SII-LancelotXie/Socio-Foundation-8B
Socio-Foundation-8B is an 8.19 billion parameter causal language model developed by SII-LancelotXie, based on the Qwen3-8B architecture. This model is specifically designed for generalizable individual behavior simulation, leveraging a hierarchical capability distillation methodology. It provides full BF16 model weights and is intended for research and applications in simulating human-like behaviors.
Loading preview...
Socio-Foundation-8B Overview
Socio-Foundation-8B is an 8.19 billion parameter causal language model built upon the Qwen3-8B architecture. Developed by SII-LancelotXie, this model is specifically engineered for generalizable individual behavior simulation. It is the accompanying model for the research paper "Socio-Foundation: A Model for Generalizable Individual Behavior Simulation via Hierarchical Capability Distillation" by Liang Wang, Wenxuan Xie, Xinyi Mou, and Zhongyu Wei.
Key Capabilities
- Individual Behavior Simulation: Designed to simulate human-like behaviors, offering a foundation for understanding and modeling social interactions.
- Hierarchical Capability Distillation: Utilizes a novel training methodology detailed in its accompanying paper, focusing on a FONTS capability taxonomy.
- Full BF16 Weights: Provides complete BF16 model weights, configuration files, and tokenizer for direct use with Hugging Face Transformers.
Good for
- Research in Social Simulation: Ideal for researchers exploring agent-based modeling and the simulation of individual and collective behaviors.
- Developing AI Agents: Suitable for creating AI agents that exhibit more nuanced and generalizable behavioral patterns.
- Understanding AI Behavior: Offers a platform to study and evaluate the complexities of AI-driven behavioral responses, as assessed through benchmarks like IndiEval.