Generated-Computers/Locals

选用指南

AI 根据 README 整理 · 2026/11/5 18:17:28

使用方式

通过命令行使用，可使用 `locals code` 启动编码助手、`locals run` 运行多代理任务、`locals chat` 聊天等。

查看 README 依据

| [Kilo Code](https://github.com/Kilo-Org/kilocode) | `locals launch kilo` | | [OpenCode](https://github.com/anomalyco/opencode) | `locals code ~/my-project` | | [OpenHands](https://github.com/OpenHands/OpenHands) | `locals launch openhands` | | [Cline](https://github.com/cline/cline) | `locals launch cline` | | [Aider](https://github.com/Aider-AI/aider) | `locals launch aider` | | [Roo Code](https://github.com/RooCodeInc/Roo-Code) | in VS Code, the settings from `locals launch editor` ([how](#use-it-in-apps-and-editors)) |

locals code ~/my-project --run "add a --verbose flag to cli.py and run the tests" ``` **Let a team of agents do a bigger task.** Several agents work on the parts of the task at the same time, each in its own copy of your project, and your tests decide when it is done. Your project is changed only when the tests pass: ```bash locals run "write wordcount.py with a function count_words(text) and tests in test_wordcount.py" \ --dir ~/my-project --check "python3 -m pytest -q" ``` **Ask a question about your code.** An agent reads the project and prints the answer: ```bash

locals run "Which function parses dates, and which format does it expect?" --dir ~/my-project ``` **Chat with the model** in the Terminal, like a private chatbot: ```bash locals chat ``` **Turn text into structured data.** Describe the fields you want in a JSON schema, for example in a file called `person.json`: ```json {"type": "object", "properties": {"name": {"type": "string"}, "age": {"type": "integer"}, "email": {"type": "string"}, "status": {"enum": ["pending", "approved", "rejected"]}, "tags": {"type": "array", "items": {"type": "string"}}},

部署要求

需要搭载 Apple 芯片（M1 或更新）的 Mac 且 macOS 14 或更高版本；16 GB 内存效果最佳，8 GB 需使用较小模型。

查看 README 依据

You need a Mac with Apple silicon (M1 or newer) and macOS 14 or newer. A Mac with 16 GB of memory gives the best results; with 8 GB, use one of the smaller models. Open the Terminal app and run: ```bash curl -LsSf https://raw.githubusercontent.com/Generated-Computers/Locals/main/install.sh | sh ``` Then open a new Terminal window (the installer adds Locals to your shell's PATH, which only new windows read) and start a coding assistant on one of your projects: ```bash locals code ~/my-project ``` After the first start, which downloads the model, Locals starts in seconds.

已知限制

在不便于复用的情况下，单个请求的速度与 mlx-lm 相当；对于超出调优内核的格式（如 mxfp4）速度无提升；在 M5 芯片上增益较小（1.5 倍而非 3 倍）。

查看 README 依据

own. A 3,000-token request that arrives during a 30,000-token one gets its first token after 7 s instead of 41. ### Where Locals is not faster - **One request alone, with nothing to reuse**: about the same as mlx-lm (64 against 61 tokens per second on the M2 Pro). It is faster alone when the reply repeats the prompt (file rewrites) or the prompt is cached. - **Prompt-bound work**: reading a prompt is limited by the GPU's arithmetic. With 8,000-token prompts and 8 requests at once the server writes 57 tokens per second; most of the time goes into reading.

- **Formats outside the tuned kernels**: 4-bit affine quantization in groups of 64 is covered; mxfp4 (a mixture-of-experts model, Mellum) ran at mlx-lm's speed. - **The M5's neural accelerators**: MLX's own kernels already use them, so the gain there is 1.5x, not 3x (before Locals' own `nax` kernel, which is experimental). ## Works with your agents <p align="center">