LLMPC: Large Language Model Predictive Control
This work addresses planning inefficiencies in LLMs for AI researchers, but it is incremental as it reframes existing methods rather than introducing a new paradigm.
The paper tackles the problem of planning with Large Language Models by framing prompting techniques as model predictive control, showing that LLMs minimize implicit cost functions, and demonstrates improved performance over few-shot prompting on planning benchmarks.
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.