File size: 3,752 Bytes
12c7e40
 
 
 
 
da66cd2
12c7e40
da66cd2
12c7e40
 
 
 
da66cd2
 
 
 
 
12c7e40
da66cd2
 
 
 
 
 
 
 
12c7e40
da66cd2
 
 
 
 
 
 
 
 
 
 
12c7e40
 
da66cd2
12c7e40
 
 
da66cd2
 
 
 
12c7e40
 
da66cd2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0530159
da66cd2
 
 
 
3a235a1
 
 
 
 
da66cd2
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
---
pipeline_tag: feature-extraction
tags:
  - eeg
  - brain-computer-interface
  - self-supervised-learning
  - mamba
  - pytorch
---

# SAMBA-100s

[Project website](https://samba.dcode.ai/) · [Code](https://github.com/Jiazhen-Hong/SAMBA) · [Paper](https://arxiv.org/abs/2511.18571) · [SAMBA-2s](https://huggingface.co/Dcode-AI/SAMBA-2s)

SAMBA-100s was pretrained on 100-second windows from the Temple University Hospital Abnormal EEG Corpus (TUAB), using 12,800 samples at 128 Hz and 16 bipolar channels. Retraining followed the original method with a rebuilt TUAB pipeline and validation split. The encoder supports other window lengths and electrode layouts.

## Download

[Download config](https://huggingface.co/Dcode-AI/SAMBA-100s/resolve/main/config.json?download=true) · [Download weights](https://huggingface.co/Dcode-AI/SAMBA-100s/resolve/main/model.safetensors?download=true)

Keep both files in the same directory. To download them together from a terminal:

```bash
pip install -U huggingface_hub
hf download Dcode-AI/SAMBA-100s config.json model.safetensors --local-dir ./SAMBA-100s
```

## Usage

Install SAMBA in a CUDA environment:

```bash
git clone https://github.com/Jiazhen-Hong/SAMBA.git
cd SAMBA
pip install -e '.[mamba,hub]'
```

Load the downloaded directory. If you skipped the download step, the following example fetches both files automatically:

```python
import torch
from huggingface_hub import snapshot_download
from samba_eeg import SambaModel

folder = snapshot_download(
    "Dcode-AI/SAMBA-100s",
    allow_patterns=["config.json", "model.safetensors"],
)
model = SambaModel.from_pretrained(folder).cuda().eval()
```

For files you already downloaded, pass their local directory to `SambaModel.from_pretrained` in place of `folder`.

Prepare EEG at 128 Hz. Use `signals` shaped `[batch, channels, samples]` and electrode `positions` shaped `[channels, 3]` or `[batch, channels, 3]`. Match the signal channel order and use the same coordinate frame and units as the target positions in `config.json`. TUAB pretraining used a 0.5–45 Hz bandpass. The decoder reconstructs EEG on a fixed 16-channel bipolar montage.

```python
signals = signals.to(device="cuda", dtype=torch.float32)
positions = positions.to(device="cuda", dtype=torch.float32)

# Skip these two lines if the signals are already normalized this way.
scale = torch.quantile(signals.abs(), 0.95, dim=-1, keepdim=True)
signals = signals / scale.clamp_min(1e-8)

with torch.inference_mode():
    features = model.encode(signals, positions)
    pooled_features = features.mean(dim=1)
```

`features` has shape `[batch, floor(samples / 8), 128]`. Use `pooled_features` for linear probing.

## Learned EEG spatial weights

SAMBA uses electrode coordinates to learn how to combine EEG channels. These maps show how the spatial weights change during training.

| Eyes open vs. closed | Driver distraction | Motor imagery |
|:---:|:---:|:---:|
| <img src="https://raw.githubusercontent.com/Jiazhen-Hong/SAMBA/main/Figures/Topo_gif/EyesState_GIF.gif" width="250" alt="Eyes open vs. closed"> | <img src="https://raw.githubusercontent.com/Jiazhen-Hong/SAMBA/main/Figures/Topo_gif/DriverDistraction_GIF.gif" width="238" alt="Driver distraction"> | <img src="https://raw.githubusercontent.com/Jiazhen-Hong/SAMBA/main/Figures/Topo_gif/MI_GIF.gif" width="245" alt="Motor imagery"> |

## Citation

```bibtex
@inproceedings{hong2026samba,
  title     = {{SAMBA}: Toward an {EEG} Foundation Model via Differential Mamba and Coordinate-Driven Spatial Embedding},
  author    = {Hong, Jiazhen and Mackellar, Geoffrey and Ghane, Soheila},
  booktitle = {NeurReps 2026: Symmetry and Geometry in Neural Representations (Proceedings Track)},
  year      = {2026}
}
```