skblv/gemma-3-27b-it-lora-surgvu-instruments

Gemma 3 27B-it LoRA — SurgVU instrument recognition

LoRA adapter plus a 17-way linear classification head on google/gemma-3-27b-it, fine-tuned for multi-label instrument presence on SurgVU frames.

This is the “Gemma 3 27B fine-tuned” row on the Instruments 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 instruments are visible in this robotic surgical frame?
Select every matching label.

- bipolar dissector
- bipolar forceps
- cadiere forceps
- clip applier
- force bipolar
- grasping retractor
- monopolar curved scissors
- needle driver
- permanent cautery hook/spatula
- potts scissors
- prograsp forceps
- stapler
- suction irrigator
- synchroseal
- tenaculum forceps
- tip-up fenestrated grasper
- vessel sealer

Files

• adapter_model.safetensors / adapter_config.json — LoRA (r=128, alpha=256)
• classifier.pt — linear head (17 tools)
• model_config.json — tool vocabulary and training hyperparameters

You need the adapter and the classification head.

Training

• 10 epochs, lr 5e-6, effective batch size 2, LoRA r=128, seed 42
• 16,287 validation frames in the published split

Evaluation

Full SurgVU validation split (95% bootstrap CI):

Metric	Value
Exact match	50.61% (49.84–51.39)
Micro-averaged F1	74.86% (74.36–75.32)

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: Zia, A., Berniker, M., Nespolo, R., et al. Surgical Visual Understanding (SurgVU) Dataset. arXiv:2501.09209, 2025.

Limitations

Research baseline only. Not a medical device.