mirror of
https://github.com/SillyTavern/Extension-DupeFinder.git
synced 2026-10-15 13:31:57 +00:00
Adds an ability to cluster characters by similarity groups to easily find duplicates.
- JavaScript 99.5%
- CSS 0.5%
| dist | ||
| src | ||
| .gitignore | ||
| LICENSE | ||
| manifest.json | ||
| package-lock.json | ||
| package.json | ||
| README.md | ||
| webpack.config.js | ||
Duplicate Finder
Adds an ability to cluster characters by similarity groups to easily find duplicates.
How to use
- Install the extension using a built-in installer.
- Find a new button in the character list view, left to a search bar. A modal window will open after you click it.
- There you can adjust the similarity threshold - the higher the number, the more "similar" the character cards should be to get grouped together.
- Select desired search method. Sentence similarity is less accurate but fast, Levenshtein distance is very slow but accurate.
- Click "Calculate" to run the search. This may take some time, especially on large character libraries.
- If any of the character clusters match the search criteria, they will be displayed in a list. You can click the "View" button to open a chosen character.
Prerequisites
SillyTavern - latest staging version preferred, or stable release >= 1.11.7.
This won't work on any older versions.
What fields are evaluated in the search?
By default, only the V1 character card fields are getting checked:
- Name
- Description
- First message
- Personality
- Scenario
- Examples of dialogue
You can include V2 card fields with the "Configure Fields" option.
Why the heck it takes so long?
Be patient.
License and credits
Initial idea and implementation by @Cohee1207
Sentence similarity method by @sinnerai
License: AGPLv3