Program-Aided Reasoners (better) Know What They Know
On the map: LLM reasoning → Reasoning benchmarks and evaluation
Abstract (arXiv)
Prior work shows that program-aided reasoning, in which large language models (LLMs) are combined with programs written in programming languages such as Python, can significantly improve accuracy on various reasoning tasks. However, while accuracy is essential, it is also important for such reasoners to "know what they know", which can be quantified through the calibration of the model. In this paper, we compare the calibration of Program Aided Language Models (PAL) and text-based Chain-of-thought (COT) prompting techniques over 5 datasets and 2 model types: LLaMA models and OpenAI models. Our results indicate that PAL leads to improved calibration in 75% of the instances. Our analysis uncovers that prompting styles that produce lesser diversity in generations also have more calibrated results, and thus we also experiment with inducing lower generation diversity using temperature scaling and find that for certain temperatures, PAL is not only more accurate but is also more calibrated than COT. Overall, we demonstrate that, in the majority of cases, program-aided reasoners better know what they know than text-based counterparts.
Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Placed in Reasoning benchmarks and evaluation, one of 17 directions in LLM reasoning. Builds on 10 of its 33 references that are in the map. First posted on arXiv 16 Nov 2023.
Builds on
10 of 33 references.
References · 10
- 2023Llama 2: Open Foundation and Fine-Tuned Chat Modelscited by 13,028
- 2019The Curious Case of Neural Text Degenerationcited by 2,916
- 2022Language Models (Mostly) Know What They Knowcited by 1,856
- 2022Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Themcited by 1,807
- 2023Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMscited by 974
- 2023Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedbackcited by 812
- 2020Calibration of Pre-trained Transformerscited by 315
- 2023Faithful Chain-of-Thought Reasoningcited by 288
- 2020Selective Question Answering under Domain Shiftcited by 238
- 2020Calibrated Language Model Fine-Tuning for In- and Out-of-Distribution Datacited by 17
Built on by
Citing papers per year: 2024: 1; 2025: 2.
3 papers.
- 2025PBEBench: A Multi-Step Programming by Examples Reasoning Benchmark inspired by Historical Linguistics29 May 2025
- 2025Code to Think, Think to Code: A Survey on Code-Enhanced Reasoning and Reasoning-Driven Code Intelligence in LLMs26 Feb 2025
- 2024Can LLMs Reason in the Wild with Programs?19 Jun 2024
Similar, not cited
Close in wording, no citation either way
- 2022PAL: Program-aided Language Models
- 2025Understanding Tool-Integrated Reasoning
- 2024General Purpose Verification for Chain of Thought Prompting
- 2025Reasoning Capabilities and Invariability of Large Language Models
- 2023Automatic Model Selection with Large Language Models for Reasoning
- 2024LLM Reasoners: New Evaluation, Library, and Analysis of Step-by-Step Reasoning with Large Language Models
- 2026Calibration Drift Under Reasoning: How Chain-of-Thought Budgets Induce Overconfidence in Large Language Models
- 2026Fragile Thoughts: How Large Language Models Handle Chain-of-Thought Perturbations