Clustering ideas with Llamafile
Working through the process of applying a local LLM to idea-clustering and labelling: - map the notes as points in a semantic space using vector embeddings; - apply k-means clustering to group nearby points in space; - map points back to groups of notes, then use a large language model to generate labels. This is interesting; I particularly like the use of local hardware
Archives
-
Kellan Elliott-McCrea of laughingmeme.org has started a new link blog! In 2024! Of course, as readers of this link blog know, link blogs never went away :)
(tags: link-blogging blogging links)