Dear Author,
I noticed that in FDS.smooth code,there exists:
"feature[labels == label] = self.calibrate_mean_var(feature[labels == label], ...)"
Does the operation an inplace operation when backwards? features receive gradient from regressor decoder and backward.This inplace operation may have error?
Looking forward for your reply!Thanks!
I found in 2021,some issues also pointed out the potential bug when apllied FDS to their work.I think we could easily use torch.masked_scatter to fix the problem like that:
features = torch.masked_scatter(features,(labels == label).unsqueeze(1).repeat(1,features.shape[1]),self.calibrate_mean_var( features[labels == label], self.running_mean_last_epoch[int(label - self.bucket_start)], self.running_var_last_epoch[int(label - self.bucket_start)], self.smoothed_mean_last_epoch[int(label - self.bucket_start)], self.smoothed_var_last_epoch[int(label - self.bucket_start)]))
Is this right?
Dear Author, I noticed that in FDS.smooth code,there exists: "feature[labels == label] = self.calibrate_mean_var(feature[labels == label], ...)" Does the operation an inplace operation when backwards? features receive gradient from regressor decoder and backward.This inplace operation may have error? Looking forward for your reply!Thanks! I found in 2021,some issues also pointed out the potential bug when apllied FDS to their work.I think we could easily use torch.masked_scatter to fix the problem like that:
features = torch.masked_scatter(features,(labels == label).unsqueeze(1).repeat(1,features.shape[1]),self.calibrate_mean_var( features[labels == label], self.running_mean_last_epoch[int(label - self.bucket_start)], self.running_var_last_epoch[int(label - self.bucket_start)], self.smoothed_mean_last_epoch[int(label - self.bucket_start)], self.smoothed_var_last_epoch[int(label - self.bucket_start)]))Is this right?