✨ GASim

This is the official implementation of the ACL 2026 paper GASim: A Graph-Accelerated Hybrid Framework for Social Simulation. GASim not only delivers a substantial 9.94× end-to-end speedup over the traditional hybrid framework, but also consumes less than 20% of baseline tokens. Our code will be released soon🔥!

🌟 Overview

Large-scale social simulators are essential for studying complex social patterns. Prior work explores hybrid methods to scale up simulations, combining large language models (LLM)-based agents with numerical agent-based models (ABM). However, this incurs high latency due to expensive memory retrieval and sequential ABM execution. To address this challenge, we propose GASim with key components as follows:

• A hybrid coordinator EDG that dynamically partitions agents into core and ordinary types.
• GOM for core agents, which rapidly retrieves memories with a lightweight graph-based memory model, alleviating the heavy latency in LLM-based retrieval process.
• GMP for ordinary agents, which updates opinions in parallel with fine-grained features and Graph Attention Network, resolving the sequential execution bottlenecks of ABMs.

---

---

📅 TODO List

• Release "Getting Started" guidelines
• Release the overall social simulation module
• Release GOM validation on LoCoMo dataset
• Release GMP training module
• Release anonymized social topic-based datasets

🙏 Acknowledgements

This repo is built upon the AgentVerse and HiSim framework. We sincerely thank the authors for their contributions to the community.