HayatoHongo/EveryonesLLM ? reverse-engineered prompt

Build me a beginner friendly project that teaches someone how to make a small GPT style language model from scratch in Google Colab.

I want it to feel like a step by step course, not just a pile of code. Start with the basics like loading text, turning it into tokens, embeddings, attention, transformer blocks, training, and then text generation. Use simple notebooks that build up chapter by chapter so a curious person can actually run each part and understand what changed. It should work with a small sample text file first, then show how to train a slightly more useful model on a tiny story style dataset, including simple speed checks on CPU and a T4 GPU in Colab.

Please make the notebooks clean, runnable, and beginner readable, with short explanations, sensible defaults, and outputs that let me see the model improving. If it makes sense, include a very simple demo where I can type a prompt and get generated text back. Look up current Colab friendly docs online if you need to.