gemma-3-27b-it-lora-dsad-anatomy

Gemma 3 27B-it LoRA — DSAD anatomy recognition

LoRA adapter plus a 12-way linear classification head on google/gemma-3-27b-it, fine-tuned for multi-label anatomical structure presence on DSAD frames.

This is the “Gemma 3 27B fine-tuned” row on the Anatomy tab of the SDSC × Chicago Booth surgical video understanding leaderboard.

Prompt example

This closed-set example mirrors the leaderboard format, not a text-input API for this checkpoint.

[surgical frame]

Which anatomical structures are visible in this laparoscopic frame?
Select every matching label.

- abdominal wall
- colon
- inferior mesenteric artery
- intestinal veins
- liver
- pancreas
- small intestine
- spleen
- stomach
- ureter
- uterus
- vesicular glands

Files

• adapter_model.safetensors / adapter_config.json — LoRA (r=128, alpha=256) on q/k/v/o_proj and out_proj
• classifier.pt — linear head over the 5376-d hidden state (12 structures)
• model_config.json — class vocabulary and training hyperparameters

You need the adapter and the classification head; this is not a generative JSON adapter.

Training

• 10 epochs, lr 5e-6, seed 42
• 7,889 training frames / 1,978 validation frames

Evaluation

Full DSAD validation split (95% bootstrap CI):

References

• Skobelev, K., Fithian, E., Baranovski, Y., et al. A Comparative Study in Surgical AI: Potential and Limitations of Data, Compute, and Scaling. arXiv:2603.27341, 2026.
• Dataset: Carstens, M., Rinner, F. M., Bodenstedt, S., et al. The Dresden Surgical Anatomy Dataset for Abdominal Organ Segmentation in Surgical Data Science. Scientific Data, 10, 3 (2023).

Limitations

Research baseline only. Not a medical device.