Comment by johndough

I've got a few. Pick whichever you like.

Factor 135066410865995223349603216278805969938881475605667027524485143851526510604859533833940287150571909441798207282164471551373680419703964191743046496589274256239341020864383202110372958725762358509643110564073501508187510676594629205563685529475213500852879416377328533906109750544334999811150056977236890927563 in less than 24 hours.

Come up with a way to sample from LLMs such that they can tell funny jokes. The jokes should not be recited jokes from elsewhere.

Implement a CUDA kernel that achieves optimal efficiency for PyTorch-like conv2d for "reasonable" shapes/strides/dilations/groups. (This task is the closest to being solved by LLMs, but they usually get stuck somewhere doing stupid things instead of considering more advanced optimization methods and still need a human to push them along).