// I need to create my own list of topic tags and the model fits the clusters to each file
This replaces the clustering-based tag taxonomy explored in 2026-08-02 Testing vault index chunking and topic clustering (by Claude): neither k-means nor HDBSCAN produced a taxonomy that felt genuinely useful, so instead of deriving tags algorithmically from the embeddings, this plan has you write the tag vocabulary yourself, then use Qwen3 14B to select every tag from that list that fits a given file. Uses Qwen3 14B and the indexing pipeline already built in 2026-08-02 Initial indexing of my vault on Mac mini (by Claude).
- // Had My topic tagscreated to consolidate into a list that makes sense
- // My topic tag library for AI model
- // Create script that reads the library and uses to tag files