HiMoE-VLA-Libero-Spatial

HiMoE-VLA: Hierarchical Mixture-of-Experts for Generalist Vision-Language-Action Policies

This repository contains the weights for HiMoE-VLA, as presented in the paper HiMoE-VLA: Hierarchical Mixture-of-Experts for Generalist Vision-Language-Action Policies.

• Paper: arXiv:2512.05693
• GitHub Repository: ZhiyingDu/HiMoE-VLA

Model Description

HiMoE-VLA is a vision–language–action (VLA) framework built to handle the pronounced heterogeneity in modern large-scale robotic datasets. It introduces a Hierarchical Mixture-of-Experts (HiMoE) action module that progressively abstracts away diverse embodiments, action spaces, sensor setups, and control frequencies across layers, enabling unified learning of shared robot behaviors.

Sample Usage

Below is an example of how to load the policy and run inference:

from moevla.policies import policy_config as _policy_config
from moevla.training import config as _config

# specific these parameter
train_config = ""
dataset_config = ""
checkpoint_dir = ""

policy = _policy_config.create_trained_policy(
    _config.get_training_config(train_config),    
    _config.get_dataset_config(dataset_config), 
    checkpoint_dir, 
    default_prompt=None
)

# Run inference on a dummy example.
example = {
    "observation/exterior_image_1_left": ...,
    "observation/wrist_image_left": ...,
    ...
    "prompt": "fold clothes"
}

action_chunk = policy.infer(example)["actions"]

For more details on installation, training, evaluation, and server-client deployment, please refer to the official GitHub repository.

Citation

If you find our work useful in your research, please consider citing our paper:

@article{du2025himoe,
  title={HiMoE-VLA: Hierarchical Mixture-of-Experts for Generalist Vision-Language-Action Policies},
  author={Du, Zhiying and Liu, Bei and Liang, Yaobo and Shen, Yichao and Cao, Haidong and Zheng, Xiangyu and Feng, Zhiyuan and Wu, Zuxuan and Yang, Jiaolong and Jiang, Yu-Gang},
  journal={arXiv preprint arXiv:2512.05693},
  year={2025}
}