Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks

TL;DR

The paper introduces 'Program of Thoughts' (PoT), a method that separates reasoning from computation in language models by generating programs for external execution, leading to improved performance on numerical reasoning tasks.

Contribution

PoT is a novel approach that disentangles reasoning from computation by using language models to generate executable programs, enhancing numerical reasoning accuracy.

Findings

PoT outperforms chain-of-thought prompting by around 12% on average.

PoT achieves state-of-the-art results on math problem datasets.

PoT attains near state-of-the-art on financial question-answering datasets.

Abstract

Recently, there has been significant progress in teaching language models to perform step-by-step reasoning to solve complex numerical reasoning tasks. Chain-of-thoughts prompting (CoT) is by far the state-of-art method for these tasks. CoT uses language models to perform both reasoning and computation in the multi-step `thought' process. To disentangle computation from reasoning, we propose `Program of Thoughts' (PoT), which uses language models (mainly Codex) to express the reasoning process as a program. The computation is relegated to an external computer, which executes the generated programs to derive the answer. We evaluate PoT on five math word problem datasets (GSM, AQuA, SVAMP, TabMWP, MultiArith) and three financial-QA datasets (FinQA, ConvFinQA, TATQA) for both few-shot and zero-shot setups. Under both few-shot and zero-shot settings, PoT can show an average performance gain over CoT by around 12\% across all the evaluated datasets. By combining PoT with self-consistency decoding, we can achieve SoTA performance on all math problem datasets and near-SoTA performance on financial datasets. All of our data and code are released in Github https://github.com/wenhuchen/Program-of-Thoughts