Pre-Act: Multi-Step Planning and Reasoning Improves Acting in LLM Agents

TL;DR

Pre-Act introduces multi-step planning and reasoning to enhance large language model agents, significantly improving their action recall and goal completion rates, especially when fine-tuned on smaller models for practical deployment.

Contribution

The paper proposes Pre-Act, a multi-step planning method that refines reasoning and actions in LLM agents, outperforming ReAct and enabling effective fine-tuning of smaller models.

Findings

Pre-Act outperforms ReAct by 70% in Action Recall.

Fine-tuned 70B Llama 3.1 surpasses GPT-4 in action accuracy.

The approach improves goal completion rates by 28%.

Abstract

The ReAct (Reasoning + Action) capability in large language models (LLMs) has become the foundation of modern agentic systems. Recent LLMs, such as DeepSeek-R1 and OpenAI o1/o3, exemplify this by emphasizing reasoning through the generation of ample intermediate tokens, which help build a strong premise before producing the final output tokens. In this paper, we introduce Pre-Act, a novel approach that enhances the agent's performance by creating a multi-step execution plan along with the detailed reasoning for the given user input. This plan incrementally incorporates previous steps and tool outputs, refining itself after each step execution until the final response is obtained. Our approach is applicable to both conversational and non-conversational agents. To measure the performance of task-oriented agents comprehensively, we propose a two-level evaluation framework: (1) turn level and (2) end-to-end. Our turn-level evaluation, averaged across five models, shows that our approach, Pre-Act, outperforms ReAct by 70% in Action Recall on the Almita dataset. While this approach is effective for larger models, smaller models crucial for practical applications, where latency and cost are key constraints, often struggle with complex reasoning tasks required for agentic systems. To address this limitation, we fine-tune relatively small models such as Llama 3.1 (8B & 70B) using the proposed Pre-Act approach. Our experiments show that the fine-tuned 70B model outperforms GPT-4, achieving a 69.5% improvement in action accuracy (turn-level) and a 28% improvement in goal completion rate (end-to-end) on the Almita (out-of-domain) dataset.