Pre-Act: Multi-Step Planning and Reasoning Improves Acting in LLM Agents
Mrinal Rawat, Ambuje Gupta, Rushil Goomer, Alessandro Di Bari, Neha Gupta, Roberto Pieraccini
TL;DR
Pre-Act enhances LLM-based agents by replacing single-step reasoning with multi-step, self-refining plans that integrate past actions and observations. A two-stage curriculum and LoRA-based fine-tuning enable smaller models to match or exceed larger proprietary systems, with strong gains in both turn-level action recall and end-to-end task completion. The paper introduces a rigorous two-level evaluation framework, including milestone-based end-to-end testing and synthetic user simulations, to comprehensively assess agent performance. Practically, Pre-Act offers a scalable path to higher reliability and efficiency for real-world agentic systems while highlighting the importance of explicit planning and structured reasoning. The results suggest smaller, well-tuned models can achieve competitive performance at lower latency and cost.
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.