LeafLoomHelp center

Prepare the semantic model and index

Your goal

Prepare the local BGE-M3 model when you need semantic retrieval.

Before you start

Select retrieval folders first. Initial download needs internet and free disk space.

Steps

  1. Open Settings → Search and index and inspect the model state and location.
  2. Explicitly start its download and watch connection, download and verification stages.
  3. Closing Settings keeps preparation running; quitting interrupts it. Retain diagnostics and retry to reuse downloaded files.
  4. Once the model is ready, wait for the separate indexing progress before semantic search.

What to expect

Reading and editing remain available. Keyword/Wiki fallback does not mean semantic preparation is complete.

If it does not work

A waiting hint appears after 30 seconds without progress. Opening the model folder does not download it. Keep original notes intact.

Next steps

View source on GitHub ↗