ViT-B/16-SAM

Reported on 5 benchmarks across 3 tasks · 1 paper · 2 SOTA

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Computer Vision3 results

Domain GeneralizationonImageNet-C
Top 1 Accuracy· 2021-06-03
56.5
best: 73.6 (FAN-L-Hybrid (IN-22k))
SOTA
When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations arXiv:2106.01548
Image ClassificationonImageNet V2
Top 1 Accuracy· uses extra data· 2021-06-03
67.5
best: 84.63 (Model soups (BASIC-L))
When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations arXiv:2106.01548
Domain GeneralizationonImageNet-R
Top-1 Error Rate· 2021-06-03
73.6
best: 3.9 (Model soups (BASIC-L))
When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations arXiv:2106.01548

Methodology2 results

Domain AdaptationonImageNet-C
Top 1 Accuracy· 2021-06-03
56.5
best: 73.6 (FAN-L-Hybrid (IN-22k))
SOTA
When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations arXiv:2106.01548
Domain AdaptationonImageNet-R
Top-1 Error Rate· 2021-06-03
73.6
best: 3.9 (Model soups (BASIC-L))
When Vision Transformers Outperform ResNets without Pre-training or Strong Data Augmentations arXiv:2106.01548