Requires CUDA 13.4; build script is Windows only.
Show HN: TurboGPT: train 22KiB transformer in 13s (github.com)
24 pointsby lostmsu3 hours ago3 comments
Inspired by minGPT for home experiments.
jey an hour ago
At that scale it seems like you could just solve the KKT conditions directly. Exaggerating but only a little