EEGMamba

An EEG Foundation Model with Mamba

🔍 About | 🔨 Setup | 🚢 Pretrain | ⛵ Finetune | 🚀 Quick Start | 🔗 Citation

💡 NOTING: This repository is implemented based on the CBraMod repository. Although in the EEGMamba paper we only evaluated 6 downstream datasets, to facilitate further evaluation by the community, we have also provided evaluation code for all the downstream datasets used in the CBraMod paper. Thank you for your support of CBraMod and EEGMamba.

🔥 NEWS: The paper "EEGMamba: An EEG Foundation Model with Mamba" has been accepted by Neural Networks!

🔍 About

We propose EEGMamba, a novel EEG foundation model based on Mamba state space model, for EEG decoding on various clinical and BCI application. The camera-ready version of the paper will be available at Neural Networks.

🔨 Setup

Install Python.

Install PyTorch.

Install Mamba

Install other requirements:

pip install -r requirements.txt

🚢 Pretrain

You can pretrain EEGMamba on our pretraining dataset or your custom pretraining dataset using the following code:

python pretrain_main.py

We have released a pretrained checkpoint on Hugginface🤗.

⛵ Finetune

You can finetune EEGMamba on our selected downstream datasets using the following code:

python finetune_main.py

🚀 Quick Start

You can fine-tune the pretrained EEGMamba on your custom downstream dataset using the following example code:

import torch
import torch.nn as nn
from models.eegmamba import EEGMamba
from einops.layers.torch import Rearrange

device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
model = EEGMamba().to(device)
model.load_state_dict(torch.load('pretrained_weights/pretrained_weights.pth', map_location=device))
model.proj_out = nn.Identity()
classifier = nn.Sequential(
  Rearrange('b c s p -> b (c s p)'),
  nn.Linear(22*4*200, 4*200),
  nn.ELU(),
  nn.Dropout(0.1),
  nn.Linear(4 * 200, 200),
  nn.ELU(),
  nn.Dropout(0.1),
  nn.Linear(200, 4),
).to(device)

# mock_eeg.shape = (batch_size, num_of_channels, time_segments, points_per_patch)
mock_eeg = torch.randn((8, 22, 4, 200)).to(device)

# logits.shape = (batch_size, num_of_classes)
logits = classifier(model(mock_eeg))

🔗 Citation

If you're using this repository in your research or applications, please cite using the following BibTeX:

@article{wang2025eegmamba,
  title={EEGMamba: An EEG foundation model with Mamba},
  author={Wang, Jiquan and Zhao, Sha and Luo, Zhiling and Zhou, Yangxuan and Li, Shijian and Pan, Gang},
  journal={Neural Networks},
  pages={107816},
  year={2025},
  publisher={Elsevier}
}

@inproceedings{wang2025cbramod,
    title={{CB}raMod: A Criss-Cross Brain Foundation Model for {EEG} Decoding},
    author={Jiquan Wang and Sha Zhao and Zhiling Luo and Yangxuan Zhou and Haiteng Jiang and Shijian Li and Tao Li and Gang Pan},
    booktitle={The Thirteenth International Conference on Learning Representations},
    year={2025},
    url={https://openreview.net/forum?id=NPNUHgHF2w}
}

⭐ Star History