MODEL_CARD.md

# Model card - tox21_snn_classifier
### Model details
- Model name: Self-Normalizing Neural Network Tox21 Baseline
- Developer: JKU Linz
- Paper URL: https://proceedings.neurips.cc/paper_files/paper/2017/hash/5d44ee6f2c3f71b73125876103c8f6c4-Abstract.html
- Model type / architecture:
    - Self-Normalizing Neural Network implemented using PyTorch.
    - Hyperparameters: https://huggingface.co/spaces/ml-jku/tox21_snn_classifier/blob/main/config/config.json
    - A multitask network is trained for all Tox21 targets.
- Inference: Access via FastAPI endpoint. Upon receiving a Tox21 prediction request, the model generates and returns predictions for all Tox21 targets simultaneously.
- Model version: v0
- Model date: 14.10.2025
- Reproducibility: Code for full training is available and enables retraining from
scratch.

### Intended use
This model serves as a baseline benchmark for evaluating and comparing toxicity prediction methods across the 12 pathway assays of the Tox21 dataset. It is not intended for clinical decision-making without experimental validation.

### Metric
Each Tox21 task is evaluated using the area under the receiver operating characteristic curve (AUC). Overall performance is reported as the mean AUC across all individual tasks.

### Training data
Tox21 training and validation sets.

### Evaluation data
Tox21 test set.