TableLLM: Enabling Tabular Data Manipulation by LLMs in Real Office Usage Scenarios

TL;DR

TableLLM is a specialized 8-billion-parameter LLM designed for effective tabular data manipulation in office scenarios, utilizing novel training strategies and benchmarks to outperform existing models.

Contribution

The paper introduces TableLLM, a purpose-built LLM for tabular data tasks, with a new training method and comprehensive evaluation benchmarks.

Findings

TableLLM outperforms existing LLMs in tabular data tasks.

The proposed training strategies improve reasoning capabilities.

Public release of model, code, and benchmarks.

Abstract

We introduce TableLLM, a robust large language model (LLM) with 8 billion parameters, purpose-built for proficiently handling tabular data manipulation tasks, whether they are embedded within documents or spreadsheets, catering to real-world office scenarios. We propose a distant supervision method for training, which comprises a reasoning process extension strategy, aiding in training LLMs to understand reasoning patterns more effectively as well as a cross-way validation strategy, ensuring the quality of the automatically generated data. To evaluate the performance of TableLLM, we have crafted benchmarks tailored to address both document and spreadsheet formats as well as constructed a well-organized evaluation pipeline capable of handling both scenarios. Thorough evaluations underscore the advantages of TableLLM when compared to various existing general-purpose and tabular data-focused LLMs. We have publicly released the model checkpoint, source code, benchmarks, and a web application for user interaction. Our codes and data are publicly available at https://github.com/TableLLM/TableLLM.