ikaru2lt

3 ms·

1. 1.▲
   Show HN: Millnew AI – On device time-series prediction AI using PyTorch & Captum(github.com)

   1 points by ikaru2lt 1mo ago|0 comments
2. 2.▲
   by ikaru2lt 1mo ago

   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 th
3. 3.▲
   Show HN: Millnew AI – Explaining time-series models using local LLMs and Captum(robinhood-demo.streamlit.app)

   5 points by ikaru2lt 1mo ago|1 comments
4. 4.▲
   by ikaru2lt 1mo ago

   Hey, I wanna ask u a question: what's the difference between ur tech and habitual AI things for linux/unix terminal?
5. 5.▲
   by ikaru2lt 1mo ago

   But how do you make it better for the GPU if the size is too large?
6. 6.▲
   by ikaru2lt 1mo ago

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