I needed to understand my options for semantic deduplication, so I went through a process of benchmarking several LMs and wrote it all up here. I hope it's usef

I needed to understand my options for semantic deduplication, so I went through a process of benchmarking several LMs and wrote it all up here. I hope it's useful to someone!