Sketch-of-Thought: Efficient LLM Reasoning

azaras · 2025-03-16T18:19:13 1742149153

It is interesting for an old and classic software developer to see how the LLMs are taught/programmed. They use markdown files:

- https://github.com/SimonAytes/SoT/blob/main/sketch_of_though...

jxjnskkzxxhx · 2025-03-19T11:59:43 1742385583

If it fullfills the purpose, then it's a good enough solution.

Bolwin · 2025-03-16T20:58:07 1742158687

This seems near identical to chain of draft, albeit more thorough and they actually did training instead of just promoting https://arxiv.org/abs/2502.18600

xyz7786 · 2025-03-20T06:29:14 1742452154

After reading both papers, I have to disagree. Chain of Draft seems "identical" to Concise Chain-of-Thought (CCoT) https://arxiv.org/abs/2401.05618. Additionally, the CoD authors seem to mention that it is cognitive inspired but don't give any actual basis for that.

Sketch-of-Thought, while similar in terms of its goal, seems to be based on cognitive science principles and has better extensibility than CoD and CCoT (they basically just say "answer in X words" and leave it at that).