Program of Thoughts Prompting: Disentangling Computation from Reasoning for Numerical Reasoning Tasks
This work addresses a bottleneck in language model-based reasoning for math and financial QA tasks, offering a novel method that improves accuracy and achieves state-of-the-art results.
The paper tackled the problem of disentangling computation from reasoning in numerical reasoning tasks by proposing Program of Thoughts (PoT), which uses language models to generate programs for reasoning and external computers for computation, resulting in an average performance gain of around 12% over Chain-of-Thoughts prompting across multiple datasets.
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