This is a pytorch-based implementation of censored regression (tobit regression) and truncated regression on batched data.
The speed would be significantly faster than using for-loops on cpu when the batch-size is large.
Pytorch >= 1.6.0
GPyTorch >= 1.1.1