Visual Question Answering on Pooled medical VQA dataset Closed
37.1ACELLaVA-NeXT-7B
Evaluation results
| Method | Links | |
|---|---|---|
| LLaVA-NeXT-7BCalibration=Uncalibrated, Confidence=Verbalized2026.04 | 37.1 | |
| Qwen3-VL-8BCalibration=Uncalibrated, Confidence=Sampling2026.04 | 21.4 | |
| Qwen3-VL-8BCalibration=Uncalibrated, Confidence=Verbalized2026.04 | 18 | |
| InternVL3-8BCalibration=Uncalibrated, Confidence=Verbalized2026.04 | 17.6 | |
| LLaVA-NeXT-7BCalibration=Uncalibrated, Confidence=Sampling2026.04 | 16.5 | |
| LLaVA-NeXT-7BCalibration=HAC-Gate, Confidence=Verbalized2026.04 | 16.3 | |
| InternVL3-8BCalibration=Uncalibrated, Confidence=Sampling2026.04 | 15.2 | |
| LLaVA-NeXT-7BCalibration=HAC-Platt, Confidence=Verbalized2026.04 | 14 | |
| LLaVA-NeXT-7BCalibration=Isotonic Regr., Confidence=Sampling2026.04 | 13.1 | |
| LLaVA-NeXT-7BCalibration=HAC-Gate, Confidence=Sampling2026.04 | 13.1 | |
| LLaVA-NeXT-7BCalibration=Platt Scaling, Confidence=Sampling2026.04 | 12.7 | |
| LLaVA-NeXT-7BCalibration=HAC-Platt, Confidence=Sampling2026.04 | 12.4 | |
| LLaVA-NeXT-7BCalibration=Platt Scaling, Confidence=Verbalized2026.04 | 12.3 | |
| InternVL3-8BCalibration=Platt Scaling, Confidence=Sampling2026.04 | 11.6 | |
| Qwen3-VL-8BCalibration=HAC-Gate, Confidence=Verbalized2026.04 | 11.5 | |
| InternVL3-8BCalibration=HAC-Platt, Confidence=Sampling2026.04 | 11.4 | |
| InternVL3-8BCalibration=HAC-Gate, Confidence=Sampling2026.04 | 11.3 | |
| LLaVA-NeXT-7BCalibration=Isotonic Regr., Confidence=Verbalized2026.04 | 11.1 | |
| Qwen3-VL-8BCalibration=Platt Scaling, Confidence=Verbalized2026.04 | 10.9 | |
| Qwen3-VL-8BCalibration=Platt Scaling, Confidence=Sampling2026.04 | 10.8 | |
| Qwen3-VL-8BCalibration=HAC-Platt, Confidence=Verbalized2026.04 | 10.8 | |
| InternVL3-8BCalibration=Isotonic Regr., Confidence=Sampling2026.04 | 10.2 | |
| Qwen3-VL-8BCalibration=HAC-Platt, Confidence=Sampling2026.04 | 10.1 | |
| Qwen3-VL-8BCalibration=HAC-Gate, Confidence=Sampling2026.04 | 10.1 | |
| Qwen3-VL-8BCalibration=Isotonic Regr., Confidence=Sampling2026.04 | 9.6 | |
| InternVL3-8BCalibration=HAC-Platt, Confidence=Verbalized2026.04 | 9.6 | |
| Qwen3-VL-8BCalibration=Isotonic Regr., Confidence=Verbalized2026.04 | 9.5 | |
| InternVL3-8BCalibration=HAC-Gate, Confidence=Verbalized2026.04 | 9.5 | |
| InternVL3-8BCalibration=Platt Scaling, Confidence=Verbalized2026.04 | 9.4 | |
| InternVL3-8BCalibration=Isotonic Regr., Confidence=Verbalized2026.04 | 9.3 |