Full threadheliophobicdude·I would be interested to know of a consistent, on-board embeddings model. Trying to reduce latency and dependence on API calls for simple vector database search will go a long wagView on HN