Local Embedding Models: BGE, Nomic, E5, and GTE

local embedding models is a useful local-model lesson because it makes one trade-off visible: private RAG, semantic search, duplicate detection, clustering, and local document assistants. The point is not to crown a permanent winner. The point is to learn how to match a model family to hardware, task, license, and risk.

Current source signal

Build the small version

Create a tiny local vector search over ten class notes, then ask which note is closest to five test questions.

1. 1Pick one exact model file or runtime tag from the current model card.
2. 2Run three short prompts: one easy, one task-specific, and one likely failure case.
3. 3Record load time, response speed, memory pressure, answer quality, and one surprising failure.
4. 4Write a one-paragraph recommendation: use it, do not use it, or use it only for a narrow job.

local_rag_stack:
  documents -> chunker
  chunks -> embedding_model
  vectors -> local_vector_index
  question -> same_embedding_model
  top_chunks -> chat_model_answer

rule: evaluate retrieval before evaluating the chat answer

The big idea: remember retrieval quality. Local model work is product design under constraints, not just downloading the model with the loudest leaderboard score.