base_model:
- Qwen/Qwen3.5-9B
license: mit
library_name: transformers
pipeline_tag: text-generation
Distillation-Expert-Qwen3.5-9B
🌐 Project Page | 💻 Code | 📄 Paper
We introduce RecursiveMAS, a multi-agent framework that scales agent collaboration through latent-space recursion. RecursiveMAS treats a multi-agent system as a unified recursive computation, where heterogeneous agents iteratively exchange, refine, and evolve their latent states across recursion rounds. In the Distillation-Style setting, the Expert Agent provides high-quality guidance to the Learner Agent, supporting recursive latent-space collaboration for improved task solving.
Model Details
| Item | Description |
|---|---|
| Model | Distillation-Expert-Qwen3.5-9B |
| Collaboration Style | Distillation-Style |
| Agent Role | Expert Agent |
| Base Model | Qwen3.5-9B |
⚠️ Note: This checkpoint is a role-specific agent in RecursiveMAS, rather than a standalone model intended for plain-text generation.
Usage
To load this agent as part of a RecursiveMAS system, you can use the loader provided in the official codebase:
from system_loader import load_mas_system
# Load the distillation-style system
mas = load_mas_system(
style="distillation",
device="cuda",
trust_remote_code=True,
)
# Access the expert and learner agents
expert = mas.agents["expert"].model
learner = mas.agents["learner"].model
For detailed usage instructions and the full inference pipeline, please refer to the GitHub repository.
Model Collections for RecursiveMAS
| Style | Model Collection |
|---|---|
| Sequential-Style | 🤗 HuggingFace |
| Mixture-Style | 🤗 HuggingFace |
| Distillation-Style | 🤗 HuggingFace |
| Deliberation-Style | 🤗 HuggingFace |
Experiment Results
Citation
@misc{recursivemas,
title={Recursive Multi-Agent Systems},
author={Xiyuan Yang and Jiaru Zou and Rui Pan and Ruizhong Qiu and Pan Lu and Shizhe Diao and Jindong Jiang and Hanghang Tong stones and Tong Zhang and Markus J. Buehler and Jingrui He and James Zou},
year={2026},
eprint={2604.25917},
archivePrefix={arXiv},
primaryClass={cs.AI},
url={https://arxiv.org/abs/2604.25917},
}