How can I reduce memory use while writing clustering results to postgres?

Question

I am running the pgsql example code on a dataset with 3 million records. I\'ve noticed dedupe creates a 100 GB temp file during pair scoring and clustering (line 292). The scrip