Science Question Answering on ScienceQA (CP/ASR)
96.93Correct Prediction RateTCAP (Ours)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| TCAP (Ours)Model=InternVL2.52026.01 | 96.93 | 0.15 | |
| Vanilla FTModel=InternVL2.52026.01 | 96.88 | 93.6 | |
| TCAP (Ours)Model=Qwen3-VL2026.01 | 96.68 | 15.62 | |
| Vanilla FTModel=Qwen3-VL2026.01 | 96.58 | 86.17 | |
| Random DropModel=InternVL2.52026.01 | 95.44 | 92.71 | |
| Random DropModel=Qwen3-VL2026.01 | 95.39 | 82.99 | |
| BYEModel=InternVL2.52026.01 | 89.49 | 91.42 | |
| TCAP (Ours)Model=LLaVA-NeXT2026.01 | 89.44 | 0.05 | |
| SampDetoxModel=Qwen3-VL2026.01 | 89.44 | 67.13 | |
| Vanilla FTModel=LLaVA-NeXT2026.01 | 89.19 | 96.03 | |
| BYEModel=Qwen3-VL2026.01 | 87.7 | 72.43 | |
| Random DropModel=LLaVA-NeXT2026.01 | 86.07 | 95.39 | |
| SampDetoxModel=LLaVA-NeXT2026.01 | 76.9 | 76.45 | |
| SampDetoxModel=InternVL2.52026.01 | 75.31 | 85.47 | |
| BYEModel=LLaVA-NeXT2026.01 | 0 | 100 |