{ "cells": [ { "cell_type": "markdown", "id": "55e0d5644a165439", "metadata": {}, "source": [ "# Chain of Thought\n", "\n", "[Chain of Thought](https://arxiv.org/pdf/2201.11903) (CoT) is a common prompt engineering technique which asks the LLM to step through its reasoning and thinking process to answer a question. In its simplest form, it can be implemented by asking asking the LLM to step through a problem step by step, but is more effective when you leverage examples and patterns of reasoning similar to your query in a few shot prompt. Chain of Thought is most effective for mathematical and reasoning tasks.\n", "\n", "
Mirascope Concepts Used
\n", "Note
\n", "Recent models will automatically explain their reasoning (to a degree) for most reasoning tasks, but explicitly asking for a step by step solution can sometimes produce better solutions and explanations.
\n", "Effective Chain of Thought Usage
\n", "