intellabs / avuc Goto Github PK

Dear authors,

Very impressive works.

I have a minor question. When I run the main_deterministic_cifar.py v.s. main_bayesian_cifar_avu.py, the deterministic net has a higher accuracy ( about ~4%) than the bayesian_cifar_avu. Is that normal?

thanks,

In the paper on arxiv you refer to an appendix.
Unfortunately I have not been able to find it. Would you be so kind to share it?

Dear author,

I found a few codes lines commented, for example,

#brier = np.mean(calib.brier_scores(target_labels, probs=pred_mean))
#ece = calib.expected_calibration_error_multiclass(pred_mean, target_labels)

which is also useful to me. Could I ask where can find the related functions?

Thanks in advance.