Copyright (C) 2025-2026 ETH Zurich, Switzerland. SPDX-License-Identifier: Apache-2.0. See LICENSE file at the root of the repository for details.
This directory contains the PyTorch nn.Module implementations for the BioFoundation model families. Hydra model settings live in ../config/model, while the canonical family metadata and batch requirements live in ../biofoundation/model_registry.py.
| Model | Signals | Summary | Resources |
|---|---|---|---|
| FEMBA | EEG | Efficient bidirectional Mamba model for long EEG sequences. | Documentation / Hugging Face |
| LUNA | EEG | Query-based channel unification for topology-agnostic EEG modeling. | Documentation / Hugging Face |
| TinyMyo | sEMG | Compact rotary Transformer designed for flexible EMG processing and edge deployment. | Documentation / Hugging Face |
| LuMamba | EEG | LUNA-style channel unification with efficient Mamba temporal modeling. | Documentation / Hugging Face |
| PanLUNA | EEG, ECG, PPG | Sensor-aware query unification for unimodal and multimodal biosignal learning. | Documentation / Hugging Face |
Use the matching pre-training or fine-tuning experiment in ../config/experiment rather than instantiating a model in isolation when starting a reproducible run.