Chest-VLM Run 12 — Study auxiliary loss (safety)
LoRA adapter fine-tuned on MedGemma 1.5 4B IT for dual-view chest X-ray structured ontology output (<Ontology v=2>).
Training objective: weighted token loss + study auxiliary loss (abnormal/indeterminate class weights ×2).
Recommended when: minimising missed pathologies (highest Normal precision, fewest false normals).
| Metric | Value |
|---|---|
| Study accuracy | 0.749 |
| Study F1 | 0.677 |
| Precision (Normal) | 0.724 |
| Finding macro F1 | 0.228 |
Usage
from peft import PeftModel
# Load base google/medgemma-1.5-4b-it, then:
model = PeftModel.from_pretrained(model, "maximehpe/chest-vlm-run12")
Or download via the project:
python script/download_model.py --run_name 12_medgemma_ontology_study_loss
Not for clinical use. Research prototype only.
- Downloads last month
- 14
Model tree for maximehpe/chest-vlm-run12
Base model
google/medgemma-1.5-4b-it