Show HN: Millnew AI – Explaining time-series models using local LLMs and Captum

Hi HN, I'm a 17-year-old AI dev from Mali. I've built this demo to experiment with LLM interpretability. Traditional LLMs tend to be baffling, as they are good at reasoning but struggle with precise math. However, if we reverse the usual approach and give the LLM the attributes of model prediction, it allows us to explain each of the AI's choices. The current web app is a live demo to showcase how the AI works. I am looking forward to your feedback and questions.