I am working with a sparse tensor of size 5440 * 2538 * 368, which only takes 21 MB of memory. However I am using a ParCube tensor factorization and there it creates a zeros matrix and hence there I get out of memory error. Unfortunately unlike other algorithms, for tensor factorization I cannot break data and work with smaller pieces. I am running this on a Linux server with almost 109GB of free memory. I have read this, but it seems that it doesn't work either.
Any help is appreciated.