# HuoziIME **Repository Path**: alegw/HuoziIME ## Basic Information - **Project Name**: HuoziIME - **Description**: No description available - **Primary Language**: Unknown - **License**: GPL-3.0 - **Default Branch**: main - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2026-07-30 - **Last Updated**: 2026-07-30 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # HuoziIME: An On-Device LLM-Enhanced Input Method for Deep Personalization
WeChat GitHub Code License Stars
## Demo Preview
## Build Environment | Component | Version | | :--- | :--- | | Android SDK | `36.1.0`, `35.0.1`, `35.0.0`, `34.0.0` | | Android NDK | `25.2.9519653` | | CMake | `3.22.1` | This project is a secondary development based on [YuyanIme](https://github.com/gurecn/YuyanIme), and it is distributed under the **GPL v3.0** license. ## Build from Source ### Option A: Android Studio (Recommended) 1. Clone the repository: ```bash git clone https://github.com/Shan-HIT/HuoziIME.git cd HuoziIME ``` 2. Open the project in Android Studio. 3. Make sure the required SDK, NDK, and CMake versions above are installed. 4. In `Build Variants`, choose one variant: - `onlineDebug` / `onlineRelease` - `offlineDebug` / `offlineRelease` 5. Build and run on your Android device. ### Option B: Command Line (If local `gradle` is configured) ```bash gradle :app:assembleOfflineDebug gradle :app:assembleOnlineDebug ``` ## Source Code & Installation | Content | Status | Note | | :--- | :---: | :--- | | **Source Code** | ![Open Source](https://img.shields.io/badge/Status-Open%20Source-success) | The core codebase is already public in this repository. | | **APK Installer** | ![Released](https://img.shields.io/badge/Status-Released-success) | v1.0.0-beta is now available in Releases. | --- ## Changelog * **v1.1.0-beta** *(Upcoming)*: Breakthrough in model post-training! Achieved major leaps in model alignment, responsiveness, and overall intelligence.(📝 A quick note: I'm currently swamped with my thesis and internship 😭, so I haven't been able to push this update out as quickly as I'd like. But it's still coming — most likely by the end of May. 🙏 Thank you for your patience!) * **v1.0.1-beta**: Fixed typing latency caused by multiple synchronization attempts and resolved the chat session clearing bug. General system optimizations. * **v1.0.0-beta**: Initial pre-release for the ACL 2026 Demo Track, including the core IME service and companion HuoziIME-Chat app. ## Acknowledgements This project stands on the shoulders of several excellent open source projects. Their ideas, code, and tooling made the development of HuoziIME possible.
| Project | Contribution | | :--- | :--- | | [**YuyanIME**](https://github.com/gurecn/YuyanIme) | The original input method foundation that inspired and enabled the secondary development of HuoziIME. | | [**YuyanSDK**](https://github.com/gurecn/yuyansdk) | Core SDK support used for building the intelligent input experience and related capabilities. | | [**llama.cpp**](https://github.com/ggml-org/llama.cpp) | Efficient on-device LLM inference support that helps bring local model capability to the IME workflow. | | [**Qwen**](https://github.com/QwenLM/Qwen3) | The open model family that provides the language intelligence foundation behind the LLM-enhanced experience. |
> We sincerely thank the maintainers and contributors of these projects for advancing open source research and engineering. --- ### Stay Tuned If you are interested in this project, please click **Star** or **Watch** on GitHub to receive the latest updates.