FBNetV5: Neural Architecture Search for Multiple Tasks in One Run

Open paper
Benchmark Model Rank Results
image-classification-on-imagenetFBNetV5–Top 1 Accuracy: 81.8%GFLOPs: 0.726
image-classification-on-imagenetFBNetV5-A-CLS–Top 1 Accuracy: 81.7%GFLOPs: 0.685
image-classification-on-imagenetFBNetV5-AC-CLS–Top 1 Accuracy: 78.4%GFLOPs: 0.280
image-classification-on-imagenetFBNetV5-AR-CLS–Top 1 Accuracy: 77.2%GFLOPs: 0.215
image-classification-on-imagenetFBNetV5-C-CLS–Top 1 Accuracy: 82.6%GFLOPs: 1
image-classification-on-imagenetFBNetV5-F-CLS–Top 1 Accuracy: 84.1%GFLOPs: 2.1
neural-architecture-search-on-imagenetFBNetV5–Top-1 Error Rate: 18.2FLOPs: 726M
neural-architecture-search-on-imagenetFBNetV5-A-CLS–Top-1 Error Rate: 18.3Accuracy: 81.7FLOPs: 685M
neural-architecture-search-on-imagenetFBNetV5-AR-CLS–Top-1 Error Rate: 22.8Accuracy: 77.2FLOPs: 215M
semantic-segmentation-on-ade20kFBNetV5–Validation mIoU: 40.4

Rank counts only results with a code link.