Notes on product, technology and applied AI

Tapirus Open 4B: fine tuning and RAG on Qwen3-4B for a domain general LLMs don't cover

Why I trained a small language model specialized in refaunation and the restoration of ecological interactions, how I combined fine tuning with RAG, and what I learned publishing the open weights on Hugging Face.

• SLM
• Fine Tuning
• RAG
• Qwen3
• Hugging Face
• Refauna