https://github.com/russellluo/llm-trans
Your personal Language Translator powered by LLMs.
https://github.com/russellluo/llm-trans
Last synced: 5 months ago
JSON representation
Your personal Language Translator powered by LLMs.
- Host: GitHub
- URL: https://github.com/russellluo/llm-trans
- Owner: RussellLuo
- Created: 2025-02-14T14:06:05.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2025-02-16T12:07:41.000Z (over 1 year ago)
- Last Synced: 2025-12-26T18:49:36.923Z (8 months ago)
- Language: Python
- Homepage:
- Size: 529 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# 🌐 LLM Translate
Your personal Language Translator powered by LLMs.
## Installation
```bash
pip install llm-trans
```
## Quick Start
Copy [settings.yaml](./settings.yaml) to your local directory, and run LLM Translate:
```bash
export OPENAI_API_KEY="your-openai-key"
llm-trans ./settings.yaml
```
> You can also add your own LLM by changing the settings as below:
>
> ```yaml
> llms:
> ...
> - title: Your LLM
> model: your-model
> api_base: your-api-base
> api_version: your-api-version
> api_key: your-api-key # plain key string, or environment variable: ${YOUR_API_KEY}
> ```
## Why?
LLM Translate is like traditional translation tools (e.g. [Google Translate][1]) but is powered by LLMs:
- LLMs are very good at language translation and are still evolving rapidly.
- Any local or online LLMs are supported.
> [!NOTE]
> **Disclaimer**
>
> LLM Translate is not a production-level project, nor is it intended (or able) to replace existing translation tools such as Google Translate and DeepL.
>
> It's just a little toy built with [Coagent][2], [Gradio][3] and LLM (at your disposal), for those who find LLM translations useful at times, but don't want to repeatedly write system prompts every time.
>
> Besides language translation, LLM Translate offers a UI similar to Google Translate, and it also has the capability of language detection (also based on LLM).
>
> Finally, if you don't want to use LLM for translation, but are interested in developing LLM applications, you can treat it as a small code example (just ~300 lines in total).
[1]: https://translate.google.com/
[2]: https://github.com/OpenCSGs/coagent
[3]: https://www.gradio.app/