To fetch them: run python scripts/download_models.py for the embedding model (offline machines can place it at backend/models/bge-m3 manually). To distribute a complete package including models across ...