Multimodal Summarization on VISTA
4.8BLEU-4CoE
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| CoEzero-shot=true, training-free=true, backbone=Qwen2.5-VL-7B-Instruct2026.03 | 4.8 | 17 | 11.5 | 16.75 | 85.87 | 29.01 | 3.82 | |
| ViTCoTzero-shot=true, training-free=true, backbone=Qwen2.5-VL-7B-Instruct2026.03 | 4.28 | 15.11 | 6.33 | 13.71 | 83.78 | 20.76 | 3.42 | |
| CoSzero-shot=true, training-free=true, backbone=LLaVA-Video / LongVA2026.03 | 4.03 | 15.28 | 3.01 | 13.16 | 83.78 | 16.97 | 2.68 | |
| CoFzero-shot=true, training-free=true, backbone=InternVL3-8B2026.03 | 3.69 | 14.2 | 2.12 | 13.91 | 83.03 | 17.62 | 2.79 | |
| TCoTzero-shot=true, training-free=true, backbone=Qwen2.5-VL-7B-Instruct2026.03 | 2.65 | 13.85 | 2.75 | 10.63 | 84.38 | 17.18 | 3.12 |