Method and device for applying computing power of intelligent computing center cloud platform based on natural language

Abstract

The invention provides a method and device for applying computing power of an intelligent computing center cloud platform based on a natural language, and relates to the technical field of intelligent computing centers, intelligent computing centers and computing power infrastructures, and the method comprises the steps: S1, receiving demand information input by a user; s2, analyzing the training target to generate configuration information; and S3, based on the configuration information and the training target, training the first deep learning model to obtain a second deep learning model. According to the invention, the demand information input by the user on the intelligent computing center cloud platform is analyzed to generate the configuration information, so that the first deep learning model is trained and / or reasoned based on the configuration information and the training target, and the second deep learning model meeting the user demand is obtained; the effect that the user can use the computing power of the intelligent computing center cloud platform without related knowledge is achieved.

Description

Technical Field

[0001] The present invention relates to the technical fields of intelligent computing centers, intelligent computing centers and computing power infrastructure, and particularly relates to a method and device for using the computing power of a cloud platform of an intelligent computing center based on natural language. Background Art

[0002] With the rapid development of artificial intelligence technology, "intelligent computing centers" and "intelligent computing centers" have emerged as the times require.

[0003] An "intelligent computing center" refers to a facility that uses large-scale heterogeneous computing power resources, including general computing power and intelligent computing power, and mainly provides the required computing power, data and algorithms for artificial intelligence applications (such as scenarios of artificial intelligence deep learning model development, model training and model inference, etc.). The intelligent computing center covers facilities, hardware, software, and can provide full-stack capabilities from underlying computing power to top-level application enabling.

[0004] The "intelligent computing center" includes but is not limited to the "intelligent computing center".

[0005] An "intelligent computing center", that is, an artificial intelligence computing center, is a type of computing power infrastructure that is based on artificial intelligence theory, adopts an artificial intelligence computing architecture, and provides computing power services, data services and algorithm services required for artificial intelligence applications.

[0006] "Computing power" is the core of "intelligent computing centers" and "intelligent computing centers". It is the ability of computer devices or computing / data centers to process information. It is the ability of computer hardware and software to cooperate to jointly execute a certain computing requirement. It is the computing ability to achieve the output of target results by processing information data. It is a new type of productive force that integrates information computing power, network carrying capacity, and data storage capacity, and mainly provides services to society through computing power infrastructure.

[0007] Since the emergence of intelligent computing centers, intelligent computing centers have been carrying out ultra-large-scale computing power operation tasks such as model training and inference. When users use computing power in an intelligent computing center, according to different purposes, users need to go through a series of complex operations. First, data related to the training target needs to be prepared, and then the required model (fine-tuning / inference) needs to be downloaded. After that, various dependency packages (python / cuda driver / various algorithm packages, etc.) need to be installed. After completing the above preparatory work, start writing / debugging the code. After the code passes, analyze and configure the resources required for this task, and then start the task. The above operations all need to be completed with the cooperation of personnel with professional knowledge, and the work is complex and has a high threshold. Therefore, how to reduce the threshold for users to use the computing power of intelligent computing centers is an urgent problem to be solved. Summary of the Invention

[0008] The present invention provides a method and apparatus for using the computing power of an intelligent computing center cloud platform based on natural language, which solves the problem that the threshold for users to use the intelligent computing center is very high in the prior art.

[0009] To solve the above problems, the present invention is implemented as follows: In a first aspect, the present invention provides a method for using the computing power of an intelligent computing center cloud platform based on natural language, the method comprising: Step S1, receiving requirement information input by a user, the requirement information including a first deep learning model that the user requests to train and / or infer on the intelligent computing center cloud platform, and a training target to be achieved for training and / or inferring the first deep learning model; Step S2, parsing the training target to generate configuration information, the configuration information including resource information required for the intelligent computing center cloud platform to train and / or infer the first deep learning model; Step S3, training and / or inferring the first deep learning model based on the configuration information and the training target to obtain a second deep learning model.

[0010] Optionally, step S1, receiving requirement information input by a user, includes: Step S11, receiving the first deep learning model input or selected by the user, and receiving natural language input by the user; Step S12, performing text processing on the natural language to obtain the training target.

[0011] Optionally, step S2, parsing the training target to generate configuration information, includes: Step S21, parsing the training target to obtain configuration environment information required for training and / or inferring the first deep learning model, the configuration environment information including a configuration environment for training the first deep learning model; Step S22, parsing the training target to obtain code information required for training and / or inferring the first deep learning model, the code information including execution code for training the first deep learning model; Wherein, the configuration information includes the configuration environment information and the code information.

[0012] Optionally, step S3, training and / or inferring the first deep learning model based on the configuration information and the training target to obtain a second deep learning model, includes: Step S31, receiving a training data set input by a user, or searching for and obtaining the training data set in a sample database based on the training target; Step S32: Based on the training data set, the configuration environment information, the code information, and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model.

[0013] Optionally, the step S32: Based on the training data set, the configuration environment information, the code information, and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model, includes: Step S321: Generate a configuration environment for training and / or inferring the first deep learning model based on the configuration environment information; Step S322: Generate code for training and / or inferring the first deep learning model based on the code information; Step S323: Call a system command to execute the code in the configuration environment, and train and / or infer the first deep learning model based on the training objective and the training data set to obtain the second deep learning model.

[0014] Optionally, the training objective includes at least one of the following: model pre-training, model fine-tuning, and model inference.

[0015] In a second aspect, an embodiment of the present invention further provides a device based on an intelligent computing center cloud platform for natural language application. The device includes: A receiving module, configured to receive requirement information input by a user. The requirement information includes a first deep learning model that the user requests to train and / or infer on the intelligent computing center cloud platform, and a training objective to be achieved for training and / or inferring the first deep learning model; An analysis module, configured to analyze the training objective to generate configuration information. The configuration information includes resource information required for training and / or inferring the first deep learning model on the intelligent computing center cloud platform; A training module, configured to train and / or infer the first deep learning model based on the configuration information and the training objective to obtain a second deep learning model.

[0016] In a third aspect, the present invention further provides an electronic device, including a processor, a memory, and a computer program stored on the memory and executable on the processor. When the computer program is executed by the processor, the steps in the method described in the first aspect above are implemented.

[0017] In a fourth aspect, the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the method described in the first aspect above are implemented.

[0018] In a fifth aspect, the present invention further provides a computer program product, including computer instructions, which, when executed by a processor, implement the steps in the method described in the first aspect above.

[0019] The present invention provides a method and apparatus for utilizing the computing power of an intelligent computing center cloud platform based on natural language, relating to the technical fields of intelligent computing centers, intelligent computing centers, and computing power infrastructure. The method includes: Step S1, receiving demand information input by a user, where the demand information includes a first deep learning model that the user requests to train and / or infer on the intelligent computing center cloud platform, and a training objective to be achieved for training and / or inferring the first deep learning model; Step S2, parsing the training objective to generate configuration information, where the configuration information includes resource information required for the intelligent computing center cloud platform to train and / or infer the first deep learning model; Step S3, based on the configuration information and the training objective, training and / or inferring the first deep learning model to obtain a second deep learning model. By parsing the demand information input by the user on the intelligent computing center cloud platform, the present invention generates configuration information, and thus trains and / or infers the first deep learning model based on the configuration information and the training objective, thereby obtaining a second deep learning model that meets the user's needs, achieving the effect that the user can utilize the computing power of the intelligent computing center cloud platform without relevant knowledge. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] To more clearly illustrate the technical solutions of the present invention, the following will briefly introduce the drawings required for the description of the present invention. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0021] Figure 1 is a flowchart of a method for utilizing the computing power of an intelligent computing center cloud platform based on natural language provided in an embodiment of the present invention; Figure 2 is a schematic diagram of the system architecture of the intelligent computing center cloud platform provided in an embodiment of the present invention; Figure 3 is a logical architecture diagram provided in an embodiment of the present invention; Figure 4 is a structural diagram of an apparatus for utilizing the computing power of an intelligent computing center cloud platform based on natural language provided in an embodiment of the present invention; Figure 5 is a structural diagram of an electronic device provided in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] Next, the technical solutions of the present invention will be clearly and completely described in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0023] The "computing power" as described in the present invention refers to: the ability of a computer device or a computing / data center to process information, the ability of computer hardware and software to cooperate to jointly execute a certain computing requirement, the computing ability to achieve the output of a target result through processing information data, a new type of productive force integrating information computing power, network carrying capacity, and data storage capacity, and mainly providing services to society through computing power infrastructure.

[0024] The "computational power" (Computational Power, CP) as described in the present invention refers to: the ability of a data center server to process data and achieve result output, a comprehensive index for measuring the computing ability of a data center, including general computing ability, supercomputing ability, and intelligent computing ability. The commonly used measurement unit is the number of floating-point operations per second (FLOPS, 1 EFLOPS = 10^18 FLOPS), and the larger the value, the stronger the comprehensive computing ability. It is estimated that 1 EFLOPS is approximately the computing power output of 5 Tianhe 2A or 500,000 mainstream server CPUs or 2 million mainstream laptops. The calculation formula is: CP = CP_general + CP_intelligent + CP_super.

[0025] The "carrying capacity" (Network Power, NP) as described in the present invention refers to: the performance of the data transmission ability of computing power facilities, a comprehensive ability including network architecture, network bandwidth, transmission delay, intelligent management and scheduling, etc., involving network transmission inside and between data centers, and a comprehensive index for measuring network transmission scheduling ability.

[0026] The "storage power" (Storage Power, SP) as described in the present invention refers to: the comprehensive ability of a data center in four aspects: data storage capacity, performance, security and reliability, and green and low-carbon, a comprehensive index for measuring the data storage ability of a data center, including external storage devices such as storage arrays and server internal storage devices. The commonly used measurement unit for storage capacity is exabyte (EB, 1 EB = 2^60 bytes), the commonly used measurement unit for performance is the number of read and write operations per second per unit capacity (IOPS / TB, Input / Output Operations Per Second / TB), and the disaster recovery ratio is an important manifestation of security and reliability.

[0027] The "computing power infrastructure" described in the present invention refers to a new type of information infrastructure that integrates information computing power, network carrying capacity, and data storage capacity, enabling centralized computing, storage, transmission, and application of information.

[0028] The "new type of information infrastructure" described in the present invention mainly includes network infrastructures such as 5G networks, fiber broadband networks, backbone networks, international communication networks, and satellite Internet, computing power infrastructures such as data centers, general computing power centers, intelligent computing centers, and supercomputing centers, and new technology facilities such as artificial intelligence, blockchain, and quantum computing.

[0029] The "computing power" described in the present invention includes general computing power, intelligent computing power, and super computing power.

[0030] The "general computing power" described in the present invention refers to the computing power provided by servers based on CPU (Central Processing Unit) chips, which is used to support basic general computing such as cloud computing and edge computing.

[0031] The "intelligent computing power" described in the present invention refers to a computing platform that is scaled and deployed based on dedicated chips such as GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), and ASIC (Application Specific Integrated Circuit) for various artificial intelligence innovation applications, such as natural language processing and machine vision.

[0032] The "super computing power" described in the present invention mainly refers to the computing power provided by high-performance computing clusters such as supercomputers. It utilizes the centralized computing resources of multiple computer systems working in parallel and processes extremely complex or data-intensive problems through a dedicated operating system. It is mainly used for computing in cutting-edge scientific fields, such as planetary simulation, drug molecule design, and gene analysis.

[0033] The "intelligent computing center cloud platform" described in the present invention refers to a cloud computing platform that comprehensively serves based on the hardware resources and software resources of the intelligent computing center.

[0034] The "intelligent computing center cloud platform" described in the present invention includes, but is not limited to, the "intelligent computing center".

[0035] The "intelligent computing center" mentioned in the present invention, namely the artificial intelligence computing center, is a type of computing power infrastructure that provides computing power services, data services, and algorithm services required for artificial intelligence applications based on artificial intelligence theory and using an artificial intelligence computing architecture.

[0036] The "computing power center" described in the present invention refers to a facility mainly composed of infrastructure such as wind, fire, water, and electricity, and IT software and hardware devices, which has computing power, carrying capacity, and storage capacity, including general data centers, intelligent computing centers, supercomputing centers, etc.

[0037] The "supercomputing center" described in the present invention refers to a supercomputing data center, which is a data center based on supercomputers or large-scale computing clusters and can provide functions such as large-scale computing, storage, and network services, and is widely used in application scenarios such as aerospace, national defense, oil exploration, climate modeling, and genome sequencing.

[0038] The "computing power resources" described in the present invention refer to technologies and facilities required for the development of the digital society, which have the capabilities of information computing, transmission, storage, and application, including but not limited to computing resources such as CPUs and GPUs, computing power resources such as switches and routers, storage resources such as storage arrays and distributed storage, security resources such as firewalls and intrusion detection systems, and support and guarantee resources such as wind, fire, water, and electricity.

[0039] The "models" described in the present invention include but are not limited to "large language models" and "multimodal large models".

[0040] The "large language model" described in the present invention refers to a large language model (LLM), which is a language model with a relatively large number of parameters, aiming to understand and generate human language, trained with a large amount of text data, and can perform a wide range of tasks including text summarization, translation, sentiment analysis, etc.

[0041] The "multimodal large model" (Multimodal Large Models) described in the present invention refers to a model trained by jointly combining multimodal information such as text, images, videos, and audio, including but not limited to multimodal large language models.

[0042] The "computing power operation task" described in the present invention refers to a specific workload or job executed on computing power resources and requiring a certain amount of computing power support, usually involving scenarios such as complex data processing, numerical calculation, model training, or simulation.

[0043] Please refer to Figure 1 , Figure 1 is a flowchart of a method for computing power based on an intelligent computing center cloud platform using natural language provided by the present invention. As Figure 1 shown, it includes the following steps: Step S1: Receive the demand information input by the user. The demand information includes the first deep learning model that the user requests to train and / or infer on the intelligent computing center cloud platform, and the training objectives to be achieved for training and / or inferring the first deep learning model.

[0044] In the present invention, the executor of the present invention is the intelligent computing center cloud platform. Specifically, when a user uses computing power on the intelligent computing center cloud platform, corresponding demand information can be generated according to different goals. The demand information includes the first deep learning model that the user requests to be trained and / or inferred on the intelligent computing center cloud platform, and the training goals that need to be achieved for training and / or inference of the first deep learning model. Using the computing power of the intelligent computing center cloud platform means consuming computing power in the intelligent computing center cloud platform to train, fine-tune, and infer the model, etc.

[0045] Among them, the first deep learning model can be a convolutional neural network, a fully connected neural network, a recurrent neural network, etc. Specifically, the convolutional neural network can be applied to tasks such as classification and regression. The fully connected neural network can be applied to tasks such as image classification, target detection and image generation. The recurrent neural network can be applied to natural language processing tasks such as text generation, machine translation and time series prediction.

[0046] The training target is a text obtained after processing the natural language input by the user, which includes the training direction and requirements of the user's request to train the first deep learning model. Specifically, the training target includes at least one of the following: model pre-training, model fine-tuning, and model reasoning.

[0047] Model pre-training refers to the process of training a large dataset to learn common features and patterns. Pre-trained models are usually used in transfer learning scenarios, that is, applying a trained model to another related task that may have a smaller amount of data. When the training goal is model pre-training, for example, the training goal entered by the user can be "pre-train a large language model with a parameter size of 7B in the data directory / data".

[0048] Model fine-tuning refers to the process of further training a pre-trained model for a specific task or data set. The main purpose of fine-tuning is to make the model better adapt to specific task requirements, thereby improving the performance of the model. When the training goal is model fine-tuning, for example, the training goal input by the user can be "Please fine-tune the XXX model into an expert model in the medical field."

[0049] Model inference refers to the process of making predictions or outputs on a machine learning model or a deep learning model. In this process, the model uses input data to generate results or make decisions. When the training goal is model inference, for example, the training goal input by the user can be "A large model has been prepared, please run it."

[0050] Step S2: Analyze the training objective to generate configuration information, where the configuration information includes the resource information required for the intelligent computing center cloud platform to train and / or infer the first deep learning model.

[0051] In the present invention, the training objective input by the user is analyzed to generate the configuration information corresponding to the first deep learning model. Specifically, the configuration information includes the resource information required for the intelligent computing center cloud platform to train and / or infer the first deep learning model. Among them, the resource information is the computing power resources required by the first deep learning model during training and / or inference. For example, the resource information may include computing resources, storage resources, network resources, monitoring and management resources, etc. Among them, the computing resources may be the usage conditions of resources such as CPU, GPU, and memory. The storage resources may be the usage conditions of storage devices such as solid-state drives and disks. The network resources may be network conditions such as bandwidth and latency. The monitoring and management resources may be the usage conditions of control devices such as monitoring tools, resource scheduling, and logs.

[0052] It should be noted that in this embodiment, the configuration information is determined according to the analysis of the training objective. In other embodiments, the user may also input the configuration information on the intelligent computing center cloud platform by themselves or select the configuration information in the intelligent computing center cloud platform. For example, the user can select the corresponding computing resources, storage resources, network resources, monitoring and management resources, etc. according to the first deep learning model by themselves.

[0053] Step S3: Based on the configuration information and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model.

[0054] In the present invention, according to the determined configuration information and the training objective, the first deep learning model is trained and / or inferred to obtain a trained second deep learning model. Among them, the second deep learning model is a model that meets the requirements for training, that is, it can continuously update the first deep learning model during training and continuously repair the problems encountered until a second deep learning model that meets the training objective is obtained. Specifically, after the first deep learning model is trained, the obtained model can be inferred, so as to realize the training process and inference process of the same model.

[0055] The present invention analyzes the demand information input by the user on the intelligent computing center cloud platform to generate configuration information, and then trains and / or infers the first deep learning model based on the configuration information and the training objective, so as to obtain a second deep learning model that meets the user's needs, realizing the effect that the user can use the computing power of the intelligent computing center cloud platform without relevant knowledge.

[0056] In some feasible embodiments, optionally, the step S1 of receiving the requirement information input by the user includes: Step S11 of receiving the first deep learning model input or selected by the user, and receiving the natural language input by the user; Step S12 of performing text processing on the natural language to obtain the training objective.

[0057] In the present invention, when the user inputs requirement information on the intelligent computing center cloud platform, the user can input the first deep learning model that needs to be trained and / or inferred on the intelligent computing center cloud platform, or can select the first deep learning model from multiple models already stored on the intelligent computing center cloud platform.

[0058] The natural language is the text content input by the user. The natural language refers to the text written in the natural language used by humans (such as Chinese, English, French, etc.). This kind of text is informal and is usually used for daily communication, literary creation, news reporting, technical documents, etc. Exemplarily, the natural language input by the user can be "Please select a suitable model, search for training data to fine-tune it, and finally obtain an expert model in the medical field". It should be noted that there is no requirement for the order of determination of the first deep learning model and the natural language in this embodiment, that is, the natural language can be input first, and then the first deep learning model can be input or selected, or the first deep learning model can be input or selected first, and then the natural language can be input.

[0059] After receiving the natural language, it is necessary to perform text processing on the natural language to extract the effective information therein, so as to obtain the training objective. Exemplarily, for example, if the natural language is "Please select a suitable model, search for training data to fine-tune it, and finally obtain an expert model in the medical field", then the training objective generated after extracting this sentence is "Search for training data to fine-tune a suitable model into an expert model in the medical field". By processing the natural language, the natural language input by a user without machine learning knowledge can be processed to obtain text that can be accurately recognized by the intelligent computing center cloud platform, thereby reducing the threshold for the user to use the computing power of the intelligent computing center cloud platform.

[0060] Optionally, the step S2 of parsing the training objective to generate configuration information includes: Step S21 of parsing the training objective to obtain the configuration environment information required for training and / or inferring the first deep learning model, where the configuration environment information includes the configuration environment for training the first deep learning model; Step S22: Analyze the training objective to obtain the code information required for training and / or inferring the first deep learning model. The code information includes the execution code for training the first deep learning model. Among them, the configuration information includes the configuration environment information and the code information.

[0061] In the present invention, after receiving the training objective, analyze the training objective to obtain the configuration information, where the configuration information includes the configuration environment information and the code information. Specifically, analyze the target training to obtain the configuration environment information required for training and / or inferring the first deep learning model, where the configuration environment information includes the configuration environment for training the first deep learning model. Specifically, the configuration environment generally includes the hardware environment, operating system, software environment, virtual environment, etc. The configuration environment is used for performance optimization, dependency management, and maintaining consistency, etc. during the training process of the first deep learning model.

[0062] Simultaneously analyzed with the configuration environment information is also the code information required for the first deep learning model, where the code information includes the execution code. Specifically, the execution code refers to the program code for implementing the model training process, which is mainly used for data loading and preprocessing, model definition, and model compilation, etc. The execution code for model training is not only used to implement and control the model training process, but also involves data processing, model construction and optimization, and model evaluation and saving. Good organization of the training code contributes to code reuse, maintainability, and readability, and also provides a basis for future experiments and improvements.

[0063] Optionally, step S3: Based on the configuration information and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model, includes: Step S31: Receive the training data set input by the user, or search for and obtain the training data set in the sample database based on the training objective; Step S32: Based on the training data set, the configuration environment information, the code information, and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model.

[0064] In the present invention, the training data set refers to the data set used for training the model in the process of machine learning and deep learning, which contains input features and corresponding target labels. The main role of the training data set is to enable the model to learn the patterns and relationships in the data, so as to make accurate predictions or classifications when encountering new data. Specifically, the user can input the training data set by themselves for training the first deep learning model, or the user can request the intelligent computing center cloud platform to search in the sample database to obtain the training data set.

[0065] After obtaining the training data set, based on the training data set, configuration environment information, code information, and training objective, train and / or infer the first deep learning model to obtain a second deep learning model. Among them, the training data set, configuration environment information, and code information are specifically used in the training process of the first deep learning model. The training objective is the training direction of the first deep learning model, that is, it defines the functions, accuracy, model size, etc. of the trained second deep learning model, which are not specifically limited in this embodiment.

[0066] As Figure 2 shown, Figure 2 It is a schematic diagram of the system architecture of the intelligent computing center cloud platform provided in this embodiment. Among them, the intelligent computing center cloud platform includes a large model for the entire system to execute. The large model includes functions such as a sandbox environment, other tools, system commands, network search, intelligent crawler, and file download. After receiving the demand information input by the user, the large model obtains relevant information for model training based on the above functions, so as to complete the training and / or inference of the model.

[0067] Specifically, the sandbox environment is the specific configuration environment for model training. System commands are used to call the configuration environment to train the model. Network search and intelligent crawlers can search for models and training data sets on publicly available websites. In the case where the user does not input a model and training data set, they can search on publicly available websites according to the training objective input by the user to obtain a suitable model or training data set. File download is to download the files searched or crawled by the network search and intelligent crawlers. Other tools can be code writing or other required functions, which are not specifically limited in this embodiment.

[0068] Optionally, step S32, based on the training data set, the configuration environment information, the code information, and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model, includes: Step S321, generate a configuration environment for training and / or inferring the first deep learning model based on the configuration environment information; Step S322, generate code for training and / or inferring the first deep learning model based on the code information; Step S323, call the system command to execute the code in the configuration environment, and train and / or infer the first deep learning model based on the training objective and the training data set to obtain the second deep learning model.

[0069] In this embodiment, as Figure 3 shown, Figure 3This is the logical architecture diagram provided in this embodiment. Here, the instruction is the requirement information input by the user, and the system is the intelligent computing center cloud platform. After the intelligent computing center cloud platform receives the user's instruction, the instruction is split into links such as "data preparation", and the AI automatically completes the work of each link. Specifically, in the data accuracy stage, if the user has already prepared the data, it is directly used. If not, the data required in the user's requirements is first analyzed, and then the data is placed in the specified directory by means of network search such as search tools, download tools / data synthesis / crawling, etc. In the model preparation stage, the training target is analyzed. If the user has already prepared the model, it is directly used. If not, the download tool is called to download from the model website. In the environment preparation stage, the environment preparation tool is called to start one or more physical machines / virtual machines / containers as the configuration environment of the model, and various required dependency packages (python / cuda driver / various algorithm packages, etc.) are installed in the environment. In the code writing stage, the large model writes the corresponding code according to the training target. For example, pre-training code, fine-tuning code, and inference code, etc. In the resource analysis stage, the execution resources required by the model are analyzed, and the execution resources required by the model are scheduled to ensure the normal training process of the model. For example, the execution resources can include storage resources, network resources, and computing power resources, etc. In the program execution stage, the code is executed in the running environment by calling the system command, and the problems encountered are continuously fixed until the model runs normally.

[0070] The present invention parses the requirement information input by the user on the intelligent computing center cloud platform to generate configuration information, and thus trains and / or infers the first deep learning model based on the configuration information and the training target, so as to obtain the second deep learning model that meets the user's requirements, realizing the effect that the user can utilize the computing power of the intelligent computing center cloud platform without relevant knowledge.

[0071] Please refer to Figure 4 , Figure 4 This is the structure diagram of a device for utilizing the computing power of the intelligent computing center cloud platform based on natural language provided by the present invention. As Figure 4 shown, the device 400 for utilizing the computing power of the intelligent computing center cloud platform based on natural language includes: A receiving module 410, configured to receive the requirement information input by the user, where the requirement information includes a first deep learning model that the user requests to train and / or infer on the intelligent computing center cloud platform, and a training target that needs to be achieved for training and / or inferring the first deep learning model; An analysis module 420, configured to analyze the training target to generate configuration information, where the configuration information includes resource information required for the intelligent computing center cloud platform to train and / or infer the first deep learning model; A training module 430, configured to train and / or infer the first deep learning model based on the configuration information and the training objective, so as to obtain a second deep learning model.

[0072] Optionally, the receiving module 410 includes: A first receiving sub-module, configured to receive the first deep learning model input or selected by a user, and receive a natural language input by the user; A processing sub-module, configured to perform text processing on the natural language to obtain the training objective.

[0073] Optionally, the parsing module 420 includes: A first parsing sub-module, configured to parse the training objective to obtain configuration environment information required for training the first deep learning model, where the configuration environment information includes a configuration environment for training and / or inferring the first deep learning model; A second parsing sub-module, configured to parse the training objective to obtain code information required for training and / or inferring the first deep learning model, where the code information includes execution code for training the first deep learning model; Wherein, the configuration information includes the configuration environment information and the code information.

[0074] Optionally, the training module 430 includes: A second receiving sub-module, configured to receive a training data set input by a user, or search for and obtain the training data set in a sample database based on the training objective; A training sub-module, configured to train and / or infer the first deep learning model based on the training data set, the configuration environment information, the code information, and the training objective, so as to obtain a second deep learning model.

[0075] Optionally, the training sub-module includes: A first generating unit, configured to generate a configuration environment for training and / or inferring the first deep learning model based on the configuration environment information; A second generating unit, configured to generate code for training and / or inferring the first deep learning model based on the code information; A training unit, configured to call a system command to execute the code in the configuration environment, and train and / or infer the first deep learning model based on the training objective and the training data set, so as to obtain the second deep learning model.

[0076] Optionally, the training objective includes at least one of the following: model pre-training, model fine-tuning, and model inference.

[0077] The present invention analyzes the demand information input by a user on the cloud platform of the intelligent computing center to generate configuration information, and then trains and / or infers a first deep learning model based on the configuration information and the training objective, so as to obtain a second deep learning model that meets the user's needs, achieving the effect that the user can utilize the computing power of the cloud platform of the intelligent computing center without relevant knowledge.

[0078] An embodiment of the present invention also provides an electronic device. Please refer to Figure 5 , the electronic device may include a processor 501, a memory 502, and a program 5021 stored on the memory 502 and executable on the processor 501.

[0079] When the program 5021 is executed by the processor 501, it can implement Figure 1 any step in the corresponding method embodiment: Step S1: Receive the demand information input by the user, where the demand information includes a first deep learning model that the user requests to train and / or infer on the cloud platform of the intelligent computing center, and a training objective required to train and / or infer the first deep learning model; Step S2: Analyze the training objective to generate configuration information, where the configuration information includes resource information required for the cloud platform of the intelligent computing center to train and / or infer the first deep learning model; Step S3: Based on the configuration information and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model.

[0080] Optionally, the step S1: Receive the demand information input by the user includes: Step S11: Receive the first deep learning model input or selected by the user, and receive the natural language input by the user; Step S12: Perform text processing on the natural language to obtain the training objective.

[0081] Optionally, the step S2: Analyze the training objective to generate configuration information includes: Step S21: Analyze the training objective to obtain configuration environment information required for training and / or inferring the first deep learning model, where the configuration environment information includes the configuration environment for training the first deep learning model; Step S22: Analyze the training objective to obtain code information required for training and / or inferring the first deep learning model, where the code information includes the execution code for training the first deep learning model; Wherein, the configuration information includes the configuration environment information and the code information.

[0082] Optionally, step S3, based on the configuration information and the training objective, trains and / or infers the first deep learning model to obtain a second deep learning model, including: Step S31, receive the training data set input by the user, or search for and obtain the training data set in the sample database based on the training objective; Step S32, based on the training data set, the configuration environment information, the code information, and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model.

[0083] Optionally, step S32, based on the training data set, the configuration environment information, the code information, and the training objective, trains and / or infers the first deep learning model to obtain a second deep learning model, including: Step S321, generate a configuration environment for training and / or inferring the first deep learning model based on the configuration environment information; Step S322, generate code for training and / or inferring the first deep learning model based on the code information; Step S323, call a system command to execute the code in the configuration environment, and train and / or infer the first deep learning model based on the training objective and the training data set to obtain the second deep learning model.

[0084] Optionally, the training objective includes at least one of the following: model pre-training, model fine-tuning, and model inference. In the present invention, by analyzing the requirement information input by the user on the intelligent computing center cloud platform, configuration information is generated, and thus the first deep learning model is trained and / or inferred based on the configuration information and the training objective, so as to obtain a second deep learning model that meets the user's requirements, achieving the effect that the user can utilize the computing power of the intelligent computing center cloud platform without relevant knowledge.

[0085] The embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements each process of the method embodiment of utilizing the computing power of the intelligent computing center cloud platform based on natural language, and can achieve the same technical effect. To avoid repetition, it will not be elaborated here. Among them, the computer-readable storage medium is, for example, a read-only memory (ROM for short), a random access memory (RAM for short), a magnetic disk, or an optical disc, etc.

[0086] Another embodiment of the present invention further provides a computer program product. The computer program product is stored in a storage medium and is executed by at least one processor to implement each process of the above method embodiment for calculating power based on the intelligent computing center cloud platform using natural language, and can achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0087] It should be noted that in this article, the term "including", "comprising" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or further includes elements inherent to such process, method, article or device. Without more limitations, an element defined by the statement "including a..." does not exclude the existence of another identical element in the process, method, article or device including that element.

[0088] Through the description of the above embodiments, those skilled in the art can clearly understand that the above method of the embodiment can be implemented by means of software plus a necessary general hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing a terminal (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in various embodiments of the present invention.

[0089] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the spirit and scope protected by the claims of the present invention, and all of them belong to the protection scope of the present invention.

Claims

1. A method for calculating power based on the computing power of the intelligent computing center cloud platform using natural language, characterized in that, The method includes: Step S1: Receive the requirement information input by the user, where the requirement information includes a first deep learning model that the user requests to train and / or infer on the intelligent computing center cloud platform, and a training objective to be achieved for training and / or inferring the first deep learning model; Step S2: Parse the training objective to generate configuration information, where the configuration information includes resource information required for the intelligent computing center cloud platform to train and / or infer the first deep learning model; Step S3: Based on the configuration information and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model.

2. The method according to claim 1, characterized in that, The step S1: Receive the requirement information input by the user includes: Step S11: Receive the first deep learning model input or selected by the user, and receive the natural language input by the user; Step S12: Perform text processing on the natural language to obtain the training objective.

3. The method according to claim 2, wherein The step S2: Parse the training objective to generate configuration information includes: Step S21: Parse the training objective to obtain configuration environment information required for training the first deep learning model, where the configuration environment information includes the configuration environment for training and / or inferring the first deep learning model; Step S22: Parse the training objective to obtain code information required for training and / or inferring the first deep learning model, where the code information includes the execution code for training the first deep learning model; Wherein, the configuration information includes the configuration environment information and the code information.

4. The method according to claim 3, wherein The step S3: Based on the configuration information and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model includes: Step S31: Receive the training data set input by the user, or search for and obtain the training data set in the sample database based on the training objective; Step S32: Based on the training data set, the configuration environment information, the code information, and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model.

5. The method according to claim 4, characterized in that, The step S32: Based on the training data set, the configuration environment information, the code information, and the training objective, train and / or infer the first deep learning model to obtain a second deep learning model includes: Step S321: Generate a configuration environment for training and / or inferring the first deep learning model based on the configuration environment information; Step S322: Generate code for training and / or inferring the first deep learning model based on the code information; Step S323: Call a system command to execute the code in the configuration environment, and train and / or infer the first deep learning model based on the training objective and the training data set to obtain the second deep learning model.

6. The method according to any one of claims 1-5, characterized in that The training objective includes at least one of the following: model pre-training, model fine-tuning, and model inference.

7. An apparatus for computing power based on the computing power of an intelligent computing center cloud platform using natural language, characterized in that, The device includes: A receiving module, configured to receive requirement information input by a user, where the requirement information includes a first deep learning model that the user requests to train and / or infer on the intelligent computing center cloud platform, and a training objective to be achieved for training and / or inferring the first deep learning model; An analysis module, configured to analyze the training objective to generate configuration information, where the configuration information includes resource information required for the intelligent computing center cloud platform to train and / or infer the first deep learning model; A training module, configured to train and / or infer the first deep learning model based on the configuration information and the training objective to obtain a second deep learning model.

8. An electronic device, characterized in that, Comprising: A processor, a memory, and a program stored on the memory and executable on the processor, where when the program is executed by the processor, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that, Comprising computer instructions, and when the computer instructions are executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images