Video Question Answering on NewsVideoQA
1.17FTL (s)DocOwl2-8B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| DocOwl2-8BToken^V=3242024.09 | 1.17 | 64.09 | |
| LongVA-7BToken^V=~2,0292024.09 | 4.29 | 50.61 | |
| LLaVA-next-interleave-7BToken^V=7292024.09 | 4.35 | 56.66 | |
| Idefics3-8BToken^V=~8382024.09 | 6.39 | 60.16 |