Closed discivigour closed 1 year ago
请问LinkPred类的forward函数中的torch.log2()函数和 s = torch.sigmoid(sum_log/self.temp) s = s * (self.inter_max - self.inter_min) + self.inter_min 到底是干什么用的呢?
Hi,
It maps the predictions to the hard concrete distribution. You can see the description in the Appendix "Approximate L0 Regularization with Hard Concrete Distribution".
请问LinkPred类的forward函数中的torch.log2()函数和 s = torch.sigmoid(sum_log/self.temp) s = s * (self.inter_max - self.inter_min) + self.inter_min 到底是干什么用的呢?