|
|
--- |
|
|
license: apache-2.0 |
|
|
datasets: |
|
|
- zhixiangwei/VLM-1B |
|
|
language: |
|
|
- en |
|
|
--- |
|
|
Finetune DFN-CLIP-L-14-39B on VLM-1B. |
|
|
|
|
|
|Dataset|Performance| |
|
|
|:----------------|---------:| |
|
|
| ImageNet 1k | 0.8143 | |
|
|
| ImageNet V2 | 0.7474 | |
|
|
| ImageNet-A | 0.688133 | |
|
|
| ImageNet-O | 0.3585 | |
|
|
| ImageNet-R | 0.908333 | |
|
|
| ImageNet Sketch | 0.685846 | |
|
|
| ObjectNet | 0.707979 | |
|
|
| IN-shifts | 0.682699 | |
|
|
| VTAB | 0.642463 | |
|
|
| MSCOCO | 0.602417 | |
|
|
| Flickr30k | 0.8466 | |
|
|
| WinoGAViL | 0.566388 | |
|
|
| Retrieval | 0.671802 | |
|
|
| Avg. | 0.663018 | |
|
|
--- |
|
|
license: apache-2.0 |
|
|
--- |
