YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
EEG-ECG-READER
EEG-ECG-READER is a LoRA/PEFT adapter fine-tuned for EEG/ECG-related biomedical time-series signal analysis and domain adaptation.
Note: This repository contains a LoRA adapter, not a standalone full model. To use it, the compatible MOMENTPipeline-based base model must be loaded first, and this adapter should then be applied on top of it.
Model Details
- Model name: EEG-ECG-READER
- Model type: LoRA adapter
- Base architecture: MOMENTPipeline-based time-series model
- Fine-tuning method: LoRA / PEFT
- PEFT version: 0.19.1
- Fine-tuned by: kdrbaserr
- Domain: EEG/ECG biomedical signal analysis
- Task: Biomedical time-series classification / domain adaptation
LoRA Configuration
| Parameter | Value |
|---|---|
| LoRA rank r | 4 |
| LoRA alpha | 8 |
| LoRA dropout | 0.1 |
| Target modules | q, k, v, o |
| Bias | none |
| Modules saved | head |
Dataset
This model was trained/fine-tuned using the following datasets:
- PhysioNet EEG Motor Movement/Imagery (EEGMMIDB): A dataset of EEG recordings during motor movement and imagery tasks. It includes 109 subjects performing various motor/imagery tasks, providing diverse data for domain adaptation studies.
Dataset details such as source, size, preprocessing steps, label definitions, and train/validation/test split will be added here.
Training Details
- Training method: LoRA fine-tuning
- Best epoch: 5
- Exact transfer: 0
- Partial transfer: 64
- Skipped transfer: 2
- Input format: TODO
- Train/validation/test split: TODO
Evaluation Results
| Metric | Score |
|---|---|
| Best AUC | 0.7886 |
| Final AUC | 0.7756 |
| Final Accuracy | 0.3792 |
| Final Macro F1 | 0.3737 |
| Final Loss | 0.3983 |
How to Use
Example loading structure:
- Load the compatible MOMENTPipeline-based base model.
- Load this LoRA adapter on top of the base model using PEFT.
- Run inference with the same preprocessing pipeline used during training.
Adapter ID:
kdrbaserr/EEG-ECG-READER
Intended Use
This model is intended for research and educational experiments involving EEG/ECG biomedical time-series signal analysis, classification, and domain adaptation.
Limitations
This model should not be used as a standalone medical diagnosis system. The outputs should be validated by qualified experts. Performance may vary depending on the input preprocessing pipeline, dataset distribution, label mapping, and the exact base model used.
Files
- adapter_model.safetensors: LoRA adapter weights
- adapter_config.json: PEFT adapter configuration
- label_map.json: Label mapping used during training/evaluation
- result.json: Evaluation/training result metadata
Citation
If you use this model, please cite the datasets, the base model, PEFT/LoRA method, and this model repository.
Acknowledgements
TODO