Semantic Search Setup Best Practices

Sub-skill of semantic-search-setup: Best Practices.

vamseeachanta Updated

File contents

Best Practices

Best Practices

  1. Use normalized embeddings - Set normalize_embeddings=True
  2. Force CPU mode - Set CUDA_VISIBLE_DEVICES="" for stability
  3. Add SQLite timeout - Use timeout=30 for concurrent access
  4. Process in batches - 100-500 chunks per batch
  5. Track progress - Save after each batch for resumability
  6. Log errors - Capture failures for debugging

vamseeachanta/workspace-hub/tree/main/.agents/skills/_archive/data/documents/semantic-search-setup/best-practices commit e8766f228a

Frequently asked questions

npx skillmds@latest add vamseeachanta/semantic-search-setup-best-practices