随笔 - 42  文章 - 0  评论 - 1  阅读 - 6929

部署ChatGLM6B模型过程

背景

大语言模型的本地化部署尝试

步骤

  1. 下载项目
    下载页:https://github.com/josStorer/selfhostedAI/releases/tag/v2
    下载selfhostedAI-chatglm-6b-int4-widoowed-nvidia.7z.001-003压缩包

  2. 解压后,更新项目:
    update.bat

  3. 安装pytorch for gpu
    pip install torch1.13.1+cu117 torchvision0.14.1+cu117 --extra-index-url https://download.pytorch.org/whl/cu117

  4. 启动
    chatglm-6b-int4.bat

  5. 启动成功

  6. 使用方法-使用chrome的插件
    下载地址:https://github.com/josStorer/chatGPTBox/releases/tag/v2.3.0
    下载chromium-without-katex-and-tiktoken.zip
    打开chrome的扩展管理,将解压的文件加载到插件系统中,在插件列表打开设置
    API模式-自定义模型
    模型名称-chatglm-6b-int4
    模型地址-http://localhost:8000/chat/completions
    就可以开始使用了

posted on   流浪的夜空  阅读(202)  评论(0编辑  收藏  举报
相关博文:
阅读排行:
· 震惊!C++程序真的从main开始吗?99%的程序员都答错了
· 单元测试从入门到精通
· 【硬核科普】Trae如何「偷看」你的代码?零基础破解AI编程运行原理
· 上周热点回顾(3.3-3.9)
· winform 绘制太阳,地球,月球 运作规律
< 2025年3月 >
23 24 25 26 27 28 1
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25 26 27 28 29
30 31 1 2 3 4 5

点击右上角即可分享
微信分享提示