Continuous thought: the horse and its blinders

Why is chain-of-thought reasoning always in language tokens? Can we build better reasoning models by removing that constraint? And what keeps them from going astray once we do?

Commenting on Bubbles has been removed. Why? If you want to talk about this post, write to the author, use the comments on their blog if it has them, or reply with a "RE:" post on your own blog. Suggestions? feedback@bubbles.town