Visual Question Answering on VisualProbe (Overall)
49.7AccuracyTTSP
Evaluation Results
| Method | Links | |
|---|---|---|
| TTSPModel=Qwen3-VL-8B-Instruct2026.04 | 49.7 | |
| RTWIModel=Qwen3-VL-8B-Instruct2026.04 | 47.8 | |
| SCModel=Qwen3-VL-8B-Instruct2026.04 | 47.2 | |
| DeepconfModel=Qwen3-VL-8B-Instruct2026.04 | 47 | |
| TRTModel=Qwen3-VL-8B-Instruct2026.04 | 45.3 | |
| Self-Ref.Model=Qwen3-VL-8B-Instruct2026.04 | 45.3 | |
| CISCModel=Qwen3-VL-8B-Instruct2026.04 | 44.6 | |
| Self-Cer.Model=Qwen3-VL-8B-Instruct2026.04 | 44.4 | |
| BaseModel=Qwen3-VL-8B-Instruct2026.04 | 40.9 |