Gemma-4-TypeScript-Coder : GGUF

This model is a specialized fine-tune of Gemma 4, engineered for TypeScript-centric web development, strict type safety, and modern full-stack architectures. It was trained using Unsloth Studio for maximum efficiency and precision.

🟦 TypeScript Mastery

This fine-tune specializes in:

• Strict Type Systems: Expertise in complex generics, utility types, and advanced interfaces.
• Modern Frameworks: High proficiency in Next.js, React, Vue 3, and Node.js.
• Visual Logic: Leverages vision-language capabilities to transform UI wireframes or screenshots directly into type-safe components.
• Best Practices: Focus on clean architecture and idiomatic TypeScript patterns.

🤝 Credits & Acknowledgments

A major shout-out to mhhmm for the typescript-instruct-20k dataset. This robust collection of instructions allowed the model to grasp the nuances of the TypeScript ecosystem effectively.

🚀 Usage & Inference

The model is provided in GGUF format, compatible with llama.cpp.

Example usage:

• Standard Text Chat: llama-cli -hf MassivDash/Gemma-4-typescript-coder --jinja
• Vision/Image Tasks: llama-mtmd-cli -hf MassivDash/Gemma-4-typescript-coder --jinja

📂 Available Model Files

• gemma-4-e2b-it.Q8_0.gguf
• gemma-4-e2b-it.BF16-mmproj.gguf

⚠️ Ollama Note for Vision Models

Important: Ollama currently requires a unified blob for vision models.

To use this with Ollama:

1. Ensure your Modelfile is in the same directory as the merged BF16 model.
2. Run: ollama create model_name -f ./Modelfile

🔗 Stay Connected

For more insights on AI development and fine-tuning, visit my blog: 👉 spaceout.pl

---

This model was trained 2x faster with Unsloth