skip to content
xi4oh4o | a programmer

Run code autopilot with Ollama locally

/ 2 min read

code-autopilot, ollama, vscode, llm

简单整理一下如何使用continue扩展配合ollama让VS Code支持本地运行的LLM实现AI代码补全

Requirements

  • Ollama.ai
  • continue.dev

在本地准备LLM运行环境

Install Ollama

On the Mac

下载 ollama app https://ollama.com/download 运行即可

On Linux / WSL

CPU Only
Terminal window
docker run -d -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama
Nvidia GPU

安装 Nvidia container toolkit Run Ollama inside a Docker container

Terminal window
docker run -d --gpus=all -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama

Install LLM Model

这里安装 starcoder:3b

Mac

Terminal window
ollama run starcoder:3b

Docker

Terminal window
docker exec -it ollama ollama run starcoder:3b

确保模型下载完成，并成功启动 模型启动

在VSCode安装 continue 扩展

VSCode Extension https://marketplace.visualstudio.com/items?itemName=Continue.continue Continue扩展

配置Tab自动补全

在VS Code扩展安装后，进入”Settings” 点击箭头编辑settings.json 配置Tab自动补全

在JSON配置中增加 tab auto complete 配置

{
...
  "tabAutocompleteModel": {
    "title": "Tab Autocomplete Model",
    "provider": "ollama",
    "model": "starcoder:3b",
    "apiBase": "https://127.0.0.1:11434"
  },
  "tabAutocompleteOptions": {
    "useCopyBuffer": false, // Determines whether the copy buffer will be considered when constructing the prompt. (Boolean)
    "maxPromptTokens": 400, // The maximum number of prompt tokens to use. A smaller number will yield faster completions, but less context. (Number)
    "prefixPercentage": 0.5 // The percentage of the input that should be dedicated to the prefix. (Number)
  },
  "continue.enableTabAutocomplete": true
}
配置Tab自动补全 JSON配置

测试

使用VS Code新建一个Python，这里我们添加一个注释来获得城市天气并输出JSON 测试Python补全 出现预览后可敲击Tab补全 测试Python Tab补全 除了starcoder-3b外，还可以选择 starcoder-1b 或者 deepseek-1b 这种更小的模型以提升速度。