Understanding graph of thoughts prompting and several other variants…
11 hours ago
Advanced prompting techniques like chain of thought [8] and tree of thought [9] prompting have drastically improved the power of huge language models (LLMs) to resolve complex, reasoning-based tasks. At a high level, forcing the LLM to construct a step-by-step response to an issue drastically improves its problem-solving capabilities. Nonetheless, all of such techniques assume that the reasoning process should follow a linear patterns that progresses from one thought to the subsequent. Notably, the reasoning process followed by humans tends to be quite different, following multiple different chains of thought and even combining insights from different thoughts to reach at a final solution. Inside this overview, we might be studying several prompting techniques that model the reasoning process as a graph structure — quite than a series or tree — that higher captures the assorted sorts of non-linear patterns that will occur when reasoning over an issue.
“Human pondering is usually characterised by its ability to make sudden leaps and connections between seemingly unrelated ideas, which may result in novel insights and solutions. This non-linear, jumping thought process is an indicator of human creativity, reasoning, and problem-solving abilities.” — from [1]
Inside this overview, we are going to explore several advanced prompting techniques for LLMs that might be used to resolve difficult multi-step reasoning problems. Luckily, we have now recently overviewed the fundamental ideas behind prompting, including:
- Prompting basics (i.e., prompt engineering, context windows, structure of a prompt, etc.) https://towardsdatascience.com/graph-based-prompting-and-reasoning-with-language-models-d6acbcd6b3d8?source=rss—-7f60cf5620c9—4
- Advanced prompting techniques (e.g., chain of thought, self-consistency, and least-to-most prompting) https://towardsdatascience.com/graph-based-prompting-and-reasoning-with-language-models-d6acbcd6b3d8?source=rss—-7f60cf5620c9—4
We’ve got covered each practical and advanced prompting techniques prior to now. All of those concepts — especially chain of thought (CoT) prompting [8], self-consistency [10], and tree of thought (ToT) prompting [9] — might be relevant for gaining an understanding of this overview. Beyond these ideas, we want to know the transformer architecture and the graph…