Medical Visual Question Answering on Path-VQA (test)
63.61BLEU-1ARMed-R
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ARMed-RModel Type=Fine-tuned VLM, Protocol=Reasoner2025.08 | 63.61 | 64.96 | 98.46 | 95.81 | 80.71 | |
| Qwen2.5-VL-3B w. SFTModel Type=Fine-tuned VLM, Protocol=SFT2025.08 | 55.43 | 56.62 | 97.74 | 94.36 | 76.04 | |
| Qwen2.5-VL-3B w. GRPOModel Type=Fine-tuned VLM, Protocol=GRPO2025.08 | 52.36 | 54.28 | 98.04 | 95.6 | 75.07 | |
| InternVL3-2BModel Type=General VLM2025.08 | 28.56 | 30.38 | 92.71 | 88.51 | 60.04 | |
| LLaVA-v1.6-7BModel Type=General VLM2025.08 | 19.27 | 20.86 | 87.12 | 83.91 | 52.79 | |
| LLaVA-v1.6-13BModel Type=General VLM2025.08 | 7.23 | 10.04 | 85.55 | 81.66 | 46.12 | |
| HuatuoGPT-V-7BModel Type=Medical VLM2025.08 | 6.59 | 10.12 | 83.1 | 80.22 | 45.01 | |
| Qwen2.5-VL-3BModel Type=Fine-tuned VLM, Protocol=Base2025.08 | 5.3 | 8.2 | 86.91 | 85.1 | 46.38 | |
| Qwen2.5-VL-7BModel Type=General VLM2025.08 | 5.25 | 8.24 | 88.45 | 85.03 | 46.74 | |
| InternVL3-8BModel Type=General VLM2025.08 | 4.53 | 6.78 | 80.66 | 78.29 | 42.57 | |
| InternVL3-14BModel Type=General VLM2025.08 | 3.7 | 6.08 | 80.37 | 77.8 | 41.99 | |
| LLaVA-Med-7BModel Type=Medical VLM2025.08 | 2.39 | 4.5 | 85.66 | 84.15 | 44.18 |