neuroencoder 1.0.0

EEG model embeddings - distilled EPI-250k

neuroencoder

EEG foundation model embeddings for Python.

pip install neuroencoder

import neuroencoder as ne
from neuroencoder import MRL

images = ne.preprocess(eeg, sfreq=256, channel_names=ch_names)

model = MRL.from_pretrained()
embeddings = model.predict(images, dim=192)

ne.explore(embeddings)   # interactive Apple Embedding Atlas
ne.plot(embeddings)      # static matplotlib UMAP

Model access

The MRL model is gated on HuggingFace. Request access, then authenticate once:

huggingface-cli login

or export HF_TOKEN=hf_... before running.

Documentation

Full docs at docs.neuroencoder.com.

• Introduction — what this is
• Quickstart — your first embedding
• Embeddings — the MRL model end-to-end
• Visualization — Apple Embedding Atlas
• Benchmarks — performance on 20 clinical EEG tasks
• API reference

Links

• Source: github.com/avocardio/neuroencoder
• Model weights: huggingface.co/Neuroencoder/epi-embedding
• Website: neuroencoder.com