LLMPC: Large Language Model Predictive Control

TL;DR

This paper introduces a unified model predictive control framework for large language models, demonstrating improved planning performance by leveraging prompting techniques as implicit cost function minimizers.

Contribution

It proposes a novel MPC-based approach for LLMs, unifying prompting techniques under a predictive control framework to enhance planning capabilities.

Findings

LLMs act as implicit planning cost function minimizers.

The proposed MPC framework improves planning performance.

Enhanced results over few-shot prompting on benchmarks.

Abstract

Recent advancements in prompting techniques for Large Language Models (LLMs) have improved their reasoning, planning, and action abilities. This paper examines these prompting techniques through the lens of model predictive control (MPC). We show that LLMs act as implicit planning cost function minimizers when planning prompts are used. We propose a unified MPC framework for planning with LLMs and demonstrate improved performance over few shot prompting on several planning benchmarks.