What Are the Odds? Language Models Are Capable of Probabilistic Reasoning

TL;DR

This paper evaluates the probabilistic reasoning abilities of language models across various tasks and contexts, revealing their capacity to infer and work with probability distributions, especially when aided by context and assumptions.

Contribution

It introduces a systematic evaluation framework and a new benchmark dataset for assessing LMs' probabilistic reasoning skills across multiple tasks.

Findings

Models can infer distribution properties with proper context.

Real-world context improves probabilistic reasoning.

Simplified assumptions can aid inference even if incorrect.

Abstract

Language models (LM) are capable of remarkably complex linguistic tasks; however, numerical reasoning is an area in which they frequently struggle. An important but rarely evaluated form of reasoning is understanding probability distributions. In this paper, we focus on evaluating the probabilistic reasoning capabilities of LMs using idealized and real-world statistical distributions. We perform a systematic evaluation of state-of-the-art LMs on three tasks: estimating percentiles, drawing samples, and calculating probabilities. We evaluate three ways to provide context to LMs 1) anchoring examples from within a distribution or family of distributions, 2) real-world context, 3) summary statistics on which to base a Normal approximation. Models can make inferences about distributions, and can be further aided by the incorporation of real-world context, example shots and simplified assumptions, even if these assumptions are incorrect or misspecified. To conduct this work, we developed a comprehensive benchmark distribution dataset with associated question-answer pairs that we have released publicly.