Plan-and-Act: Improving Planning of Agents for Long-Horizon Tasks [Paper]

TL;DR

We introduce Plan-and-Act, a framework that enables accurate and reliable long-horizon task solving with explicit planning. We additionally introduce a synthetic data generation method for training the planner.

Abstract: Large language models (LLMs) have shown remarkable advancements in enabling language agents to tackle simple tasks. However, applying them for complex, multi-step, long-horizon tasks remains a challenge. Recent work has found success by separating high-level planning from low-level execution, which enables the model to effectively balance high-level planning objectives and low-level execution details. However, generating accurate plans remains difficult since LLMs are not inherently trained for this task. To address this, we propose Plan-and-Act, a novel framework that incorporates explicit planning into LLM-based agents and introduces a scalable method to enhance plan generation through a novel synthetic data generation method. Plan-and-Act consists of a Planner model which generates structured, high-level plans to achieve user goals, and an Executor model that translates these plans into environment-specific actions. To train the Planner effectively, we introduce a synthetic data generation method that annotates ground-truth trajectories with feasible plans, augmented with diverse and extensive examples to enhance generalization. We evaluate Plan-and-Act using web navigation as a representative long-horizon planning environment, demonstrating a state-of-the-art 57.58% success rate on the WebArena-Lite benchmark as well as a text-only state-of-the-art 81.36% success rate on WebVoyager.

• Dataset: Plan-and-Act Synthetic Data
• Planner Model: plan-and-act-planner-70b
• Executor Model: plan-and-act-actor-70b

Installation

1. Install Dependencies

pip install -r requirements.txt

2. Set up WebArena Environment

This code is designed to work with the WebArena-Lite environment. Follow these steps:

1. Clone and set up WebArena-Lite:

   git clone https://github.com/THUDM/VisualAgentBench.git
   cd VisualAgentBench/VAB-WebArena-Lite
   # Follow the installation instructions in the WebArena-Lite README

2. Place this repository in the WebArena directory:

   # Copy this Plan-and-Act repository into the WebArena-Lite directory
   cp -r /path/to/plan-and-act-repo ./plan_and_act

3. Follow the WebArena-Lite setup instructions from https://github.com/THUDM/VisualAgentBench/tree/main/VAB-WebArena-Lite to launch the WebArena environment.

Usage

Running Plan-and-Act

• Basic execution: python run_plan_and_act.py
• With replanning: python run_plan_and_act_with_replanning.py
• Parallel execution: Use the provided shell scripts for batch processing

Framework Components

For detailed information about specific components, see:

• Core Framework - Models, utilities, and core framework components
• Inference Components - Planning and execution inference modules
• Data Generation - Synthetic data generation for training

Citation

If you find our work helpful, please consider citing us:

@inproceedings{
erdogan2025planandact,
title={Plan-and-Act: Improving Planning of Agents for Long-Horizon Tasks},
author={Lutfi Eren Erdogan and Hiroki Furuta and Sehoon Kim and Nicholas Lee and Suhong Moon and Gopala Anumanchipalli and Kurt Keutzer and Amir Gholami},
booktitle={Forty-second International Conference on Machine Learning},
year={2025},
url={https://openreview.net/forum?id=ybA4EcMmUZ}
}