Improving Extreme Weather Forecasting with Deep Learning and Small Language Models

This project extends on my prior work on extreme weather forecasting using deep learning by incorporating a small language model to interpret and summarize the model outputs. In addition to generating forecasts using models like MetNet 2.0, I explored how small language models can translate complex prediction data into easily understandable results for faster decision-making. This approach really highlights the potential of combining numerical prediction systems with AI driven explanation tools to improve efficiency, accessibility and real world usability.

Intern: Aaron Zhong

Mentor: Janet Zhang (AMDS)