Artistic interface |
Experience intelligent Vietnamese typing with context-aware predictions and intuitive radial tone selection |
Revolutionizing Vietnamese text input on iOS with cutting-edge AI technology
V7 Vietnamese Keyboard is not just another input method – it's a revolutionary AI-powered Vietnamese typing experience that transforms how Vietnamese speakers communicate on mobile devices. Built with state-of-the-art machine learning technology and designed for the modern user, V7 brings unprecedented intelligence and efficiency to Vietnamese text input.
- 🧠 Custom GPT-2 Language Model - 10M parameters optimized for Vietnamese with >70% top-1 prediction accuracy
- 📚 Advanced Tokenization - 21,869 Vietnamese-specific tokens with greedy merge algorithm for optimal word boundaries
- 💾 Bias Vector Learning - Personalization system that adapts to your vocabulary and writing style across sessions
- ⎈︎ Radial Menu Interface - Revolutionary gesture-based tone mark selection
- 🎪 Context-Aware Punctuation - Smart punctuation menu accessible via spacebar radial gesture
- 🌓 Adaptive Design - Modern translucent UI that responds to iOS appearance modes
- ⚡ CoreML Optimized - Native iOS inference with Neural Engine for near real-time response
Tone Marks: ◌́ (sắc) · ◌̀ (huyền) · ◌̉ (hỏi) · ◌̃ (ngã) · ◌̣ (nặng) · ◌ (ngang-không dấu)
| Component | Details |
|---|---|
| Model | GPT-2 (10M params) |
| Vocabulary | 21,869 Vietnamese tokens |
| Context | 32 tokens with intelligent truncation |
| Accuracy | >70% top-1, >90% top-5 |
| Performance | Real time prediction |
v7-ios/
├── v7Keyboard/ # Main app container
├── v7KeyboardExtension/ # Keyboard extension core
│ ├── KeyboardViewController.swift # Main UI controller
│ ├── V7GPTModel.swift # CoreML model wrapper
│ ├── V7GPTTokenizer.swift # Vietnamese tokenization
│ ├── BiasVectorManager.swift # Personalization engine
│ ├── RadialMenu.swift # Gesture-based tone input
│ ├── CacheManager.swift # Prediction caching
│ └── Constants.swift # Configuration
This project is licensed under the MIT License - see the LICENSE file for details.
- karpathy/nanoGPT - Base model architecture and training framework
- EthanSK - iOS keyboard extension foundation


