Mini-project
Agriculture & Food
Fine-tuned crop advisory model
Fine-tune a small multilingual model on advisories, and evaluate it with farmers' real questions.
The brief
A farmer app needs answers in local languages on cheap phones and servers. Fine-tune a small multilingual model on official crop advisories, and evaluate it with real farmers' questions against a large hosted model: accuracy, made-up advice, speed and cost. Good looks like: a clean training set, a test set of real questions, and an honest look at unsafe advice.
What you will build
- Build question and answer pairs from official advisories
- Fine-tune a small multilingual model with LoRA
- Test both models on real farmer questions
- Check for unsafe or made-up advice
- Recommend one, with its limits
Free, open data to build it with
Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.