90 lines
2.9 KiB
YAML
90 lines
2.9 KiB
YAML
Collections:
|
|
- Name: OFA
|
|
Metadata:
|
|
Architecture:
|
|
- ResNet
|
|
- Transformer
|
|
Training Data:
|
|
- CC12M
|
|
- CC3M
|
|
- SBU
|
|
- COCO
|
|
- VG
|
|
- VQAv2
|
|
- GQA
|
|
- RefCOCO
|
|
- OpenImages
|
|
- Object365
|
|
- YFCC100M
|
|
- ImageNet-21K
|
|
- Pile
|
|
Paper:
|
|
Title: 'OFA: Unifying Architectures, Tasks, and Modalities Through a Simple
|
|
Sequence-to-Sequence Learning Framework'
|
|
URL: https://arxiv.org/abs/2202.03052
|
|
README: configs/ofa/README.md
|
|
|
|
Models:
|
|
- Name: ofa-base_3rdparty-finetuned_refcoco
|
|
Metadata:
|
|
FLOPs: null
|
|
Parameters: 182238536
|
|
In Collection: OFA
|
|
Results:
|
|
- Task: Visual Grounding
|
|
Dataset: RefCOCO
|
|
Metrics:
|
|
Accuracy (testA): 90.49
|
|
Accuracy (testB): 83.63
|
|
Weights: https://download.openmmlab.com/mmclassification/v1/ofa/ofa-base_3rdparty_refcoco_20230418-2797d3ab.pth
|
|
Config: configs/ofa/ofa-base_finetuned_refcoco.py
|
|
Converted From:
|
|
Weights: https://ofa-beijing.oss-cn-beijing.aliyuncs.com/checkpoints/refcoco_base_best.pt
|
|
Code: https://github.com/OFA-Sys/OFA
|
|
- Name: ofa-base_3rdparty-finetuned_vqa
|
|
Metadata:
|
|
FLOPs: null
|
|
Parameters: 182238536
|
|
In Collection: OFA
|
|
Results:
|
|
- Task: Visual Question Answering
|
|
Dataset: VQAv2
|
|
Metrics:
|
|
Accuracy: 78.00 # Report from the official repo
|
|
Weights: https://download.openmmlab.com/mmclassification/v1/ofa/ofa-base_3rdparty_coco-vqa_20230418-f38539a5.pth
|
|
Config: configs/ofa/ofa-base_finetuned_vqa.py
|
|
Converted From:
|
|
Weights: https://ofa-beijing.oss-cn-beijing.aliyuncs.com/checkpoints/vqa_large_best.pt
|
|
Code: https://github.com/OFA-Sys/OFA
|
|
- Name: ofa-base_3rdparty-finetuned_caption
|
|
Metadata:
|
|
FLOPs: null
|
|
Parameters: 182238536
|
|
In Collection: OFA
|
|
Results:
|
|
- Task: Image Caption
|
|
Dataset: COCO
|
|
Metrics:
|
|
BLEU-4: 42.64
|
|
CIDER: 144.50
|
|
Weights: https://download.openmmlab.com/mmclassification/v1/ofa/ofa-base_3rdparty_coco-caption_20230418-de18914e.pth
|
|
Config: configs/ofa/ofa-base_finetuned_caption.py
|
|
Converted From:
|
|
Weights: https://ofa-beijing.oss-cn-beijing.aliyuncs.com/checkpoints/caption_base_best.pt
|
|
Code: https://github.com/OFA-Sys/OFA
|
|
- Name: ofa-base_3rdparty-zeroshot_vqa
|
|
Metadata:
|
|
FLOPs: null
|
|
Parameters: 182238536
|
|
In Collection: OFA
|
|
Results:
|
|
- Task: Visual Question Answering
|
|
Dataset: VQAv2
|
|
Metrics:
|
|
Accuracy: 58.32
|
|
Weights: https://download.openmmlab.com/mmclassification/v1/ofa/ofa-base_3rdparty_pretrain_20230418-dccfc07f.pth
|
|
Config: configs/ofa/ofa-base_zeroshot_vqa.py
|
|
Converted From:
|
|
Weights: https://ofa-beijing.oss-cn-beijing.aliyuncs.com/checkpoints/ofa_base.pt
|
|
Code: https://github.com/OFA-Sys/OFA
|