---
title: "Comment by suzie on \"Top 4 Lessons I Learned in 2024:\" | Hashnode"
description: "hello !! :D We are students studying AI in Korea. We are participating in an AI education program and are currently working on the final project making fake new"
url: "https://hashnode.com/posts/top-4-lessons-i-learned-in-2024/677a9fd9491a1758c13d93d0/comment/6785bc1489ad36c797fc7478"
---

[Discussion](https://hashnode.com/posts/top-4-lessons-i-learned-in-2024/677a9fd9491a1758c13d93d0)

Comment [Top 4 Lessons I Learned in 2024:](https://hashnode.com/posts/top-4-lessons-i-learned-in-2024/677a9fd9491a1758c13d93d0)

[suzie](https://hashnode.com/@suzie789)

[Jan 14, 2025](https://hashnode.com/posts/top-4-lessons-i-learned-in-2024/677a9fd9491a1758c13d93d0/comment/6785bc1489ad36c797fc7478)

hello !! :D We are students studying AI in Korea. We are participating in an AI education program and are currently working on the final project making fake news detector

While looking for a basic model at Hugging Face, HEM's code was light and fast. I would like to tune this model or use it as a basic model. the license was not written, so i sent this email.

We don't have any plans, but after creating the program, we have to submit it to the AI ​​education program, and it has to be a commercially viable license.

If this is a commercially impossible license, can you tell me how you fine-tuned the VIT model? Other models are not good at classifying the latest fake images. I am very impressed that even the images created with SORA.AI were classified perfectly.

I read your last post. It's really amazing I will be cheering you on from afar :) Have a happy 2025 !!!