This new, dead simple prompt technique boosts accuracy on LLMs by up to 76% on non-reasoning tasks

Most modern LLMs are trained as "causal" language models. This means they process text strictly from left to right. When the ...