Transformer models can implement any computable algorithm because MLPs handle arbitrary computations and attention provides writable memory. The author built a compiler that generates Llama weights to execute programs without training.
Transformer models can implement any computable algorithm because MLPs handle arbitrary computations and attention provides writable memory. The author built a compiler that generates Llama weights to execute programs without training.
@damiav.bsky.socialI’m *very* proud of my circuit to divide two 8-bit unsigned integers :3
I think it took me like 1.5 to 2 hours to design and build but it works!
It was made in Turing Complete (store.steampowered.com/app/1444480/...)
#TuringCompleteGame #TuringComplete
