Hi,
Is the function torch.quantile differentiable? Can I use it in a custom loss function?
Thanks!
Hi,
Is the function torch.quantile differentiable? Can I use it in a custom loss function?
Thanks!
Yep. If you want to check if something is differentiable you can just call it with a tensor that requires grad and it’ll spit out a warning if the function does not support autograd
Very helpful, many thanks!