Structured Inference with Large Language Gibbs

Sanghyeok Choi, Henry Gouk, Esmeralda S. Whitammer

Keywords: LLM, structured inference, Gibbs sampling, consistent reasoning, Bayesian structure learning, causal inference

TL;DR: We argue and demonstrate that approximate Gibbs sampling using LLM conditionals as transition kernels enables a variety of reasoning and inference tasks.

Abstract:

We propose Large Language Gibbs, a scheme for structured probabilistic inference that uses conditional distributions of a large language model (LLM) as transition operators.
Rather than sampling structured objects through single-pass autoregressive generation, we iteratively resample individual variables conditioned on others using an LLM's next-token conditionals.
This approach avoids order-dependent biases and produces a stationary distribution that reflects a compromise between all local conditionals.
We apply this approach to sampling from synthetic distributions, consistent reasoning tasks, and Bayesian structure learning.
The results suggest that the use of LLM conditionals in MCMC as a practical alternative to one-pass generation for structured probabilistic inference under a world prior accessible through noisy LLM conditionals.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 146