data_text_search / output /36de65711121889ccdcb768b85e97e386d8fe4bd
seanpedrickcase's picture
Changed embedding model to MiniLM-L6 as faster. Compressed embeddings are now int8. General improvements to API mode
ea0dd40