Fig. 2: Comparison of average Log Norm and Weighted Alpha quality metrics for CV models. | Nature Communications

Fig. 2: Comparison of average Log Norm and Weighted Alpha quality metrics for CV models.

From: Predicting trends in the quality of state-of-the-art neural networks without access to training or testing data

Fig. 2

Comparison of average Log Norms (in (a), (b), and (d)) and Weighted Alpha (in (c)) quality metrics versus reported test accuracy for pretrained VGG models: VGG11, VGG13, VGG16, and VGG19, with and without Batch Normalization (BN), trained on ImageNet, available in pyTorch (v1.4). Metrics fit by linear regression, RMSE, R2, and the Kendal-tau rank correlation metric reported.

Back to article page