Do you think make
masked_scatter_() a torch function (torch.masked_scatter()) is a good idea? As it is now an in_place function, sometimes the “in-place operations can be only used on variables that don’t share storage with any other variables, but detected that there are x objects sharing it” will occur.
Hi, you can always do the following to do an out of place operation:
new_t = t.clone().masked_scatter_(...)