Plan-over-Graph: Towards Parallelable LLM Agent Schedule
This addresses the problem of inefficient sequential planning in LLM agents for real-life tasks, offering a novel approach to enable parallelism, though it is incremental in advancing agent scheduling.
The paper tackles the challenge of parallel scheduling for LLM-based task planning by introducing a plan-over-graph paradigm that decomposes tasks into subtask graphs for parallel execution, showing significant performance improvements on both API-based and open-sourced LLMs.
Large Language Models (LLMs) have demonstrated exceptional abilities in reasoning for task planning. However, challenges remain under-explored for parallel schedules. This paper introduces a novel paradigm, plan-over-graph, in which the model first decomposes a real-life textual task into executable subtasks and constructs an abstract task graph. The model then understands this task graph as input and generates a plan for parallel execution. To enhance the planning capability of complex, scalable graphs, we design an automated and controllable pipeline to generate synthetic graphs and propose a two-stage training scheme. Experimental results show that our plan-over-graph method significantly improves task performance on both API-based LLMs and trainable open-sourced LLMs. By normalizing complex tasks as graphs, our method naturally supports parallel execution, demonstrating global efficiency. The code and data are available at https://github.com/zsq259/Plan-over-Graph.