Entity Recognition on Public Thyroid Ultrasound Datasets OOD (test)
62.4Hamming AccuracyEchoVLM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| EchoVLMParameter Count=11B2025.09 | 62.4 | 57.77 | 49.99 | 49.08 | |
| LingshuParameter Count=7B2025.09 | 62.03 | 58.1 | 49.38 | 49.04 | |
| HuatuoGPT-VisionParameter Count=7B2025.09 | 58.18 | 60.34 | 40.63 | 43.18 | |
| LLaVA-OneVisionParameter Count=7B2025.09 | 55.52 | 59.68 | 37.09 | 38.11 | |
| EchoVLMParameter Count=3B2025.09 | 53.24 | 60.31 | 32.58 | 33.88 | |
| MedgemmaParameter Count=4B2025.09 | 52.79 | 61.97 | 32.33 | 33.58 | |
| Qwen2-VLParameter Count=7B2025.09 | 52.1 | 61.67 | 30.95 | 31.04 | |
| Gemma-3Parameter Count=12B2025.09 | 51.3 | 60.18 | 30.31 | 30.28 | |
| Qwen2.5-VLParameter Count=7B2025.09 | 49.36 | 60.39 | 28.99 | 26.9 | |
| LLaVA-MedParameter Count=7B2025.09 | 49.2 | 64.89 | 26.41 | 30.13 | |
| LLaVA1.5Parameter Count=7B2025.09 | 48.76 | 64.22 | 25.67 | 29.9 | |
| LLaVA-NeXTParameter Count=13B2025.09 | 46.46 | 57.66 | 24.49 | 19.97 |