A Plant-Specific AI Trained to Answer Real Gardening Questions. Built From Scratch and Fine-Tuned to Learn.
SproutBrain is the language-model heart of the Sprout system. A local AI that learns about plants the way you do: through questions, trial, and care.
SproutBrain started as a question: *“What if a robot could talk about plants like a gardener, not a search engine?”*
Version 1 was trained on around 300 handcrafted Q&A examples covering everything from asparagus seasons to beet spacing.
It isn’t fast. It isn’t huge. But it knows what mulch is for and can explain why basil pouts in dry air. That’s enough for now.
Fine-tuned small LLM (tinyLlama)
Supervised fine-tuning using Hugging Face + Colab
Uses analogies, garden metaphors, and gentle confidence
Intended for offline/local use via Raspberry Pi
SproutBrain V1 proves small language models can learn real plant knowledge. Next, I'm connecting with live SproutNodes to combine knowledge with real-time soil data.