Video Captioning and Summarization on Dream 1k
20.8Rouge-LPercep. LM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Percep. LMModel Size=8B2026.03 | 20.8 | 19.7 | |
| Percep. LMModel Size=3B2026.03 | 20.4 | 19 | |
| Qwen2-VLModel Size=7B2026.03 | 19.8 | 18.1 | |
| v-LCMvariant=v-LCM2026.03 | 19.8 | 19.2 | |
| Percep. LMModel Size=1B2026.03 | 19.3 | 15.5 | |
| Qwen2-VLModel Size=2B2026.03 | 18.5 | 13.9 | |
| LCMvariant=LCM2026.03 | 18.5 | 16.6 | |
| InterVL2Model Size=1B2026.03 | 17.2 | 19.5 | |
| InternVL-2.5Model Size=8B2026.03 | 17 | 17 | |
| InterVL2Model Size=8B2026.03 | 16.4 | 19.4 | |
| Qwen2.5-VLModel Size=3B2026.03 | 15.9 | 8.6 | |
| Qwen2.5-VLModel Size=7B2026.03 | 15.7 | 10.5 | |
| InternVL-2.5Model Size=1B2026.03 | 15.6 | 18.6 | |
| InternVL-2.5Model Size=4B2026.03 | 13.3 | 14.8 | |
| InterVL2Model Size=4B2026.03 | 12.2 | 14.7 |