There was a bug in the Python classification scripts when deduplication was used with NBC or hybrid classification methods. This bug resulted in inaccurate tally outputs, and is fixed in this release.