Open source

Code

EEGSimpleNet - Simple Interpretable EEG Classification Neural Network

Compact EEG classification network with built-in interpretability tooling for analyzing learned spatial and temporal patterns.

Low latency EEG filtering

Real-time EEG filtering and rhythm tracking with both classical baselines and neural-network-based approaches, designed for low-latency closed-loop setups.

Cube++ dataset tools and benchmarks

Reference code, download information, and evaluation scripts for the Cube++ illumination estimation dataset.

SIGNAL dataset tools for brain-LLM alignment

Dataset tooling, preprocessing, and analysis code for the SIGNAL EEG language benchmark made for brain-model alignment studies.

Stimulation-Free ECoG Speech Mapping

Preprocessing and machine-learning pipelines for stimulation-free speech mapping from ECoG signals, including evaluation workflows aligned with clinical ground truth.

Datasets

SIGNAL dataset

Controlled EEG dataset for language processing with grammatical and semantic incongruencies.

ECoG Speech Mapping Dataset (ECM+ESM)

ECM+ESM speech mapping dataset from 14 patients, enabling stimulation-free mapping research with ECoG features and ESM ground truth.

Synthetic and real EEG datasets for low-latency filtering and forecasting

Real and synthetic EEG datasets designed for low-latency rhythm tracking, filtering, and forecasting in closed-loop paradigms.

Cube++ Illumination Estimation Dataset

Large-scale RAW dataset for illumination estimation with SpyderCube ground truth, semantic labels, and support for single- and two-illuminant setups.