nanoGPT

nanoGPT is a project that helps users understand how GPT works by representing the internal mechanisms of a transformer in a spreadsheet. It is based on Andrej Karpathy's NanoGPT structure and includes roughly 85,000 parameters. The spreadsheet is color-coded, with purple representing parameters, green representing values, and orange representing intermediate values. The nanoGPT pipeline is packed in a spreadsheet, allowing users to visualize the entire structure and data flow. The spreadsheet is fully interactive and configurable, making it easier for users to understand the mechanisms and calculations involved. The project is useful for visual thinkers who want to understand how GPT works, and it can be used to build more complicated models. The spreadsheet does not contain actual trained weights and parameters, so users should not expect it to calculate correct results without updating the parameters.