Changed embedding model to MiniLM-L6 as faster. Compressed embeddings are now int8. General improvements to API mode
ea0dd40
seanpedrickcase
commited on