Visual Question Answering on VQA-RAD Closed
70.2AUROCInternVL3-2B
Evaluation Results
| Method | Links | |
|---|---|---|
| InternVL3-2BCalibration Method=Hist. Binning, Confidence Type=Sampling2026.04 | 70.2 | |
| InternVL3-2BCalibration Method=Isotonic Regr., Confidence Type=Sampling2026.04 | 69.7 | |
| InternVL3-2BCalibration Method=Uncalibrated, Confidence Type=Sampling2026.04 | 68.7 | |
| InternVL3-2BCalibration Method=Platt Scaling, Confidence Type=Sampling2026.04 | 68.7 | |
| InternVL3-2BCalibration Method=HAC-Gate, Confidence Type=Sampling2026.04 | 68.1 | |
| InternVL3-2BCalibration Method=HAC-Platt, Confidence Type=Sampling2026.04 | 68 | |
| InternVL3-2BCalibration Method=Uncalibrated, Confidence Type=Verbalized2026.04 | 67.5 | |
| InternVL3-2BCalibration Method=Platt Scaling, Confidence Type=Verbalized2026.04 | 67.5 | |
| InternVL3-2BCalibration Method=Isotonic Regr., Confidence Type=Verbalized2026.04 | 67 | |
| InternVL3-38BCalibration Method=Uncalibrated, Confidence Type=Sampling2026.04 | 66.6 | |
| InternVL3-38BCalibration Method=Platt Scaling, Confidence Type=Sampling2026.04 | 66.6 | |
| InternVL3-38BCalibration Method=HAC-Platt, Confidence Type=Sampling2026.04 | 66.6 | |
| InternVL3-38BCalibration Method=HAC-Gate, Confidence Type=Sampling2026.04 | 66.6 | |
| InternVL3-2BCalibration Method=HAC-Platt, Confidence Type=Verbalized2026.04 | 66.3 | |
| InternVL3-2BCalibration Method=HAC-Gate, Confidence Type=Verbalized2026.04 | 66.2 | |
| InternVL3-38BCalibration Method=Isotonic Regr., Confidence Type=Sampling2026.04 | 64.5 | |
| InternVL3-2BCalibration Method=Hist. Binning, Confidence Type=Verbalized2026.04 | 64.3 | |
| LLaVA-NeXT-34BCalibration Method=Uncalibrated, Confidence Type=Sampling2026.04 | 62.5 | |
| LLaVA-NeXT-34BCalibration Method=Platt Scaling, Confidence Type=Sampling2026.04 | 62.5 | |
| LLaVA-NeXT-34BCalibration Method=HAC-Platt, Confidence Type=Sampling2026.04 | 62.4 | |
| LLaVA-NeXT-34BCalibration Method=HAC-Gate, Confidence Type=Sampling2026.04 | 62.4 | |
| Qwen3-VL-2BCalibration Method=HAC-Platt, Confidence Type=Sampling2026.04 | 62.1 | |
| Qwen3-VL-2BCalibration Method=HAC-Gate, Confidence Type=Sampling2026.04 | 62.1 | |
| LLaVA-NeXT-7BCalibration Method=Uncalibrated, Confidence Type=Sampling2026.04 | 61.8 | |
| LLaVA-NeXT-7BCalibration Method=Platt Scaling, Confidence Type=Sampling2026.04 | 61.8 | |
| LLaVA-NeXT-7BCalibration Method=HAC-Platt, Confidence Type=Sampling2026.04 | 61.8 | |
| LLaVA-NeXT-7BCalibration Method=HAC-Gate, Confidence Type=Sampling2026.04 | 61.8 | |
| InternVL3-38BCalibration Method=Hist. Binning, Confidence Type=Sampling2026.04 | 61.7 | |
| InternVL3-8BCalibration Method=Uncalibrated, Confidence Type=Sampling2026.04 | 61.6 | |
| InternVL3-8BCalibration Method=Platt Scaling, Confidence Type=Sampling2026.04 | 61.6 | |
| Qwen3-VL-2BCalibration Method=HAC-Gate, Confidence Type=Verbalized2026.04 | 61.2 | |
| InternVL3-8BCalibration Method=Isotonic Regr., Confidence Type=Sampling2026.04 | 61 | |
| LLaVA-NeXT-34BCalibration Method=Hist. Binning, Confidence Type=Sampling2026.04 | 60.7 | |
| Qwen3-VL-2BCalibration Method=HAC-Platt, Confidence Type=Verbalized2026.04 | 60.1 | |
| InternVL3-8BCalibration Method=HAC-Gate, Confidence Type=Sampling2026.04 | 60 | |
| InternVL3-38BCalibration Method=HAC-Gate, Confidence Type=Verbalized2026.04 | 60 | |
| LLaVA-NeXT-34BCalibration Method=Isotonic Regr., Confidence Type=Sampling2026.04 | 60 | |
| InternVL3-38BCalibration Method=Uncalibrated, Confidence Type=Verbalized2026.04 | 59.8 | |
| InternVL3-38BCalibration Method=Platt Scaling, Confidence Type=Verbalized2026.04 | 59.8 | |
| InternVL3-38BCalibration Method=HAC-Platt, Confidence Type=Verbalized2026.04 | 59.8 | |
| InternVL3-8BCalibration Method=HAC-Platt, Confidence Type=Sampling2026.04 | 59.5 | |
| Qwen3-VL-8BCalibration Method=HAC-Gate, Confidence Type=Verbalized2026.04 | 59.2 | |
| Qwen3-VL-8BCalibration Method=Hist. Binning, Confidence Type=Verbalized2026.04 | 59.1 | |
| Qwen3-VL-32BCalibration Method=HAC-Platt, Confidence Type=Verbalized2026.04 | 58.8 | |
| Qwen3-VL-32BCalibration Method=HAC-Gate, Confidence Type=Verbalized2026.04 | 58.8 | |
| LLaVA-NeXT-7BCalibration Method=Isotonic Regr., Confidence Type=Sampling2026.04 | 58.8 | |
| Qwen3-VL-8BCalibration Method=Isotonic Regr., Confidence Type=Verbalized2026.04 | 58 | |
| Qwen3-VL-32BCalibration Method=HAC-Platt, Confidence Type=Sampling2026.04 | 57.6 | |
| LLaVA-NeXT-7BCalibration Method=Hist. Binning, Confidence Type=Sampling2026.04 | 57.6 | |
| Qwen3-VL-8BCalibration Method=Uncalibrated, Confidence Type=Verbalized2026.04 | 57.5 | |
| Qwen3-VL-8BCalibration Method=Platt Scaling, Confidence Type=Verbalized2026.04 | 57.5 | |
| InternVL3-8BCalibration Method=Hist. Binning, Confidence Type=Sampling2026.04 | 57 | |
| Qwen3-VL-2BCalibration Method=Hist. Binning, Confidence Type=Sampling2026.04 | 56.6 | |
| Qwen3-VL-8BCalibration Method=HAC-Platt, Confidence Type=Sampling2026.04 | 56.6 | |
| LLaVA-NeXT-7BCalibration Method=Isotonic Regr., Confidence Type=Verbalized2026.04 | 56.5 | |
| Qwen3-VL-8BCalibration Method=HAC-Platt, Confidence Type=Verbalized2026.04 | 56.2 | |
| InternVL3-38BCalibration Method=Isotonic Regr., Confidence Type=Verbalized2026.04 | 56.2 | |
| Qwen3-VL-2BCalibration Method=Isotonic Regr., Confidence Type=Sampling2026.04 | 56.1 | |
| LLaVA-NeXT-7BCalibration Method=Hist. Binning, Confidence Type=Verbalized2026.04 | 56.1 | |
| Qwen3-VL-2BCalibration Method=Uncalibrated, Confidence Type=Sampling2026.04 | 56 | |
| Qwen3-VL-2BCalibration Method=Platt Scaling, Confidence Type=Sampling2026.04 | 56 | |
| Qwen3-VL-32BCalibration Method=Uncalibrated, Confidence Type=Verbalized2026.04 | 55.9 | |
| Qwen3-VL-32BCalibration Method=Platt Scaling, Confidence Type=Verbalized2026.04 | 55.9 | |
| Qwen3-VL-32BCalibration Method=Isotonic Regr., Confidence Type=Verbalized2026.04 | 55.4 | |
| LLaVA-NeXT-34BCalibration Method=HAC-Gate, Confidence Type=Verbalized2026.04 | 55.3 | |
| Qwen3-VL-32BCalibration Method=Uncalibrated, Confidence Type=Sampling2026.04 | 55.1 | |
| Qwen3-VL-32BCalibration Method=Platt Scaling, Confidence Type=Sampling2026.04 | 55.1 | |
| LLaVA-NeXT-7BCalibration Method=HAC-Gate, Confidence Type=Verbalized2026.04 | 55.1 | |
| LLaVA-NeXT-34BCalibration Method=HAC-Platt, Confidence Type=Verbalized2026.04 | 55.1 | |
| LLaVA-NeXT-7BCalibration Method=Uncalibrated, Confidence Type=Verbalized2026.04 | 54.9 | |
| LLaVA-NeXT-7BCalibration Method=Platt Scaling, Confidence Type=Verbalized2026.04 | 54.9 | |
| LLaVA-NeXT-34BCalibration Method=Uncalibrated, Confidence Type=Verbalized2026.04 | 54.9 | |
| LLaVA-NeXT-34BCalibration Method=Platt Scaling, Confidence Type=Verbalized2026.04 | 54.9 | |
| LLaVA-NeXT-34BCalibration Method=Isotonic Regr., Confidence Type=Verbalized2026.04 | 54.8 | |
| Qwen3-VL-8BCalibration Method=Hist. Binning, Confidence Type=Sampling2026.04 | 54.6 | |
| Qwen3-VL-32BCalibration Method=HAC-Gate, Confidence Type=Sampling2026.04 | 54.1 | |
| Qwen3-VL-8BCalibration Method=Isotonic Regr., Confidence Type=Sampling2026.04 | 53.9 | |
| InternVL3-38BCalibration Method=Hist. Binning, Confidence Type=Verbalized2026.04 | 53.9 | |
| LLaVA-NeXT-34BCalibration Method=Hist. Binning, Confidence Type=Verbalized2026.04 | 53.8 | |
| Qwen3-VL-8BCalibration Method=Uncalibrated, Confidence Type=Sampling2026.04 | 53.7 | |
| Qwen3-VL-8BCalibration Method=Platt Scaling, Confidence Type=Sampling2026.04 | 53.7 | |
| Qwen3-VL-8BCalibration Method=HAC-Gate, Confidence Type=Sampling2026.04 | 53.4 | |
| Qwen3-VL-32BCalibration Method=Isotonic Regr., Confidence Type=Sampling2026.04 | 52.2 | |
| Qwen3-VL-2BCalibration Method=Uncalibrated, Confidence Type=Verbalized2026.04 | 51.8 | |
| Qwen3-VL-2BCalibration Method=Platt Scaling, Confidence Type=Verbalized2026.04 | 51.8 | |
| Qwen3-VL-2BCalibration Method=Isotonic Regr., Confidence Type=Verbalized2026.04 | 51.5 | |
| Qwen3-VL-32BCalibration Method=Hist. Binning, Confidence Type=Sampling2026.04 | 51.5 | |
| InternVL3-8BCalibration Method=HAC-Gate, Confidence Type=Verbalized2026.04 | 51.2 | |
| Qwen3-VL-32BCalibration Method=Hist. Binning, Confidence Type=Verbalized2026.04 | 51.1 | |
| Qwen3-VL-2BCalibration Method=Hist. Binning, Confidence Type=Verbalized2026.04 | 50.4 | |
| InternVL3-8BCalibration Method=Uncalibrated, Confidence Type=Verbalized2026.04 | 50.3 | |
| InternVL3-8BCalibration Method=Platt Scaling, Confidence Type=Verbalized2026.04 | 50.3 | |
| InternVL3-8BCalibration Method=HAC-Platt, Confidence Type=Verbalized2026.04 | 49.9 | |
| LLaVA-NeXT-7BCalibration Method=HAC-Platt, Confidence Type=Verbalized2026.04 | 48.2 | |
| InternVL3-8BCalibration Method=Isotonic Regr., Confidence Type=Verbalized2026.04 | 47.6 | |
| InternVL3-8BCalibration Method=Hist. Binning, Confidence Type=Verbalized2026.04 | 44.9 |