GitHub - dipanjanpathak/mini-gpt-model: This project is a minimal GPT implementation focused on a question-answering system. The model was trained from scratch on the WikiText-2 dataset without leveraging transfer learning or pretrained models, resulting in limited performance. It serves as an educational experiment to explore the fundamentals of training language models from scratch.

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
README.md		README.md
mini-GPT-model.ipynb		mini-GPT-model.ipynb

Repository files navigation

About

This project is a minimal GPT implementation focused on a question-answering system. The model was trained from scratch on the WikiText-2 dataset without leveraging transfer learning or pretrained models, resulting in limited performance. It serves as an educational experiment to explore the fundamentals of training language models from scratch.