Share your thoughts, 1 month free Claude Pro on usSee more

Surgical Action on MedVidBench

24.4SA Accuracy

Qwen2.5VL-7B_MedGRPO

Updated 1mo ago

Evaluation Results

Method	Links
Qwen2.5VL-7B_MedGRPO 2025.12		24.4
Qwen2.5VL-7B_SFT 2025.12		21.8
Gemini-2.5-flash 2025.12		10.7
GPT-4.1 2025.12		8.7
Qwen2.5VL-7B 2025.12		1