README.md

metadata

title: Story GPT Python
colorFrom: yellow
colorTo: red
sdk: gradio
app_file: app.py
pinned: false
license: mit

Story GPT Python

This is a tiny story-writing GPT-style language model project written in Python from scratch.

What it includes

• Word-level tokenizer
• Causal transformer decoder with self-attention
• Story-focused local training corpus
• Structured local story composer for cleaner paragraph output
• Local CPU training loop
• Checkpoint save and load
• Gradio user interface

Important

• No external pretrained LLM is used
• This is a small educational GPT-like story model
• The first generate or train call will initialize and train the model locally