Skeleton-of-Thought

What is Skeleton-of-Thought?

Skeleton-of-Thought is a latency optimization technique where the model first generates an outline of the answer, then expands each section in parallel.

Where did the term "Skeleton-of-Thought" come from?

Technique to speed up long-form generation.

How is "Skeleton-of-Thought" used today?

Reduces user wait time for complex reasoned responses.

Related Terms