Closed djghosh13 closed 1 year ago
Finishing off support for all datasets currently supported through the original interface:
You can test the new dataset with: (replacing model with whatever you like)
clip_benchmark eval --dataset wds/voc2007_multilabel --dataset_root "https://huggingface.co/datasets/clip-benchmark/wds_voc2007_multilabel/tree/main" --model "ViT-B-32" --pretrained "openai" --output results_voc2007_multilabel.json
Finishing off support for all datasets currently supported through the original interface:
You can test the new dataset with: (replacing model with whatever you like)