400k parameter LLM trained on just Shakespeare’s Hamlet. Very slow due to the size of the model and Scratch's limitations, but you can speed it up with TurboWarp- turbowarp.org/1375412980/ Under 1k blocks! Shouldn't take a lot of time to load, but it does have quite a few list elements so may not be the best
Do note this isn't anywhere near high-quality generation. This is because the model only has ~0.4M parameters; Scratch can't fit more without struggling to save, at least with the list configuration I used. I did however manage to fit in twice as many as I could last time!
