SpreadsheetLLM: Encoding Spreadsheets for Large Language Models

TL;DR

SpreadsheetLLM introduces an innovative encoding framework for large language models to better understand and reason about spreadsheets, significantly improving performance in table detection, compression, and question answering tasks.

Contribution

The paper presents SheetCompressor, a novel encoding method that effectively compresses spreadsheets for LLMs, enabling enhanced understanding and reasoning capabilities.

Findings

Outperforms vanilla serialization by 25.6% in table detection

Achieves 78.9% F1 score in spreadsheet QA tasks

Provides an average compression ratio of 25 times

Abstract

Spreadsheets are characterized by their extensive two-dimensional grids, flexible layouts, and varied formatting options, which pose significant challenges for large language models (LLMs). In response, we introduce SpreadsheetLLM, pioneering an efficient encoding method designed to unleash and optimize LLMs' powerful understanding and reasoning capability on spreadsheets. Initially, we propose a vanilla serialization approach that incorporates cell addresses, values, and formats. However, this approach was limited by LLMs' token constraints, making it impractical for most applications. To tackle this challenge, we develop SheetCompressor, an innovative encoding framework that compresses spreadsheets effectively for LLMs. It comprises three modules: structural-anchor-based compression, inverse index translation, and data-format-aware aggregation. It significantly improves performance in the spreadsheet table detection task, outperforming the vanilla approach by 25.6% in GPT4's in-context learning setting. Moreover, fine-tuned LLM with SheetCompressor has an average compression ratio of 25 times, and achieves a state-of-the-art 78.9% F1 score, surpassing the best existing models by 12.3%. Finally, we propose Chain of Spreadsheet for downstream tasks of spreadsheet understanding and validate it in a new and demanding spreadsheet QA task. We methodically leverage the inherent layout and structure of spreadsheets, demonstrating that SpreadsheetLLM is highly effective across a variety of spreadsheet tasks.