InstructExcel: A Benchmark for Natural Language Instruction in Excel
This provides a benchmark for evaluating LLMs on spreadsheet tasks, but it is incremental as it builds on existing automation features.
The authors tackled the problem of whether large language models can generate Excel OfficeScripts from natural language instructions by introducing InstructExcel, a benchmark with over 10k samples, and found that GPT-4 outperforms GPT-3.5 with improvements from more examples and dynamic prompting.
With the evolution of Large Language Models (LLMs) we can solve increasingly more complex NLP tasks across various domains, including spreadsheets. This work investigates whether LLMs can generate code (Excel OfficeScripts, a TypeScript API for executing many tasks in Excel) that solves Excel specific tasks provided via natural language user instructions. To do so we introduce a new large-scale benchmark, InstructExcel, created by leveraging the 'Automate' feature in Excel to automatically generate OfficeScripts from users' actions. Our benchmark includes over 10k samples covering 170+ Excel operations across 2,000 publicly available Excel spreadsheets. Experiments across various zero-shot and few-shot settings show that InstructExcel is a hard benchmark for state of the art models like GPT-4. We observe that (1) using GPT-4 over GPT-3.5, (2) providing more in-context examples, and (3) dynamic prompting can help improve performance on this benchmark.