A arxiv.org Let's Think Dot by Dot: Hidden Computation in Transformer Language Models 159 points Jimmc414 2 years ago 32 comments
