issues
search
defenseunicorns
/
leapfrogai-backend-rag
Apache License 2.0
4
stars
1
forks
source link
Adds local model caching on docker container build
#24
Closed
CollectiveUnicorn
closed
9 months ago
CollectiveUnicorn
commented
9 months ago
Adds local embedding model caching for chroma db on docker container build. This allows using the embedding function offline.
Changes model to
instructor-xl
to match embeddings endpoint
Disables chromadb telemetry that causes persistent errors when offline
Sets env variables so that tiktoken with gpt-2 can be cached locally
Adds parallel upload support via threading
Persists db to local drive @
~/.db/
Adds initial tests
instructor-xl
to match embeddings endpoint~/.db/