This is my concept for a compact AI model (15–25M parameters) trained from scratch using Hugging Face, focused solely on NCERT Class 12 content — textbooks, notes, and Q&A — to create a personalized, offline-friendly AI tutor.
The idea is that this model could later be converted into a Windows or Android app, similar to Google’s recent AI Edge Gallery LLMs running locally on devices. It aims to make learning accessible on low-resource hardware without internet dependence.
While initially focused on Class 12 NCERT, this concept can be extended to other classes and syllabuses to support more students with tailored AI-powered learning.
Open to collaboration and contributions!
(hejhdiss@gmail.com)
Top comments (0)