DEPO: Dual-Efficiency Preference Optimization for LLM Agents

TL;DR

DEPO introduces a dual-efficiency framework for LLM agents, optimizing both token usage and step count to improve interaction efficiency without sacrificing performance.

Contribution

The paper proposes a novel dual-efficiency preference optimization method for LLM agents, balancing response succinctness and action steps to enhance efficiency.

Findings

Token usage reduced by up to 60.9%

Steps decreased by up to 26.9%

Performance improved by up to 29.3%

Abstract

Recent advances in large language models (LLMs) have greatly improved their reasoning and decision-making abilities when deployed as agents. Richer reasoning, however, often comes at the cost of longer chain of thought (CoT), hampering interaction efficiency in real-world scenarios. Nevertheless, there still lacks systematic definition of LLM agent efficiency, hindering targeted improvements. To this end, we introduce dual-efficiency, comprising (i) step-level efficiency, which minimizes tokens per step, and (ii) trajectory-level efficiency, which minimizes the number of steps to complete a task. Building on this definition, we propose DEPO, a dual-efficiency preference optimization method that jointly rewards succinct responses and fewer action steps. Experiments on WebShop and BabyAI show that DEPO cuts token usage by up to 60.9% and steps by up to 26.9%, while achieving up to a 29.3% improvement in performance. DEPO also generalizes to three out-of-domain math benchmarks and retains its efficiency gains when trained on only 25% of the data. Our project page is at https://opencausalab.github.io/DEPO.