful or not, and if errors occur in the execution process, calling a fine-tuned baichuan-7B model to update the vulnerability environment building strategy; if the current task is successfully executed, continuing to execute the subsequent tasks on the task chain until all the tasks on the task chain are completed, and building the vulnerability environment is successful. The invention can reduce the technical threshold of building the vulnerability environment, greatly improve the efficiency of building the vulnerability environment and relieve the problem of hysteresis of the vulnerability environment.
Description
Technical Field
The invention relates to the technical field of network security and automation, in particular to a vulnerability environment automatic building method based on a large language model.
Background
Large language models (Large Language Models, LLMs) refer to deep learning models trained using large amounts of text data that can generate natural language text or understand the meaning of language text. Large language models, which are trained on large-scale text corpora and contain billions of level parameters, are a series of artificial intelligence models that aim to understand and generate human language. They train on a large amount of text data and can perform a wide range of tasks including text summarization, translation, emotion analysis, etc. The large language model can process various natural language tasks, such as text classification, question-answering, dialogue and the like, and is an important path to artificial intelligence.
With more and more network attacks in the information age, the network security situation is more and more severe, and the network security problem becomes an important issue of increasing attention, and after the occurrence of the vulnerability security event, the analysis and verification of the vulnerability are very important. Building a vulnerability environment is a main way to analyze and verify vulnerabilities.
In the prior art, two main ways are used for constructing the vulnerability environment, but different defects exist:
mode 1: self-building
The disadvantage of this approach is that: the steps for constructing the vulnerability environment are very complicated, so that not only is professional knowledge needed, but also a great deal of time is consumed in the construction process.
And manually writing a set-up environment script, wherein the script file is used for calling the service according to the name of the vulnerability environment and a preset calling model, designating a basic mirror image according to the service name, and setting up the vulnerability environment through the configured script file. The proposal firstly requires relevant technicians to write the script, and if problems occur in the process of executing the script, the vulnerability environment construction task is terminated.
Mode 2: pulling vulnerability environment through open source vulnerability shooting range
The disadvantage of this approach is that: the vulnerability environment in an open source vulnerability range often has a problem of hysteresis.
How to apply a large language model to the construction of a vulnerability environment to solve the technical defects existing in the prior art is one of the technical problems which are solved at present.
Disclosure of Invention
The object of the present invention is to solve at least one of the technical drawbacks.
Therefore, the invention aims to provide a large language model-based vulnerability environment automatic building method, which realizes the function of automatically building the vulnerability environment by using a large language model and Langcain technology.
In order to achieve the above object, an embodiment of the present invention provides a method for automatically building a vulnerability environment based on a large language model, including the following steps:
step S1: the latest vulnerability information is regularly crawled by utilizing a crawler technology, and the vulnerability information is stored in a vector database after being cut;
step S2: vectorizing the input of a user, then comparing vector similarity in a vector database to obtain relevant information of vulnerability environment construction, and generating a promt as the input of the trimmed baichuan-7B model according to a promt template;
step S3: generating a vulnerability environment construction strategy according to the input promt by the trimmed baichuan-7B model;
step S4: the vulnerability environment building strategy generates a task chain through a langchain-agent technology, judges which tool needs to be called to execute a command, and generates the command by a trimmed baichuan-7B model;
step S5: judging whether the command is successful or not, if the error reporting occurs in the execution process, calling the trimmed baichuan-7B model to update the vulnerability environment construction strategy, and if the current task is successfully executed, continuing to execute the subsequent tasks on the task chain until all tasks on the task chain are completed, wherein the vulnerability environment construction is successful.
Further, the vulnerability information includes: vulnerability details, vulnerability concept verification POC information and vulnerability environment construction information.
Further, in the step S1, the vulnerability information is processed, including the following steps:
and cutting the document of the obtained vulnerability information, and storing the obtained vulnerability information into the vector database after Embedding the vulnerability information.
Further, the fine-tuned baichuan-7B model uses baichuan-7B as a basic model, and performs fine tuning on the basic model, including:
introducing a trainable low-rank decomposition matrix in a privately deployed mode, and freezing pre-training weights at the same time; the training data are from belle0.5M instruction fine tuning data and a loophole environment building script history file.
Further, in the step S4, a LangChain agent technology is adopted to process the vulnerability environment building policy.
Further, in the step S5, the trimmed baichuan-7B model is called to update the vulnerability environment construction policy, the updated vulnerability environment construction policy is returned to the step S3, and the vulnerability environment construction policy is disassembled after passing through the promt template.
Further, in the step S5, after the current task is successfully executed, whether the task chain is finished is determined, if not, the trimmed baichuan-7B model is returned, and the subsequent tasks on the task chain are continuously executed according to the generated instruction until the task chain is judged to be finished.
Further, the task chain comprises a plurality of tasks, each task corresponds to a control instruction, and the execution logic of each task is configured with a preset sequence.
According to the embodiment of the invention, the vulnerability environment automatic building method based on the large language model has the following beneficial effects:
(1) By using a large language model technology, the traditional manual vulnerability construction mode is changed into an automatic vulnerability construction mode. Namely, the steps and the commands needed for building the vulnerability environment can be automatically generated through the large language model, and the problem of low efficiency of manually building the vulnerability environment is solved.
(2) And analyzing the environment information required by the loopholes by crawling the latest loophole information, and automatically generating a loophole environment deployment task chain.
(3) The method combines a large language model and task chain ideas, reduces labor cost, improves the efficiency of building the vulnerability environment, and improves the instantaneity of vulnerability verification.
(4) The adopted fine-tuned basic model is baichuan-7B, and the method has better understanding capability in the vertical field.
(5) The Langchain technology is adopted in the problem of building the vulnerability environment.
(6) The method has the fault tolerance which is not built in the traditional script mode, and when a task encounters a fault, the fault tolerance can be judged according to a task chain to try to solve the problem in other modes.
(7) By analyzing the latest vulnerability information and automatically building the vulnerability environment, the problem of mirror image hysteresis of the vulnerability environment of the traditional open source shooting range is solved.
(8) By combining a large language model and a LangChain technology, the personnel for constructing the vulnerability environment do not need to know the dependency relationship of a plurality of services or related commands, and the technical threshold for constructing the vulnerability environment is effectively reduced.
(9) When a fault is reported in the script executing process, the vulnerability environment construction strategy can be updated through the trimmed baichuan-7B model, so that the normal execution of the vulnerability environment construction task is ensured, and the vulnerability environment construction efficiency is greatly improved.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The foregoing and/or additional aspects and advantages of the invention will become apparent and may be better understood from the following description of embodiments taken in conjunction with the accompanying drawings in which:
FIG. 1 is a flow chart of a large language model based vulnerability environment automation building method according to an embodiment of the invention;
FIG. 2 is a flow chart of processing vulnerability information and generating a Prompt as an input to a trimmed baichuan-7B model in accordance with an embodiment of the present invention;
FIG. 3 is a schematic diagram of Self-Attention mechanism according to an embodiment of the present invention;
FIG. 4 is a flow chart of vulnerability environment setup according to an embodiment of the present invention.
Detailed Description
Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative and intended to explain the present invention and should not be construed as limiting the invention.
The invention provides a large language model-based automatic vulnerability environment building method, which realizes the function of automatically building the vulnerability environment by using a large language model and a Langchain technology and is suitable for network security personnel to build the vulnerability environment.
As shown in fig. 1, the method for automatically building a vulnerability environment based on a large language model according to the embodiment of the invention comprises the following steps:
step S1, the latest vulnerability information is periodically crawled by utilizing a crawler technology, and the vulnerability information is stored in a vector database after being processed.
In an embodiment of the present invention, vulnerability information includes: vulnerability details, vulnerability concept verification POC information, vulnerability environment construction and other information. It should be noted that, the vulnerability information is not limited to the above examples, and may be added according to the need, which is not described herein.
When a user needs to build a certain vulnerability environment, the user only needs to input and build an XX vulnerability environment, the system carries out vectorization according to the user input, then obtains a content block of vulnerability environment building information from a vector database through a vector similarity comparison algorithm, processes a promt template, and finally gives the model to generate a final vulnerability environment building strategy.
Referring to fig. 2, the process of the crawled latest vulnerability information includes the following steps: firstly, cutting a document of the obtained vulnerability information, and then storing the obtained vulnerability information into a vector database after Embedding the vulnerability information.
Step S2: vectorizing the input of a user, comparing the vector similarity in a vector database, acquiring relevant information of vulnerability environment construction, and generating a promt according to a promt template to serve as the input of the trimmed baichuan-7B model.
Where promt is a natural language description that is used to provide context and information about the desired output to the model. By using appropriate cues, the model may be guided to generate content related to the desired output.
Step S3: and generating a vulnerability environment construction strategy according to the input promt by the fine-tuned baichuan-7B model.
The fine tuning process of the baichuan-7B model will be described first.
Referring to fig. 3, in the present invention, fine tuning is performed for Wq, wv of Self-Attention of each layer, and since a large amount of resources are consumed in a full-scale fine tuning (w=w0+Δw) manner, a manner of freezing W0 is adopted in the scheme, only Δw is trained, and the parameter magnitude is reduced by a low-rank decomposition manner at the time of training Δw. The dimension of the original parameter is K, the magnitude of the parameter after low-rank decomposition is 2 x r x K, and r is far smaller than K, so that the magnitude of the fine tuning parameter is greatly reduced. In the scheme, the value of r is finally 6 after the downstream task performance.
The trimmed baichuan-7B model is described below:
the fine-tuned baichuan-7B model takes baichuan-7B as a basic model, and baichuan-7B is a 70 hundred million-parameter Chinese-English mixed large-scale pre-training language model consisting of Chinese pinyin, chinese words, chinese characters and English words and phrases. And performing fine tuning on the basic model to obtain a fine-tuned baichuan-7B model.
Specifically, a privatized deployment mode is adopted, fine adjustment is carried out on a baichuan-7B basic model, a trainable low-rank decomposition matrix is introduced, pre-training weights are frozen at the same time, and training data come from belle0.5M instruction fine adjustment data and a loophole environment construction script history file. The model has better understanding ability in the vertical field after fine adjustment.
In the step, a loophole environment construction strategy is generated according to input by utilizing the trimmed baichuan-7B model.
It should be noted that, the core information in building the concrete policy of the vulnerability environment is from the vector database. For example, after the user inputs the name of the vulnerability, the system generates a strategy corresponding to the establishment of the vulnerability environment not directly through the capability of the large model, but recalls the environment establishment information of the vulnerability stored in the vector database, processes the environment establishment information through a template and then gives the environment establishment information to the model to generate a final vulnerability environment establishment strategy.
Step S4: the vulnerability environment building strategy generates a task chain through a langchain-agent technology, judges which tool needs to be called to execute a command, and generates the command by the trimmed baichuan-7B model.
The control instructions form tasks on the task chain, and the actions are logically sequential and are related in sequence, for example, XXX software is installed first and then XXX software is installed, so that each task on the task chain corresponds to an execution sequence.
The vulnerability environment construction strategy is implemented by installing XXX software and then installing XXX software, and the fine-tuned baichuan-7B model is used for splitting the strategy into a plurality of actions through semantics and definitely calling the agency of the control instruction. That is, the function of the trimmed baichuan-7B model is to break this policy down into individual actions by semantics and explicitly call which commands those agents use.
Referring to fig. 4, in this step, a LangChain agent technology is used to process the vulnerability environment construction policy. Specifically, the control instructions are generated by the trimmed baichuan-7B model, and the specific application to invoke to execute the control commands is scheduled by LangChain proxy technology.
Firstly, a loophole environment construction strategy is generated, then the loophole environment construction strategy is disassembled into an action through a Prompt template, then the action is generated into an instruction through a trimmed baichuan-7B model, and then a command line tool is called to execute the instruction.
Step S5: judging whether the command is successful or not, if the error reporting occurs in the execution process, calling the trimmed baichuan-7B model to update the vulnerability environment construction strategy, and if the current task is successfully executed, continuing to execute the subsequent tasks on the task chain until all tasks on the task chain are completed, wherein the vulnerability environment construction is successful.
Specifically, if an error occurs in the execution process, the fine-tuned baichuan-7B model is called to update the vulnerability environment construction strategy, the updated vulnerability environment construction strategy is returned to the step S3, and the vulnerability environment construction strategy is disassembled through a promt template. And inputting error reporting information into the trimmed baichuan-7B model so as to obtain a solution, and then updating the vulnerability environment building strategy. For example, if a task is in error during execution, error reporting information is input into the trimmed baichuan-7B model, for example, whether a method can skip the task is found, so that a solution is generated, and the vulnerability environment building strategy is updated.
And after judging that the current task is successfully executed, judging whether the task chain is ended, if not, returning to the trimmed baichuan-7B model, and continuing to execute the subsequent tasks on the task chain according to the generated instruction until the task chain is judged to be ended.
It should be noted that the task chain includes a plurality of tasks, each task corresponds to a control instruction, and a preset sequence is configured for execution logic of each task.
According to the embodiment of the invention, the vulnerability environment automatic building method based on the large language model has the following beneficial effects:
(1) By using a large language model technology, the traditional manual vulnerability construction mode is changed into an automatic vulnerability construction mode. Namely, the steps and the commands needed for building the vulnerability environment can be automatically generated through the large language model, and the problem of low efficiency of manually building the vulnerability environment is solved.
(2) And analyzing the environment information required by the loopholes by crawling the latest loophole information, and automatically generating a loophole environment deployment task chain.
(3) The method combines a large language model and task chain ideas, reduces labor cost, improves the efficiency of building the vulnerability environment, and improves the instantaneity of vulnerability verification.
(4) The adopted fine-tuned basic model is baichuan-7B, and the method has better understanding capability in the vertical field.
(5) The Langchain technology is adopted in the problem of building the vulnerability environment.
(6) The method has the fault tolerance which is not built in the traditional script mode, and when a task encounters a fault, the fault tolerance can be judged according to a task chain to try to solve the problem in other modes.
(7) By analyzing the latest vulnerability information and automatically building the vulnerability environment, the problem of mirror image hysteresis of the vulnerability environment of the traditional open source shooting range is solved.
(8) By combining a large language model and a LangChain technology, the personnel for constructing the vulnerability environment do not need to know the dependency relationship of a plurality of services or related commands, and the technical threshold for constructing the vulnerability environment is effectively reduced.
(9) When a fault is reported in the script executing process, the vulnerability environment construction strategy can be updated through the trimmed baichuan-7B model, so that the normal execution of the vulnerability environment construction task is ensured, and the vulnerability environment construction efficiency is greatly improved.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
Although embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives, and variations may be made in the above embodiments by those skilled in the art without departing from the spirit and principles of the invention. The scope of the invention is defined by the appended claims and equivalents thereof.
Claims (8)
1. The automatic vulnerability environment building method based on the large language model is characterized by comprising the following steps of:
step S1: the latest vulnerability information is regularly crawled by utilizing a crawler technology, and the vulnerability information is stored in a vector database after being cut;
step S2: vectorizing the input of a user, then comparing vector similarity in a vector database to obtain relevant information of vulnerability environment construction, and generating a promt as the input of the trimmed baichuan-7B model according to a promt template;
step S3: generating a vulnerability environment construction strategy according to the input promt by the trimmed baichuan-7B model;
step S4: the vulnerability environment building strategy generates a task chain through a langchain-agent technology, judges which tool needs to be called to execute a command, and generates the command by a trimmed baichuan-7B model;
step S5: judging whether the command is successful or not, if the error reporting occurs in the execution process, calling the trimmed baichuan-7B model to update the vulnerability environment construction strategy, and if the current task is successfully executed, continuing to execute the subsequent tasks on the task chain until all tasks on the task chain are completed, wherein the vulnerability environment construction is successful.
2. The automated vulnerability environment building method based on large language model of claim 1, wherein the vulnerability information comprises: vulnerability details, vulnerability concept verification POC information and vulnerability environment construction information.
3. The automated vulnerability environment building method based on large language model of claim 1, wherein in step S1, the vulnerability information is processed, comprising the steps of:
and cutting the document of the obtained vulnerability information, and storing the obtained vulnerability information into the vector database after Embedding the vulnerability information.
4. The automated vulnerability environment building method based on large language model of claim 1, wherein the fine-tuned baichuan-7B model uses baichuan-7B as a basic model, and fine-tuning is performed on the basic model, comprising:
introducing a trainable low-rank decomposition matrix in a privately deployed mode, and freezing pre-training weights at the same time; the training data are from belle0.5M instruction fine tuning data and a loophole environment building script history file.
5. The automated vulnerability environment building method based on large language model of claim 1, wherein in step S4, langChain agent technology is adopted to process the vulnerability environment building policy.
6. The automated vulnerability environment building method based on large language model of claim 1, wherein in step S5, the fine-tuned baichuan-7B model is invoked to update the vulnerability environment building policy, the updated vulnerability environment building policy is returned to step S3, passed through the promt template, and disassembled.
7. The automated vulnerability environment building method based on large language model of claim 1, wherein in step S5, after judging that the current task is executed successfully, judging whether the task chain is finished, if not, returning to the trimmed baichuan-7B model, and continuing to execute the subsequent tasks on the task chain according to the generated instruction until judging that the task chain is finished.
8. The automated vulnerability environment building method of claim 1 or 7, wherein the task chain comprises a plurality of tasks, each task corresponding to a control instruction, and the execution logic of each task is configured with a preset order.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311318082.XA CN117370985A (en) | 2023-10-12 | 2023-10-12 | Vulnerability environment automatic building method based on large language model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311318082.XA CN117370985A (en) | 2023-10-12 | 2023-10-12 | Vulnerability environment automatic building method based on large language model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117370985A true CN117370985A (en) | 2024-01-09 |
Family
ID=89395817
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311318082.XA Pending CN117370985A (en) | 2023-10-12 | 2023-10-12 | Vulnerability environment automatic building method based on large language model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117370985A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117610026A (en) * | 2024-01-22 | 2024-02-27 | 广州大学 | A honey-spot vulnerability generation method based on large language model |
| CN118034661A (en) * | 2024-04-12 | 2024-05-14 | 清华大学 | Intelligent task application system of large language model |
-
2023
- 2023-10-12 CN CN202311318082.XA patent/CN117370985A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117610026A (en) * | 2024-01-22 | 2024-02-27 | 广州大学 | A honey-spot vulnerability generation method based on large language model |
| CN117610026B (en) * | 2024-01-22 | 2024-04-26 | 广州大学 | Honey point vulnerability generation method based on large language model |
| CN118034661A (en) * | 2024-04-12 | 2024-05-14 | 清华大学 | Intelligent task application system of large language model |
| CN118034661B (en) * | 2024-04-12 | 2024-06-11 | 清华大学 | Large language model intelligent task application system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN117370985A (en) | Vulnerability environment automatic building method based on large language model | |
| KR102422729B1 (en) | Learning Data Augmentation Policy | |
| CN118760427B (en) | A low-code application development method and system based on AI-assisted generation model | |
| JP7571167B2 (en) | Improved performance of neural networks using automatically discovered failure cases | |
| Katz et al. | Synthesizing, correcting and improving code, using model checking-based genetic programming | |
| CN110321430B (en) | Domain name recognition and domain name recognition model generation method, device and storage medium | |
| KR102377645B1 (en) | AI(Artificial Intelligence) service providing system and method for non-AI experts | |
| Leduc et al. | Synthesis method for hierarchical interface-based supervisory control | |
| Inverardi et al. | Towards an assume-guarantee theory for adaptable systems | |
| US20030217352A1 (en) | Script processing apparatus, script processing method, and script processing program | |
| CN1048460A (en) | Computer-aided decision-making device and method | |
| Paler | Surfbraid: A concept tool for preparing and resource estimating quantum circuits protected by the surface code | |
| Staub | xBIT: an easy to use scanning tool with machine learning abilities | |
| CN118349705A (en) | Network topology structure generation method and device | |
| CN118536491A (en) | Low-resource unified information extraction method based on high-efficiency fine adjustment of large model parameters | |
| CN118364909A (en) | Model reasoning optimization method and device, electronic equipment and storage medium | |
| CN117251150A (en) | Training and industry code processing method and device of industry code model and integrated machine | |
| CN111079932A (en) | Intelligent honeypot system based on reward feedback | |
| CN117291259A (en) | Operator optimization method and device, electronic equipment and storage medium | |
| CN116340495A (en) | Question-answering task processing method and device, storage medium and computer equipment | |
| EP4471691A1 (en) | Agents for executing complex computing tasks in a computing platform | |
| Noll et al. | The Erlang verification tool | |
| Ngo et al. | Command and control of discrete-event systems: Towards online hierarchical control based on feasible system decomposition | |
| CN118034661B (en) | Large language model intelligent task application system | |
| US20250069595A1 (en) | Conversational Artificial Intelligence Runtime System Using A Human-Like Instructions |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |