Image Captioning on Image Captioning Dataset
39.7BLEU-4Baseline
Evaluation Results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| BaselineSurrogate Model=None (Clean Data), Transfer Direction=ITR -> IC2026.04 | 39.7 | 31 | 60 | 133.3 | 23.8 | — | — | — | — | — | |
| Co-AttackSurrogate Model=ALBEF, Transfer Direction=ITR -> IC2026.04 | 37.4 | 29.8 | 58.4 | 125.5 | 22.8 | — | — | — | — | — | |
| SGASurrogate Model=ALBEF, Transfer Direction=ITR -> IC2026.04 | 34.8 | 28.4 | 56.3 | 116 | 21.4 | — | — | — | — | — | |
| DRASurrogate Model=ALBEF, Transfer Direction=ITR -> IC2026.04 | 34.8 | 28.4 | 56.4 | 115.9 | 21.4 | — | — | — | — | — | |
| HQA-VLAttackSurrogate Model=ALBEF, Transfer Direction=ITR -> IC2026.04 | 31.8 | 26.8 | 54.1 | 104.6 | 19.8 | — | — | — | — | — | |
| FES-RAG-top5Model=MM-RAIT-Qwen2.5VL2026.04 | 14.96 | — | 46.57 | 171.51 | — | 27.83 | 21.44 | 17.58 | 48.13 | 29.35 | |
| InfoGain-MM-top3Model=MM-RAIT-Qwen2.5VL2026.04 | 11.36 | — | 41.24 | 136.44 | — | 23.47 | 17.46 | 13.74 | 42.91 | 24.1 | |
| FES-RAG-top5Model=Qwen2.5-VL2026.04 | 8.99 | — | 37.02 | 110.99 | — | 20.51 | 14.77 | 11.3 | 39.89 | 21.23 | |
| FES-RAG-top5Model=InternVL3.52026.04 | 8.07 | — | 37.18 | 108.18 | — | 18.6 | 13.26 | 9.94 | 39.09 | 20.4 | |
| Jina-Reranker-top3Model=MM-RAIT-Qwen2.5VL2026.04 | 7.42 | — | 35.74 | 96.66 | — | 19.39 | 13.05 | 9.47 | 38.05 | 18.31 | |
| FES-RAG-top5Model=LLaVA-NeXT2026.04 | 6.87 | — | 34.01 | 92.82 | — | 16.16 | 11.26 | 8.33 | 36.23 | 18.53 | |
| Naïve-MRAG-top3Model=MM-RAIT-Qwen2.5VL2026.04 | 5.89 | — | 34.57 | 84.41 | — | 17.15 | 11.41 | 7.9 | 36.09 | 16.89 | |
| None-MRAGModel=MM-RAIT-Qwen2.5VL2026.04 | 5.68 | — | 32.58 | 79.41 | — | 16.18 | 10.46 | 7.56 | 34.16 | 15.43 | |
| InfoGain-MM-top3Model=Qwen2.5-VL2026.04 | 5.58 | — | 28.5 | 73.66 | — | 15.31 | 9.77 | 7.1 | 31.24 | 16.29 | |
| InfoGain-MM-top3Model=InternVL3.52026.04 | 4.79 | — | 28.29 | 64 | — | 15.12 | 8.64 | 5.84 | 32.04 | 15.38 | |
| InfoGain-MM-top3Model=LLaVA-NeXT2026.04 | 4.43 | — | 25.96 | 57.78 | — | 13.17 | 8.18 | 5.95 | 29.85 | 12.84 | |
| Naïve-MRAG-top3Model=Qwen2.5-VL2026.04 | 3.7 | — | 24.55 | 46.64 | — | 11.75 | 6.92 | 4.93 | 26.99 | 10.77 | |
| Jina-Reranker-top3Model=Qwen2.5-VL2026.04 | 3.49 | — | 25.43 | 45.73 | — | 12.62 | 7.16 | 4.56 | 27.82 | 11.86 | |
| None-MRAGModel=InternVL3.52026.04 | 3.07 | — | 21.47 | 45.09 | — | 11.76 | 6.69 | 4.37 | 23.66 | 9.42 | |
| Jina-Reranker-top3Model=LLaVA-NeXT2026.04 | 3.02 | — | 22.91 | 44.6 | — | 10.05 | 5.98 | 4.1 | 25.79 | 10.68 | |
| Jina-Reranker-top3Model=InternVL3.52026.04 | 2.99 | — | 24.11 | 40.73 | — | 11.43 | 6.63 | 3.94 | 27.91 | 11.98 | |
| Naïve-MRAG-top3Model=LLaVA-NeXT2026.04 | 2.84 | — | 21.87 | 35.48 | — | 9.3 | 5.52 | 3.85 | 23.99 | 9.74 | |
| Naïve-MRAG-top3Model=InternVL3.52026.04 | 2.75 | — | 24.33 | 41.46 | — | 10.96 | 6.24 | 3.61 | 27.12 | 12.43 | |
| None-MRAGModel=Qwen2.5-VL2026.04 | 2.03 | — | 18.2 | 19.51 | — | 8.97 | 4.46 | 2.84 | 20.05 | 6.64 | |
| None-MRAGModel=LLaVA-NeXT2026.04 | 1.63 | — | 17.33 | 18.98 | — | 6.42 | 3.08 | 2.02 | 19.17 | 5.34 |