Accelerate local language-model generation by drafting multiple possible next-token blocks in parallel and checking them with the main model.

Soon you can unlock how this was done.

Behind this: the tool used · the method · what actually resulted.