Om-codex/NLP-Architectures ? reverse-engineered prompt

Build me a Jupyter notebook based NLP learning project that explains and demonstrates core text models from scratch.

I want a clean repo where I can open the notebooks and actually follow along with how attention, self attention, transformers, and simple decoder only GPT style text generation work. Include small working examples for things like text preprocessing, embeddings or vector representation, sequence to sequence with attention, basic question answering, summarization, named entity recognition, and a simple text generation demo. If it helps, use the saved model files already in the project and make the notebooks run end to end with sample data.

Please keep the code easy to read and educational, with short explanations in the notebooks so someone learning NLP can understand what each part is doing. Organize the project so the different topics feel separated but connected, and make sure the examples are practical enough to run without a huge setup. If you need to check current best practices or docs online, go ahead.