---
license: apache-2.0
---

## A Solution for ORena SAVE FOCUS Challenge—FRAME Track 

- Team: frame_surgvqa
- Team Members: xq20667402 (Qing XU), Zhuochen_Li (Zhuochen LI)


### Model

- We use a finetuned and compressed Qwen3.8-27B model
- Please view our model at `path/to/lzc2001/orena_focus_frame_solution/Qwen3.8-27B-FOCUS-Merged-W8A16-V2`


### Self-Generated Annotations

- We use two Qwen models to generate annotations for unlabeled [lapchole-focus-vqa](https://huggingface.co/datasets/orena-dkfz/lapchole-focus-vqa) videos
- The generated annotations contain 4763 QAs
- Will be released by us once [lapchole-focus-vqa](https://huggingface.co/datasets/orena-dkfz/lapchole-focus-vqa) is made publicly available