ikaru2lt

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.

Hey, I wanna ask u a question: what's the difference between ur tech and habitual AI things for linux/unix terminal?

But how do you make it better for the GPU if the size is too large?

It's really impressive to see the quality of the datasets again; in most AIs, a bad dataset means that nothing works.