Get Started
OPEN SOURCE

Download NEST

Get started with pre-trained models and install via pip or download source code

Installation

Install via pip

Recommended. PyPI release coming soon — use GitHub install for now.

pip install nest-eeg
PyPI release coming soon
Install from source

Recommended for now — always up to date with latest changes.

git clone https://github.com/wazder/NEST.git
cd NEST && pip install -e .

Pre-trained Models

Checkpoints will be released on GitHub Releases and Hugging Face Hub simultaneously

Model weights are not yet released. Watch the GitHub Releases page for updates.
Model
Params
WER
BLEU
Size
Download
nest-base
ZuCo trained • 105 channels • recommended
45M
26.1%
0.74
~180MB
GitHub HF Hub
nest-large
Extended training • higher capacity
125M
23.8%
0.78
~500MB
GitHub HF Hub
nest-tiny
Lightweight • edge deployment • fast inference
12M
31.2%
0.68
~48MB
GitHub HF Hub
HUGGING FACE HUB

Load with from_pretrained()

Once released, all NEST model variants will be available on Hugging Face Hub for one-line loading. Weights are downloaded and cached automatically.

View on Hugging Face
PYTHON
# pip install nest-eeg  (coming soon)
from nest import NESTDecoder

# Downloads from HF Hub automatically
decoder = NESTDecoder.from_pretrained(
    "wazder/nest-base"
)

text = decoder.decode(eeg_epochs)
print(text)

Requirements

Python
3.8+
PyTorch
2.0+
CUDA
11.8+
GPU Memory
8GB+
Apple Silicon devices/systems supported via Metal Performance Shaders

Ready to decode brain signals?

Read Documentation View on GitHub