Summary of Tic: Translate-infer-compile For Accurate “text to Plan” Using Llms and Logical Representations, by Sudhir Agarwal and Anu Sreepathy

TIC: Translate-Infer-Compile for accurate “text to plan” using LLMs and Logical Representations

by Sudhir Agarwal, Anu Sreepathy

First submitted to arxiv on: 9 Feb 2024

GrooveSquid.com Paper Summaries

GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!

Summary difficulty	Written by	Summary
High	Paper authors	High Difficulty Summary Read the original abstract here
Medium	GrooveSquid.com (original content)	Medium Difficulty Summary This paper presents a novel approach to natural language planning task requests by leveraging the strengths of both Large Language Models (LLMs) and classical planning tools. Specifically, an LLM is used to generate a Planning Domain Definition Language (PDDL) representation of the task request, which is then fed into a classical planner to compute a plan. The authors’ approach involves three stages: translating natural language descriptions into a logically interpretable intermediate representation using an LLM, inferring additional information from this representation using a logic reasoner, and compiling the target PDDL from the base and inferred information. By using an LLM only to generate the intermediate representation, the authors are able to significantly reduce errors and achieve high accuracy on task PDDL generation for all seven domains of their evaluation dataset.
Low	GrooveSquid.com (original content)	Low Difficulty Summary This paper helps us understand how computers can make plans when given a description in everyday language. Currently, there are two main approaches: using special computer programs called Large Language Models (LLMs) that are great at understanding language but not so good at making plans, or using classical planning tools that are excellent at making plans but need input in a specific format. The authors of this paper came up with a new idea to combine the strengths of both methods. They use an LLM to help translate the everyday language into a special format that the planner can understand, and then use the planner to create a plan. This approach helps reduce mistakes and is more accurate than previous attempts.