Typhoon-Gemma3-KordTranslate-EN-TH-4B-GGUF

🇹🇭🇺🇸 Typhoon Gemma 3 KordTranslate EN-TH 4B (GGUF)

<p align="center"> <img src="https://i.ibb.co/q3H1WPgj/logo.png" width="220" alt="KordTranslate Logo"> </p>

Typhoon Gemma 3 KordTranslate EN-TH 4B (GGUF) is the official GGUF quantized release of Typhoon Gemma 3 KordTranslate EN-TH 4B, optimized for fast and efficient local inference with llama.cpp and compatible applications including LM Studio, Ollama, KoboldCpp, Jan, and other GGUF-supported runtimes.

The original model is instruction-tuned specifically for high-quality English ↔ Thai machine translation, while preserving the multilingual understanding inherited from the Typhoon Gemma 3 base model.

---

✨ Features

• 🇺🇸 English → Thai translation
• 🇹🇭 Thai → English translation
• Instruction-tuned for translation tasks
• Optimized GGUF quantizations for local inference
• CPU and GPU compatible
• Supports llama.cpp ecosystem
• Multiple quantization variants for different hardware

---

📋 Model Details

---

📦 Available Quantizations

Recommended Models

• ⭐ Q5_K_M — Best overall choice for most users.
• 💻 Q4_K_M — Great for CPU inference and laptops.
• 🎯 Q8_0 — Best quantized quality.
• 🚀 F16 — Best accuracy if sufficient VRAM/RAM is available.

---

🎯 Intended Use

This model is optimized for:

• English → Thai translation
• Thai → English translation
• Localization
• Offline translation
• Educational applications
• Chatbot translation
• API integration
• Batch document translation

---

📝 Prompt Templates

For the best translation quality, use the following prompts.

English → Thai

Translate the following English text to Thai.

### English:
[Your English text here]

Thai → English

Translate the following Thai text to English.

### Thai:
[Your Thai text here]

---

🚀 Running with llama.cpp

./llama-cli \
-m Typhoon-Gemma3-KordTranslate-EN-TH-4B.Q5_K_M.gguf \
-p "Translate the following English text to Thai.

### English:
Artificial intelligence is transforming the world."

---

🚀 Running with LM Studio

1. Download your preferred GGUF file.
2. Import the model into LM Studio.
3. Load the model.
4. Use the recommended prompt template for translation.

---

🚀 Running with Ollama

Create a Modelfile:

FROM Typhoon-Gemma3-KordTranslate-EN-TH-4B.Q5_K_M.gguf

Create the model:

ollama create kordtranslate -f Modelfile

Run it:

ollama run kordtranslate

Example prompt:

Translate the following Thai text to English.

### Thai:
สวัสดีครับ ยินดีที่ได้รู้จัก

---

💻 Hardware Recommendations

Actual memory usage depends on the inference engine, context length, GPU offloading, and batch size.

---

🏋️ Training

This repository contains GGUF quantized versions of the original model.

The original model was fine-tuned from typhoon-ai/typhoon2.1-gemma3-4b using Supervised Fine-Tuning (SFT) on bilingual English–Thai translation datasets.

No additional fine-tuning has been performed for this release. Only weight quantization has been applied to improve inference efficiency while maintaining translation quality.

---

⚠️ Quantization Notes

GGUF quantization significantly reduces model size and memory requirements while maintaining excellent translation performance.

Compared with the original full-precision model:

• Lower-bit quantizations trade a small amount of quality for reduced memory usage and faster inference.
• Q4_K_M and Q5_K_M provide excellent quality for most translation tasks.
• Q8_0 delivers translation quality very close to the original model.
• F16 preserve the original model weights without quantization.

---

⚠️ Limitations

• Translation quality may decrease for highly specialized domains such as medical, legal, or scientific texts.
• Proper nouns and newly coined terminology may require manual verification.
• The model may occasionally hallucinate or paraphrase ambiguous inputs.
• Human review is recommended for production-critical translations.

---

🙏 Acknowledgments

Special thanks to:

• Typhoon AI for the excellent typhoon2.1-gemma3-4b base model.
• KordAI for developing and fine-tuning the translation model.
• Unsloth for enabling efficient supervised fine-tuning.
• llama.cpp for the GGUF format and inference ecosystem.
• The open-source AI community for advancing multilingual language models.

---

📖 Citation

If you use this model in your work, please cite the original model.

@misc{kordtranslate2026,
  title={Typhoon Gemma3 KordTranslate EN-TH 4B},
  author={KordAI},
  year={2026},
  publisher={Hugging Face},
  howpublished={https://huggingface.co/KordAI/Typhoon-Gemma3-KordTranslate-EN-TH-4B}
}

---

❤️ Support

If you find this project useful, please consider:

• ⭐ Starring the repository
• ❤️ Liking the model on Hugging Face
• 📢 Sharing it with the community

Happy translating! 🇹🇭🇺🇸