Large model online development environment building system and method
By designing a large model online development environment to build a system, a variety of development configuration and resource management options are provided, which solves the problems of complex resource allocation and low development efficiency in large model development, and achieves rapid configuration and efficient resource utilization.
Patent Information
- Application Number
- CN202510360628.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-26
- Publication Date
- 2025-07-01
AI Technical Summary
In large-scale model development, it is difficult for the existing technology to quickly configure and manage complex development environments, resulting in high resource occupation, low development efficiency, and inconvenient environment configuration between different devices.
A large-model online development environment construction system was designed, including development configuration modules, model square modules and online development modules, providing a variety of Python language versions, AI framework selection and resource settings, and using the Conda virtual environment and Jupyterlab development environment to realize online development and resource management.
By building a system through the online development environment, the resource allocation time is significantly reduced, development efficiency is improved, resource utilization is optimized, and convenient environment migration between devices is realized, avoiding the problem of repeated configuration of the environment.
Smart Images
Figure CN120234036A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of large model data, and in particular to a system and method for building an online development environment for large models. Background Art
[0002] In the development of machine learning and artificial intelligence (AI) models, code development is an indispensable part. The designed algorithms ultimately need to be coded and run on a computer to detect the effectiveness of the algorithms. With the development of the industry, the number of models has increased and the configuration has become increasingly complex. The emergence of large models has raised higher requirements for the specifications of CPU and GPU resources. How to make programming development more convenient, reduce the time investment in configuration, and improve development efficiency is an urgent problem to be solved: 1. Configuration of various frameworks: There is a wide variety of algorithm frameworks, and different large models use different frameworks. Code development requires the configuration of multiple development environments, which is time-consuming and laborious; 2. Higher resource requirements for large models: The development of large models has higher requirements for CPU and GPU resources. It is difficult for ordinary computers to meet their resource requirements, which raises the development threshold; 3. Excessive memory occupation: For the development of different large models, configuring different environments will occupy a large amount of memory, and the load on the server is too high. Summary of the Invention
[0003] The purpose of the present invention is to propose a system and method for building an online development environment for large models in view of the above technical problems.
[0004] An online development environment building system for large models includes a development configuration module, a model square module, and an online development module; The development configuration module optimizes the online development page, provides different Python language versions and Pytorch model frameworks, and sets development environment resources; The model square module provides a variety of large model selections; The online development module provides a Conda virtual environment for online development of large models.
[0005] Furthermore, for an online development environment building system for large models, the development configuration module includes a development tool selection sub-module, a Python language selection sub-module, an AI framework selection sub-module, and a resource setting sub-module The development tool selection sub-module selects VScode or Jupyter development tools for development; The Python language selection sub-module selects different versions of the Python language; The AI framework selection sub-module selects different model frameworks; The resource setting sub-module sets the number of CPUs, memory, GPUs, and GPU memory storage space used by the model to ensure controllable resources.
[0006] Furthermore, a large model online development environment construction system, the online development module includes a Miniforge tool sub-module and a Jupyterlab development environment sub-module; The Miniforge tool sub-module is a Python environment and package management tool; The Jupyterlab development environment sub-module provides an interactive development environment that enables users to create and manage data science projects, supporting notebooks, code, and data.
[0007] Furthermore, a large model online development environment construction system, the Miniforge tool sub-module includes a lightweight design sub-module and a flexibility design sub-module; The lightweight design sub-module only contains conda and necessary dependency packages to suit lightweight environment management; The flexibility design sub-module reduces the installation volume and supports the operation of mamba and PyPy.
[0008] Furthermore, a large model online development environment construction system, the Jupyterlab development environment sub-module includes a tab interface sub-module, a browser support sub-module, and an interactive display sub-module; The tab interface sub-module implements a multi-tab interface that supports parallel viewing and editing of multiple files; The browser support sub-module supports an integrated terminal, text editor, and file browser; The interactive display sub-module supports the interactive display of rich text, code, and data.
[0009] A method for constructing a large model online development environment includes the following sub-steps: S1: Manage development tools; S11: Select development tools such as VScode and Jupyter; S2: Manage large models; S3: Manage development configurations; S4: Manage the online environment.
[0010] Furthermore, a method for constructing a large model online development environment, S2 includes the following sub-steps: S21: Select a large model to use; S22: Obtain the large models pre-set in the model square; S23: Configure the large model used for development.
[0011] Further, a method for building an online development environment for large models, where S4 includes the following sub-steps: S41: Initialize the online development environment according to the development tool, large model, and development configuration; S42: Open the running development environment for development; S43: Stop the development environment and release resources after the development is completed.
[0012] Advantages of the present invention: Through a system and method for building an online development environment for large models, it is possible to configure the resource environment required for large model development more quickly than local development, reduce the time for configuring resources, and improve development efficiency; configure and start at any time when needed, release resources after the development is completed, and optimize the resource utilization method and efficiency; achieve convenient migration, develop on different devices in the same online environment, and avoid the problem of reconfiguring the environment after changing devices. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] Figure 1 is a structural diagram of a system for building an online development environment for large models.
[0014] Figure 2 is a flowchart of a method for building an online development environment for large models. DETAILED DESCRIPTION OF THE INVENTION
[0015] For a clearer understanding of the technical features, objectives, and effects of the present invention, the specific embodiments of the present invention will now be described with reference to the accompanying drawings.
[0016] As shown in the attached Figure 1 figures, a system for building an online development environment for large models includes a development configuration module, a model square module, and an online development module; The development configuration module optimizes the page for online development, provides different Python language versions, Pytorch model frameworks, and sets the development environment resources; The model square module provides a variety of large model selections; The online development module provides a Conda virtual environment for online development of large models.
[0017] Further, in a system for building an online development environment for large models, the development configuration module includes a development tool selection sub-module, a Python language selection sub-module, an AI framework selection sub-module, and a resource setting sub-module The development tool selection sub-module selects VScode or Jupyter development tools for development; The Python language selection sub-module selects different versions of the Python language; The AI framework selection sub-module selects different model frameworks; The resource setting sub-module sets the number of CPUs, memory, GPUs, and GPU memory storage space used by the model to ensure controllable resources.
[0018] Furthermore, a large model online development environment building system, the online development module includes a Miniforge tool sub-module and a Jupyterlab development environment sub-module; The Miniforge tool sub-module is a Python environment and package management tool; The Jupyterlab development environment sub-module provides an interactive development environment that enables users to create and manage data science projects and supports notebooks, code, and data.
[0019] Furthermore, a large model online development environment building system, the Miniforge tool sub-module includes a lightweight design sub-module and a flexibility design sub-module; The lightweight design sub-module only contains conda and necessary dependency packages to suit lightweight environment management; The flexibility design sub-module reduces the installation volume and supports the operation of mamba and PyPy.
[0020] Furthermore, a large model online development environment building system, the Jupyterlab development environment sub-module includes a tab interface sub-module, a browser support sub-module, and an interactive display sub-module; The tab interface sub-module implements a multi-tab interface and supports parallel viewing and editing of multiple files; The browser support sub-module supports an integrated terminal, text editor, and file browser; The interactive display sub-module supports the interactive display of rich text, code, and data.
[0021] A method for building a large model online development environment includes the following sub-steps: S1: Manage development tools; S11: Select development tools such as VScode and Jupyter; S2: Manage large models; S3: Manage development configurations; S4: Manage the online environment.
[0022] Furthermore, a method for building a large model online development environment, the S2 includes the following sub-steps: S21: Select a large model to use; S22: Obtain the large models pre-set in the model square; S23: Configure the large model for development use.
[0023] Further, a method for setting up an online development environment for a large model, where S4 includes the following sub-steps: S41: Initialize the online development environment according to the development tools, large model, and development configuration; S42: Open the running development environment for development; S43: Stop the development environment and release resources after development is completed.
[0024] Specific Example 1: HAMI Database HAMi, formerly known as k8s-vGPU-scheduler, is middleware for managing heterogeneous devices in Kubernetes. It manages different types of heterogeneous devices, such as GPUs, NPUs, etc., shares heterogeneous devices among Pods, and makes better scheduling decisions based on device topology information and scheduling policies. Its aim is to eliminate the gap between different heterogeneous devices and provide users with a unified management interface without changing the application.
[0025] The HAMI database includes device multiplexing capabilities and device resource isolation capabilities; The device multiplexing capabilities specify video memory to apply for computing power devices or specify the computing power usage ratio to apply for computing power devices; The device resource isolation capabilities support hard isolation of device resources.
[0026] Specific Example 2: Docker Open Platform Docker is developed based on Linux Container technology and is an open platform for developing, delivering, and running application programs. It packages the application service itself, as well as the environment, configuration, resource files, and dependencies required to run the service, into a single version of a software image package that is independent of the operating system distribution.
[0027] Main features are: Lightweight: Docker containers share the operating system kernel, do not require a full operating system to run, are faster to start and shut down, and consume less resources. On the same host, multiple Docker containers can be run to improve resource utilization; Good isolation: Docker containers can provide good isolation, where each container is independent and does not affect each other. This isolation allows multiple instances of an application to run securely on the same host without interfering with each other; Fast deployment and iteration: Docker containers can package the application program and its dependencies together to form an independent container. It enables fast deployment, testing, and debugging of application programs, saving a large amount of time and effort. At the same time, it also supports fast iteration and version control, allowing easy rollback to a previous version or upgrade to a new version.
[0028] Specific Embodiment 3: MinIO Object Storage System MinIO is a high-performance object storage system that is compatible with the Amazon S3 API, which means it can be seamlessly integrated with widely used cloud storage services. MinIO is designed for private cloud and edge computing environments, capable of handling large-scale datasets, and supports multiple data protection mechanisms such as Erasure Coding to ensure high availability and durability of data.
[0029] The main features are as follows: High performance: MinIO can achieve very high throughput and low latency on standard hardware, making it particularly suitable for large-scale data applications that require fast read and write speeds, such as machine learning and big data analysis; Compatibility: It is fully compatible with the Amazon S3 API, which means that all tools, applications, and services using S3 can directly interface with MinIO. This compatibility makes it easy to migrate existing S3-based workloads; Distributed architecture: MinIO can be deployed as a single-node or distributed multi-node cluster. In the distributed mode, data redundancy is provided through Erasure Coding to ensure data is not lost even if multiple nodes fail; Security: Supports TLS / SSL encrypted transmission to ensure data security during transmission, provides authentication and authorization mechanisms, including policy-based access control. Server-side encryption is supported, which can encrypt data stored in MinIO; Easy to deploy and manage: MinIO can be easily deployed in containerized environments such as Kubernetes. Command-line tools and graphical interfaces are provided for management and monitoring; Cross-platform support: Supports multiple operating systems such as Linux, Windows, and macOS, and can also run on ARM architectures, suitable for edge computing scenarios; Open source: MinIO is open-source software released under the GNU Affero General Public License v3.0 (AGPLv3) license.
[0030] The open-source community is active, with a large number of contributors and support resources; Rich plugins and tools: The community provides many tools and plugins for operations such as backup, synchronization, and migration, and supports monitoring tools such as Prometheus and Grafana, facilitating performance monitoring and alarm setting.
[0031] Specific Embodiment 4: Qwen1.5-14B-Chat Qwen-1.5-14B-Chat is a version in the Qwen series of large language models launched by Alibaba Cloud. It is optimized for dialogue and chat scenarios. Based on the Transformer architecture, this model has 14 billion parameters and aims to provide high-quality natural language understanding and generation capabilities, suitable for a variety of application scenarios.
[0032] It has the following characteristics: Large-scale parameters: With 14 billion parameters, Qwen-1.5-14B-Chat can capture complex language patterns and subtle semantic differences. A large number of parameters helps improve the model's ability to understand long texts and the quality of generation.
[0033] Dialogue optimization: The model is specially trained to handle multi-turn conversations and performs excellently in maintaining context coherence. It can understand and respond to users' questions or statements while maintaining the fluency and relevance of the conversation.
[0034] Breadth and depth of knowledge: The training data covers a wide range of topic areas, from general common sense to knowledge in professional fields. Through a large amount of pre-training and fine-tuning, the model can demonstrate a profound knowledge background on different topics.
[0035] Multi-language support: Qwen series models generally support multiple languages. Although mainly optimized for Chinese, they can also handle input and output in English and some other languages, making it more flexible and practical in international applications.
[0036] Security and compliance: It follows strict data privacy protection principles, considers the requirements of security and compliance in design, takes measures to avoid generating harmful, false or biased content, and strives to ensure the security and reliability of the output.
[0037] This solution uses a system and method for building an online development environment for large models, which can configure the resource environment required for large model development more quickly than local development, reduce the time for configuring resources, and improve development efficiency; it can be configured and started at any time when needed, and release resources after development is completed, optimizing the resource utilization method and efficiency; it realizes convenient migration, and different devices can be developed in the same online environment, avoiding the problem of reconfiguring the environment after changing devices.
[0038] The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and what is described in the above embodiments and the specification is only to illustrate the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.
Claims
1. A large model online development environment construction system, characterized in that: Including development configuration module, model square module, online development module; The development configuration module optimizes the online development page, provides different Python language versions, Pytorch model frameworks, and sets development environment resources; The model square module provides a variety of large model options; The online development module provides a Conda virtual environment for developing large models online.
2. A large model online development environment construction system according to claim 1, characterized in that: The development configuration module includes a development tool selection submodule, a Python language selection submodule, an AI framework selection submodule, and a resource setting submodule. The development tool selection submodule selects VScode or Jupyter development tools for development; The Python language selection submodule selects different versions of the Python language; The AI framework selection submodule selects different model frameworks; The resource setting submodule sets the number of CPUs, memory, GPUs, and storage space of GPU memory used by the model to ensure that resources are controllable.
3. A large model online development environment construction system according to claim 1, characterized in that: The online development module includes a Miniforge tool submodule and a Jupyterlab development environment submodule; The Miniforge tool submodule is a Python environment and package management tool; The Jupyterlab development environment submodule provides an interactive development environment that enables users to create and manage data science projects, supporting notebooks, codes, and data.
4. A large model online development environment construction system according to claim 3, characterized in that: The Miniforge tool submodules include a lightweight design submodule and a flexibility design submodule; The lightweight design submodule only contains conda and necessary dependency packages to be suitable for lightweight environment management; The flexible design submodule reduces the installation volume and supports the operation of mamba and PyPy.
5. A large model online development environment construction system according to claim 3, characterized in that: The Jupyterlab development environment submodule includes a tab interface submodule, a browser support submodule, and an interactive display submodule; The tabbed interface submodule implements a multi-tabbed interface, supporting the viewing and editing of multiple files in parallel; The browser support submodule supports an integrated terminal, a text editor, and a file browser; The interactive display submodule supports interactive display of rich text, code and data.
6. A method for building a large model online development environment, implemented based on a large model online development environment building system according to any one of claims 1 to 5, characterized in that: It includes the following sub-steps: S1: Management development tools; S11: Choose VScode or Jupyter development tools; S2: Manage large models; S3: manage development configuration; S4: Manage online environments.
7. A method for building a large model online development environment according to claim 6, characterized in that: The S2 comprises the following sub-steps: S21: Choose to use the large model; S22: Obtaining the large model preset in the model square; S23: Configure the large model used for development.
8. A method for building a large model online development environment according to claim 6, characterized in that: The S4 comprises the following sub-steps: S41: Initialize the online development environment according to the development tools, large models, and development configuration; S42: Open the running development environment for development; S43: After development is completed, stop the development environment and release resources.