Thanks for your comprehensive overview of GoT, ToT, CoT. I agree with your idea that this is paper is not qualified for being a promo technique. Since it actually looks like a fine-tuned to trained framework to solve any reasoning task.
However, I do think we might face such an emerging need to use GoT (also I think this might be the reason why Jacek mentioned this).
And I found this work is more like a pure prompting approach. “*Graph of Thoughts: Solving Elaborate Problems with Large Language Models”. It uses a casual pre-trained LLM and does not require any fine-tuning.
Any other good reference on GoT, ToT?