| license: mit | |
| base_model: emilyalsentzer/Bio_ClinicalBERT | |
| pipeline_tag: text-classification | |
| tags: [multi-label, clinical, overdose] | |
| # DrugDetectionBERT | |
| Bio_ClinicalBERT fine-tuned for multi-label drug detection in death investigation narratives. | |
| Labels: Methamphetamine, Heroin, Cocaine, Fentanyl, Alcohol, Prescription.opioids, | |
| Any Opioids, Benzodiazepines, Others, Any Drugs. | |
| Sigmoid outputs. Per-label thresholds in `best_thresholds.json` — do not use 0.5. | |
| ```python | |
| from transformers import AutoTokenizer, AutoModelForSequenceClassification | |
| tok = AutoTokenizer.from_pretrained("Elemets/DrugDetectionBERT") | |
| model = AutoModelForSequenceClassification.from_pretrained("Elemets/DrugDetectionBERT") | |