"Dictionary gains are mostly effective in the first few KB." https://facebook.github.io/zstd/index.html https://facebook.github.io/zstd/index.html Pretrained

"Dictionary gains are mostly effective in the first few KB." https://facebook.github.io/zstd/index.html https://facebook.github.io/zstd/index.html Pretrained dictionaries have never been intended to help with book sized or bigger compression. zstd automatically learns the most efficient dictionary it can within a few kilobytes. Pretrained dictionaries are only useful when you're independently compressing very small records.