Danbooru character wildcards sorted by gender and popularity/post count版本v0.9 (ID: 1221550)

Danbooru Character tags as wildcards sorted by post count and gender

Someone asked for this and I was bored enough to make it. e621 list soon, maybe idk
The zip archive you can download from here includes multiple versions and the 3 main ones are female, male and indeterminate.

For each version there's versions with only the top X amount of tags in them. They are all sorted by post count so the top 100 female text file for example means it contains the top 100 most popular danbooru characters.

The files suffixed with "full" contain characters with ~6+ posts associated with them and most of those with that low posts are in indeterminate. I don't believe the "full" ones are actually useful but I added it anyway, so if you want the most characters that still work on - let's say NoobAI models - choose top 5k-30k suffixed files (5-10k for male/indeterminate | 5-30k female). Otherwise I recommend whatever you think is a good threshold to stop at

NOTE About the filtering:

I'm fairly confident that the female and male text files have very few false positives but there are some missing/not sorted ones, and in that case the missing tags will be in indeterminate.txt.
That's why these need testing and if you find any character tag that should be in another file, comment here, or message me here on civit or make a github issue.

Known issues:
- Character tags that should implicate other character tags are not filtered because they aren't marked as implications on danbooru (yet? usually low post count chars issue)
- non human creatures (pokemon, kirby, etc) will likely be falsely in male/female, kirby is in male for example while he's androgynous, will be fixed in later versions, androgynous is hard to filter out because it's not tagged enough so if you find a false positive, please tell me

GitHub repository for wildcard archive + a csv with all the tags in 1 file and post count for easy post-process, and also, the csvs for e621 and danbooru tag autocomplete
And the GitHub repository for the tag autocomplete csv creation script thing which is outdated because im slacking off on updating it but i will soon:tm:

I got some plans to expand this in the future with wildcards that include tags that appear the most throughout the posts

描述:

If there's anything wrong in one of the files, comment so I can change it
See description for more details

Known issues:
- Character tags that should implicate other character tags are not filtered because they aren't marked as implications on danbooru (yet? usually low post count chars issue)
- non human creatures (pokemon, kirby, etc) will likely be falsely in male/female, kirby is in male for example while he's androgynous, will be fixed in later versions, androgynous is hard to filter out because it's not tagged enough so if you find a false positive, please tell me

训练词语: