Virtual Crop Manager using Large Language Models
Supervisor
Suitable for
Abstract
Co-supervised by Julian Godding (j.godding@gardin.co.uk)
Large language models have made extraordinary progress over the past year but fail to deliver scientifically accurate information in a reliable way. This project will look at fine-tuning existing approaches to NLP Q&A ranking using academic literature to create a virtual plant science assistant. A proof-of-concept reinforcement learning architecture should be incorporated to improve the model’s accuracy and reliability over time as it is used by plant scientists for improved crop management.