https://paperswithcode.com/paper/thought-propagation-an-analogical-approach-to * [ ] [ ] * Browse State-of-the-Art * Datasets * Methods * More Newsletter RC2022 About Trends Portals Libraries * * * Sign In Subscribe to the PwC Newsletter x Stay informed on the latest trending ML papers with code, research developments, libraries, methods, and datasets. Read previous issues [ ] Subscribe Join the community x You need to log in to edit. You can create a new account if you don't have one. Or, discuss a change on Slack. Edit Social Preview x Description (*) Default ( ) Custom [No code available ye] [ ] [ ] [ ] [ ] [ ] Image (*) Default ( ) Custom ( ) None [2310] File is too large Upload an image to customize your repository's social media preview. Images should be at least 640x320px (1280x640px for best display). [ ] Close Save Add a new code entry for this paper x GitHub, GitLab or BitBucket URL:* [ ] [ ] Official code from paper authors Submit Remove a code repository from this paper x Mark the official implementation from paper authors x --------------------------------------------------------------------- There is no official implementation (*) --------------------------------------------------------------------- Multiple official implementations [ ] Submit Add a new evaluation result row x Task:* [--------- ] Not in the list? Add a task. Dataset:* [---------] Model name:* [ ] Metric name:* [---------] [ ] Higher is better (for the metric) Metric value:* [ ] [ ] Uses extra training data Data evaluated on [2023-10-06 ] Submit Add a new evaluation result row x TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK REMOVE Add a task x Attached tasks: * PROMPT ENGINEERING Add: [---------] Not in the list? Create a new task. New task name: [ ] Top-level area: [--------- ] Parent task (if any): [---------] Description: [ ] [ ] [ ] Submit Remove a task x * [5caee997-0] Prompt Engineering - Add a method x Add: [---------] Not in the list? Create a new method. New method name (e.g. ReLU): [ ] New method full name (e.g. Rectified Linear Unit): [ ] Paper where method was first introduced: [---------] Method category (e.g. Activation Functions): If no match, add something for now then you can add a new category afterwards. [---------] Markdown description (optional; $\LaTeX$ enabled): You can edit this later, so feel free to start with something succinct. [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] Submit Remove a method x Edit Datasets x Add or remove datasets introduced in this paper: Add or remove other datasets used in this paper: [*]ALFWorld Paper introduces a new dataset? Add a new dataset here Save Thought Propagation: An Analogical Approach to Complex Reasoning with Large Language Models 6 Oct 2023 * Junchi Yu, Ran He, Rex Ying * Edit social preview Large Language Models (LLMs) have achieved remarkable success in reasoning tasks with the development of prompting methods. However, existing prompting approaches cannot reuse insights of solving similar problems and suffer from accumulated errors in multi-step reasoning, since they prompt LLMs to reason \textit{from scratch}. To address these issues, we propose \textbf{\textit{Thought Propagation} (TP)}, which explores the analogous problems and leverages their solutions to enhance the complex reasoning ability of LLMs. These analogous problems are related to the input one, with reusable solutions and problem-solving strategies. Thus, it is promising to propagate insights of solving previous analogous problems to inspire new problem-solving. To achieve this, TP first prompts LLMs to propose and solve a set of analogous problems that are related to the input one. Then, TP reuses the results of analogous problems to directly yield a new solution or derive a knowledge-intensive plan for execution to amend the initial solution obtained from scratch. TP is compatible with existing prompting approaches, allowing plug-and-play generalization and enhancement in a wide range of tasks without much labor in task-specific prompt engineering. Experiments across three challenging tasks demonstrate TP enjoys a substantial improvement over the baselines by an average of 12\% absolute increase in finding the optimal solutions in Shortest-path Reasoning, 13\% improvement of human preference in Creative Writing, and 15\% enhancement in the task completion rate of LLM-Agent Planning. PDF Abstract Code Edit Add Remove Mark official --------------------------------------------------------------------- No code implementations yet. Submit your code now Tasks Edit Add Remove --------------------------------------------------------------------- [5caee997-0] Prompt Engineering Datasets Edit --------------------------------------------------------------------- [dataset-00] ALFWorld Results from the Paper Edit --------------------------------------------------------------------- Submit results from this paper to get state-of-the-art GitHub badges and help the community compare results to other papers. Methods Edit Add Remove --------------------------------------------------------------------- No methods listed for this paper. Add relevant methods here Contact us on: hello@paperswithcode.com . Papers With Code is a free resource with all data licensed under CC-BY-SA. Terms Data policy Cookies policy from [3Wqt0fQd1Y]