Visual Question Answering on TextVQA (ANLS, Token, FTL)
0.23FTL (s)DocOwl2
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| DocOwl2Size=8B2024.09 | 0.23 | 324 | 66.7 | |
| TextMonkeySize=9B2024.09 | 0.5 | 768 | 65.9 | |
| DocOwl 1.5Size=8B2024.09 | 0.56 | — | 68.6 | |
| InternVL 2Size=8B2024.09 | 1.01 | — | 77.4 | |
| IXC 2.5Size=7B2024.09 | 1.11 | — | 78.2 |